diff options
603 files changed, 25271 insertions, 17262 deletions
diff --git a/CREDITS.TXT b/CREDITS.TXT index f6467ab..e58b85f 100644 --- a/CREDITS.TXT +++ b/CREDITS.TXT @@ -335,3 +335,8 @@ D: Bunches of stuff N: Bob Wilson E: bob.wilson@acm.org D: Advanced SIMD (NEON) support in the ARM backend + +N: Wesley Peck +E: peckw@wesleypeck.com +W: http://wesleypeck.com/ +D: MicroBlaze backend @@ -30,8 +30,8 @@ ifeq ($(BUILD_DIRS_ONLY),1) DIRS := lib/System lib/Support utils OPTIONAL_DIRS := else - DIRS := lib/System lib/Support utils lib/VMCore lib tools/llvm-config \ - tools runtime docs unittests + DIRS := lib/System lib/Support utils lib/VMCore lib tools/llvm-shlib \ + tools/llvm-config tools runtime docs unittests OPTIONAL_DIRS := projects bindings endif @@ -43,12 +43,9 @@ EXTRA_DIST := test unittests llvm.spec include win32 Xcode include $(LEVEL)/Makefile.config -# llvm-gcc4 doesn't need runtime libs. llvm-gcc4 is the only supported one. -# FIXME: Remove runtime entirely once we have an understanding of where -# libprofile etc should go. -#ifeq ($(LLVMGCC_MAJVERS),4) -# DIRS := $(filter-out runtime, $(DIRS)) -#endif +ifneq ($(ENABLE_SHARED),1) + DIRS := $(filter-out tools/llvm-shlib, $(DIRS)) +endif ifeq ($(MAKECMDGOALS),libs-only) DIRS := $(filter-out tools runtime docs, $(DIRS)) diff --git a/Makefile.config.in b/Makefile.config.in index 2cc69dc..aca21e5 100644 --- a/Makefile.config.in +++ b/Makefile.config.in @@ -183,25 +183,21 @@ TARGETS_TO_BUILD=@TARGETS_TO_BUILD@ # want to override the value set by configure. LLVMGCCDIR := @LLVMGCCDIR@ -# Determine the target for which LLVM should generate code. -ifeq (@LLVMGCC_MAJVERS@,3) -LLVMGCCARCH := @target@/3.4-llvm -else -LLVMGCCARCH := @target@/@LLVMGCC_VERSION@ -endif - -# Determine the path where the library executables are -LLVMGCCLIBEXEC := @LLVMGCCLIBEXEC@ - # Full pathnames of LLVM C/C++ front-end 'cc1' and 'cc1plus' binaries: LLVMGCC := @LLVMGCC@ LLVMGXX := @LLVMGXX@ LLVMCC1 := @LLVMCC1@ LLVMCC1PLUS := @LLVMCC1PLUS@ -LLVMGCC_VERSION := @LLVMGCC_VERSION@ -LLVMGCC_MAJVERS := @LLVMGCC_MAJVERS@ LLVMGCC_LANGS := @LLVMGCC_LANGS@ +# Information on Clang, if configured. +CLANGPATH := @CLANGPATH@ +CLANGXXPATH := @CLANGXXPATH@ +ENABLE_BUILT_CLANG := @ENABLE_BUILT_CLANG@ + +# The LLVM capable compiler to use. +LLVMCC_OPTION := @LLVMCC_OPTION@ + # Path to directory where object files should be stored during a build. # Set OBJ_ROOT to "." if you do not want to use a separate place for # object files. @@ -266,6 +262,9 @@ ENABLE_THREADS := @ENABLE_THREADS@ # Do we want to build with position independent code? ENABLE_PIC := @ENABLE_PIC@ +# Do we want to build a shared library and link the tools with it? +ENABLE_SHARED := @ENABLE_SHARED@ + # Use -fvisibility-inlines-hidden? ENABLE_VISIBILITY_INLINES_HIDDEN := @ENABLE_VISIBILITY_INLINES_HIDDEN@ @@ -276,6 +275,9 @@ ENABLE_VISIBILITY_INLINES_HIDDEN := @ENABLE_VISIBILITY_INLINES_HIDDEN@ # Enable JIT for this platform TARGET_HAS_JIT = @TARGET_HAS_JIT@ +# Environment variable to set to change the runtime shared library search path. +SHLIBPATH_VAR = @SHLIBPATH_VAR@ + # Shared library extension for host platform. SHLIBEXT = @SHLIBEXT@ diff --git a/Makefile.rules b/Makefile.rules index 761cc81..b5b3525 100644 --- a/Makefile.rules +++ b/Makefile.rules @@ -493,7 +493,27 @@ ExmplDir := $(PROJ_OBJ_ROOT)/$(BuildMode)/examples LLVMLibDir := $(LLVM_OBJ_ROOT)/$(BuildMode)/lib LLVMToolDir := $(LLVM_OBJ_ROOT)/$(BuildMode)/bin LLVMExmplDir:= $(LLVM_OBJ_ROOT)/$(BuildMode)/examples -CFERuntimeLibDir := $(LLVMGCCDIR)/lib + +#-------------------------------------------------------------------- +# LLVM Capable Compiler +#-------------------------------------------------------------------- + +ifeq ($(LLVMCC_OPTION),llvm-gcc) + LLVMCC := $(LLVMGCC) + LLVMCXX := $(LLVMGXX) +else + ifeq ($(LLVMCC_OPTION),clang) + ifneq ($(CLANGPATH),) + LLVMCC := $(CLANGPATH) + LLVMCXX := $(CLANGXXPATH) + else + ifeq ($(ENABLE_BUILT_CLANG),1) + LLVMCC := $(LLVMToolDir)/clang + LLVMCXX := $(LLVMToolDir)/clang++ + endif + endif + endif +endif #-------------------------------------------------------------------- # Full Paths To Compiled Tools and Utilities @@ -529,16 +549,6 @@ endif ifndef LBUGPOINT LBUGPOINT := $(LLVMToolDir)/bugpoint$(EXEEXT) endif -ifndef LUPGRADE -LUPGRADE := $(LLVMToolDir)/llvm-upgrade$(EXEEXT) -endif -ifeq ($(LLVMGCC_MAJVERS),3) -LLVMGCCWITHPATH := PATH="$(LLVMToolDir):$(PATH)" $(LLVMGCC) -LLVMGXXWITHPATH := PATH="$(LLVMToolDir):$(PATH)" $(LLVMGXX) -else -LLVMGCCWITHPATH := $(LLVMGCC) -LLVMGXXWITHPATH := $(LLVMGXX) -endif #-------------------------------------------------------------------- # Adjust to user's request @@ -613,11 +623,12 @@ endif ifneq ($(HOST_OS),Darwin) ifneq ($(DARWIN_MAJVERS),4) ifdef TOOLNAME -ifdef EXAMPLE_TOOL - LD.Flags += $(RPATH) -Wl,$(ExmplDir) $(RDYNAMIC) -else - LD.Flags += $(RPATH) -Wl,$(ToolDir) $(RDYNAMIC) -endif + LD.Flags += $(RPATH) -Wl,'$$ORIGIN/../lib' + ifdef EXAMPLE_TOOL + LD.Flags += $(RPATH) -Wl,$(ExmplDir) $(RDYNAMIC) + else + LD.Flags += $(RPATH) -Wl,$(ToolDir) $(RDYNAMIC) + endif endif endif endif @@ -710,14 +721,12 @@ else $(TargetCommonOpts) $(CompileCommonOpts) $(LD.Flags) $(Strip) endif -BCCompile.C = $(LLVMGCCWITHPATH) $(CPP.Flags) $(C.Flags) $(CFLAGS) \ - $(CPPFLAGS) \ +BCCompile.C = $(LLVMCC) $(CPP.Flags) $(C.Flags) $(CFLAGS) $(CPPFLAGS) \ $(TargetCommonOpts) $(CompileCommonOpts) Preprocess.C = $(CC) $(CPP.Flags) $(C.Flags) $(CPPFLAGS) \ $(TargetCommonOpts) $(CompileCommonOpts) -E -BCCompile.CXX = $(LLVMGXXWITHPATH) $(CPP.Flags) $(CXX.Flags) $(CXXFLAGS) \ - $(CPPFLAGS) \ +BCCompile.CXX = $(LLVMCXX) $(CPP.Flags) $(CXX.Flags) $(CXXFLAGS) $(CPPFLAGS) \ $(TargetCommonOpts) $(CompileCommonOpts) ProgInstall = $(INSTALL) $(Install.StripFlag) -m 0755 @@ -952,11 +961,16 @@ $(LLVM_CONFIG): $(ToolDir)/$(strip $(TOOLNAME))$(EXEEXT): $(LLVM_CONFIG) +ifeq ($(ENABLE_SHARED), 1) +LLVMLibsOptions += -lLLVM-$(LLVMVersion) +LLVMLibsPaths += $(LibDir)/libLLVM-$(LLVMVersion)$(SHLIBEXT) +else LLVMLibsOptions += $(shell $(LLVM_CONFIG) --libs $(LINK_COMPONENTS)) LLVMLibsPaths += $(LLVM_CONFIG) \ $(shell $(LLVM_CONFIG) --libfiles $(LINK_COMPONENTS)) endif endif +endif ############################################################################### # Library Build Rules: Four ways to build a library @@ -971,12 +985,12 @@ endif #--------------------------------------------------------- ifdef MODULE_NAME -ifeq ($(strip $(LLVMGCC)),) -$(warning Modules require llvm-gcc but no llvm-gcc is available ****) +ifeq ($(strip $(LLVMCC)),) +$(warning Modules require LLVM capable compiler but none is available ****) else Module := $(LibDir)/$(MODULE_NAME).bc -LinkModule := $(LLVMLD) -L$(CFERuntimeLibDir) -r +LinkModule := $(LLVMLD) -r ifdef EXPORTED_SYMBOL_FILE @@ -1097,15 +1111,14 @@ endif # targets for building them. #--------------------------------------------------------- ifdef BYTECODE_LIBRARY -ifeq ($(strip $(LLVMGCC)),) -$(warning Bytecode libraries require llvm-gcc which could not be found ****) +ifeq ($(strip $(LLVMCC)),) +$(warning Bytecode libraries require LLVM capable compiler but none is available ****) else all-local:: $(LibName.BCA) ifdef EXPORTED_SYMBOL_FILE -BCLinkLib = $(LLVMLD) -L$(CFERuntimeLibDir) \ - -internalize-public-api-file=$(EXPORTED_SYMBOL_FILE) +BCLinkLib = $(LLVMLD) -internalize-public-api-file=$(EXPORTED_SYMBOL_FILE) $(LibName.BCA): $(ObjectsBC) $(LibDir)/.dir $(LLVMLD) \ $(LLVMToolDir)/llvm-ar @@ -1162,11 +1175,13 @@ endif # If neither BUILD_ARCHIVE or LOADABLE_MODULE are specified, default to # building an archive. #--------------------------------------------------------- +ifndef NO_BUILD_ARCHIVE ifndef BUILD_ARCHIVE ifndef LOADABLE_MODULE BUILD_ARCHIVE = 1 endif endif +endif #--------------------------------------------------------- # Archive Library Targets: @@ -1382,19 +1397,19 @@ BC_DEPEND_OPTIONS = -MMD -MP -MF "$(ObjDir)/$*.bc.d.tmp" \ BC_DEPEND_MOVEFILE = then $(MV) -f "$(ObjDir)/$*.bc.d.tmp" "$(ObjDir)/$*.bc.d"; \ else $(RM) "$(ObjDir)/$*.bc.d.tmp"; exit 1; fi -$(ObjDir)/%.ll: %.cpp $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGXX) +$(ObjDir)/%.ll: %.cpp $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMCXX) $(Echo) "Compiling $*.cpp for $(BuildMode) build (bytecode)" $(Verb) if $(BCCompile.CXX) $(BC_DEPEND_OPTIONS) \ $< -o $(ObjDir)/$*.ll -S -emit-llvm ; \ $(BC_DEPEND_MOVEFILE) -$(ObjDir)/%.ll: %.cc $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGXX) +$(ObjDir)/%.ll: %.cc $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMCXX) $(Echo) "Compiling $*.cc for $(BuildMode) build (bytecode)" $(Verb) if $(BCCompile.CXX) $(BC_DEPEND_OPTIONS) \ $< -o $(ObjDir)/$*.ll -S -emit-llvm ; \ $(BC_DEPEND_MOVEFILE) -$(ObjDir)/%.ll: %.c $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGCC) +$(ObjDir)/%.ll: %.c $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMCC) $(Echo) "Compiling $*.c for $(BuildMode) build (bytecode)" $(Verb) if $(BCCompile.C) $(BC_DEPEND_OPTIONS) \ $< -o $(ObjDir)/$*.ll -S -emit-llvm ; \ @@ -1415,15 +1430,15 @@ $(ObjDir)/%.o: %.c $(ObjDir)/.dir $(BUILT_SOURCES) $(Echo) "Compiling $*.c for $(BuildMode) build" $(PIC_FLAG) $(Compile.C) $< -o $@ -$(ObjDir)/%.ll: %.cpp $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGXX) +$(ObjDir)/%.ll: %.cpp $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMCXX) $(Echo) "Compiling $*.cpp for $(BuildMode) build (bytecode)" $(BCCompile.CXX) $< -o $@ -S -emit-llvm -$(ObjDir)/%.ll: %.cc $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGXX) +$(ObjDir)/%.ll: %.cc $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMCXX) $(Echo) "Compiling $*.cc for $(BuildMode) build (bytecode)" $(BCCompile.CXX) $< -o $@ -S -emit-llvm -$(ObjDir)/%.ll: %.c $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMGCC) +$(ObjDir)/%.ll: %.c $(ObjDir)/.dir $(BUILT_SOURCES) $(LLVMCC) $(Echo) "Compiling $*.c for $(BuildMode) build (bytecode)" $(BCCompile.C) $< -o $@ -S -emit-llvm @@ -10,3 +10,6 @@ the license agreement found in LICENSE.txt. Please see the HTML documentation provided in docs/index.html for further assistance with LLVM. + +If you're writing a package for LLVM, see docs/Packaging.html for our +suggestions. diff --git a/autoconf/ExportMap.map b/autoconf/ExportMap.map index 43e310e..17b185f 100644 --- a/autoconf/ExportMap.map +++ b/autoconf/ExportMap.map @@ -1,4 +1,7 @@ { global: main; + __progname; + environ; + local: *; }; diff --git a/autoconf/configure.ac b/autoconf/configure.ac index 40f77d8..a5cb788 100644 --- a/autoconf/configure.ac +++ b/autoconf/configure.ac @@ -80,6 +80,7 @@ do llvm-tv) AC_CONFIG_SUBDIRS([projects/llvm-tv]) ;; llvm-poolalloc) AC_CONFIG_SUBDIRS([projects/llvm-poolalloc]) ;; poolalloc) AC_CONFIG_SUBDIRS([projects/poolalloc]) ;; + safecode) AC_CONFIG_SUBDIRS([projects/safecode]) ;; llvm-kernel) AC_CONFIG_SUBDIRS([projects/llvm-kernel]) ;; *) AC_MSG_WARN([Unknown project (${i}) won't be configured automatically]) @@ -184,7 +185,7 @@ AC_CACHE_CHECK([type of operating system we're going to host on], llvm_cv_link_all_option="-Wl,--whole-archive" llvm_cv_no_link_all_option="-Wl,--no-whole-archive" llvm_cv_os_type="Haiku" - llvm_cv_platform_type="Unix" ;; + llvm_cv_platform_type="Unix" ;; *-unknown-eabi*) llvm_cv_link_all_option="-Wl,--whole-archive" llvm_cv_no_link_all_option="-Wl,--no-whole-archive" @@ -236,7 +237,7 @@ AC_CACHE_CHECK([type of operating system we're going to target], *-*-mingw*) llvm_cv_target_os_type="MingW" ;; *-*-haiku*) - llvm_cv_target_os_type="Haiku" ;; + llvm_cv_target_os_type="Haiku" ;; *-unknown-eabi*) llvm_cv_target_os_type="Freestanding" ;; *) @@ -291,6 +292,7 @@ AC_CACHE_CHECK([target architecture],[llvm_cv_target_arch], msp430-*) llvm_cv_target_arch="MSP430" ;; s390x-*) llvm_cv_target_arch="SystemZ" ;; bfin-*) llvm_cv_target_arch="Blackfin" ;; + mblaze-*) llvm_cv_target_arch="MBlaze" ;; *) llvm_cv_target_arch="Unknown" ;; esac]) @@ -427,6 +429,7 @@ else MSP430) AC_SUBST(TARGET_HAS_JIT,0) ;; SystemZ) AC_SUBST(TARGET_HAS_JIT,0) ;; Blackfin) AC_SUBST(TARGET_HAS_JIT,0) ;; + MBlaze) AC_SUBST(TARGET_HAS_JIT,0) ;; *) AC_SUBST(TARGET_HAS_JIT,0) ;; esac fi @@ -470,6 +473,18 @@ esac AC_DEFINE_UNQUOTED([ENABLE_PIC],$ENABLE_PIC, [Define if position independent code is enabled]) +dnl Allow building a shared library and linking tools against it. +AC_ARG_ENABLE(shared, + AS_HELP_STRING([--enable-shared], + [Build a shared library and link tools against it (default is NO)]),, + enableval=default) +case "$enableval" in + yes) AC_SUBST(ENABLE_SHARED,[1]) ;; + no) AC_SUBST(ENABLE_SHARED,[0]) ;; + default) AC_SUBST(ENABLE_SHARED,[0]) ;; + *) AC_MSG_ERROR([Invalid setting for --enable-shared. Use "yes" or "no"]) ;; +esac + dnl Allow specific targets to be specified for building (or not) TARGETS_TO_BUILD="" AC_ARG_ENABLE([targets],AS_HELP_STRING([--enable-targets], @@ -481,7 +496,7 @@ if test "$enableval" = host-only ; then enableval=host fi case "$enableval" in - all) TARGETS_TO_BUILD="X86 Sparc PowerPC Alpha ARM Mips CellSPU PIC16 XCore MSP430 SystemZ Blackfin CBackend MSIL CppBackend" ;; + all) TARGETS_TO_BUILD="X86 Sparc PowerPC Alpha ARM Mips CellSPU PIC16 XCore MSP430 SystemZ Blackfin CBackend MSIL CppBackend MBlaze" ;; *)for a_target in `echo $enableval|sed -e 's/,/ /g' ` ; do case "$a_target" in x86) TARGETS_TO_BUILD="X86 $TARGETS_TO_BUILD" ;; @@ -500,6 +515,7 @@ case "$enableval" in cbe) TARGETS_TO_BUILD="CBackend $TARGETS_TO_BUILD" ;; msil) TARGETS_TO_BUILD="MSIL $TARGETS_TO_BUILD" ;; cpp) TARGETS_TO_BUILD="CppBackend $TARGETS_TO_BUILD" ;; + mblaze) TARGETS_TO_BUILD="MBlaze $TARGETS_TO_BUILD" ;; host) case "$llvm_cv_target_arch" in x86) TARGETS_TO_BUILD="X86 $TARGETS_TO_BUILD" ;; x86_64) TARGETS_TO_BUILD="X86 $TARGETS_TO_BUILD" ;; @@ -508,6 +524,7 @@ case "$enableval" in Alpha) TARGETS_TO_BUILD="Alpha $TARGETS_TO_BUILD" ;; ARM) TARGETS_TO_BUILD="ARM $TARGETS_TO_BUILD" ;; Mips) TARGETS_TO_BUILD="Mips $TARGETS_TO_BUILD" ;; + MBlaze) TARGETS_TO_BUILD="MBlaze $TARGETS_TO_BUILD" ;; CellSPU|SPU) TARGETS_TO_BUILD="CellSPU $TARGETS_TO_BUILD" ;; PIC16) TARGETS_TO_BUILD="PIC16 $TARGETS_TO_BUILD" ;; XCore) TARGETS_TO_BUILD="XCore $TARGETS_TO_BUILD" ;; @@ -612,6 +629,56 @@ if test -n "$LLVMGXX" && test -z "$LLVMGCC"; then AC_MSG_ERROR([Invalid llvm-gcc. Use --with-llvmgcc when --with-llvmgxx is used]); fi +dnl Allow a specific Clang compiler to be used with this LLVM config. +AC_ARG_WITH(clang, + AS_HELP_STRING([--with-clang], + [Specify location of clang compiler (default is --with-built-clang)]), + [],[with_clang=default]) + +dnl Enable use of the built Clang. +AC_ARG_WITH(built-clang, + AS_HELP_STRING([--with-built-clang], + [Use the compiled Clang as the LLVM compiler (default=check)]), + [],[with_built_clang=check]) + +dnl Select the Clang compiler option. +dnl +dnl If --with-clang is given, always honor that; otherwise honor +dnl --with-built-clang, or check if we have the clang sources. +AC_MSG_CHECKING([clang compiler]) +WITH_CLANGPATH="" +WITH_BUILT_CLANG=0 +if test "$with_clang" != "default"; then + WITH_CLANGPATH="$with_clang" + if ! test -x "$WITH_CLANGPATH"; then + AC_MSG_ERROR([invalid --with-clang, path does not specify an executable]) + fi +elif test "$with_built_clang" = "yes"; then + WITH_BUILT_CLANG=1 +elif test "$with_built_clang" = "no"; then + WITH_BUILT_CLANG=0 +else + if test "$with_built_clang" != "check"; then + AC_MSG_ERROR([invalid value for --with-built-clang.]) + fi + + if test -f ${srcdir}/tools/clang/README.txt; then + WITH_BUILT_CLANG=1 + fi +fi + +if ! test -z "$WITH_CLANGPATH"; then + AC_MSG_RESULT([$WITH_CLANGPATH]) + WITH_CLANGXXPATH=`"$WITH_CLANGPATH" --print-prog-name=clang++` +elif test "$WITH_BUILT_CLANG" = "1"; then + AC_MSG_RESULT([built]) +else + AC_MSG_RESULT([none]) +fi +AC_SUBST(CLANGPATH,$WITH_CLANGPATH) +AC_SUBST(CLANGXXPATH,$WITH_CLANGXXPATH) +AC_SUBST(ENABLE_BUILT_CLANG,$WITH_BUILT_CLANG) + dnl Override the option to use for optimized builds. AC_ARG_WITH(optimize-option, AS_HELP_STRING([--with-optimize-option], @@ -946,6 +1013,29 @@ else AC_SUBST(LLVMGXXCOMMAND,$LLVMGXXCOMMAND) fi +dnl Select the LLVM capable compiler to use, we default to using llvm-gcc if +dnl found, otherwise clang if available. +AC_ARG_WITH(llvmcc, + AS_HELP_STRING([--with-llvmcc=<name>], + [Choose the LLVM capable compiler to use (llvm-gcc, clang, or none; default=check)]), + [],[with_llvmcc=check]) +AC_MSG_CHECKING([LLVM capable compiler]) +if test "$with_llvmcc" != "check"; then + if (test "$with_llvmcc" != "llvm-gcc" && + test "$with_llvmcc" != "clang" && + test "$with_llvmcc" != "none"); then + AC_MSG_ERROR([invalid value for --with-llvmcc, expected 'llvm-gcc', 'clang', or 'none'.]) + fi + WITH_LLVMCC="$with_llvmcc" +elif test -n "$LLVMGCC"; then + WITH_LLVMCC=llvm-gcc +elif test -n "$WITH_CLANGPATH" || test "$WITH_BUILT_CLANG" -ne "0"; then + WITH_LLVMCC=clang +else + WITH_LLVMCC=none +fi +AC_MSG_RESULT([$WITH_LLVMCC]) +AC_SUBST(LLVMCC_OPTION,$WITH_LLVMCC) AC_MSG_CHECKING([tool compatibility]) @@ -994,7 +1084,7 @@ fi dnl Tool compatibility is okay if we make it here. AC_MSG_RESULT([ok]) -dnl Check optional compiler flags. +dnl Check optional compiler flags. AC_MSG_CHECKING([optional compiler flags]) CXX_FLAG_CHECK(NO_VARIADIC_MACROS, [-Wno-variadic-macros]) CXX_FLAG_CHECK(NO_MISSING_FIELD_INITIALIZERS, [-Wno-missing-field-initializers]) @@ -1250,12 +1340,6 @@ if test "$llvm_cv_llvmgcc_sanity" = "yes" ; then AC_SUBST(LLVMCC1PLUS,$llvmcc1pluspath) llvmgccdir=`echo "$llvmcc1path" | sed 's,/libexec/.*,,'` AC_SUBST(LLVMGCCDIR,$llvmgccdir) - llvmgcclibexec=`echo "$llvmcc1path" | sed 's,/cc1,,'` - AC_SUBST(LLVMGCCLIBEXEC,$llvmgcclibexec) - llvmgccversion=[`"$LLVMGCC" -dumpversion 2>&1 | sed 's/^\([0-9.]*\).*/\1/'`] - llvmgccmajvers=[`echo $llvmgccversion | sed 's/^\([0-9]\).*/\1/'`] - AC_SUBST(LLVMGCC_VERSION,$llvmgccversion) - AC_SUBST(LLVMGCC_MAJVERS,$llvmgccmajvers) llvmgcclangs=[`"$LLVMGCC" -v --help 2>&1 | grep '^Configured with:' | sed 's/^.*--enable-languages=\([^ ]*\).*/\1/'`] AC_SUBST(LLVMGCC_LANGS,$llvmgcclangs) AC_MSG_RESULT([ok]) @@ -1265,6 +1349,10 @@ dnl Propagate the shared library extension that the libltdl checks did to dnl the Makefiles so we can use it there too AC_SUBST(SHLIBEXT,$libltdl_cv_shlibext) +dnl Propagate the run-time library path variable that the libltdl +dnl checks found to the Makefiles so we can use it there too +AC_SUBST(SHLIBPATH_VAR,$libltdl_cv_shlibpath_var) + # Translate the various configuration directories and other basic # information into substitutions that will end up in Makefile.config.in # that these configured values can be used by the makefiles @@ -1275,7 +1363,7 @@ eval LLVM_PREFIX="${prefix}"; eval LLVM_BINDIR="${prefix}/bin"; eval LLVM_LIBDIR="${prefix}/lib"; eval LLVM_DATADIR="${prefix}/share/llvm"; -eval LLVM_DOCSDIR="${prefix}/docs/llvm"; +eval LLVM_DOCSDIR="${prefix}/share/doc/llvm"; eval LLVM_ETCDIR="${prefix}/etc/llvm"; eval LLVM_INCLUDEDIR="${prefix}/include"; eval LLVM_INFODIR="${prefix}/info"; diff --git a/autoconf/m4/cxx_flag_check.m4 b/autoconf/m4/cxx_flag_check.m4 new file mode 100644 index 0000000..ab09f2a --- /dev/null +++ b/autoconf/m4/cxx_flag_check.m4 @@ -0,0 +1,2 @@ +AC_DEFUN([CXX_FLAG_CHECK], + [AC_SUBST($1, `$CXX $2 -fsyntax-only -xc /dev/null 2>/dev/null && echo $2`)]) diff --git a/autoconf/m4/huge_val.m4 b/autoconf/m4/huge_val.m4 index fd94c11..5fffbfc 100644 --- a/autoconf/m4/huge_val.m4 +++ b/autoconf/m4/huge_val.m4 @@ -5,6 +5,7 @@ AC_DEFUN([AC_HUGE_VAL_CHECK],[ AC_CACHE_CHECK([for HUGE_VAL sanity], [ac_cv_huge_val_sanity],[ AC_LANG_PUSH([C++]) + ac_save_CXXFLAGS=$CXXFLAGS CXXFLAGS=-pedantic AC_RUN_IFELSE( AC_LANG_PROGRAM( @@ -12,6 +13,7 @@ AC_DEFUN([AC_HUGE_VAL_CHECK],[ [double x = HUGE_VAL; return x != x; ]), [ac_cv_huge_val_sanity=yes],[ac_cv_huge_val_sanity=no], [ac_cv_huge_val_sanity=yes]) + CXXFLAGS=$ac_save_CXXFLAGS AC_LANG_POP([C++]) ]) AC_SUBST(HUGE_VAL_SANITY,$ac_cv_huge_val_sanity) diff --git a/bindings/ocaml/bitreader/bitreader_ocaml.c b/bindings/ocaml/bitreader/bitreader_ocaml.c index 5fd9f85..ef72ce2 100644 --- a/bindings/ocaml/bitreader/bitreader_ocaml.c +++ b/bindings/ocaml/bitreader/bitreader_ocaml.c @@ -46,17 +46,16 @@ static void llvm_raise(value Prototype, char *Message) { /*===-- Modules -----------------------------------------------------------===*/ /* Llvm.llcontext -> Llvm.llmemorybuffer -> Llvm.llmodule */ -CAMLprim value llvm_get_module_provider(LLVMContextRef C, - LLVMMemoryBufferRef MemBuf) { +CAMLprim value llvm_get_module(LLVMContextRef C, LLVMMemoryBufferRef MemBuf) { CAMLparam0(); CAMLlocal2(Variant, MessageVal); char *Message; - LLVMModuleProviderRef MP; - if (LLVMGetBitcodeModuleProviderInContext(C, MemBuf, &MP, &Message)) + LLVMModuleRef M; + if (LLVMGetBitcodeModuleInContext(C, MemBuf, &M, &Message)) llvm_raise(llvm_bitreader_error_exn, Message); - CAMLreturn((value) MemBuf); + CAMLreturn((value) M); } /* Llvm.llcontext -> Llvm.llmemorybuffer -> Llvm.llmodule */ diff --git a/bindings/ocaml/bitreader/llvm_bitreader.ml b/bindings/ocaml/bitreader/llvm_bitreader.ml index 88587cb..8b9d01d 100644 --- a/bindings/ocaml/bitreader/llvm_bitreader.ml +++ b/bindings/ocaml/bitreader/llvm_bitreader.ml @@ -13,9 +13,8 @@ exception Error of string external register_exns : exn -> unit = "llvm_register_bitreader_exns" let _ = register_exns (Error "") -external get_module_provider : Llvm.llcontext -> Llvm.llmemorybuffer -> - Llvm.llmoduleprovider - = "llvm_get_module_provider" +external get_module : Llvm.llcontext -> Llvm.llmemorybuffer -> Llvm.llmodule + = "llvm_get_module" external parse_bitcode : Llvm.llcontext -> Llvm.llmemorybuffer -> Llvm.llmodule = "llvm_parse_bitcode" diff --git a/bindings/ocaml/bitreader/llvm_bitreader.mli b/bindings/ocaml/bitreader/llvm_bitreader.mli index 5648b35..5e22409 100644 --- a/bindings/ocaml/bitreader/llvm_bitreader.mli +++ b/bindings/ocaml/bitreader/llvm_bitreader.mli @@ -14,14 +14,12 @@ exception Error of string -(** [get_module_provider context mb] reads the bitcode for a new - module provider [m] from the memory buffer [mb] in the context [context]. - Returns [m] if successful, or raises [Error msg] otherwise, where [msg] is a - description of the error encountered. See the function - [llvm::getBitcodeModuleProvider]. *) -external get_module_provider : Llvm.llcontext -> Llvm.llmemorybuffer -> - Llvm.llmoduleprovider - = "llvm_get_module_provider" +(** [get_module context mb] reads the bitcode for a new module [m] from the + memory buffer [mb] in the context [context]. Returns [m] if successful, or + raises [Error msg] otherwise, where [msg] is a description of the error + encountered. See the function [llvm::getBitcodeModule]. *) +external get_module : Llvm.llcontext -> Llvm.llmemorybuffer -> Llvm.llmodule + = "llvm_get_module" (** [parse_bitcode context mb] parses the bitcode for a new module [m] from the memory buffer [mb] in the context [context]. Returns [m] if successful, or diff --git a/bindings/ocaml/executionengine/executionengine_ocaml.c b/bindings/ocaml/executionengine/executionengine_ocaml.c index 072d583..1d3e57a 100644 --- a/bindings/ocaml/executionengine/executionengine_ocaml.c +++ b/bindings/ocaml/executionengine/executionengine_ocaml.c @@ -168,41 +168,31 @@ CAMLprim value llvm_genericvalue_as_nativeint(value GenVal) { /*--... Operations on execution engines ....................................--*/ -/* llmoduleprovider -> ExecutionEngine.t */ -CAMLprim LLVMExecutionEngineRef llvm_ee_create(LLVMModuleProviderRef MP) { +/* llmodule -> ExecutionEngine.t */ +CAMLprim LLVMExecutionEngineRef llvm_ee_create(LLVMModuleRef M) { LLVMExecutionEngineRef Interp; char *Error; - if (LLVMCreateExecutionEngine(&Interp, MP, &Error)) + if (LLVMCreateExecutionEngineForModule(&Interp, M, &Error)) llvm_raise(llvm_ee_error_exn, Error); return Interp; } -/* llmoduleprovider -> ExecutionEngine.t */ +/* llmodule -> ExecutionEngine.t */ CAMLprim LLVMExecutionEngineRef -llvm_ee_create_interpreter(LLVMModuleProviderRef MP) { +llvm_ee_create_interpreter(LLVMModuleRef M) { LLVMExecutionEngineRef Interp; char *Error; - if (LLVMCreateInterpreter(&Interp, MP, &Error)) + if (LLVMCreateInterpreterForModule(&Interp, M, &Error)) llvm_raise(llvm_ee_error_exn, Error); return Interp; } -/* llmoduleprovider -> ExecutionEngine.t */ +/* llmodule -> int -> ExecutionEngine.t */ CAMLprim LLVMExecutionEngineRef -llvm_ee_create_jit(LLVMModuleProviderRef MP) { +llvm_ee_create_jit(LLVMModuleRef M, value OptLevel) { LLVMExecutionEngineRef JIT; char *Error; - if (LLVMCreateJITCompiler(&JIT, MP, 3, &Error)) - llvm_raise(llvm_ee_error_exn, Error); - return JIT; -} - -/* llmoduleprovider -> ExecutionEngine.t */ -CAMLprim LLVMExecutionEngineRef -llvm_ee_create_fast_jit(LLVMModuleProviderRef MP) { - LLVMExecutionEngineRef JIT; - char *Error; - if (LLVMCreateJITCompiler(&JIT, MP, 0, &Error)) + if (LLVMCreateJITCompilerForModule(&JIT, M, Int_val(OptLevel), &Error)) llvm_raise(llvm_ee_error_exn, Error); return JIT; } @@ -213,19 +203,18 @@ CAMLprim value llvm_ee_dispose(LLVMExecutionEngineRef EE) { return Val_unit; } -/* llmoduleprovider -> ExecutionEngine.t -> unit */ -CAMLprim value llvm_ee_add_mp(LLVMModuleProviderRef MP, - LLVMExecutionEngineRef EE) { - LLVMAddModuleProvider(EE, MP); +/* llmodule -> ExecutionEngine.t -> unit */ +CAMLprim value llvm_ee_add_mp(LLVMModuleRef M, LLVMExecutionEngineRef EE) { + LLVMAddModule(EE, M); return Val_unit; } -/* llmoduleprovider -> ExecutionEngine.t -> llmodule */ -CAMLprim LLVMModuleRef llvm_ee_remove_mp(LLVMModuleProviderRef MP, +/* llmodule -> ExecutionEngine.t -> llmodule */ +CAMLprim LLVMModuleRef llvm_ee_remove_mp(LLVMModuleRef M, LLVMExecutionEngineRef EE) { LLVMModuleRef RemovedModule; char *Error; - if (LLVMRemoveModuleProvider(EE, MP, &RemovedModule, &Error)) + if (LLVMRemoveModule(EE, M, &RemovedModule, &Error)) llvm_raise(llvm_ee_error_exn, Error); return RemovedModule; } @@ -237,7 +226,7 @@ CAMLprim value llvm_ee_find_function(value Name, LLVMExecutionEngineRef EE) { LLVMValueRef Found; if (LLVMFindFunction(EE, String_val(Name), &Found)) CAMLreturn(Val_unit); - Option = alloc(1, 1); + Option = alloc(1, 0); Field(Option, 0) = Val_op(Found); CAMLreturn(Option); } diff --git a/bindings/ocaml/executionengine/llvm_executionengine.ml b/bindings/ocaml/executionengine/llvm_executionengine.ml index c9e8f18..95faa58 100644 --- a/bindings/ocaml/executionengine/llvm_executionengine.ml +++ b/bindings/ocaml/executionengine/llvm_executionengine.ml @@ -56,19 +56,17 @@ module ExecutionEngine = struct call into LLVM. *) let _ = register_exns (Error "") - external create: Llvm.llmoduleprovider -> t + external create: Llvm.llmodule -> t = "llvm_ee_create" - external create_interpreter: Llvm.llmoduleprovider -> t + external create_interpreter: Llvm.llmodule -> t = "llvm_ee_create_interpreter" - external create_jit: Llvm.llmoduleprovider -> t + external create_jit: Llvm.llmodule -> int -> t = "llvm_ee_create_jit" - external create_fast_jit: Llvm.llmoduleprovider -> t - = "llvm_ee_create_fast_jit" external dispose: t -> unit = "llvm_ee_dispose" - external add_module_provider: Llvm.llmoduleprovider -> t -> unit + external add_module: Llvm.llmodule -> t -> unit = "llvm_ee_add_mp" - external remove_module_provider: Llvm.llmoduleprovider -> t -> Llvm.llmodule + external remove_module: Llvm.llmodule -> t -> Llvm.llmodule = "llvm_ee_remove_mp" external find_function: string -> t -> Llvm.llvalue option = "llvm_ee_find_function" diff --git a/bindings/ocaml/executionengine/llvm_executionengine.mli b/bindings/ocaml/executionengine/llvm_executionengine.mli index 6c2fdfb..ac6665b 100644 --- a/bindings/ocaml/executionengine/llvm_executionengine.mli +++ b/bindings/ocaml/executionengine/llvm_executionengine.mli @@ -85,48 +85,38 @@ module ExecutionEngine: sig invoking a static compiler and generating a native executable. *) type t - (** [create mp] creates a new execution engine, taking ownership of the - module provider [mp] if successful. Creates a JIT if possible, else falls - back to an interpreter. Raises [Error msg] if an error occurrs. The - execution engine is not garbage collected and must be destroyed with - [dispose ee]. See the function [llvm::EngineBuilder::create]. *) - val create: Llvm.llmoduleprovider -> t - - (** [create_interpreter mp] creates a new interpreter, taking ownership of the - module provider [mp] if successful. Raises [Error msg] if an error - occurrs. The execution engine is not garbage collected and must be - destroyed with [dispose ee]. + (** [create m] creates a new execution engine, taking ownership of the + module [m] if successful. Creates a JIT if possible, else falls back to an + interpreter. Raises [Error msg] if an error occurrs. The execution engine + is not garbage collected and must be destroyed with [dispose ee]. See the function [llvm::EngineBuilder::create]. *) - val create_interpreter: Llvm.llmoduleprovider -> t + val create: Llvm.llmodule -> t - (** [create_jit mp] creates a new JIT (just-in-time compiler), taking - ownership of the module provider [mp] if successful. This function creates - a JIT which favors code quality over compilation speed. Raises [Error msg] - if an error occurrs. The execution engine is not garbage collected and - must be destroyed with [dispose ee]. + (** [create_interpreter m] creates a new interpreter, taking ownership of the + module [m] if successful. Raises [Error msg] if an error occurrs. The + execution engine is not garbage collected and must be destroyed with + [dispose ee]. See the function [llvm::EngineBuilder::create]. *) - val create_jit: Llvm.llmoduleprovider -> t + val create_interpreter: Llvm.llmodule -> t - (** [create_fast_jit mp] creates a new JIT (just-in-time compiler) which - favors compilation speed over code quality. It takes ownership of the - module provider [mp] if successful. Raises [Error msg] if an error - occurrs. The execution engine is not garbage collected and must be - destroyed with [dispose ee]. + (** [create_jit m optlevel] creates a new JIT (just-in-time compiler), taking + ownership of the module [m] if successful with the desired optimization + level [optlevel]. Raises [Error msg] if an error occurrs. The execution + engine is not garbage collected and must be destroyed with [dispose ee]. See the function [llvm::EngineBuilder::create]. *) - val create_fast_jit: Llvm.llmoduleprovider -> t - + val create_jit : Llvm.llmodule -> int -> t + (** [dispose ee] releases the memory used by the execution engine and must be invoked to avoid memory leaks. *) val dispose: t -> unit - (** [add_module_provider mp ee] adds the module provider [mp] to the execution - engine [ee]. *) - val add_module_provider: Llvm.llmoduleprovider -> t -> unit + (** [add_module m ee] adds the module [m] to the execution engine [ee]. *) + val add_module: Llvm.llmodule -> t -> unit - (** [remove_module_provider mp ee] removes the module provider [mp] from the - execution engine [ee], disposing of [mp] and the module referenced by - [mp]. Raises [Error msg] if an error occurs. *) - val remove_module_provider: Llvm.llmoduleprovider -> t -> Llvm.llmodule + (** [remove_module m ee] removes the module [m] from the execution engine + [ee], disposing of [m] and the module referenced by [mp]. Raises + [Error msg] if an error occurs. *) + val remove_module: Llvm.llmodule -> t -> Llvm.llmodule (** [find_function n ee] finds the function named [n] defined in any of the modules owned by the execution engine [ee]. Returns [None] if the function diff --git a/bindings/ocaml/llvm/llvm.ml b/bindings/ocaml/llvm/llvm.ml index 7e4acbf..407c1fc 100644 --- a/bindings/ocaml/llvm/llvm.ml +++ b/bindings/ocaml/llvm/llvm.ml @@ -13,9 +13,9 @@ type llmodule type lltype type lltypehandle type llvalue +type lluse type llbasicblock type llbuilder -type llmoduleprovider type llmemorybuffer module TypeKind = struct @@ -35,6 +35,7 @@ module TypeKind = struct | Opaque | Vector | Metadata + | Union end module Linkage = struct @@ -147,6 +148,7 @@ type ('a, 'b) llrev_pos = external create_context : unit -> llcontext = "llvm_create_context" external dispose_context : llcontext -> unit = "llvm_dispose_context" external global_context : unit -> llcontext = "llvm_global_context" +external mdkind_id : llcontext -> string -> int = "llvm_mdkind_id" (*===-- Modules -----------------------------------------------------------===*) external create_module : llcontext -> string -> llmodule = "llvm_create_module" @@ -163,6 +165,8 @@ external define_type_name : string -> lltype -> llmodule -> bool = "llvm_add_type_name" external delete_type_name : string -> llmodule -> unit = "llvm_delete_type_name" +external type_by_name : llmodule -> string -> lltype option + = "llvm_type_by_name" external dump_module : llmodule -> unit = "llvm_dump_module" (*===-- Types -------------------------------------------------------------===*) @@ -198,9 +202,15 @@ external param_types : lltype -> lltype array = "llvm_param_types" external struct_type : llcontext -> lltype array -> lltype = "llvm_struct_type" external packed_struct_type : llcontext -> lltype array -> lltype = "llvm_packed_struct_type" -external element_types : lltype -> lltype array = "llvm_element_types" +external struct_element_types : lltype -> lltype array + = "llvm_struct_element_types" external is_packed : lltype -> bool = "llvm_is_packed" +(*--... Operations on union types ..........................................--*) +external union_type : llcontext -> lltype array -> lltype = "llvm_union_type" +external union_element_types : lltype -> lltype array + = "llvm_union_element_types" + (*--... Operations on pointer, vector, and array types .....................--*) external array_type : lltype -> int -> lltype = "llvm_array_type" external pointer_type : lltype -> lltype = "llvm_pointer_type" @@ -229,15 +239,63 @@ external type_of : llvalue -> lltype = "llvm_type_of" external value_name : llvalue -> string = "llvm_value_name" external set_value_name : string -> llvalue -> unit = "llvm_set_value_name" external dump_value : llvalue -> unit = "llvm_dump_value" +external replace_all_uses_with : llvalue -> llvalue -> unit + = "LLVMReplaceAllUsesWith" + +(*--... Operations on uses .................................................--*) +external use_begin : llvalue -> lluse option = "llvm_use_begin" +external use_succ : lluse -> lluse option = "llvm_use_succ" +external user : lluse -> llvalue = "llvm_user" +external used_value : lluse -> llvalue = "llvm_used_value" + +let iter_uses f v = + let rec aux = function + | None -> () + | Some u -> + f u; + aux (use_succ u) + in + aux (use_begin v) + +let fold_left_uses f init v = + let rec aux init u = + match u with + | None -> init + | Some u -> aux (f init u) (use_succ u) + in + aux init (use_begin v) + +let fold_right_uses f v init = + let rec aux u init = + match u with + | None -> init + | Some u -> f u (aux (use_succ u) init) + in + aux (use_begin v) init + + +(*--... Operations on users ................................................--*) +external operand : llvalue -> int -> llvalue = "llvm_operand" (*--... Operations on constants of (mostly) any type .......................--*) external is_constant : llvalue -> bool = "llvm_is_constant" external const_null : lltype -> llvalue = "LLVMConstNull" external const_all_ones : (*int|vec*)lltype -> llvalue = "LLVMConstAllOnes" +external const_pointer_null : lltype -> llvalue = "LLVMConstPointerNull" external undef : lltype -> llvalue = "LLVMGetUndef" external is_null : llvalue -> bool = "llvm_is_null" external is_undef : llvalue -> bool = "llvm_is_undef" +(*--... Operations on instructions .........................................--*) +external has_metadata : llvalue -> bool = "llvm_has_metadata" +external metadata : llvalue -> int -> llvalue option = "llvm_metadata" +external set_metadata : llvalue -> int -> llvalue -> unit = "llvm_set_metadata" +external clear_metadata : llvalue -> int -> unit = "llvm_clear_metadata" + +(*--... Operations on metadata .......,.....................................--*) +external mdstring : llcontext -> string -> llvalue = "llvm_mdstring" +external mdnode : llcontext -> llvalue array -> llvalue = "llvm_mdnode" + (*--... Operations on scalar constants .....................................--*) external const_int : lltype -> int -> llvalue = "llvm_const_int" external const_of_int64 : lltype -> Int64.t -> bool -> llvalue @@ -257,19 +315,27 @@ external const_struct : llcontext -> llvalue array -> llvalue external const_packed_struct : llcontext -> llvalue array -> llvalue = "llvm_const_packed_struct" external const_vector : llvalue array -> llvalue = "llvm_const_vector" +external const_union : lltype -> llvalue -> llvalue = "LLVMConstUnion" (*--... Constant expressions ...............................................--*) external align_of : lltype -> llvalue = "LLVMAlignOf" external size_of : lltype -> llvalue = "LLVMSizeOf" external const_neg : llvalue -> llvalue = "LLVMConstNeg" +external const_nsw_neg : llvalue -> llvalue = "LLVMConstNSWNeg" +external const_nuw_neg : llvalue -> llvalue = "LLVMConstNUWNeg" external const_fneg : llvalue -> llvalue = "LLVMConstFNeg" external const_not : llvalue -> llvalue = "LLVMConstNot" external const_add : llvalue -> llvalue -> llvalue = "LLVMConstAdd" external const_nsw_add : llvalue -> llvalue -> llvalue = "LLVMConstNSWAdd" +external const_nuw_add : llvalue -> llvalue -> llvalue = "LLVMConstNUWAdd" external const_fadd : llvalue -> llvalue -> llvalue = "LLVMConstFAdd" external const_sub : llvalue -> llvalue -> llvalue = "LLVMConstSub" +external const_nsw_sub : llvalue -> llvalue -> llvalue = "LLVMConstNSWSub" +external const_nuw_sub : llvalue -> llvalue -> llvalue = "LLVMConstNUWSub" external const_fsub : llvalue -> llvalue -> llvalue = "LLVMConstFSub" external const_mul : llvalue -> llvalue -> llvalue = "LLVMConstMul" +external const_nsw_mul : llvalue -> llvalue -> llvalue = "LLVMConstNSWMul" +external const_nuw_mul : llvalue -> llvalue -> llvalue = "LLVMConstNUWMul" external const_fmul : llvalue -> llvalue -> llvalue = "LLVMConstFMul" external const_udiv : llvalue -> llvalue -> llvalue = "LLVMConstUDiv" external const_sdiv : llvalue -> llvalue -> llvalue = "LLVMConstSDiv" @@ -325,6 +391,10 @@ external const_extractvalue : llvalue -> int array -> llvalue = "llvm_const_extractvalue" external const_insertvalue : llvalue -> llvalue -> int array -> llvalue = "llvm_const_insertvalue" +external const_inline_asm : lltype -> string -> string -> bool -> bool -> + llvalue + = "llvm_const_inline_asm" +external block_address : llvalue -> llbasicblock -> llvalue = "LLVMBlockAddress" (*--... Operations on global variables, functions, and aliases (globals) ...--*) external global_parent : llvalue -> llmodule = "LLVMGetGlobalParent" @@ -344,8 +414,14 @@ external set_global_constant : bool -> llvalue -> unit (*--... Operations on global variables .....................................--*) external declare_global : lltype -> string -> llmodule -> llvalue = "llvm_declare_global" +external declare_qualified_global : lltype -> string -> int -> llmodule -> + llvalue + = "llvm_declare_qualified_global" external define_global : string -> llvalue -> llmodule -> llvalue = "llvm_define_global" +external define_qualified_global : string -> llvalue -> int -> llmodule -> + llvalue + = "llvm_define_qualified_global" external lookup_global : string -> llmodule -> llvalue option = "llvm_lookup_global" external delete_global : llvalue -> unit = "llvm_delete_global" @@ -403,6 +479,10 @@ let rec fold_right_global_range f i e init = let fold_right_globals f m init = fold_right_global_range f (global_end m) (At_start m) init +(*--... Operations on aliases ..............................................--*) +external add_alias : llmodule -> lltype -> llvalue -> string -> llvalue + = "llvm_add_alias" + (*--... Operations on functions ............................................--*) external declare_function : string -> lltype -> llmodule -> llvalue = "llvm_declare_function" @@ -680,6 +760,17 @@ let position_before i = position_builder (Before i) let position_at_end bb = position_builder (At_end bb) +(*--... Metadata ...........................................................--*) +external set_current_debug_location : llbuilder -> llvalue -> unit + = "llvm_set_current_debug_location" +external clear_current_debug_location : llbuilder -> unit + = "llvm_clear_current_debug_location" +external current_debug_location : llbuilder -> llvalue option + = "llvm_current_debug_location" +external set_inst_debug_location : llbuilder -> llvalue -> unit + = "llvm_set_inst_debug_location" + + (*--... Terminators ........................................................--*) external build_ret_void : llbuilder -> llvalue = "llvm_build_ret_void" external build_ret : llvalue -> llbuilder -> llvalue = "llvm_build_ret" @@ -692,6 +783,10 @@ external build_switch : llvalue -> llbasicblock -> int -> llbuilder -> llvalue = "llvm_build_switch" external add_case : llvalue -> llvalue -> llbasicblock -> unit = "llvm_add_case" +external build_indirect_br : llvalue -> int -> llbuilder -> llvalue + = "llvm_build_indirect_br" +external add_destination : llvalue -> llbasicblock -> unit + = "llvm_add_destination" external build_invoke : llvalue -> llvalue array -> llbasicblock -> llbasicblock -> string -> llbuilder -> llvalue = "llvm_build_invoke_bc" "llvm_build_invoke_nat" @@ -703,14 +798,24 @@ external build_add : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_add" external build_nsw_add : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_nsw_add" +external build_nuw_add : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nuw_add" external build_fadd : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_fadd" external build_sub : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_sub" +external build_nsw_sub : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nsw_sub" +external build_nuw_sub : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nuw_sub" external build_fsub : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_fsub" external build_mul : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_mul" +external build_nsw_mul : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nsw_mul" +external build_nuw_mul : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nuw_mul" external build_fmul : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_fmul" external build_udiv : llvalue -> llvalue -> string -> llbuilder -> llvalue @@ -741,19 +846,20 @@ external build_xor : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_xor" external build_neg : llvalue -> string -> llbuilder -> llvalue = "llvm_build_neg" +external build_nsw_neg : llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nsw_neg" +external build_nuw_neg : llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nuw_neg" +external build_fneg : llvalue -> string -> llbuilder -> llvalue + = "llvm_build_fneg" external build_not : llvalue -> string -> llbuilder -> llvalue = "llvm_build_not" (*--... Memory .............................................................--*) -external build_malloc : lltype -> string -> llbuilder -> llvalue - = "llvm_build_malloc" -external build_array_malloc : lltype -> llvalue -> string -> llbuilder -> - llvalue = "llvm_build_array_malloc" external build_alloca : lltype -> string -> llbuilder -> llvalue = "llvm_build_alloca" external build_array_alloca : lltype -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_array_alloca" -external build_free : llvalue -> llbuilder -> llvalue = "llvm_build_free" external build_load : llvalue -> string -> llbuilder -> llvalue = "llvm_build_load" external build_store : llvalue -> llvalue -> llbuilder -> llvalue @@ -841,14 +947,6 @@ external build_is_not_null : llvalue -> string -> llbuilder -> llvalue external build_ptrdiff : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_ptrdiff" -(*===-- Module providers --------------------------------------------------===*) - -module ModuleProvider = struct - external create : llmodule -> llmoduleprovider - = "LLVMCreateModuleProviderForExistingModule" - external dispose : llmoduleprovider -> unit = "llvm_dispose_module_provider" -end - (*===-- Memory buffers ----------------------------------------------------===*) @@ -865,7 +963,7 @@ module PassManager = struct type 'a t type any = [ `Module | `Function ] external create : unit -> [ `Module ] t = "llvm_passmanager_create" - external create_function : llmoduleprovider -> [ `Function ] t + external create_function : llmodule -> [ `Function ] t = "LLVMCreateFunctionPassManager" external run_module : llmodule -> [ `Module ] t -> bool = "llvm_passmanager_run_module" @@ -897,11 +995,14 @@ let rec string_of_lltype ty = | TypeKind.Pointer -> (string_of_lltype (element_type ty)) ^ "*" | TypeKind.Struct -> let s = "{ " ^ (concat2 ", " ( - Array.map string_of_lltype (element_types ty) + Array.map string_of_lltype (struct_element_types ty) )) ^ " }" in if is_packed ty then "<" ^ s ^ ">" else s + | TypeKind.Union -> "union { " ^ (concat2 ", " ( + Array.map string_of_lltype (union_element_types ty) + )) ^ " }" | TypeKind.Array -> "[" ^ (string_of_int (array_length ty)) ^ " x " ^ (string_of_lltype (element_type ty)) ^ "]" | TypeKind.Vector -> "<" ^ (string_of_int (vector_size ty)) ^ diff --git a/bindings/ocaml/llvm/llvm.mli b/bindings/ocaml/llvm/llvm.mli index 4f28ae6..aa5ea76 100644 --- a/bindings/ocaml/llvm/llvm.mli +++ b/bindings/ocaml/llvm/llvm.mli @@ -39,6 +39,9 @@ type lltypehandle This type covers a wide range of subclasses. *) type llvalue +(** Used to store users and usees of values. See the [llvm::Use] class. *) +type lluse + (** A basic block in LLVM IR. See the [llvm::BasicBlock] class. *) type llbasicblock @@ -46,10 +49,6 @@ type llbasicblock class. *) type llbuilder -(** Used to provide a module to JIT or interpreter. - See the [llvm::ModuleProvider] class. *) -type llmoduleprovider - (** Used to efficiently handle large buffers of read-only binary data. See the [llvm::MemoryBuffer] class. *) type llmemorybuffer @@ -73,6 +72,7 @@ module TypeKind : sig | Opaque | Vector | Metadata + | Union end (** The linkage of a global value, accessed with {!linkage} and @@ -220,6 +220,11 @@ external dispose_context : llcontext -> unit = "llvm_dispose_context" (** See the function [llvm::getGlobalContext]. *) external global_context : unit -> llcontext = "llvm_global_context" +(** [mdkind_id context name] returns the MDKind ID that corresponds to the + name [name] in the context [context]. See the function + [llvm::LLVMContext::getMDKindID]. *) +external mdkind_id : llcontext -> string -> int = "llvm_mdkind_id" + (** {6 Modules} *) @@ -268,6 +273,11 @@ external define_type_name : string -> lltype -> llmodule -> bool external delete_type_name : string -> llmodule -> unit = "llvm_delete_type_name" +(** [type_by_name m n] returns the type in the module [m] named [n], or [None] + if it does not exist. See the method [llvm::Module::getTypeByName]. *) +external type_by_name : llmodule -> string -> lltype option + = "llvm_type_by_name" + (** [dump_module m] prints the .ll representation of the module [m] to standard error. See the method [llvm::Module::dump]. *) external dump_module : llmodule -> unit = "llvm_dump_module" @@ -381,15 +391,29 @@ external struct_type : llcontext -> lltype array -> lltype external packed_struct_type : llcontext -> lltype array -> lltype = "llvm_packed_struct_type" -(** [element_types sty] returns the constituent types of the struct type [sty]. - See the method [llvm::StructType::getElementType]. *) -external element_types : lltype -> lltype array = "llvm_element_types" +(** [struct_element_types sty] returns the constituent types of the struct type + [sty]. See the method [llvm::StructType::getElementType]. *) +external struct_element_types : lltype -> lltype array + = "llvm_struct_element_types" (** [is_packed sty] returns [true] if the structure type [sty] is packed, [false] otherwise. See the method [llvm::StructType::isPacked]. *) external is_packed : lltype -> bool = "llvm_is_packed" +(** {7 Operations on union types} *) + +(** [union_type context tys] returns the union type in the context [context] + containing the types in the array [tys]. See the method + [llvm::UnionType::get] *) +external union_type : llcontext -> lltype array -> lltype = "llvm_union_type" + +(** [union_element_types uty] returns the constituent types of the union type + [uty]. See the method [llvm::UnionType::getElementType]. *) +external union_element_types : lltype -> lltype array + = "llvm_union_element_types" + + (** {7 Operations on pointer, vector, and array types} *) (** [array_type ty n] returns the array type containing [n] elements of type @@ -482,6 +506,50 @@ external set_value_name : string -> llvalue -> unit = "llvm_set_value_name" error. See the method [llvm::Value::dump]. *) external dump_value : llvalue -> unit = "llvm_dump_value" +(** [replace_all_uses_with old new] replaces all uses of the value [old] + * with the value [new]. See the method [llvm::Value::replaceAllUsesWith]. *) +external replace_all_uses_with : llvalue -> llvalue -> unit + = "LLVMReplaceAllUsesWith" + + +(* {6 Uses} *) + +(** [use_begin v] returns the first position in the use list for the value [v]. + [use_begin] and [use_succ] can e used to iterate over the use list in order. + See the method [llvm::Value::use_begin]. *) +external use_begin : llvalue -> lluse option = "llvm_use_begin" + +(** [use_succ u] returns the use list position succeeding [u]. + See the method [llvm::use_value_iterator::operator++]. *) +external use_succ : lluse -> lluse option = "llvm_use_succ" + +(** [user u] returns the user of the use [u]. + See the method [llvm::Use::getUser]. *) +external user : lluse -> llvalue = "llvm_user" + +(** [used_value u] returns the usee of the use [u]. + See the method [llvm::Use::getUsedValue]. *) +external used_value : lluse -> llvalue = "llvm_used_value" + +(** [iter_uses f v] applies function [f] to each of the users of the value [v] + in order. Tail recursive. *) +val iter_uses : (lluse -> unit) -> llvalue -> unit + +(** [fold_left_uses f init v] is [f (... (f init u1) ...) uN] where + [u1,...,uN] are the users of the value [v]. Tail recursive. *) +val fold_left_uses : ('a -> lluse -> 'a) -> 'a -> llvalue -> 'a + +(** [fold_right_uses f v init] is [f u1 (... (f uN init) ...)] where + [u1,...,uN] are the users of the value [v]. Not tail recursive. *) +val fold_right_uses : (lluse -> 'a -> 'a) -> llvalue -> 'a -> 'a + + +(* {6 Users} *) + +(** [operand v i] returns the operand at index [i] for the value [v]. See the + method [llvm::User::getOperand]. *) +external operand : llvalue -> int -> llvalue = "llvm_operand" + (** {7 Operations on constants of (mostly) any type} *) @@ -497,6 +565,10 @@ external const_null : lltype -> llvalue = "LLVMConstNull" [ty]. See the method [llvm::Constant::getAllOnesValue]. *) external const_all_ones : (*int|vec*)lltype -> llvalue = "LLVMConstAllOnes" +(** [const_pointer_null ty] returns the constant null (zero) pointer of the type + [ty]. See the method [llvm::ConstantPointerNull::get]. *) +external const_pointer_null : lltype -> llvalue = "LLVMConstPointerNull" + (** [undef ty] returns the undefined value of the type [ty]. See the method [llvm::UndefValue::get]. *) external undef : lltype -> llvalue = "LLVMGetUndef" @@ -510,6 +582,39 @@ external is_null : llvalue -> bool = "llvm_is_null" external is_undef : llvalue -> bool = "llvm_is_undef" +(** {7 Operations on instructions} *) + +(** [has_metadata i] returns whether or not the instruction [i] has any + metadata attached to it. See the function + [llvm::Instruction::hasMetadata]. *) +external has_metadata : llvalue -> bool = "llvm_has_metadata" + +(** [metadata i kind] optionally returns the metadata associated with the + kind [kind] in the instruction [i] See the function + [llvm::Instruction::getMetadata]. *) +external metadata : llvalue -> int -> llvalue option = "llvm_metadata" + +(** [set_metadata i kind md] sets the metadata [md] of kind [kind] in the + instruction [i]. See the function [llvm::Instruction::setMetadata]. *) +external set_metadata : llvalue -> int -> llvalue -> unit = "llvm_set_metadata" + +(** [clear_metadata i kind] clears the metadata of kind [kind] in the + instruction [i]. See the function [llvm::Instruction::setMetadata]. *) +external clear_metadata : llvalue -> int -> unit = "llvm_clear_metadata" + + +(** {7 Operations on metadata} *) + +(** [mdstring c s] returns the MDString of the string [s] in the context [c]. + See the method [llvm::MDNode::get]. *) +external mdstring : llcontext -> string -> llvalue = "llvm_mdstring" + +(** [mdnode c elts] returns the MDNode containing the values [elts] in the + context [c]. + See the method [llvm::MDNode::get]. *) +external mdnode : llcontext -> llvalue array -> llvalue = "llvm_mdnode" + + (** {7 Operations on scalar constants} *) (** [const_int ty i] returns the integer constant of type [ty] and value [i]. @@ -577,6 +682,10 @@ external const_packed_struct : llcontext -> llvalue array -> llvalue values [elts]. See the method [llvm::ConstantVector::get]. *) external const_vector : llvalue array -> llvalue = "llvm_const_vector" +(** [const_union ty v] returns the union constant of type [union_type tys] and + containing the value [v]. See the method [llvm::ConstantUnion::get]. *) +external const_union : lltype -> llvalue -> llvalue = "LLVMConstUnion" + (** {7 Constant expressions} *) @@ -596,6 +705,16 @@ external size_of : lltype -> llvalue = "LLVMSizeOf" See the method [llvm::ConstantExpr::getNeg]. *) external const_neg : llvalue -> llvalue = "LLVMConstNeg" +(** [const_nsw_neg c] returns the arithmetic negation of the constant [c] with + no signed wrapping. The result is undefined if the negation overflows. + See the method [llvm::ConstantExpr::getNSWNeg]. *) +external const_nsw_neg : llvalue -> llvalue = "LLVMConstNSWNeg" + +(** [const_nuw_neg c] returns the arithmetic negation of the constant [c] with + no unsigned wrapping. The result is undefined if the negation overflows. + See the method [llvm::ConstantExpr::getNUWNeg]. *) +external const_nuw_neg : llvalue -> llvalue = "LLVMConstNUWNeg" + (** [const_fneg c] returns the arithmetic negation of the constant float [c]. See the method [llvm::ConstantExpr::getFNeg]. *) external const_fneg : llvalue -> llvalue = "LLVMConstFNeg" @@ -613,6 +732,11 @@ external const_add : llvalue -> llvalue -> llvalue = "LLVMConstAdd" See the method [llvm::ConstantExpr::getNSWAdd]. *) external const_nsw_add : llvalue -> llvalue -> llvalue = "LLVMConstNSWAdd" +(** [const_nuw_add c1 c2] returns the constant sum of two constants with no + unsigned wrapping. The result is undefined if the sum overflows. + See the method [llvm::ConstantExpr::getNSWAdd]. *) +external const_nuw_add : llvalue -> llvalue -> llvalue = "LLVMConstNUWAdd" + (** [const_fadd c1 c2] returns the constant sum of two constant floats. See the method [llvm::ConstantExpr::getFAdd]. *) external const_fadd : llvalue -> llvalue -> llvalue = "LLVMConstFAdd" @@ -621,6 +745,16 @@ external const_fadd : llvalue -> llvalue -> llvalue = "LLVMConstFAdd" constants. See the method [llvm::ConstantExpr::getSub]. *) external const_sub : llvalue -> llvalue -> llvalue = "LLVMConstSub" +(** [const_nsw_sub c1 c2] returns the constant difference of two constants with + no signed wrapping. The result is undefined if the sum overflows. + See the method [llvm::ConstantExpr::getNSWSub]. *) +external const_nsw_sub : llvalue -> llvalue -> llvalue = "LLVMConstNSWSub" + +(** [const_nuw_sub c1 c2] returns the constant difference of two constants with + no unsigned wrapping. The result is undefined if the sum overflows. + See the method [llvm::ConstantExpr::getNSWSub]. *) +external const_nuw_sub : llvalue -> llvalue -> llvalue = "LLVMConstNUWSub" + (** [const_fsub c1 c2] returns the constant difference, [c1 - c2], of two constant floats. See the method [llvm::ConstantExpr::getFSub]. *) external const_fsub : llvalue -> llvalue -> llvalue = "LLVMConstFSub" @@ -629,6 +763,16 @@ external const_fsub : llvalue -> llvalue -> llvalue = "LLVMConstFSub" See the method [llvm::ConstantExpr::getMul]. *) external const_mul : llvalue -> llvalue -> llvalue = "LLVMConstMul" +(** [const_nsw_mul c1 c2] returns the constant product of two constants with + no signed wrapping. The result is undefined if the sum overflows. + See the method [llvm::ConstantExpr::getNSWMul]. *) +external const_nsw_mul : llvalue -> llvalue -> llvalue = "LLVMConstNSWMul" + +(** [const_nuw_mul c1 c2] returns the constant product of two constants with + no unsigned wrapping. The result is undefined if the sum overflows. + See the method [llvm::ConstantExpr::getNSWMul]. *) +external const_nuw_mul : llvalue -> llvalue -> llvalue = "LLVMConstNUWMul" + (** [const_fmul c1 c2] returns the constant product of two constants floats. See the method [llvm::ConstantExpr::getFMul]. *) external const_fmul : llvalue -> llvalue -> llvalue = "LLVMConstFMul" @@ -858,6 +1002,16 @@ external const_extractvalue : llvalue -> int array -> llvalue external const_insertvalue : llvalue -> llvalue -> int array -> llvalue = "llvm_const_insertvalue" +(** [const_inline_asm ty asm con side align] inserts a inline assembly string. + See the method [llvm::InlineAsm::get]. *) +external const_inline_asm : lltype -> string -> string -> bool -> bool -> + llvalue + = "llvm_const_inline_asm" + +(** [block_address f bb] returns the address of the basic block [bb] in the + function [f]. See the method [llvm::BasicBlock::get]. *) +external block_address : llvalue -> llbasicblock -> llvalue = "LLVMBlockAddress" + (** {7 Operations on global variables, functions, and aliases (globals)} *) @@ -907,19 +1061,36 @@ external set_alignment : int -> llvalue -> unit = "llvm_set_alignment" (** {7 Operations on global variables} *) (** [declare_global ty name m] returns a new global variable of type [ty] and - with name [name] in module [m]. If such a global variable already exists, - it is returned. If the type of the existing global differs, then a bitcast - to [ty] is returned. *) + with name [name] in module [m] in the default address space (0). If such a + global variable already exists, it is returned. If the type of the existing + global differs, then a bitcast to [ty] is returned. *) external declare_global : lltype -> string -> llmodule -> llvalue = "llvm_declare_global" +(** [declare_qualified_global ty name addrspace m] returns a new global variable + of type [ty] and with name [name] in module [m] in the address space + [addrspace]. If such a global variable already exists, it is returned. If + the type of the existing global differs, then a bitcast to [ty] is + returned. *) +external declare_qualified_global : lltype -> string -> int -> llmodule -> + llvalue + = "llvm_declare_qualified_global" + (** [define_global name init m] returns a new global with name [name] and - initializer [init] in module [m]. If the named global already exists, it is - renamed. + initializer [init] in module [m] in the default address space (0). If the + named global already exists, it is renamed. See the constructor of [llvm::GlobalVariable]. *) external define_global : string -> llvalue -> llmodule -> llvalue = "llvm_define_global" +(** [define_qualified_global name init addrspace m] returns a new global with + name [name] and initializer [init] in module [m] in the address space + [addrspace]. If the named global already exists, it is renamed. + See the constructor of [llvm::GlobalVariable]. *) +external define_qualified_global : string -> llvalue -> int -> llmodule -> + llvalue + = "llvm_define_qualified_global" + (** [lookup_global name m] returns [Some g] if a global variable with name [name] exists in module [m]. If no such global exists, returns [None]. See the [llvm::GlobalVariable] constructor. *) @@ -1008,6 +1179,15 @@ external is_thread_local : llvalue -> bool = "llvm_is_thread_local" external set_thread_local : bool -> llvalue -> unit = "llvm_set_thread_local" +(** {7 Operations on aliases} *) + +(** [add_alias m t a n] inserts an alias in the module [m] with the type [t] and + the aliasee [a] with the name [n]. + See the constructor for [llvm::GlobalAlias]. *) +external add_alias : llmodule -> lltype -> llvalue -> string -> llvalue + = "llvm_add_alias" + + (** {7 Operations on functions} *) (** [declare_function name ty m] returns a new function of type [ty] and @@ -1397,6 +1577,30 @@ external insertion_block : llbuilder -> llbasicblock = "llvm_insertion_block" external insert_into_builder : llvalue -> string -> llbuilder -> unit = "llvm_insert_into_builder" +(** {7 Metadata} *) + +(** [set_current_debug_location b md] sets the current debug location [md] in + the builder [b]. + See the method [llvm::IRBuilder::SetDebugLocation]. *) +external set_current_debug_location : llbuilder -> llvalue -> unit + = "llvm_set_current_debug_location" + +(** [clear_current_debug_location b] clears the current debug location in the + builder [b]. *) +external clear_current_debug_location : llbuilder -> unit + = "llvm_clear_current_debug_location" + +(** [current_debug_location b] returns the current debug location, or None + if none is currently set. + See the method [llvm::IRBuilder::GetDebugLocation]. *) +external current_debug_location : llbuilder -> llvalue option + = "llvm_current_debug_location" + +(** [set_inst_debug_location b i] sets the current debug location of the builder + [b] to the instruction [i]. + See the method [llvm::IRBuilder::SetInstDebugLocation]. *) +external set_inst_debug_location : llbuilder -> llvalue -> unit + = "llvm_set_inst_debug_location" (** {7 Terminators} *) @@ -1446,6 +1650,20 @@ external build_switch : llvalue -> llbasicblock -> int -> llbuilder -> llvalue external add_case : llvalue -> llvalue -> llbasicblock -> unit = "llvm_add_case" +(** [build_indirect_br addr count b] creates a + [indirectbr %addr] + instruction at the position specified by the instruction builder [b] with + space reserved for [count] destinations. + See the method [llvm::LLVMBuilder::CreateIndirectBr]. *) +external build_indirect_br : llvalue -> int -> llbuilder -> llvalue + = "llvm_build_indirect_br" + +(** [add_destination br bb] adds the basic block [bb] as a possible branch + location for the indirectbr instruction [br]. + See the method [llvm::IndirectBrInst::addDestination]. **) +external add_destination : llvalue -> llbasicblock -> unit + = "llvm_add_destination" + (** [build_invoke fn args tobb unwindbb name b] creates an [%name = invoke %fn(args) to %tobb unwind %unwindbb] instruction at the position specified by the instruction builder [b]. @@ -1483,6 +1701,13 @@ external build_add : llvalue -> llvalue -> string -> llbuilder -> llvalue external build_nsw_add : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_nsw_add" +(** [build_nuw_add x y name b] creates a + [%name = nuw add %x, %y] + instruction at the position specified by the instruction builder [b]. + See the method [llvm::LLVMBuilder::CreateNUWAdd]. *) +external build_nuw_add : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nuw_add" + (** [build_fadd x y name b] creates a [%name = fadd %x, %y] instruction at the position specified by the instruction builder [b]. @@ -1497,6 +1722,20 @@ external build_fadd : llvalue -> llvalue -> string -> llbuilder -> llvalue external build_sub : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_sub" +(** [build_nsw_sub x y name b] creates a + [%name = nsw sub %x, %y] + instruction at the position specified by the instruction builder [b]. + See the method [llvm::LLVMBuilder::CreateNSWSub]. *) +external build_nsw_sub : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nsw_sub" + +(** [build_nuw_sub x y name b] creates a + [%name = nuw sub %x, %y] + instruction at the position specified by the instruction builder [b]. + See the method [llvm::LLVMBuilder::CreateNUWSub]. *) +external build_nuw_sub : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nuw_sub" + (** [build_fsub x y name b] creates a [%name = fsub %x, %y] instruction at the position specified by the instruction builder [b]. @@ -1511,6 +1750,20 @@ external build_fsub : llvalue -> llvalue -> string -> llbuilder -> llvalue external build_mul : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_mul" +(** [build_nsw_mul x y name b] creates a + [%name = nsw mul %x, %y] + instruction at the position specified by the instruction builder [b]. + See the method [llvm::LLVMBuilder::CreateNSWMul]. *) +external build_nsw_mul : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nsw_mul" + +(** [build_nuw_mul x y name b] creates a + [%name = nuw mul %x, %y] + instruction at the position specified by the instruction builder [b]. + See the method [llvm::LLVMBuilder::CreateNUWMul]. *) +external build_nuw_mul : llvalue -> llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nuw_mul" + (** [build_fmul x y name b] creates a [%name = fmul %x, %y] instruction at the position specified by the instruction builder [b]. @@ -1617,6 +1870,30 @@ external build_xor : llvalue -> llvalue -> string -> llbuilder -> llvalue external build_neg : llvalue -> string -> llbuilder -> llvalue = "llvm_build_neg" +(** [build_nsw_neg x name b] creates a + [%name = nsw sub 0, %x] + instruction at the position specified by the instruction builder [b]. + [-0.0] is used for floating point types to compute the correct sign. + See the method [llvm::LLVMBuilder::CreateNeg]. *) +external build_nsw_neg : llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nsw_neg" + +(** [build_nuw_neg x name b] creates a + [%name = nuw sub 0, %x] + instruction at the position specified by the instruction builder [b]. + [-0.0] is used for floating point types to compute the correct sign. + See the method [llvm::LLVMBuilder::CreateNeg]. *) +external build_nuw_neg : llvalue -> string -> llbuilder -> llvalue + = "llvm_build_nuw_neg" + +(** [build_fneg x name b] creates a + [%name = fsub 0, %x] + instruction at the position specified by the instruction builder [b]. + [-0.0] is used for floating point types to compute the correct sign. + See the method [llvm::LLVMBuilder::CreateFNeg]. *) +external build_fneg : llvalue -> string -> llbuilder -> llvalue + = "llvm_build_fneg" + (** [build_xor x name b] creates a [%name = xor %x, -1] instruction at the position specified by the instruction builder [b]. @@ -1628,20 +1905,6 @@ external build_not : llvalue -> string -> llbuilder -> llvalue (** {7 Memory} *) -(** [build_malloc ty name b] creates a - [%name = malloc %ty] - instruction at the position specified by the instruction builder [b]. - See the method [llvm::LLVMBuilder::CreateAlloca]. *) -external build_malloc : lltype -> string -> llbuilder -> llvalue - = "llvm_build_malloc" - -(** [build_array_malloc ty n name b] creates a - [%name = malloc %ty, %n] - instruction at the position specified by the instruction builder [b]. - See the method [llvm::LLVMBuilder::CreateMalloc]. *) -external build_array_malloc : lltype -> llvalue -> string -> llbuilder -> - llvalue = "llvm_build_array_malloc" - (** [build_alloca ty name b] creates a [%name = alloca %ty] instruction at the position specified by the instruction builder [b]. @@ -1656,12 +1919,6 @@ external build_alloca : lltype -> string -> llbuilder -> llvalue external build_array_alloca : lltype -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_array_alloca" -(** [build_free v b] creates a - [free %v] - instruction at the position specified by the instruction builder [b]. - See the method [llvm::LLVMBuilder::CreateFree]. *) -external build_free : llvalue -> llbuilder -> llvalue = "llvm_build_free" - (** [build_load v name b] creates a [%name = load %v] instruction at the position specified by the instruction builder [b]. @@ -1938,20 +2195,6 @@ external build_is_not_null : llvalue -> string -> llbuilder -> llvalue external build_ptrdiff : llvalue -> llvalue -> string -> llbuilder -> llvalue = "llvm_build_ptrdiff" -(** {6 Module providers} *) - -module ModuleProvider : sig - (** [create_module_provider m] encapsulates [m] in a module provider and takes - ownership of the module. See the constructor - [llvm::ExistingModuleProvider::ExistingModuleProvider]. *) - external create : llmodule -> llmoduleprovider - = "LLVMCreateModuleProviderForExistingModule" - - (** [dispose_module_provider mp] destroys the module provider [mp] as well as - the contained module. *) - external dispose : llmoduleprovider -> unit = "llvm_dispose_module_provider" -end - (** {6 Memory buffers} *) @@ -1983,12 +2226,12 @@ module PassManager : sig See the constructor of [llvm::PassManager]. *) external create : unit -> [ `Module ] t = "llvm_passmanager_create" - (** [PassManager.create_function mp] constructs a new function-by-function - pass pipeline over the module provider [mp]. It does not take ownership of - [mp]. This type of pipeline is suitable for code generation and JIT - compilation tasks. + (** [PassManager.create_function m] constructs a new function-by-function + pass pipeline over the module [m]. It does not take ownership of [m]. + This type of pipeline is suitable for code generation and JIT compilation + tasks. See the constructor of [llvm::FunctionPassManager]. *) - external create_function : llmoduleprovider -> [ `Function ] t + external create_function : llmodule -> [ `Function ] t = "LLVMCreateFunctionPassManager" (** [run_module m pm] initializes, executes on the module [m], and finalizes @@ -2018,7 +2261,7 @@ module PassManager : sig external finalize : [ `Function ] t -> bool = "llvm_passmanager_finalize" (** Frees the memory of a pass pipeline. For function pipelines, does not free - the module provider. + the module. See the destructor of [llvm::BasePassManager]. *) external dispose : [< any ] t -> unit = "llvm_passmanager_dispose" end diff --git a/bindings/ocaml/llvm/llvm_ocaml.c b/bindings/ocaml/llvm/llvm_ocaml.c index 86492d7..d526a05 100644 --- a/bindings/ocaml/llvm/llvm_ocaml.c +++ b/bindings/ocaml/llvm/llvm_ocaml.c @@ -110,6 +110,13 @@ CAMLprim LLVMContextRef llvm_global_context(value Unit) { return LLVMGetGlobalContext(); } +/* llcontext -> string -> int */ +CAMLprim value llvm_mdkind_id(LLVMContextRef C, value Name) { + unsigned MDKindID = LLVMGetMDKindIDInContext(C, String_val(Name), + caml_string_length(Name)); + return Val_int(MDKindID); +} + /*===-- Modules -----------------------------------------------------------===*/ /* llcontext -> string -> llmodule */ @@ -157,6 +164,18 @@ CAMLprim value llvm_delete_type_name(value Name, LLVMModuleRef M) { return Val_unit; } +/* llmodule -> string -> lltype option */ +CAMLprim value llvm_type_by_name(LLVMModuleRef M, value Name) { + CAMLparam1(Name); + LLVMTypeRef T; + if ((T = LLVMGetTypeByName(M, String_val(Name)))) { + value Option = alloc(1, 0); + Field(Option, 0) = (value) T; + CAMLreturn(Option); + } + CAMLreturn(Val_int(0)); +} + /* llmodule -> unit */ CAMLprim value llvm_dump_module(LLVMModuleRef M) { LLVMDumpModule(M); @@ -283,7 +302,7 @@ CAMLprim LLVMTypeRef llvm_packed_struct_type(LLVMContextRef C, } /* lltype -> lltype array */ -CAMLprim value llvm_element_types(LLVMTypeRef StructTy) { +CAMLprim value llvm_struct_element_types(LLVMTypeRef StructTy) { value Tys = alloc(LLVMCountStructElementTypes(StructTy), 0); LLVMGetStructElementTypes(StructTy, (LLVMTypeRef *) Tys); return Tys; @@ -294,6 +313,21 @@ CAMLprim value llvm_is_packed(LLVMTypeRef StructTy) { return Val_bool(LLVMIsPackedStruct(StructTy)); } +/*--... Operations on union types ..........................................--*/ + +/* llcontext -> lltype array -> lltype */ +CAMLprim LLVMTypeRef llvm_union_type(LLVMContextRef C, value ElementTypes) { + return LLVMUnionTypeInContext(C, (LLVMTypeRef *) ElementTypes, + Wosize_val(ElementTypes)); +} + +/* lltype -> lltype array */ +CAMLprim value llvm_union_element_types(LLVMTypeRef UnionTy) { + value Tys = alloc(LLVMCountUnionElementTypes(UnionTy), 0); + LLVMGetUnionElementTypes(UnionTy, (LLVMTypeRef *) Tys); + return Tys; +} + /*--... Operations on array, pointer, and vector types .....................--*/ /* lltype -> int -> lltype */ @@ -406,6 +440,13 @@ CAMLprim value llvm_dump_value(LLVMValueRef Val) { return Val_unit; } +/*--... Operations on users ................................................--*/ + +/* llvalue -> int -> llvalue */ +CAMLprim LLVMValueRef llvm_operand(LLVMValueRef V, value I) { + return LLVMGetOperand(V, Int_val(I)); +} + /*--... Operations on constants of (mostly) any type .......................--*/ /* llvalue -> bool */ @@ -423,6 +464,52 @@ CAMLprim value llvm_is_undef(LLVMValueRef Val) { return Val_bool(LLVMIsUndef(Val)); } +/*--... Operations on instructions .........................................--*/ + +/* llvalue -> bool */ +CAMLprim value llvm_has_metadata(LLVMValueRef Val) { + return Val_bool(LLVMHasMetadata(Val)); +} + +/* llvalue -> int -> llvalue option */ +CAMLprim value llvm_metadata(LLVMValueRef Val, value MDKindID) { + CAMLparam1(MDKindID); + LLVMValueRef MD; + if ((MD = LLVMGetMetadata(Val, Int_val(MDKindID)))) { + value Option = alloc(1, 0); + Field(Option, 0) = (value) MD; + CAMLreturn(Option); + } + CAMLreturn(Val_int(0)); +} + +/* llvalue -> int -> llvalue -> unit */ +CAMLprim value llvm_set_metadata(LLVMValueRef Val, value MDKindID, + LLVMValueRef MD) { + LLVMSetMetadata(Val, Int_val(MDKindID), MD); + return Val_unit; +} + +/* llvalue -> int -> unit */ +CAMLprim value llvm_clear_metadata(LLVMValueRef Val, value MDKindID) { + LLVMSetMetadata(Val, Int_val(MDKindID), NULL); + return Val_unit; +} + + +/*--... Operations on metadata .............................................--*/ + +/* llcontext -> string -> llvalue */ +CAMLprim LLVMValueRef llvm_mdstring(LLVMContextRef C, value S) { + return LLVMMDStringInContext(C, String_val(S), caml_string_length(S)); +} + +/* llcontext -> llvalue array -> llvalue */ +CAMLprim LLVMValueRef llvm_mdnode(LLVMContextRef C, value ElementVals) { + return LLVMMDNodeInContext(C, (LLVMValueRef*) Op_val(ElementVals), + Wosize_val(ElementVals)); +} + /*--... Operations on scalar constants .....................................--*/ /* lltype -> int -> llvalue */ @@ -561,6 +648,14 @@ CAMLprim LLVMValueRef llvm_const_insertvalue(LLVMValueRef Aggregate, CAMLreturnT(LLVMValueRef, result); } +/* lltype -> string -> string -> bool -> bool -> llvalue */ +CAMLprim LLVMValueRef llvm_const_inline_asm(LLVMTypeRef Ty, value Asm, + value Constraints, value HasSideEffects, + value IsAlignStack) { + return LLVMConstInlineAsm(Ty, String_val(Asm), String_val(Constraints), + Bool_val(HasSideEffects), Bool_val(IsAlignStack)); +} + /*--... Operations on global variables, functions, and aliases (globals) ...--*/ /* llvalue -> bool */ @@ -612,6 +707,42 @@ CAMLprim value llvm_set_alignment(value Bytes, LLVMValueRef Global) { return Val_unit; } +/*--... Operations on uses .................................................--*/ + +/* llvalue -> lluse option */ +CAMLprim value llvm_use_begin(LLVMValueRef Val) { + CAMLparam0(); + LLVMUseRef First; + if ((First = LLVMGetFirstUse(Val))) { + value Option = alloc(1, 0); + Field(Option, 0) = (value) First; + CAMLreturn(Option); + } + CAMLreturn(Val_int(0)); +} + +/* lluse -> lluse option */ +CAMLprim value llvm_use_succ(LLVMUseRef U) { + CAMLparam0(); + LLVMUseRef Next; + if ((Next = LLVMGetNextUse(U))) { + value Option = alloc(1, 0); + Field(Option, 0) = (value) Next; + CAMLreturn(Option); + } + CAMLreturn(Val_int(0)); +} + +/* lluse -> llvalue */ +CAMLprim LLVMValueRef llvm_user(LLVMUseRef UR) { + return LLVMGetUser(UR); +} + +/* lluse -> llvalue */ +CAMLprim LLVMValueRef llvm_used_value(LLVMUseRef UR) { + return LLVMGetUsedValue(UR); +} + /*--... Operations on global variables .....................................--*/ DEFINE_ITERATORS(global, Global, LLVMModuleRef, LLVMValueRef, @@ -629,6 +760,20 @@ CAMLprim LLVMValueRef llvm_declare_global(LLVMTypeRef Ty, value Name, return LLVMAddGlobal(M, Ty, String_val(Name)); } +/* lltype -> string -> int -> llmodule -> llvalue */ +CAMLprim LLVMValueRef llvm_declare_qualified_global(LLVMTypeRef Ty, value Name, + value AddressSpace, + LLVMModuleRef M) { + LLVMValueRef GlobalVar; + if ((GlobalVar = LLVMGetNamedGlobal(M, String_val(Name)))) { + if (LLVMGetElementType(LLVMTypeOf(GlobalVar)) != Ty) + return LLVMConstBitCast(GlobalVar, + LLVMPointerType(Ty, Int_val(AddressSpace))); + return GlobalVar; + } + return LLVMAddGlobal(M, Ty, String_val(Name)); +} + /* string -> llmodule -> llvalue option */ CAMLprim value llvm_lookup_global(value Name, LLVMModuleRef M) { CAMLparam1(Name); @@ -650,6 +795,19 @@ CAMLprim LLVMValueRef llvm_define_global(value Name, LLVMValueRef Initializer, return GlobalVar; } +/* string -> llvalue -> int -> llmodule -> llvalue */ +CAMLprim LLVMValueRef llvm_define_qualified_global(value Name, + LLVMValueRef Initializer, + value AddressSpace, + LLVMModuleRef M) { + LLVMValueRef GlobalVar = LLVMAddGlobalInAddressSpace(M, + LLVMTypeOf(Initializer), + String_val(Name), + Int_val(AddressSpace)); + LLVMSetInitializer(GlobalVar, Initializer); + return GlobalVar; +} + /* llvalue -> unit */ CAMLprim value llvm_delete_global(LLVMValueRef GlobalVar) { LLVMDeleteGlobal(GlobalVar); @@ -692,6 +850,13 @@ CAMLprim value llvm_set_global_constant(value Flag, LLVMValueRef GlobalVar) { return Val_unit; } +/*--... Operations on aliases ..............................................--*/ + +CAMLprim LLVMValueRef llvm_add_alias(LLVMModuleRef M, LLVMTypeRef Ty, + LLVMValueRef Aliasee, value Name) { + return LLVMAddAlias(M, Ty, Aliasee, String_val(Name)); +} + /*--... Operations on functions ............................................--*/ DEFINE_ITERATORS(function, Function, LLVMModuleRef, LLVMValueRef, @@ -978,7 +1143,7 @@ CAMLprim value llvm_position_builder(value Pos, value B) { } /* llbuilder -> llbasicblock */ -CAMLprim LLVMBasicBlockRef llvm_insertion_block(LLVMBuilderRef B) { +CAMLprim LLVMBasicBlockRef llvm_insertion_block(value B) { LLVMBasicBlockRef InsertBlock = LLVMGetInsertBlock(Builder_val(B)); if (!InsertBlock) raise_not_found(); @@ -986,12 +1151,44 @@ CAMLprim LLVMBasicBlockRef llvm_insertion_block(LLVMBuilderRef B) { } /* llvalue -> string -> llbuilder -> unit */ -CAMLprim value llvm_insert_into_builder(LLVMValueRef I, value Name, - LLVMBuilderRef B) { - LLVMInsertIntoBuilderWithName(B, I, String_val(Name)); +CAMLprim value llvm_insert_into_builder(LLVMValueRef I, value Name, value B) { + LLVMInsertIntoBuilderWithName(Builder_val(B), I, String_val(Name)); + return Val_unit; +} + +/*--... Metadata ...........................................................--*/ + +/* llbuilder -> llvalue -> unit */ +CAMLprim value llvm_set_current_debug_location(value B, LLVMValueRef V) { + LLVMSetCurrentDebugLocation(Builder_val(B), V); + return Val_unit; +} + +/* llbuilder -> unit */ +CAMLprim value llvm_clear_current_debug_location(value B) { + LLVMSetCurrentDebugLocation(Builder_val(B), NULL); return Val_unit; } +/* llbuilder -> llvalue option */ +CAMLprim value llvm_current_debug_location(value B) { + CAMLparam0(); + LLVMValueRef L; + if ((L = LLVMGetCurrentDebugLocation(Builder_val(B)))) { + value Option = alloc(1, 0); + Field(Option, 0) = (value) L; + CAMLreturn(Option); + } + CAMLreturn(Val_int(0)); +} + +/* llbuilder -> llvalue -> unit */ +CAMLprim value llvm_set_inst_debug_location(value B, LLVMValueRef V) { + LLVMSetInstDebugLocation(Builder_val(B), V); + return Val_unit; +} + + /*--... Terminators ........................................................--*/ /* llbuilder -> llvalue */ @@ -1038,6 +1235,20 @@ CAMLprim value llvm_add_case(LLVMValueRef Switch, LLVMValueRef OnVal, return Val_unit; } +/* llvalue -> llbasicblock -> llbuilder -> llvalue */ +CAMLprim LLVMValueRef llvm_build_indirect_br(LLVMValueRef Addr, + value EstimatedDests, + value B) { + return LLVMBuildIndirectBr(Builder_val(B), Addr, EstimatedDests); +} + +/* llvalue -> llvalue -> llbasicblock -> unit */ +CAMLprim value llvm_add_destination(LLVMValueRef IndirectBr, + LLVMBasicBlockRef Dest) { + LLVMAddDestination(IndirectBr, Dest); + return Val_unit; +} + /* llvalue -> llvalue array -> llbasicblock -> llbasicblock -> string -> llbuilder -> llvalue */ CAMLprim LLVMValueRef llvm_build_invoke_nat(LLVMValueRef Fn, value Args, @@ -1082,6 +1293,12 @@ CAMLprim LLVMValueRef llvm_build_nsw_add(LLVMValueRef LHS, LLVMValueRef RHS, } /* llvalue -> llvalue -> string -> llbuilder -> llvalue */ +CAMLprim LLVMValueRef llvm_build_nuw_add(LLVMValueRef LHS, LLVMValueRef RHS, + value Name, value B) { + return LLVMBuildNUWAdd(Builder_val(B), LHS, RHS, String_val(Name)); +} + +/* llvalue -> llvalue -> string -> llbuilder -> llvalue */ CAMLprim LLVMValueRef llvm_build_fadd(LLVMValueRef LHS, LLVMValueRef RHS, value Name, value B) { return LLVMBuildFAdd(Builder_val(B), LHS, RHS, String_val(Name)); @@ -1094,6 +1311,18 @@ CAMLprim LLVMValueRef llvm_build_sub(LLVMValueRef LHS, LLVMValueRef RHS, } /* llvalue -> llvalue -> string -> llbuilder -> llvalue */ +CAMLprim LLVMValueRef llvm_build_nsw_sub(LLVMValueRef LHS, LLVMValueRef RHS, + value Name, value B) { + return LLVMBuildNSWSub(Builder_val(B), LHS, RHS, String_val(Name)); +} + +/* llvalue -> llvalue -> string -> llbuilder -> llvalue */ +CAMLprim LLVMValueRef llvm_build_nuw_sub(LLVMValueRef LHS, LLVMValueRef RHS, + value Name, value B) { + return LLVMBuildNUWSub(Builder_val(B), LHS, RHS, String_val(Name)); +} + +/* llvalue -> llvalue -> string -> llbuilder -> llvalue */ CAMLprim LLVMValueRef llvm_build_fsub(LLVMValueRef LHS, LLVMValueRef RHS, value Name, value B) { return LLVMBuildFSub(Builder_val(B), LHS, RHS, String_val(Name)); @@ -1106,6 +1335,18 @@ CAMLprim LLVMValueRef llvm_build_mul(LLVMValueRef LHS, LLVMValueRef RHS, } /* llvalue -> llvalue -> string -> llbuilder -> llvalue */ +CAMLprim LLVMValueRef llvm_build_nsw_mul(LLVMValueRef LHS, LLVMValueRef RHS, + value Name, value B) { + return LLVMBuildNSWMul(Builder_val(B), LHS, RHS, String_val(Name)); +} + +/* llvalue -> llvalue -> string -> llbuilder -> llvalue */ +CAMLprim LLVMValueRef llvm_build_nuw_mul(LLVMValueRef LHS, LLVMValueRef RHS, + value Name, value B) { + return LLVMBuildNUWMul(Builder_val(B), LHS, RHS, String_val(Name)); +} + +/* llvalue -> llvalue -> string -> llbuilder -> llvalue */ CAMLprim LLVMValueRef llvm_build_fmul(LLVMValueRef LHS, LLVMValueRef RHS, value Name, value B) { return LLVMBuildFMul(Builder_val(B), LHS, RHS, String_val(Name)); @@ -1196,25 +1437,31 @@ CAMLprim LLVMValueRef llvm_build_neg(LLVMValueRef X, } /* llvalue -> string -> llbuilder -> llvalue */ -CAMLprim LLVMValueRef llvm_build_not(LLVMValueRef X, - value Name, value B) { - return LLVMBuildNot(Builder_val(B), X, String_val(Name)); +CAMLprim LLVMValueRef llvm_build_nsw_neg(LLVMValueRef X, + value Name, value B) { + return LLVMBuildNSWNeg(Builder_val(B), X, String_val(Name)); } -/*--... Memory .............................................................--*/ +/* llvalue -> string -> llbuilder -> llvalue */ +CAMLprim LLVMValueRef llvm_build_nuw_neg(LLVMValueRef X, + value Name, value B) { + return LLVMBuildNUWNeg(Builder_val(B), X, String_val(Name)); +} -/* lltype -> string -> llbuilder -> llvalue */ -CAMLprim LLVMValueRef llvm_build_malloc(LLVMTypeRef Ty, - value Name, value B) { - return LLVMBuildMalloc(Builder_val(B), Ty, String_val(Name)); +/* llvalue -> string -> llbuilder -> llvalue */ +CAMLprim LLVMValueRef llvm_build_fneg(LLVMValueRef X, + value Name, value B) { + return LLVMBuildFNeg(Builder_val(B), X, String_val(Name)); } -/* lltype -> llvalue -> string -> llbuilder -> llvalue */ -CAMLprim LLVMValueRef llvm_build_array_malloc(LLVMTypeRef Ty, LLVMValueRef Size, - value Name, value B) { - return LLVMBuildArrayMalloc(Builder_val(B), Ty, Size, String_val(Name)); +/* llvalue -> string -> llbuilder -> llvalue */ +CAMLprim LLVMValueRef llvm_build_not(LLVMValueRef X, + value Name, value B) { + return LLVMBuildNot(Builder_val(B), X, String_val(Name)); } +/*--... Memory .............................................................--*/ + /* lltype -> string -> llbuilder -> llvalue */ CAMLprim LLVMValueRef llvm_build_alloca(LLVMTypeRef Ty, value Name, value B) { @@ -1227,11 +1474,6 @@ CAMLprim LLVMValueRef llvm_build_array_alloca(LLVMTypeRef Ty, LLVMValueRef Size, return LLVMBuildArrayAlloca(Builder_val(B), Ty, Size, String_val(Name)); } -/* llvalue -> llbuilder -> llvalue */ -CAMLprim LLVMValueRef llvm_build_free(LLVMValueRef Pointer, value B) { - return LLVMBuildFree(Builder_val(B), Pointer); -} - /* llvalue -> string -> llbuilder -> llvalue */ CAMLprim LLVMValueRef llvm_build_load(LLVMValueRef Pointer, value Name, value B) { @@ -1510,14 +1752,6 @@ CAMLprim LLVMValueRef llvm_build_ptrdiff(LLVMValueRef LHS, LLVMValueRef RHS, return LLVMBuildPtrDiff(Builder_val(B), LHS, RHS, String_val(Name)); } -/*===-- Module Providers --------------------------------------------------===*/ - -/* llmoduleprovider -> unit */ -CAMLprim value llvm_dispose_module_provider(LLVMModuleProviderRef MP) { - LLVMDisposeModuleProvider(MP); - return Val_unit; -} - /*===-- Memory buffers ----------------------------------------------------===*/ @@ -689,12 +689,16 @@ TARGET_HAS_JIT ENABLE_DOXYGEN ENABLE_THREADS ENABLE_PIC +ENABLE_SHARED TARGETS_TO_BUILD LLVM_ENUM_TARGETS LLVM_ENUM_ASM_PRINTERS LLVM_ENUM_ASM_PARSERS LLVM_ENUM_DISASSEMBLERS ENABLE_CBE_PRINTF_A +CLANGPATH +CLANGXXPATH +ENABLE_BUILT_CLANG OPTIMIZE_OPTION EXTRA_OPTIONS BINUTILS_INCDIR @@ -754,6 +758,7 @@ LLVMGCCCOMMAND LLVMGXXCOMMAND LLVMGCC LLVMGXX +LLVMCC_OPTION NO_VARIADIC_MACROS NO_MISSING_FIELD_INITIALIZERS USE_UDIS86 @@ -764,11 +769,9 @@ MMAP_FILE LLVMCC1 LLVMCC1PLUS LLVMGCCDIR -LLVMGCCLIBEXEC -LLVMGCC_VERSION -LLVMGCC_MAJVERS LLVMGCC_LANGS SHLIBEXT +SHLIBPATH_VAR LLVM_PREFIX LLVM_BINDIR LLVM_LIBDIR @@ -810,6 +813,7 @@ projects/llvm-java projects/llvm-tv projects/llvm-poolalloc projects/poolalloc +projects/safecode projects/llvm-kernel' # Initialize some variables set by options. @@ -1401,6 +1405,8 @@ Optional Features: --enable-threads Use threads if available (default is YES) --enable-pic Build LLVM with Position Independent Code (default is YES) + --enable-shared Build a shared library and link tools against it + (default is NO) --enable-targets Build specific host targets: all or target1,target2,... Valid targets are: host, x86, x86_64, sparc, powerpc, alpha, arm, mips, spu, @@ -1426,6 +1432,10 @@ Optional Packages: searches PATH) --with-llvmgxx Specify location of llvm-g++ driver (default searches PATH) + --with-clang Specify location of clang compiler (default is + --with-built-clang) + --with-built-clang Use the compiled Clang as the LLVM compiler + (default=check) --with-optimize-option Select the compiler options to use for optimized builds --with-extra-options Specify additional options to compile LLVM with @@ -1442,6 +1452,8 @@ Optional Packages: --with-binutils-include Specify path to binutils/include/ containing plugin-api.h file for gold plugin. --with-tclinclude directory where tcl headers are + --with-llvmcc=<name> Choose the LLVM capable compiler to use (llvm-gcc, + clang, or none; default=check) --with-udis86=<path> Use udis86 external x86 disassembler library --with-oprofile=<prefix> Tell OProfile >= 0.9.4 how to symbolize JIT output @@ -1968,6 +1980,8 @@ do ;; poolalloc) subdirs="$subdirs projects/poolalloc" ;; + safecode) subdirs="$subdirs projects/safecode" + ;; llvm-kernel) subdirs="$subdirs projects/llvm-kernel" ;; *) @@ -2320,6 +2334,7 @@ else msp430-*) llvm_cv_target_arch="MSP430" ;; s390x-*) llvm_cv_target_arch="SystemZ" ;; bfin-*) llvm_cv_target_arch="Blackfin" ;; + mblaze-*) llvm_cv_target_arch="MBlaze" ;; *) llvm_cv_target_arch="Unknown" ;; esac fi @@ -4785,6 +4800,8 @@ else ;; Blackfin) TARGET_HAS_JIT=0 ;; + MBlaze) TARGET_HAS_JIT=0 + ;; *) TARGET_HAS_JIT=0 ;; esac @@ -4857,6 +4874,25 @@ cat >>confdefs.h <<_ACEOF _ACEOF +# Check whether --enable-shared was given. +if test "${enable_shared+set}" = set; then + enableval=$enable_shared; +else + enableval=default +fi + +case "$enableval" in + yes) ENABLE_SHARED=1 + ;; + no) ENABLE_SHARED=0 + ;; + default) ENABLE_SHARED=0 + ;; + *) { { echo "$as_me:$LINENO: error: Invalid setting for --enable-shared. Use \"yes\" or \"no\"" >&5 +echo "$as_me: error: Invalid setting for --enable-shared. Use \"yes\" or \"no\"" >&2;} + { (exit 1); exit 1; }; } ;; +esac + TARGETS_TO_BUILD="" # Check whether --enable-targets was given. if test "${enable_targets+set}" = set; then @@ -4869,7 +4905,7 @@ if test "$enableval" = host-only ; then enableval=host fi case "$enableval" in - all) TARGETS_TO_BUILD="X86 Sparc PowerPC Alpha ARM Mips CellSPU PIC16 XCore MSP430 SystemZ Blackfin CBackend MSIL CppBackend" ;; + all) TARGETS_TO_BUILD="X86 Sparc PowerPC Alpha ARM Mips CellSPU PIC16 XCore MSP430 SystemZ Blackfin CBackend MSIL CppBackend MBlaze" ;; *)for a_target in `echo $enableval|sed -e 's/,/ /g' ` ; do case "$a_target" in x86) TARGETS_TO_BUILD="X86 $TARGETS_TO_BUILD" ;; @@ -4888,6 +4924,7 @@ case "$enableval" in cbe) TARGETS_TO_BUILD="CBackend $TARGETS_TO_BUILD" ;; msil) TARGETS_TO_BUILD="MSIL $TARGETS_TO_BUILD" ;; cpp) TARGETS_TO_BUILD="CppBackend $TARGETS_TO_BUILD" ;; + mblaze) TARGETS_TO_BUILD="MBlaze $TARGETS_TO_BUILD" ;; host) case "$llvm_cv_target_arch" in x86) TARGETS_TO_BUILD="X86 $TARGETS_TO_BUILD" ;; x86_64) TARGETS_TO_BUILD="X86 $TARGETS_TO_BUILD" ;; @@ -4896,6 +4933,7 @@ case "$enableval" in Alpha) TARGETS_TO_BUILD="Alpha $TARGETS_TO_BUILD" ;; ARM) TARGETS_TO_BUILD="ARM $TARGETS_TO_BUILD" ;; Mips) TARGETS_TO_BUILD="Mips $TARGETS_TO_BUILD" ;; + MBlaze) TARGETS_TO_BUILD="MBlaze $TARGETS_TO_BUILD" ;; CellSPU|SPU) TARGETS_TO_BUILD="CellSPU $TARGETS_TO_BUILD" ;; PIC16) TARGETS_TO_BUILD="PIC16 $TARGETS_TO_BUILD" ;; XCore) TARGETS_TO_BUILD="XCore $TARGETS_TO_BUILD" ;; @@ -5029,6 +5067,69 @@ echo "$as_me: error: Invalid llvm-gcc. Use --with-llvmgcc when --with-llvmgxx is fi +# Check whether --with-clang was given. +if test "${with_clang+set}" = set; then + withval=$with_clang; +else + with_clang=default +fi + + + +# Check whether --with-built-clang was given. +if test "${with_built_clang+set}" = set; then + withval=$with_built_clang; +else + with_built_clang=check +fi + + +{ echo "$as_me:$LINENO: checking clang compiler" >&5 +echo $ECHO_N "checking clang compiler... $ECHO_C" >&6; } +WITH_CLANGPATH="" +WITH_BUILT_CLANG=0 +if test "$with_clang" != "default"; then + WITH_CLANGPATH="$with_clang" + if ! test -x "$WITH_CLANGPATH"; then + { { echo "$as_me:$LINENO: error: invalid --with-clang, path does not specify an executable" >&5 +echo "$as_me: error: invalid --with-clang, path does not specify an executable" >&2;} + { (exit 1); exit 1; }; } + fi +elif test "$with_built_clang" = "yes"; then + WITH_BUILT_CLANG=1 +elif test "$with_built_clang" = "no"; then + WITH_BUILT_CLANG=0 +else + if test "$with_built_clang" != "check"; then + { { echo "$as_me:$LINENO: error: invalid value for --with-built-clang." >&5 +echo "$as_me: error: invalid value for --with-built-clang." >&2;} + { (exit 1); exit 1; }; } + fi + + if test -f ${srcdir}/tools/clang/README.txt; then + WITH_BUILT_CLANG=1 + fi +fi + +if ! test -z "$WITH_CLANGPATH"; then + { echo "$as_me:$LINENO: result: $WITH_CLANGPATH" >&5 +echo "${ECHO_T}$WITH_CLANGPATH" >&6; } + WITH_CLANGXXPATH=`"$WITH_CLANGPATH" --print-prog-name=clang++` +elif test "$WITH_BUILT_CLANG" = "1"; then + { echo "$as_me:$LINENO: result: built" >&5 +echo "${ECHO_T}built" >&6; } +else + { echo "$as_me:$LINENO: result: none" >&5 +echo "${ECHO_T}none" >&6; } +fi +CLANGPATH=$WITH_CLANGPATH + +CLANGXXPATH=$WITH_CLANGXXPATH + +ENABLE_BUILT_CLANG=$WITH_BUILT_CLANG + + + # Check whether --with-optimize-option was given. if test "${with_optimize_option+set}" = set; then withval=$with_optimize_option; @@ -11035,7 +11136,7 @@ else lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2 lt_status=$lt_dlunknown cat > conftest.$ac_ext <<EOF -#line 11038 "configure" +#line 11139 "configure" #include "confdefs.h" #if HAVE_DLFCN_H @@ -12805,6 +12906,36 @@ else fi +# Check whether --with-llvmcc was given. +if test "${with_llvmcc+set}" = set; then + withval=$with_llvmcc; +else + with_llvmcc=check +fi + +{ echo "$as_me:$LINENO: checking LLVM capable compiler" >&5 +echo $ECHO_N "checking LLVM capable compiler... $ECHO_C" >&6; } +if test "$with_llvmcc" != "check"; then + if (test "$with_llvmcc" != "llvm-gcc" && + test "$with_llvmcc" != "clang" && + test "$with_llvmcc" != "none"); then + { { echo "$as_me:$LINENO: error: invalid value for --with-llvmcc, expected 'llvm-gcc', 'clang', or 'none'." >&5 +echo "$as_me: error: invalid value for --with-llvmcc, expected 'llvm-gcc', 'clang', or 'none'." >&2;} + { (exit 1); exit 1; }; } + fi + WITH_LLVMCC="$with_llvmcc" +elif test -n "$LLVMGCC"; then + WITH_LLVMCC=llvm-gcc +elif test -n "$WITH_CLANGPATH" || test "$WITH_BUILT_CLANG" -ne "0"; then + WITH_LLVMCC=clang +else + WITH_LLVMCC=none +fi +{ echo "$as_me:$LINENO: result: $WITH_LLVMCC" >&5 +echo "${ECHO_T}$WITH_LLVMCC" >&6; } +LLVMCC_OPTION=$WITH_LLVMCC + + { echo "$as_me:$LINENO: checking tool compatibility" >&5 echo $ECHO_N "checking tool compatibility... $ECHO_C" >&6; } @@ -16413,6 +16544,7 @@ ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5' ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5' ac_compiler_gnu=$ac_cv_cxx_compiler_gnu + ac_save_CXXFLAGS=$CXXFLAGS CXXFLAGS=-pedantic if test "$cross_compiling" = yes; then ac_cv_huge_val_sanity=yes @@ -16465,6 +16597,7 @@ rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$a fi + CXXFLAGS=$ac_save_CXXFLAGS ac_ext=c ac_cpp='$CPP $CPPFLAGS' ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5' @@ -19576,15 +19709,6 @@ echo $ECHO_N "checking llvm-gcc component support... $ECHO_C" >&6; } llvmgccdir=`echo "$llvmcc1path" | sed 's,/libexec/.*,,'` LLVMGCCDIR=$llvmgccdir - llvmgcclibexec=`echo "$llvmcc1path" | sed 's,/cc1,,'` - LLVMGCCLIBEXEC=$llvmgcclibexec - - llvmgccversion=`"$LLVMGCC" -dumpversion 2>&1 | sed 's/^\([0-9.]*\).*/\1/'` - llvmgccmajvers=`echo $llvmgccversion | sed 's/^\([0-9]\).*/\1/'` - LLVMGCC_VERSION=$llvmgccversion - - LLVMGCC_MAJVERS=$llvmgccmajvers - llvmgcclangs=`"$LLVMGCC" -v --help 2>&1 | grep '^Configured with:' | sed 's/^.*--enable-languages=\([^ ]*\).*/\1/'` LLVMGCC_LANGS=$llvmgcclangs @@ -19595,6 +19719,9 @@ fi SHLIBEXT=$libltdl_cv_shlibext +SHLIBPATH_VAR=$libltdl_cv_shlibpath_var + + # Translate the various configuration directories and other basic # information into substitutions that will end up in Makefile.config.in # that these configured values can be used by the makefiles @@ -19605,7 +19732,7 @@ eval LLVM_PREFIX="${prefix}"; eval LLVM_BINDIR="${prefix}/bin"; eval LLVM_LIBDIR="${prefix}/lib"; eval LLVM_DATADIR="${prefix}/share/llvm"; -eval LLVM_DOCSDIR="${prefix}/docs/llvm"; +eval LLVM_DOCSDIR="${prefix}/share/doc/llvm"; eval LLVM_ETCDIR="${prefix}/etc/llvm"; eval LLVM_INCLUDEDIR="${prefix}/include"; eval LLVM_INFODIR="${prefix}/info"; @@ -20652,16 +20779,16 @@ TARGET_HAS_JIT!$TARGET_HAS_JIT$ac_delim ENABLE_DOXYGEN!$ENABLE_DOXYGEN$ac_delim ENABLE_THREADS!$ENABLE_THREADS$ac_delim ENABLE_PIC!$ENABLE_PIC$ac_delim +ENABLE_SHARED!$ENABLE_SHARED$ac_delim TARGETS_TO_BUILD!$TARGETS_TO_BUILD$ac_delim LLVM_ENUM_TARGETS!$LLVM_ENUM_TARGETS$ac_delim LLVM_ENUM_ASM_PRINTERS!$LLVM_ENUM_ASM_PRINTERS$ac_delim LLVM_ENUM_ASM_PARSERS!$LLVM_ENUM_ASM_PARSERS$ac_delim LLVM_ENUM_DISASSEMBLERS!$LLVM_ENUM_DISASSEMBLERS$ac_delim ENABLE_CBE_PRINTF_A!$ENABLE_CBE_PRINTF_A$ac_delim -OPTIMIZE_OPTION!$OPTIMIZE_OPTION$ac_delim -EXTRA_OPTIONS!$EXTRA_OPTIONS$ac_delim -BINUTILS_INCDIR!$BINUTILS_INCDIR$ac_delim -ENABLE_LLVMC_DYNAMIC!$ENABLE_LLVMC_DYNAMIC$ac_delim +CLANGPATH!$CLANGPATH$ac_delim +CLANGXXPATH!$CLANGXXPATH$ac_delim +ENABLE_BUILT_CLANG!$ENABLE_BUILT_CLANG$ac_delim _ACEOF if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 97; then @@ -20703,6 +20830,10 @@ _ACEOF ac_delim='%!_!# ' for ac_last_try in false false false false false :; do cat >conf$$subs.sed <<_ACEOF +OPTIMIZE_OPTION!$OPTIMIZE_OPTION$ac_delim +EXTRA_OPTIONS!$EXTRA_OPTIONS$ac_delim +BINUTILS_INCDIR!$BINUTILS_INCDIR$ac_delim +ENABLE_LLVMC_DYNAMIC!$ENABLE_LLVMC_DYNAMIC$ac_delim ENABLE_LLVMC_DYNAMIC_PLUGINS!$ENABLE_LLVMC_DYNAMIC_PLUGINS$ac_delim CXX!$CXX$ac_delim CXXFLAGS!$CXXFLAGS$ac_delim @@ -20758,6 +20889,7 @@ LLVMGCCCOMMAND!$LLVMGCCCOMMAND$ac_delim LLVMGXXCOMMAND!$LLVMGXXCOMMAND$ac_delim LLVMGCC!$LLVMGCC$ac_delim LLVMGXX!$LLVMGXX$ac_delim +LLVMCC_OPTION!$LLVMCC_OPTION$ac_delim NO_VARIADIC_MACROS!$NO_VARIADIC_MACROS$ac_delim NO_MISSING_FIELD_INITIALIZERS!$NO_MISSING_FIELD_INITIALIZERS$ac_delim USE_UDIS86!$USE_UDIS86$ac_delim @@ -20768,11 +20900,9 @@ MMAP_FILE!$MMAP_FILE$ac_delim LLVMCC1!$LLVMCC1$ac_delim LLVMCC1PLUS!$LLVMCC1PLUS$ac_delim LLVMGCCDIR!$LLVMGCCDIR$ac_delim -LLVMGCCLIBEXEC!$LLVMGCCLIBEXEC$ac_delim -LLVMGCC_VERSION!$LLVMGCC_VERSION$ac_delim -LLVMGCC_MAJVERS!$LLVMGCC_MAJVERS$ac_delim LLVMGCC_LANGS!$LLVMGCC_LANGS$ac_delim SHLIBEXT!$SHLIBEXT$ac_delim +SHLIBPATH_VAR!$SHLIBPATH_VAR$ac_delim LLVM_PREFIX!$LLVM_PREFIX$ac_delim LLVM_BINDIR!$LLVM_BINDIR$ac_delim LLVM_LIBDIR!$LLVM_LIBDIR$ac_delim @@ -20793,7 +20923,7 @@ LIBOBJS!$LIBOBJS$ac_delim LTLIBOBJS!$LTLIBOBJS$ac_delim _ACEOF - if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 88; then + if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 91; then break elif $ac_last_try; then { { echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5 diff --git a/docs/AdvancedGetElementPtr.html b/docs/AdvancedGetElementPtr.html new file mode 100644 index 0000000..1d37278 --- /dev/null +++ b/docs/AdvancedGetElementPtr.html @@ -0,0 +1,362 @@ +<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" + "http://www.w3.org/TR/html4/strict.dtd"> +<html> +<head> + <meta http-equiv="Content-Type" content="text/html; charset=utf-8"> + <title>The Revenge Of The Often Misunderstood GEP Instruction</title> + <link rel="stylesheet" href="llvm.css" type="text/css"> + <style type="text/css"> + TABLE { text-align: left; border: 1px solid black; border-collapse: collapse; margin: 0 0 0 0; } + </style> +</head> +<body> + +<div class="doc_title"> + The Revenge Of The Often Misunderstood GEP Instruction +</div> + +<!-- *********************************************************************** --> +<div class="doc_section"><a name="intro"><b>Introduction</b></a></div> +<!-- *********************************************************************** --> +<div class="doc_text"> + <p>GEP was mysterious and wily at first, but it turned out that the basic + workings were fairly comprehensible. However the dragon was merely subdued; + now it's back, and it has more fundamental complexity to confront. This + document seeks to uncover misunderstandings of the GEP operator that tend + to persist past initial confusion about the funky "extra 0" thing. Here we + show that the GEP instruction is really not quite as simple as it seems, + even after the initial confusion is overcome.</p> +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>How is GEP different from ptrtoint, arithmetic, + and inttoptr?</b></a> +</div> +<div class="doc_text"> + <p>It's very similar; there are only subtle differences.</p> + + <p>With ptrtoint, you have to pick an integer type. One approach is to pick i64; + this is safe on everything LLVM supports (LLVM internally assumes pointers + are never wider than 64 bits in many places), and the optimizer will actually + narrow the i64 arithmetic down to the actual pointer size on targets which + don't support 64-bit arithmetic in most cases. However, there are some cases + where it doesn't do this. With GEP you can avoid this problem. + + <p>Also, GEP carries additional pointer aliasing rules. It's invalid to take a + GEP from one object, address into a different separately allocated + object, and dereference it. IR producers (front-ends) must follow this rule, + and consumers (optimizers, specifically alias analysis) benefit from being + able to rely on it.</p> + + <p>And, GEP is more concise in common cases.</p> + + <p>However, for the underlying integer computation implied, there + is no difference.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>I'm writing a backend for a target which needs custom + lowering for GEP. How do I do this?</b></a> +</div> +<div class="doc_text"> + <p>You don't. The integer computation implied by a GEP is target-independent. + Typically what you'll need to do is make your backend pattern-match + expressions trees involving ADD, MUL, etc., which are what GEP is lowered + into. This has the advantage of letting your code work correctly in more + cases.</p> + + <p>GEP does use target-dependent parameters for the size and layout of data + types, which targets can customize.</p> + + <p>If you require support for addressing units which are not 8 bits, you'll + need to fix a lot of code in the backend, with GEP lowering being only a + small piece of the overall picture.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>Why do struct member indices always use i32?</b></a> +</div> +<div class="doc_text"> + <p>The specific type i32 is probably just a historical artifact, however it's + wide enough for all practical purposes, so there's been no need to change it. + It doesn't necessarily imply i32 address arithmetic; it's just an identifier + which identifies a field in a struct. Requiring that all struct indices be + the same reduces the range of possibilities for cases where two GEPs are + effectively the same but have distinct operand types.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>How does VLA addressing work with GEPs?</b></a> +</div> +<div class="doc_text"> + <p>GEPs don't natively support VLAs. LLVM's type system is entirely static, + and GEP address computations are guided by an LLVM type.</p> + + <p>VLA indices can be implemented as linearized indices. For example, an + expression like X[a][b][c], must be effectively lowered into a form + like X[a*m+b*n+c], so that it appears to the GEP as a single-dimensional + array reference.</p> + + <p>This means if you want to write an analysis which understands array + indices and you want to support VLAs, your code will have to be + prepared to reverse-engineer the linearization. One way to solve this + problem is to use the ScalarEvolution library, which always presents + VLA and non-VLA indexing in the same manner.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>What happens if an array index is out of bounds?</b></a> +</div> +<div class="doc_text"> + <p>There are two senses in which an array index can be out of bounds.</p> + + <p>First, there's the array type which comes from the (static) type of + the first operand to the GEP. Indices greater than the number of elements + in the corresponding static array type are valid. There is no problem with + out of bounds indices in this sense. Indexing into an array only depends + on the size of the array element, not the number of elements.</p> + + <p>A common example of how this is used is arrays where the size is not known. + It's common to use array types with zero length to represent these. The + fact that the static type says there are zero elements is irrelevant; it's + perfectly valid to compute arbitrary element indices, as the computation + only depends on the size of the array element, not the number of + elements. Note that zero-sized arrays are not a special case here.</p> + + <p>This sense is unconnected with <tt>inbounds</tt> keyword. The + <tt>inbounds</tt> keyword is designed to describe low-level pointer + arithmetic overflow conditions, rather than high-level array + indexing rules. + + <p>Analysis passes which wish to understand array indexing should not + assume that the static array type bounds are respected.</p> + + <p>The second sense of being out of bounds is computing an address that's + beyond the actual underlying allocated object.</p> + + <p>With the <tt>inbounds</tt> keyword, the result value of the GEP is + undefined if the address is outside the actual underlying allocated + object and not the address one-past-the-end.</p> + + <p>Without the <tt>inbounds</tt> keyword, there are no restrictions + on computing out-of-bounds addresses. Obviously, performing a load or + a store requires an address of allocated and sufficiently aligned + memory. But the GEP itself is only concerned with computing addresses.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>Can array indices be negative?</b></a> +</div> +<div class="doc_text"> + <p>Yes. This is basically a special case of array indices being out + of bounds.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>Can I compare two values computed with GEPs?</b></a> +</div> +<div class="doc_text"> + <p>Yes. If both addresses are within the same allocated object, or + one-past-the-end, you'll get the comparison result you expect. If either + is outside of it, integer arithmetic wrapping may occur, so the + comparison may not be meaningful.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>Can I do GEP with a different pointer type than the type of + the underlying object?</b></a> +</div> +<div class="doc_text"> + <p>Yes. There are no restrictions on bitcasting a pointer value to an arbitrary + pointer type. The types in a GEP serve only to define the parameters for the + underlying integer computation. They need not correspond with the actual + type of the underlying object.</p> + + <p>Furthermore, loads and stores don't have to use the same types as the type + of the underlying object. Types in this context serve only to specify + memory size and alignment. Beyond that there are merely a hint to the + optimizer indicating how the value will likely be used.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>Can I cast an object's address to integer and add it + to null?</b></a> +</div> +<div class="doc_text"> + <p>You can compute an address that way, but if you use GEP to do the add, + you can't use that pointer to actually access the object, unless the + object is managed outside of LLVM.</p> + + <p>The underlying integer computation is sufficiently defined; null has a + defined value -- zero -- and you can add whatever value you want to it.</p> + + <p>However, it's invalid to access (load from or store to) an LLVM-aware + object with such a pointer. This includes GlobalVariables, Allocas, and + objects pointed to by noalias pointers.</p> + + <p>If you really need this functionality, you can do the arithmetic with + explicit integer instructions, and use inttoptr to convert the result to + an address. Most of GEP's special aliasing rules do not apply to pointers + computed from ptrtoint, arithmetic, and inttoptr sequences.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>Can I compute the distance between two objects, and add + that value to one address to compute the other address?</b></a> +</div> +<div class="doc_text"> + <p>As with arithmetic on null, You can use GEP to compute an address that + way, but you can't use that pointer to actually access the object if you + do, unless the object is managed outside of LLVM.</p> + + <p>Also as above, ptrtoint and inttoptr provide an alternative way to do this + which do not have this restriction.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="lead0"><b>Can I do type-based alias analysis on LLVM IR?</b></a> +</div> +<div class="doc_text"> + <p>You can't do type-based alias analysis using LLVM's built-in type system, + because LLVM has no restrictions on mixing types in addressing, loads or + stores.</p> + + <p>It would be possible to add special annotations to the IR, probably using + metadata, to describe a different type system (such as the C type system), + and do type-based aliasing on top of that. This is a much bigger + undertaking though.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="lead0"><b>What's an uglygep?</b></a> +</div> +<div class="doc_text"> + <p>Some LLVM optimizers operate on GEPs by internally lowering them into + more primitive integer expressions, which allows them to be combined + with other integer expressions and/or split into multiple separate + integer expressions. If they've made non-trivial changes, translating + back into LLVM IR can involve reverse-engineering the structure of + the addressing in order to fit it into the static type of the original + first operand. It isn't always possibly to fully reconstruct this + structure; sometimes the underlying addressing doesn't correspond with + the static type at all. In such cases the optimizer instead will emit + a GEP with the base pointer casted to a simple address-unit pointer, + using the name "uglygep". This isn't pretty, but it's just as + valid, and it's sufficient to preserve the pointer aliasing guarantees + that GEP provides.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="lead0"><b>Can GEP index into vector elements?</b></a> +</div> +<div class="doc_text"> + <p>Sort of. This hasn't always been forcefully disallowed, though it's + not recommended. It leads to awkward special cases in the optimizers. + In the future, it may be outright disallowed.</p> + + <p>Instead, you should cast your pointer types and use arrays instead of + vectors for addressing. Arrays have the same in-memory representation + as vectors, so the addressing is interchangeable.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="lead0"><b>Can GEP index into unions?</b></a> +</div> +<div class="doc_text"> + <p>Unknown.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="lead0"><b>What happens if a GEP computation overflows?</b></a> +</div> +<div class="doc_text"> + <p>If the GEP has the <tt>inbounds</tt> keyword, the result value is + undefined.</p> + + <p>Otherwise, the result value is the result from evaluating the implied + two's complement integer computation. However, since there's no + guarantee of where an object will be allocated in the address space, + such values have limited meaning.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="lead0"><b>What effect do address spaces have on GEPs?</b></a> +</div> +<div class="doc_text"> + <p>None, except that the address space qualifier on the first operand pointer + type always matches the address space qualifier on the result type.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="lead0"><b>Why is GEP designed this way?</b></a> +</div> +<div class="doc_text"> + <p>The design of GEP has the following goals, in rough unofficial + order of priority:</p> + <ul> + <li>Support C, C-like languages, and languages which can be + conceptually lowered into C (this covers a lot).</li> + <li>Support optimizations such as those that are common in + C compilers.</li> + <li>Provide a consistent method for computing addresses so that + address computations don't need to be a part of load and + store instructions in the IR.</li> + <li>Support non-C-like languages, to the extent that it doesn't + interfere with other goals.</li> + <li>Minimize target-specific information in the IR.</li> + </ul> +</div> + +<!-- *********************************************************************** --> + +<hr> +<address> + <a href="http://jigsaw.w3.org/css-validator/check/referer"><img + src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a> + <a href="http://validator.w3.org/check/referer"><img + src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a> + <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br/> + Last modified: $Date: 2010-02-18 19:40:29 +0100 (Thu, 18 Feb 2010) $ +</address> +</body> +</html> + diff --git a/docs/AliasAnalysis.html b/docs/AliasAnalysis.html index 6903ede..ef0f414 100644 --- a/docs/AliasAnalysis.html +++ b/docs/AliasAnalysis.html @@ -403,7 +403,7 @@ implementing, you just override the interfaces you can improve.</p> href="#basic-aa">basicaa</a></tt> and <a href="#no-aa"><tt>no-aa</tt></a> passes) every alias analysis pass chains to another alias analysis implementation (for example, the user can specify "<tt>-basicaa -ds-aa --anders-aa -licm</tt>" to get the maximum benefit from the three alias +-licm</tt>" to get the maximum benefit from both alias analyses). The alias analysis class automatically takes care of most of this for methods that you don't override. For methods that you do override, in code paths that return a conservative MayAlias or Mod/Ref result, simply return @@ -705,25 +705,6 @@ non-address taken globals), but is very quick analysis.</p> <!-- _______________________________________________________________________ --> <div class="doc_subsubsection"> - <a name="anders-aa">The <tt>-anders-aa</tt> pass</a> -</div> - -<div class="doc_text"> - -<p>The <tt>-anders-aa</tt> pass implements the well-known "Andersen's algorithm" -for interprocedural alias analysis. This algorithm is a subset-based, -flow-insensitive, context-insensitive, and field-insensitive alias analysis that -is widely believed to be fairly precise. Unfortunately, this algorithm is also -O(N<sup>3</sup>). The LLVM implementation currently does not implement any of -the refinements (such as "online cycle elimination" or "offline variable -substitution") to improve its efficiency, so it can be quite slow in common -cases. -</p> - -</div> - -<!-- _______________________________________________________________________ --> -<div class="doc_subsubsection"> <a name="steens-aa">The <tt>-steens-aa</tt> pass</a> </div> @@ -855,7 +836,7 @@ pointer.</p> <div class="doc_text"> <p>These passes are useful for evaluating the various alias analysis -implementations. You can use them with commands like '<tt>opt -anders-aa -ds-aa +implementations. You can use them with commands like '<tt>opt -ds-aa -aa-eval foo.bc -disable-output -stats</tt>'.</p> </div> @@ -949,7 +930,7 @@ analysis directly.</p> <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-11-22 17:01:44 +0100 (Sun, 22 Nov 2009) $ + Last modified: $Date: 2010-03-01 20:24:17 +0100 (Mon, 01 Mar 2010) $ </address> </body> diff --git a/docs/CodeGenerator.html b/docs/CodeGenerator.html index 2eb7abc..ad6222d 100644 --- a/docs/CodeGenerator.html +++ b/docs/CodeGenerator.html @@ -1041,9 +1041,9 @@ ret <div class="doc_code"> <pre> -%t1 = add float %W, %X -%t2 = mul float %t1, %Y -%t3 = add float %t2, %Z +%t1 = fadd float %W, %X +%t2 = fmul float %t1, %Y +%t3 = fadd float %t2, %Z </pre> </div> @@ -2116,7 +2116,7 @@ MOVSX32rm16 -> movsx, 32-bit register, 16-bit memory <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-01-11 19:53:47 +0100 (Mon, 11 Jan 2010) $ + Last modified: $Date: 2010-03-02 02:11:08 +0100 (Tue, 02 Mar 2010) $ </address> </body> diff --git a/docs/CodingStandards.html b/docs/CodingStandards.html index f93e1ea..8dc5678 100644 --- a/docs/CodingStandards.html +++ b/docs/CodingStandards.html @@ -194,9 +194,9 @@ something sane goes a long ways towards avoiding writing documentation.</p> <b>Method information</b> <p>Methods defined in a class (as well as any global functions) should also be -documented properly. A quick note about what it does any a description of the +documented properly. A quick note about what it does and a description of the borderline behaviour is all that is necessary here (unless something -particularly tricky or insideous is going on). The hope is that people can +particularly tricky or insidious is going on). The hope is that people can figure out how to use your interfaces without reading the code itself... that is the goal metric.</p> @@ -1346,7 +1346,7 @@ something.</p> <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-10-12 16:46:08 +0200 (Mon, 12 Oct 2009) $ + Last modified: $Date: 2010-02-26 21:18:32 +0100 (Fri, 26 Feb 2010) $ </address> </body> diff --git a/docs/CommandGuide/FileCheck.pod b/docs/CommandGuide/FileCheck.pod index 32516ad..433979a 100644 --- a/docs/CommandGuide/FileCheck.pod +++ b/docs/CommandGuide/FileCheck.pod @@ -25,7 +25,7 @@ match. The file to verify is always read from standard input. =over -=item B<--help> +=item B<-help> Print a summary of command line options. diff --git a/docs/CommandGuide/bugpoint.pod b/docs/CommandGuide/bugpoint.pod index 204ea4d..7afeea1 100644 --- a/docs/CommandGuide/bugpoint.pod +++ b/docs/CommandGuide/bugpoint.pod @@ -85,7 +85,7 @@ mis-management. Continually randomize the specified passes and run them on the test program until a bug is found or the user kills B<bugpoint>. -=item B<--help> +=item B<-help> Print a summary of command line options. @@ -99,9 +99,9 @@ it runs, to come from that file. Load the dynamic object F<plugin> into B<bugpoint> itself. This object should register new optimization passes. Once loaded, the object will add new command line options to enable various optimizations. To see the new complete list of -optimizations, use the B<--help> and B<--load> options together; for example: +optimizations, use the B<-help> and B<--load> options together; for example: - bugpoint --load myNewPass.so --help + bugpoint --load myNewPass.so -help =item B<--mlimit> F<megabytes> diff --git a/docs/CommandGuide/index.html b/docs/CommandGuide/index.html index 14635dd..d0212b9 100644 --- a/docs/CommandGuide/index.html +++ b/docs/CommandGuide/index.html @@ -17,7 +17,7 @@ for all of the LLVM tools. These pages describe how to use the LLVM commands and what their options are. Note that these pages do not describe all of the options available for all tools. To get a complete listing, pass the -<tt>--help</tt> (general options) or <tt>--help-hidden</tt> (general+debugging +<tt>-help</tt> (general options) or <tt>-help-hidden</tt> (general+debugging options) arguments to the tool you are interested in.</p> </div> @@ -148,7 +148,7 @@ options) arguments to the tool you are interested in.</p> src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a> <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-10-19 05:54:13 +0200 (Mon, 19 Oct 2009) $ + Last modified: $Date: 2010-02-18 15:08:13 +0100 (Thu, 18 Feb 2010) $ </address> </body> diff --git a/docs/CommandGuide/llc.pod b/docs/CommandGuide/llc.pod index 8adfb68..ac24aab 100644 --- a/docs/CommandGuide/llc.pod +++ b/docs/CommandGuide/llc.pod @@ -38,7 +38,7 @@ Other B<llc> options are as follows: =over -=item B<--help> +=item B<-help> Print a summary of command line options. @@ -56,7 +56,7 @@ string. =item B<-march>=I<arch> Specify the architecture for which to generate assembly, overriding the target -encoded in the input file. See the output of B<llc --help> for a list of +encoded in the input file. See the output of B<llc -help> for a list of valid architectures. By default this is inferred from the target triple or autodetected to the current architecture. diff --git a/docs/CommandGuide/lli.pod b/docs/CommandGuide/lli.pod index e9fdf74..6d1e1c6 100644 --- a/docs/CommandGuide/lli.pod +++ b/docs/CommandGuide/lli.pod @@ -73,7 +73,7 @@ architecture which is not compatible with the current system. =item B<-march>=I<arch> Specify the architecture for which to generate assembly, overriding the target -encoded in the bitcode file. See the output of B<llc --help> for a list of +encoded in the bitcode file. See the output of B<llc -help> for a list of valid architectures. By default this is inferred from the target triple or autodetected to the current architecture. @@ -170,7 +170,7 @@ Instruction schedulers available (before register allocation): =item B<-regalloc>=I<allocator> -Register allocator to use: (default = linearscan) +Register allocator to use (default=linearscan) =bigblock: Big-block register allocator =linearscan: linear scan register allocator =local - local register allocator @@ -186,7 +186,7 @@ Choose relocation model from: =item B<-spiller> -Spiller to use: (default: local) +Spiller to use (default=local) =simple: simple spiller =local: local spiller diff --git a/docs/CommandGuide/llvm-as.pod b/docs/CommandGuide/llvm-as.pod index 045a924..185c009 100644 --- a/docs/CommandGuide/llvm-as.pod +++ b/docs/CommandGuide/llvm-as.pod @@ -50,7 +50,7 @@ Enable binary output on terminals. Normally, B<llvm-as> will refuse to write raw bitcode output if the output stream is a terminal. With this option, B<llvm-as> will write raw bitcode regardless of the output device. -=item B<--help> +=item B<-help> Print a summary of command line options. diff --git a/docs/CommandGuide/llvm-bcanalyzer.pod b/docs/CommandGuide/llvm-bcanalyzer.pod index f60b513..b0bc0cd 100644 --- a/docs/CommandGuide/llvm-bcanalyzer.pod +++ b/docs/CommandGuide/llvm-bcanalyzer.pod @@ -43,7 +43,7 @@ Causes B<llvm-bcanalyzer> to verify the module produced by reading the bitcode. This ensures that the statistics generated are based on a consistent module. -=item B<--help> +=item B<-help> Print a summary of command line options. diff --git a/docs/CommandGuide/llvm-config.pod b/docs/CommandGuide/llvm-config.pod index 06f10de..4e38dae 100644 --- a/docs/CommandGuide/llvm-config.pod +++ b/docs/CommandGuide/llvm-config.pod @@ -30,7 +30,7 @@ To link against the JIT: Print the version number of LLVM. -=item B<--help> +=item B<-help> Print a summary of B<llvm-config> arguments. diff --git a/docs/CommandGuide/llvm-dis.pod b/docs/CommandGuide/llvm-dis.pod index 2b83290..5b2f4ef 100644 --- a/docs/CommandGuide/llvm-dis.pod +++ b/docs/CommandGuide/llvm-dis.pod @@ -33,7 +33,7 @@ Enable binary output on terminals. Normally, B<llvm-dis> will refuse to write raw bitcode output if the output stream is a terminal. With this option, B<llvm-dis> will write raw bitcode regardless of the output device. -=item B<--help> +=item B<-help> Print a summary of command line options. diff --git a/docs/CommandGuide/llvm-extract.pod b/docs/CommandGuide/llvm-extract.pod index 02f38ad..d4baab7 100644 --- a/docs/CommandGuide/llvm-extract.pod +++ b/docs/CommandGuide/llvm-extract.pod @@ -42,7 +42,7 @@ specified multiple times to extract multiple functions at once. Extract the global variable named I<global-name> from the LLVM bitcode. May be specified multiple times to extract multiple global variables at once. -=item B<--help> +=item B<-help> Print a summary of command line options. diff --git a/docs/CommandGuide/llvm-link.pod b/docs/CommandGuide/llvm-link.pod index e1a1267..8d06cc9 100644 --- a/docs/CommandGuide/llvm-link.pod +++ b/docs/CommandGuide/llvm-link.pod @@ -51,7 +51,7 @@ Write output in LLVM intermediate language (instead of bitcode). If specified, B<llvm-link> prints a human-readable version of the output bitcode file to standard error. -=item B<--help> +=item B<-help> Print a summary of command line options. diff --git a/docs/CommandGuide/llvm-nm.pod b/docs/CommandGuide/llvm-nm.pod index 995ac08..a580d3f 100644 --- a/docs/CommandGuide/llvm-nm.pod +++ b/docs/CommandGuide/llvm-nm.pod @@ -77,7 +77,7 @@ Use POSIX.2 output format. Alias for B<--format=posix>. Use BSD output format. Alias for B<--format=bsd>. -=item B<--help> +=item B<-help> Print a summary of command-line options and their meanings. diff --git a/docs/CommandGuide/llvm-prof.pod b/docs/CommandGuide/llvm-prof.pod index 381387d..9541b05 100644 --- a/docs/CommandGuide/llvm-prof.pod +++ b/docs/CommandGuide/llvm-prof.pod @@ -18,7 +18,7 @@ where the program hotspots are. This program is often used in conjunction with the F<utils/profile.pl> script. This script automatically instruments a program, runs it with the JIT, then runs B<llvm-prof> to format a report. To get more information about -F<utils/profile.pl>, execute it with the B<--help> option. +F<utils/profile.pl>, execute it with the B<-help> option. =head1 OPTIONS diff --git a/docs/CommandGuide/llvm-ranlib.pod b/docs/CommandGuide/llvm-ranlib.pod index 130edb0..53cd34b 100644 --- a/docs/CommandGuide/llvm-ranlib.pod +++ b/docs/CommandGuide/llvm-ranlib.pod @@ -6,7 +6,7 @@ llvm-ranlib - Generate index for LLVM archive =head1 SYNOPSIS -B<llvm-ranlib> [--version] [--help] <archive-file> +B<llvm-ranlib> [--version] [-help] <archive-file> =head1 DESCRIPTION @@ -30,7 +30,7 @@ Specifies the archive-file to which the symbol table is added or updated. Print the version of B<llvm-ranlib> and exit without building a symbol table. -=item F<--help> +=item F<-help> Print usage help for B<llvm-ranlib> and exit without building a symbol table. diff --git a/docs/CommandGuide/llvmc.pod b/docs/CommandGuide/llvmc.pod index e5e0651..d237ca4 100644 --- a/docs/CommandGuide/llvmc.pod +++ b/docs/CommandGuide/llvmc.pod @@ -77,11 +77,11 @@ Store temporary files in the given directory. This directory is deleted on exit unless I<--save-temps> is specified. If I<--save-temps=obj> is also specified, I<--temp-dir> is given the precedence. -=item B<--help> +=item B<-help> Print a summary of command-line options and exit. -=item B<--help-hidden> +=item B<-help-hidden> Print a summary of command-line options and exit. Print help even for options intended for developers. diff --git a/docs/CommandGuide/tblgen.pod b/docs/CommandGuide/tblgen.pod index c8244af..d127492 100644 --- a/docs/CommandGuide/tblgen.pod +++ b/docs/CommandGuide/tblgen.pod @@ -26,7 +26,7 @@ to read as input. =over -=item B<--help> +=item B<-help> Print a summary of command line options. diff --git a/docs/CommandLine.html b/docs/CommandLine.html index 7e6e2f2..092fd2b 100644 --- a/docs/CommandLine.html +++ b/docs/CommandLine.html @@ -44,7 +44,7 @@ <li><a href="#modifiers">Option Modifiers</a> <ul> - <li><a href="#hiding">Hiding an option from <tt>--help</tt> + <li><a href="#hiding">Hiding an option from <tt>-help</tt> output</a></li> <li><a href="#numoccurrences">Controlling the number of occurrences required and allowed</a></li> @@ -161,7 +161,7 @@ you declare it. <a href="#customparser">Custom parsers</a> are no problem.</li> <li>Labor Saving: The CommandLine library cuts down on the amount of grunt work that you, the user, have to do. For example, it automatically provides a -<tt>--help</tt> option that shows the available command line options for your +<tt>-help</tt> option that shows the available command line options for your tool. Additionally, it does most of the basic correctness checking for you.</li> @@ -239,14 +239,14 @@ href="#list">"<tt>cl::list</tt> template</a>), and tell the CommandLine library that the data type that we are parsing is a string.</p> <p>The second and third parameters (which are optional) are used to specify what -to output for the "<tt>--help</tt>" option. In this case, we get a line that +to output for the "<tt>-help</tt>" option. In this case, we get a line that looks like this:</p> <div class="doc_code"><pre> USAGE: compiler [options] OPTIONS: - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) <b>-o <filename> - Specify output filename</b> </pre></div> @@ -308,14 +308,14 @@ the CommandLine library will automatically issue an error if the argument is not specified, which shifts all of the command line option verification code out of your application into the library. This is just one example of how using flags can alter the default behaviour of the library, on a per-option basis. By -adding one of the declarations above, the <tt>--help</tt> option synopsis is now +adding one of the declarations above, the <tt>-help</tt> option synopsis is now extended to:</p> <div class="doc_code"><pre> USAGE: compiler [options] <b><input file></b> OPTIONS: - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) -o <filename> - Specify output filename </pre></div> @@ -346,8 +346,8 @@ declaring options of boolean type like this:</p> ("<tt>Force</tt>", "<tt>Quiet</tt>", and "<tt>Quiet2</tt>") to recognize these options. Note that the "<tt>-q</tt>" option is specified with the "<a href="#cl::Hidden"><tt>cl::Hidden</tt></a>" flag. This modifier prevents it -from being shown by the standard "<tt>--help</tt>" output (note that it is still -shown in the "<tt>--help-hidden</tt>" output).</p> +from being shown by the standard "<tt>-help</tt>" output (note that it is still +shown in the "<tt>-help-hidden</tt>" output).</p> <p>The CommandLine library uses a <a href="#builtinparsers">different parser</a> for different data types. For example, in the string case, the argument passed @@ -372,7 +372,7 @@ href="#doubleparser">double</a>, and <a href="#intparser">int</a> parsers work like you would expect, using the '<tt>strtol</tt>' and '<tt>strtod</tt>' C library calls to parse the string value into the specified data type.</p> -<p>With the declarations above, "<tt>compiler --help</tt>" emits this:</p> +<p>With the declarations above, "<tt>compiler -help</tt>" emits this:</p> <div class="doc_code"><pre> USAGE: compiler [options] <input file> @@ -381,10 +381,10 @@ OPTIONS: <b>-f - Enable binary output on terminals</b> -o - Override output filename <b>-quiet - Don't print informational messages</b> - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) </pre></div> -<p>and "<tt>compiler --help-hidden</tt>" prints this:</p> +<p>and "<tt>compiler -help-hidden</tt>" prints this:</p> <div class="doc_code"><pre> USAGE: compiler [options] <input file> @@ -394,7 +394,7 @@ OPTIONS: -o - Override output filename <b>-q - Don't print informational messages</b> -quiet - Don't print informational messages - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) </pre></div> <p>This brief example has shown you how to use the '<tt><a @@ -438,7 +438,7 @@ the <tt><a href="#cl::aliasopt">cl::aliasopt</a></tt> modifier) whenever it is specified. Because aliases do not hold state, the only thing the program has to query is the <tt>Quiet</tt> variable now. Another nice feature of aliases is that they automatically hide themselves from the <tt>-help</tt> output -(although, again, they are still visible in the <tt>--help-hidden +(although, again, they are still visible in the <tt>-help-hidden output</tt>).</p> <p>Now the application code can simply use:</p> @@ -531,7 +531,7 @@ OPTIONS: -O2 - Enable default optimizations -O3 - Enable expensive optimizations</b> -f - Enable binary output on terminals - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) -o <filename> - Specify output filename -quiet - Don't print informational messages </pre></div> @@ -599,7 +599,7 @@ enum DebugLev { <p>This definition defines an enumerated command line variable of type "<tt>enum DebugLev</tt>", which works exactly the same way as before. The difference here is just the interface exposed to the user of your program and the help output by -the "<tt>--help</tt>" option:</p> +the "<tt>-help</tt>" option:</p> <div class="doc_code"><pre> USAGE: compiler [options] <input file> @@ -615,7 +615,7 @@ OPTIONS: =quick - enable quick debug information =detailed - enable detailed debug information</b> -f - Enable binary output on terminals - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) -o <filename> - Specify output filename -quiet - Don't print informational messages </pre></div> @@ -706,7 +706,7 @@ checking we have to do.</p> <div class="doc_text"> <p>Instead of collecting sets of options in a list, it is also possible to -gather information for enum values in a <b>bit vector</b>. The represention used by +gather information for enum values in a <b>bit vector</b>. The representation used by the <a href="#bits"><tt>cl::bits</tt></a> class is an <tt>unsigned</tt> integer. An enum value is represented by a 0/1 in the enum's ordinal value bit position. 1 indicating that the enum was specified, 0 otherwise. As each @@ -794,7 +794,7 @@ USAGE: compiler [options] <input file> OPTIONS: ... - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) -o <filename> - Specify output filename </pre></div> @@ -835,14 +835,14 @@ Using the CommandLine library, this would be specified as:</p> <a href="#cl::opt">cl::opt</a><string> Filename(<a href="#cl::Positional">cl::Positional</a>, <a href="#cl::desc">cl::desc</a>("<i><input file></i>"), <a href="#cl::init">cl::init</a>("<i>-</i>")); </pre></div> -<p>Given these two option declarations, the <tt>--help</tt> output for our grep +<p>Given these two option declarations, the <tt>-help</tt> output for our grep replacement would look like this:</p> <div class="doc_code"><pre> USAGE: spiffygrep [options] <b><regular expression> <input file></b> OPTIONS: - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) </pre></div> <p>... and the resultant program could be used just like the standard @@ -872,7 +872,7 @@ Note that the system <tt>grep</tt> has the same problem:</p> <div class="doc_code"><pre> $ spiffygrep '-foo' test.txt - Unknown command line argument '-foo'. Try: spiffygrep --help' + Unknown command line argument '-foo'. Try: spiffygrep -help' $ grep '-foo' test.txt grep: illegal option -- f @@ -986,7 +986,7 @@ shell itself. Using the CommandLine library, we would specify this as:</p> USAGE: spiffysh [options] <b><input script> <program arguments>...</b> OPTIONS: - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) <b>-x - Enable trace output</b> </pre></div> @@ -1098,11 +1098,11 @@ This option is specified in simple double quotes: </li> <li><a name="cl::desc">The <b><tt>cl::desc</tt></b></a> attribute specifies a -description for the option to be shown in the <tt>--help</tt> output for the +description for the option to be shown in the <tt>-help</tt> output for the program.</li> <li><a name="cl::value_desc">The <b><tt>cl::value_desc</tt></b></a> attribute -specifies a string that can be used to fine tune the <tt>--help</tt> output for +specifies a string that can be used to fine tune the <tt>-help</tt> output for a command line option. Look <a href="#value_desc_example">here</a> for an example.</li> @@ -1130,7 +1130,7 @@ the string-to-value mapping to be used by the generic parser. It takes a <b>clEnumValEnd terminated</b> list of (option, value, description) triplets that specify the option name, the value mapped to, and the description shown in the -<tt>--help</tt> for the tool. Because the generic parser is used most +<tt>-help</tt> for the tool. Because the generic parser is used most frequently with enum values, two macros are often useful: <ol> @@ -1175,13 +1175,13 @@ obviously).</li> <p>Option modifiers are the flags and expressions that you pass into the constructors for <tt><a href="#cl::opt">cl::opt</a></tt> and <tt><a href="#cl::list">cl::list</a></tt>. These modifiers give you the ability to -tweak how options are parsed and how <tt>--help</tt> output is generated to fit +tweak how options are parsed and how <tt>-help</tt> output is generated to fit your application well.</p> <p>These options fall into five main categories:</p> <ol> -<li><a href="#hiding">Hiding an option from <tt>--help</tt> output</a></li> +<li><a href="#hiding">Hiding an option from <tt>-help</tt> output</a></li> <li><a href="#numoccurrences">Controlling the number of occurrences required and allowed</a></li> <li><a href="#valrequired">Controlling whether or not a value must be @@ -1200,14 +1200,14 @@ usually shouldn't have to worry about these.</p> <!-- _______________________________________________________________________ --> <div class="doc_subsubsection"> - <a name="hiding">Hiding an option from <tt>--help</tt> output</a> + <a name="hiding">Hiding an option from <tt>-help</tt> output</a> </div> <div class="doc_text"> <p>The <tt>cl::NotHidden</tt>, <tt>cl::Hidden</tt>, and <tt>cl::ReallyHidden</tt> modifiers are used to control whether or not an option -appears in the <tt>--help</tt> and <tt>--help-hidden</tt> output for the +appears in the <tt>-help</tt> and <tt>-help-hidden</tt> output for the compiled program:</p> <ul> @@ -1219,8 +1219,8 @@ in both help listings.</li> <li><a name="cl::Hidden">The <b><tt>cl::Hidden</tt></b></a> modifier (which is the default for <tt><a href="#cl::alias">cl::alias</a></tt> options) indicates that -the option should not appear in the <tt>--help</tt> output, but should appear in -the <tt>--help-hidden</tt> output.</li> +the option should not appear in the <tt>-help</tt> output, but should appear in +the <tt>-help-hidden</tt> output.</li> <li><a name="cl::ReallyHidden">The <b><tt>cl::ReallyHidden</tt></b></a> modifier indicates that the option should not appear in any help output.</li> @@ -1508,7 +1508,7 @@ available.</p> <p>The <tt>cl::ParseCommandLineOptions</tt> function requires two parameters (<tt>argc</tt> and <tt>argv</tt>), but may also take an optional third parameter which holds <a href="#description">additional extra text</a> to emit when the -<tt>--help</tt> option is invoked, and a fourth boolean parameter that enables +<tt>-help</tt> option is invoked, and a fourth boolean parameter that enables <a href="#response">response files</a>.</p> </div> @@ -1535,7 +1535,7 @@ does.</p> not be available, it can't just look in <tt>argv[0]</tt>), the name of the environment variable to examine, the optional <a href="#description">additional extra text</a> to emit when the -<tt>--help</tt> option is invoked, and the boolean +<tt>-help</tt> option is invoked, and the boolean switch that controls whether <a href="#response">response files</a> should be read.</p> @@ -1689,7 +1689,7 @@ the conversion from string to data.</p> <div class="doc_text"> <p>The <tt>cl::extrahelp</tt> class is a nontemplated class that allows extra -help text to be printed out for the <tt>--help</tt> option.</p> +help text to be printed out for the <tt>-help</tt> option.</p> <div class="doc_code"><pre> <b>namespace</b> cl { @@ -1906,7 +1906,7 @@ MFS(<i>"max-file-size"</i>, <a href="#cl::desc">cl::desc</a>(<i>"Maximum file si <div class="doc_code"><pre> OPTIONS: - -help - display available options (--help-hidden for more) + -help - display available options (-help-hidden for more) ... <b>-max-file-size=<size> - Maximum file size to accept</b> </pre></div> @@ -1972,7 +1972,7 @@ tutorial.</p> <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-10-12 16:46:08 +0200 (Mon, 12 Oct 2009) $ + Last modified: $Date: 2010-02-26 21:18:32 +0100 (Fri, 26 Feb 2010) $ </address> </body> diff --git a/docs/CompilerDriver.html b/docs/CompilerDriver.html index b69b8c9..a0eed43 100644 --- a/docs/CompilerDriver.html +++ b/docs/CompilerDriver.html @@ -132,7 +132,7 @@ directory with the compilation graph description in Graphviz format (identical to the file used by the <tt class="docutils literal"><span class="pre">--view-graph</span></tt> option). The <tt class="docutils literal"><span class="pre">-o</span></tt> option can be used to set the output file name. Hidden option, useful for debugging LLVMC plugins.</li> -<li><tt class="docutils literal"><span class="pre">--help</span></tt>, <tt class="docutils literal"><span class="pre">--help-hidden</span></tt>, <tt class="docutils literal"><span class="pre">--version</span></tt> - These options have +<li><tt class="docutils literal"><span class="pre">-help</span></tt>, <tt class="docutils literal"><span class="pre">-help-hidden</span></tt>, <tt class="docutils literal"><span class="pre">--version</span></tt> - These options have their standard meaning.</li> </ul> </div> @@ -325,7 +325,7 @@ aliased option name. Usage example: <tt class="docutils literal"><span class="pr <li><p class="first">Possible option properties:</p> <blockquote> <ul class="simple"> -<li><tt class="docutils literal"><span class="pre">help</span></tt> - help string associated with this option. Used for <tt class="docutils literal"><span class="pre">--help</span></tt> +<li><tt class="docutils literal"><span class="pre">help</span></tt> - help string associated with this option. Used for <tt class="docutils literal"><span class="pre">-help</span></tt> output.</li> <li><tt class="docutils literal"><span class="pre">required</span></tt> - this option must be specified exactly once (or, in case of the list options without the <tt class="docutils literal"><span class="pre">multi_val</span></tt> property, at least @@ -338,7 +338,7 @@ it is synonymous with <tt class="docutils literal"><span class="pre">required</s for list options in conjunction with <tt class="docutils literal"><span class="pre">multi_val</span></tt>. Incompatible with <tt class="docutils literal"><span class="pre">required</span></tt> and <tt class="docutils literal"><span class="pre">one_or_more</span></tt>.</li> <li><tt class="docutils literal"><span class="pre">hidden</span></tt> - the description of this option will not appear in -the <tt class="docutils literal"><span class="pre">--help</span></tt> output (but will appear in the <tt class="docutils literal"><span class="pre">--help-hidden</span></tt> +the <tt class="docutils literal"><span class="pre">-help</span></tt> output (but will appear in the <tt class="docutils literal"><span class="pre">-help-hidden</span></tt> output).</li> <li><tt class="docutils literal"><span class="pre">really_hidden</span></tt> - the option will not be mentioned in any help output.</li> @@ -748,7 +748,7 @@ the <tt class="docutils literal"><span class="pre">Base</span></tt> plugin behav <a href="mailto:foldr@codedgers.com">Mikhail Glushenkov</a><br /> <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br /> -Last modified: $Date: 2009-12-23 13:49:51 +0100 (Wed, 23 Dec 2009) $ +Last modified: $Date: 2010-02-18 15:08:13 +0100 (Thu, 18 Feb 2010) $ </address></div> </div> </div> diff --git a/docs/DeveloperPolicy.html b/docs/DeveloperPolicy.html index 4986606..1f9451d 100644 --- a/docs/DeveloperPolicy.html +++ b/docs/DeveloperPolicy.html @@ -520,7 +520,7 @@ Changes</a></div> <p>We intend to keep LLVM perpetually open source and to use a liberal open source license. The current license is the <a href="http://www.opensource.org/licenses/UoI-NCSA.php">University of - llinois/NCSA Open Source License</a>, which boils down to this:</p> + Illinois/NCSA Open Source License</a>, which boils down to this:</p> <ul> <li>You can freely distribute LLVM.</li> @@ -601,7 +601,7 @@ Changes</a></div> Written by the <a href="mailto:llvm-oversight@cs.uiuc.edu">LLVM Oversight Group</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-10-10 23:37:16 +0200 (Sat, 10 Oct 2009) $ + Last modified: $Date: 2010-02-26 21:18:32 +0100 (Fri, 26 Feb 2010) $ </address> </body> </html> diff --git a/docs/FAQ.html b/docs/FAQ.html index 0074638..31fc0c0 100644 --- a/docs/FAQ.html +++ b/docs/FAQ.html @@ -437,7 +437,7 @@ Stop. <div class="answer"> <p>The <tt>GNUmakefile</tt> in the top-level directory of LLVM-GCC is a special <tt>Makefile</tt> used by Apple to invoke the <tt>build_gcc</tt> script after - setting up a special environment. This has the unforunate side-effect that + setting up a special environment. This has the unfortunate side-effect that trying to build LLVM-GCC with srcdir == objdir in a "non-Apple way" invokes the <tt>GNUmakefile</tt> instead of <tt>Makefile</tt>. Because the environment isn't set up correctly to do this, the build fails.</p> @@ -931,7 +931,7 @@ F.i: src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a> <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-10-12 16:46:08 +0200 (Mon, 12 Oct 2009) $ + Last modified: $Date: 2010-02-26 00:41:41 +0100 (Fri, 26 Feb 2010) $ </address> </body> diff --git a/docs/GetElementPtr.html b/docs/GetElementPtr.html index d5863e8..f6dd71f 100644 --- a/docs/GetElementPtr.html +++ b/docs/GetElementPtr.html @@ -17,7 +17,7 @@ <ol> <li><a href="#intro">Introduction</a></li> - <li><a href="#questions">The Questions</a> + <li><a href="#addresses">Address Computation</a> <ol> <li><a href="#extra_index">Why is the extra 0 index required?</a></li> <li><a href="#deref">What is dereferenced by GEP?</a></li> @@ -25,6 +25,30 @@ subsequent ones?</a></li> <li><a href="#lead0">Why don't GEP x,0,0,1 and GEP x,1 alias? </a></li> <li><a href="#trail0">Why do GEP x,1,0,0 and GEP x,1 alias? </a></li> + <li><a href="#vectors">Can GEP index into vector elements?</a> + <li><a href="#unions">Can GEP index into unions?</a> + <li><a href="#addrspace">What effect do address spaces have on GEPs?</a> + <li><a href="#int">How is GEP different from ptrtoint, arithmetic, and inttoptr?</a></li> + <li><a href="#be">I'm writing a backend for a target which needs custom lowering for GEP. How do I do this?</a> + <li><a href="#vla">How does VLA addressing work with GEPs?</a> + </ol></li> + <li><a href="#rules">Rules</a> + <ol> + <li><a href="#bounds">What happens if an array index is out of bounds?</a> + <li><a href="#negative">Can array indices be negative?</a> + <li><a href="#compare">Can I compare two values computed with GEPs?</a> + <li><a href="#types">Can I do GEP with a different pointer type than the type of the underlying object?</a> + <li><a href="#null">Can I cast an object's address to integer and add it to null?</a> + <li><a href="#ptrdiff">Can I compute the distance between two objects, and add that value to one address to compute the other address?</a> + <li><a href="#tbaa">Can I do type-based alias analysis on LLVM IR?</a> + <li><a href="#overflow">What happens if a GEP computation overflows?</a> + <li><a href="#check">How can I tell if my front-end is following the rules?</a> + </ol></li> + <li><a href="#rationale">Rationale</a> + <ol> + <li><a href="#goals">Why is GEP designed this way?</a></li> + <li><a href="#i32">Why do struct member indices always use i32?</a></li> + <li><a href="#uglygep">What's an uglygep?</a> </ol></li> <li><a href="#summary">Summary</a></li> </ol> @@ -37,9 +61,10 @@ <!-- *********************************************************************** --> <div class="doc_section"><a name="intro"><b>Introduction</b></a></div> <!-- *********************************************************************** --> + <div class="doc_text"> <p>This document seeks to dispel the mystery and confusion surrounding LLVM's - GetElementPtr (GEP) instruction. Questions about the wiley GEP instruction are + GetElementPtr (GEP) instruction. Questions about the wily GEP instruction are probably the most frequently occurring questions once a developer gets down to coding with LLVM. Here we lay out the sources of confusion and show that the GEP instruction is really quite simple. @@ -47,22 +72,14 @@ </div> <!-- *********************************************************************** --> -<div class="doc_section"><a name="questions"><b>The Questions</b></a></div> +<div class="doc_section"><a name="addresses"><b>Address Computation</b></a></div> <!-- *********************************************************************** --> <div class="doc_text"> <p>When people are first confronted with the GEP instruction, they tend to relate it to known concepts from other programming paradigms, most notably C - array indexing and field selection. However, GEP is a little different and - this leads to the following questions; all of which are answered in the - following sections.</p> - <ol> - <li><a href="#firstptr">What is the first index of the GEP instruction?</a> - </li> - <li><a href="#extra_index">Why is the extra 0 index required?</a></li> - <li><a href="#deref">What is dereferenced by GEP?</a></li> - <li><a href="#lead0">Why don't GEP x,0,0,1 and GEP x,1 alias? </a></li> - <li><a href="#trail0">Why do GEP x,1,0,0 and GEP x,1 alias? </a></li> - </ol> + array indexing and field selection. GEP closely resembles C array indexing + and field selection, however it's is a little different and this leads to + the following questions.</p> </div> <!-- *********************************************************************** --> @@ -85,7 +102,7 @@ X = &Foo->F; <p>it is natural to think that there is only one index, the selection of the field <tt>F</tt>. However, in this example, <tt>Foo</tt> is a pointer. That - pointer must be indexed explicitly in LLVM. C, on the other hand, indexs + pointer must be indexed explicitly in LLVM. C, on the other hand, indices through it transparently. To arrive at the same address location as the C code, you would provide the GEP instruction with two index operands. The first operand indexes through the pointer; the second operand indexes the @@ -155,7 +172,7 @@ entry: <div class="doc_code"> <pre> -%MyVar = unintialized global i32 +%MyVar = uninitialized global i32 ... %idx1 = getelementptr i32* %MyVar, i64 0 %idx2 = getelementptr i32* %MyVar, i64 1 @@ -210,7 +227,7 @@ idx3 = (char*) &MyVar + 8 field of the structure <tt>%MyStruct</tt>. When people first look at it, they wonder why the <tt>i64 0</tt> index is needed. However, a closer inspection of how globals and GEPs work reveals the need. Becoming aware of the following - facts will dispell the confusion:</p> + facts will dispel the confusion:</p> <ol> <li>The type of <tt>%MyStruct</tt> is <i>not</i> <tt>{ float*, i32 }</tt> but rather <tt>{ float*, i32 }*</tt>. That is, <tt>%MyStruct</tt> is a @@ -297,8 +314,8 @@ idx3 = (char*) &MyVar + 8 <div class="doc_code"> <pre> %MyVar = global { [10 x i32 ] } -%idx1 = getlementptr { [10 x i32 ] }* %MyVar, i64 0, i32 0, i64 1 -%idx2 = getlementptr { [10 x i32 ] }* %MyVar, i64 1 +%idx1 = getelementptr { [10 x i32 ] }* %MyVar, i64 0, i32 0, i64 1 +%idx2 = getelementptr { [10 x i32 ] }* %MyVar, i64 1 </pre> </div> @@ -326,8 +343,8 @@ idx3 = (char*) &MyVar + 8 <div class="doc_code"> <pre> %MyVar = global { [10 x i32 ] } -%idx1 = getlementptr { [10 x i32 ] }* %MyVar, i64 1, i32 0, i64 0 -%idx2 = getlementptr { [10 x i32 ] }* %MyVar, i64 1 +%idx1 = getelementptr { [10 x i32 ] }* %MyVar, i64 1, i32 0, i64 0 +%idx2 = getelementptr { [10 x i32 ] }* %MyVar, i64 1 </pre> </div> @@ -337,6 +354,352 @@ idx3 = (char*) &MyVar + 8 </div> <!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="vectors"><b>Can GEP index into vector elements?</b></a> +</div> +<div class="doc_text"> + <p>This hasn't always been forcefully disallowed, though it's not recommended. + It leads to awkward special cases in the optimizers, and fundamental + inconsistency in the IR. In the future, it will probably be outright + disallowed.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="unions"><b>Can GEP index into unions?</b></a> +</div> +<div class="doc_text"> + <p>Unknown.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="addrspace"><b>What effect do address spaces have on GEPs?</b></a> +</div> +<div class="doc_text"> + <p>None, except that the address space qualifier on the first operand pointer + type always matches the address space qualifier on the result type.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="int"><b>How is GEP different from ptrtoint, arithmetic, + and inttoptr?</b></a> +</div> +<div class="doc_text"> + <p>It's very similar; there are only subtle differences.</p> + + <p>With ptrtoint, you have to pick an integer type. One approach is to pick i64; + this is safe on everything LLVM supports (LLVM internally assumes pointers + are never wider than 64 bits in many places), and the optimizer will actually + narrow the i64 arithmetic down to the actual pointer size on targets which + don't support 64-bit arithmetic in most cases. However, there are some cases + where it doesn't do this. With GEP you can avoid this problem. + + <p>Also, GEP carries additional pointer aliasing rules. It's invalid to take a + GEP from one object, address into a different separately allocated + object, and dereference it. IR producers (front-ends) must follow this rule, + and consumers (optimizers, specifically alias analysis) benefit from being + able to rely on it. See the <a href="#rules">Rules</a> section for more + information.</p> + + <p>And, GEP is more concise in common cases.</p> + + <p>However, for the underlying integer computation implied, there + is no difference.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="be"><b>I'm writing a backend for a target which needs custom + lowering for GEP. How do I do this?</b></a> +</div> +<div class="doc_text"> + <p>You don't. The integer computation implied by a GEP is target-independent. + Typically what you'll need to do is make your backend pattern-match + expressions trees involving ADD, MUL, etc., which are what GEP is lowered + into. This has the advantage of letting your code work correctly in more + cases.</p> + + <p>GEP does use target-dependent parameters for the size and layout of data + types, which targets can customize.</p> + + <p>If you require support for addressing units which are not 8 bits, you'll + need to fix a lot of code in the backend, with GEP lowering being only a + small piece of the overall picture.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="vla"><b>How does VLA addressing work with GEPs?</b></a> +</div> +<div class="doc_text"> + <p>GEPs don't natively support VLAs. LLVM's type system is entirely static, + and GEP address computations are guided by an LLVM type.</p> + + <p>VLA indices can be implemented as linearized indices. For example, an + expression like X[a][b][c], must be effectively lowered into a form + like X[a*m+b*n+c], so that it appears to the GEP as a single-dimensional + array reference.</p> + + <p>This means if you want to write an analysis which understands array + indices and you want to support VLAs, your code will have to be + prepared to reverse-engineer the linearization. One way to solve this + problem is to use the ScalarEvolution library, which always presents + VLA and non-VLA indexing in the same manner.</p> +</div> + +<!-- *********************************************************************** --> +<div class="doc_section"><a name="rules"><b>Rules</b></a></div> +<!-- *********************************************************************** --> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="bounds"><b>What happens if an array index is out of bounds?</b></a> +</div> +<div class="doc_text"> + <p>There are two senses in which an array index can be out of bounds.</p> + + <p>First, there's the array type which comes from the (static) type of + the first operand to the GEP. Indices greater than the number of elements + in the corresponding static array type are valid. There is no problem with + out of bounds indices in this sense. Indexing into an array only depends + on the size of the array element, not the number of elements.</p> + + <p>A common example of how this is used is arrays where the size is not known. + It's common to use array types with zero length to represent these. The + fact that the static type says there are zero elements is irrelevant; it's + perfectly valid to compute arbitrary element indices, as the computation + only depends on the size of the array element, not the number of + elements. Note that zero-sized arrays are not a special case here.</p> + + <p>This sense is unconnected with <tt>inbounds</tt> keyword. The + <tt>inbounds</tt> keyword is designed to describe low-level pointer + arithmetic overflow conditions, rather than high-level array + indexing rules. + + <p>Analysis passes which wish to understand array indexing should not + assume that the static array type bounds are respected.</p> + + <p>The second sense of being out of bounds is computing an address that's + beyond the actual underlying allocated object.</p> + + <p>With the <tt>inbounds</tt> keyword, the result value of the GEP is + undefined if the address is outside the actual underlying allocated + object and not the address one-past-the-end.</p> + + <p>Without the <tt>inbounds</tt> keyword, there are no restrictions + on computing out-of-bounds addresses. Obviously, performing a load or + a store requires an address of allocated and sufficiently aligned + memory. But the GEP itself is only concerned with computing addresses.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="negative"><b>Can array indices be negative?</b></a> +</div> +<div class="doc_text"> + <p>Yes. This is basically a special case of array indices being out + of bounds.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="compare"><b>Can I compare two values computed with GEPs?</b></a> +</div> +<div class="doc_text"> + <p>Yes. If both addresses are within the same allocated object, or + one-past-the-end, you'll get the comparison result you expect. If either + is outside of it, integer arithmetic wrapping may occur, so the + comparison may not be meaningful.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="types"><b>Can I do GEP with a different pointer type than the type of + the underlying object?</b></a> +</div> +<div class="doc_text"> + <p>Yes. There are no restrictions on bitcasting a pointer value to an arbitrary + pointer type. The types in a GEP serve only to define the parameters for the + underlying integer computation. They need not correspond with the actual + type of the underlying object.</p> + + <p>Furthermore, loads and stores don't have to use the same types as the type + of the underlying object. Types in this context serve only to specify + memory size and alignment. Beyond that there are merely a hint to the + optimizer indicating how the value will likely be used.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="null"><b>Can I cast an object's address to integer and add it + to null?</b></a> +</div> +<div class="doc_text"> + <p>You can compute an address that way, but if you use GEP to do the add, + you can't use that pointer to actually access the object, unless the + object is managed outside of LLVM.</p> + + <p>The underlying integer computation is sufficiently defined; null has a + defined value -- zero -- and you can add whatever value you want to it.</p> + + <p>However, it's invalid to access (load from or store to) an LLVM-aware + object with such a pointer. This includes GlobalVariables, Allocas, and + objects pointed to by noalias pointers.</p> + + <p>If you really need this functionality, you can do the arithmetic with + explicit integer instructions, and use inttoptr to convert the result to + an address. Most of GEP's special aliasing rules do not apply to pointers + computed from ptrtoint, arithmetic, and inttoptr sequences.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="ptrdiff"><b>Can I compute the distance between two objects, and add + that value to one address to compute the other address?</b></a> +</div> +<div class="doc_text"> + <p>As with arithmetic on null, You can use GEP to compute an address that + way, but you can't use that pointer to actually access the object if you + do, unless the object is managed outside of LLVM.</p> + + <p>Also as above, ptrtoint and inttoptr provide an alternative way to do this + which do not have this restriction.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="tbaa"><b>Can I do type-based alias analysis on LLVM IR?</b></a> +</div> +<div class="doc_text"> + <p>You can't do type-based alias analysis using LLVM's built-in type system, + because LLVM has no restrictions on mixing types in addressing, loads or + stores.</p> + + <p>It would be possible to add special annotations to the IR, probably using + metadata, to describe a different type system (such as the C type system), + and do type-based aliasing on top of that. This is a much bigger + undertaking though.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="overflow"><b>What happens if a GEP computation overflows?</b></a> +</div> +<div class="doc_text"> + <p>If the GEP has the <tt>inbounds</tt> keyword, the result value is + undefined.</p> + + <p>Otherwise, the result value is the result from evaluating the implied + two's complement integer computation. However, since there's no + guarantee of where an object will be allocated in the address space, + such values have limited meaning.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="check"><b>How can I tell if my front-end is following the + rules?</b></a> +</div> +<div class="doc_text"> + <p>There is currently no checker for the getelementptr rules. Currently, + the only way to do this is to manually check each place in your front-end + where GetElementPtr operators are created.</p> + + <p>It's not possible to write a checker which could find all rule + violations statically. It would be possible to write a checker which + works by instrumenting the code with dynamic checks though. Alternatively, + it would be possible to write a static checker which catches a subset of + possible problems. However, no such checker exists today.</p> + +</div> + +<!-- *********************************************************************** --> +<div class="doc_section"><a name="rationale"><b>Rationale</b></a></div> +<!-- *********************************************************************** --> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="goals"><b>Why is GEP designed this way?</b></a> +</div> +<div class="doc_text"> + <p>The design of GEP has the following goals, in rough unofficial + order of priority:</p> + <ul> + <li>Support C, C-like languages, and languages which can be + conceptually lowered into C (this covers a lot).</li> + <li>Support optimizations such as those that are common in + C compilers.</li> + <li>Provide a consistent method for computing addresses so that + address computations don't need to be a part of load and + store instructions in the IR.</li> + <li>Support non-C-like languages, to the extent that it doesn't + interfere with other goals.</li> + <li>Minimize target-specific information in the IR.</li> + </ul> +</div> + +<!-- *********************************************************************** --> +<div class="doc_subsection"> + <a name="i32"><b>Why do struct member indices always use i32?</b></a> +</div> +<div class="doc_text"> + <p>The specific type i32 is probably just a historical artifact, however it's + wide enough for all practical purposes, so there's been no need to change it. + It doesn't necessarily imply i32 address arithmetic; it's just an identifier + which identifies a field in a struct. Requiring that all struct indices be + the same reduces the range of possibilities for cases where two GEPs are + effectively the same but have distinct operand types.</p> + +</div> + +<!-- *********************************************************************** --> + +<div class="doc_subsection"> + <a name="uglygep"><b>What's an uglygep?</b></a> +</div> +<div class="doc_text"> + <p>Some LLVM optimizers operate on GEPs by internally lowering them into + more primitive integer expressions, which allows them to be combined + with other integer expressions and/or split into multiple separate + integer expressions. If they've made non-trivial changes, translating + back into LLVM IR can involve reverse-engineering the structure of + the addressing in order to fit it into the static type of the original + first operand. It isn't always possibly to fully reconstruct this + structure; sometimes the underlying addressing doesn't correspond with + the static type at all. In such cases the optimizer instead will emit + a GEP with the base pointer casted to a simple address-unit pointer, + using the name "uglygep". This isn't pretty, but it's just as + valid, and it's sufficient to preserve the pointer aliasing guarantees + that GEP provides.</p> + +</div> + +<!-- *********************************************************************** --> <div class="doc_section"><a name="summary"><b>Summary</b></a></div> <!-- *********************************************************************** --> @@ -365,7 +728,7 @@ idx3 = (char*) &MyVar + 8 <a href="http://validator.w3.org/check/referer"><img src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br/> - Last modified: $Date: 2009-10-12 16:46:08 +0200 (Mon, 12 Oct 2009) $ + Last modified: $Date: 2010-02-25 19:16:03 +0100 (Thu, 25 Feb 2010) $ </address> </body> </html> diff --git a/docs/GettingStarted.html b/docs/GettingStarted.html index 44fbf21..d301e4c 100644 --- a/docs/GettingStarted.html +++ b/docs/GettingStarted.html @@ -1193,10 +1193,16 @@ $ ./hello.bc </div> <p> -This allows you to execute LLVM bitcode files directly. Thanks to Jack -Cummings for pointing this out! +This allows you to execute LLVM bitcode files directly. On Debian, you +can also use this command instead of the 'echo' command above:</p> </p> +<div class="doc_code"> +<pre> +$ sudo update-binfmts --install llvm /path/to/lli --magic 'BC' +</pre> +</div> + </div> <!-- *********************************************************************** --> @@ -1363,7 +1369,7 @@ end to compile.</p> <p>The <b>tools</b> directory contains the executables built out of the libraries above, which form the main part of the user interface. You can -always get help for a tool by typing <tt>tool_name --help</tt>. The +always get help for a tool by typing <tt>tool_name -help</tt>. The following is a brief introduction to the most important tools. More detailed information is in the <a href="CommandGuide/index.html">Command Guide</a>.</p> @@ -1434,7 +1440,7 @@ information is in the <a href="CommandGuide/index.html">Command Guide</a>.</p> <dt><tt><b>opt</b></tt></dt> <dd><tt>opt</tt> reads LLVM bitcode, applies a series of LLVM to LLVM transformations (which are specified on the command line), and then outputs - the resultant bitcode. The '<tt>opt --help</tt>' command is a good way to + the resultant bitcode. The '<tt>opt -help</tt>' command is a good way to get a list of the program transformations available in LLVM.<br> <dd><tt>opt</tt> can also be used to run a specific analysis on an input LLVM bitcode file and print out the results. It is primarily useful for @@ -1667,7 +1673,7 @@ out:</p> <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.x10sys.com/rspencer/">Reid Spencer</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-11 22:51:51 +0100 (Thu, 11 Feb 2010) $ + Last modified: $Date: 2010-02-18 15:08:13 +0100 (Thu, 18 Feb 2010) $ </address> </body> </html> diff --git a/docs/LangRef.html b/docs/LangRef.html index b31d22f..f3f73fa 100644 --- a/docs/LangRef.html +++ b/docs/LangRef.html @@ -848,7 +848,7 @@ define i32 @main() { <i>; i32()* </i> <div class="doc_text"> -<p>LLVM function definitions consist of the "<tt>define</tt>" keyord, an +<p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an optional <a href="#linkage">linkage type</a>, an optional <a href="#visibility">visibility style</a>, an optional <a href="#callingconv">calling convention</a>, a return type, an optional @@ -1277,7 +1277,7 @@ target datalayout = "<i>layout specification</i>" <ul> <li><tt>E</tt> - big endian</li> - <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li> + <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li> <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li> <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li> <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li> @@ -1664,7 +1664,7 @@ Classifications</a> </div> which indicates that the function takes a variable number of arguments. Variable argument functions can access their arguments with the <a href="#int_varargs">variable argument handling intrinsic</a> - functions. '<tt><returntype></tt>' is a any type except + functions. '<tt><returntype></tt>' is any type except <a href="#t_label">label</a>.</p> <h5>Examples:</h5> @@ -1674,12 +1674,11 @@ Classifications</a> </div> <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt> </td> </tr><tr class="layout"> - <td class="left"><tt>float (i16 signext, i32 *) * + <td class="left"><tt>float (i16, i32 *) * </tt></td> <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes - an <tt>i16</tt> that should be sign extended and a - <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning - <tt>float</tt>. + an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>, + returning <tt>float</tt>. </td> </tr><tr class="layout"> <td class="left"><tt>i32 (i8*, ...)</tt></td> @@ -1792,7 +1791,7 @@ Classifications</a> </div> and the alignment requirements of the union as a whole will be the largest alignment requirement of any member.</p> -<p>Unions members are accessed using '<tt><a href="#i_load">load</a></tt> and +<p>Union members are accessed using '<tt><a href="#i_load">load</a></tt> and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. Since all members are at offset zero, the getelementptr instruction does @@ -1827,10 +1826,13 @@ Classifications</a> </div> <div class="doc_text"> <h5>Overview:</h5> -<p>As in many languages, the pointer type represents a pointer or reference to - another object, which must live in memory. Pointer types may have an optional - address space attribute defining the target-specific numbered address space - where the pointed-to object resides. The default address space is zero.</p> +<p>The pointer type is used to specify memory locations. + Pointers are commonly used to reference objects in memory.</p> + +<p>Pointer types may have an optional address space attribute defining the + numbered address space where the pointed-to object resides. The default + address space is number zero. The semantics of non-zero address + spaces are target-specific.</p> <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p> @@ -2885,9 +2887,10 @@ IfUnequal: function to be invoked. </li> <li>'<tt>function args</tt>': argument list whose types match the function - signature argument types. If the function signature indicates the - function accepts a variable number of arguments, the extra arguments can - be specified.</li> + signature argument types and parameter attributes. All arguments must be + of <a href="#t_firstclass">first class</a> type. If the function + signature indicates the function accepts a variable number of arguments, + the extra arguments can be specified.</li> <li>'<tt>normal label</tt>': the label reached when the called function executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li> @@ -4074,8 +4077,9 @@ Instruction</a> </div> <h5>Syntax:</h5> <pre> - <result> = load <ty>* <pointer>[, align <alignment>] - <result> = volatile load <ty>* <pointer>[, align <alignment>] + <result> = load <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] + <result> = volatile load <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] + !<index> = !{ i32 1 } </pre> <h5>Overview:</h5> @@ -4088,16 +4092,24 @@ Instruction</a> </div> marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the number or order of execution of this <tt>load</tt> with other volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt> - instructions. </p> + instructions.</p> -<p>The optional constant "align" argument specifies the alignment of the +<p>The optional constant <tt>align</tt> argument specifies the alignment of the operation (that is, the alignment of the memory address). A value of 0 or an - omitted "align" argument means that the operation has the preferential + omitted <tt>align</tt> argument means that the operation has the preferential alignment for the target. It is the responsibility of the code emitter to ensure that the alignment information is correct. Overestimating the - alignment results in an undefined behavior. Underestimating the alignment may + alignment results in undefined behavior. Underestimating the alignment may produce less efficient code. An alignment of 1 is always safe.</p> +<p>The optional <tt>!nontemporal</tt> metadata must reference a single + metatadata name <index> corresponding to a metadata node with + one <tt>i32</tt> entry of value 1. The existence of + the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer + and code generator that this load is not expected to be reused in the cache. + The code generator may select special instructions to save cache bandwidth, + such as the <tt>MOVNT</tt> instruction on x86.</p> + <h5>Semantics:</h5> <p>The location of memory pointed to is loaded. If the value being loaded is of scalar type then the number of bytes read does not exceed the minimum number @@ -4124,8 +4136,8 @@ Instruction</a> </div> <h5>Syntax:</h5> <pre> - store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i> - volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i> + store <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] <i>; yields {void}</i> + volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] <i>; yields {void}</i> </pre> <h5>Overview:</h5> @@ -4150,6 +4162,15 @@ Instruction</a> </div> alignment results in an undefined behavior. Underestimating the alignment may produce less efficient code. An alignment of 1 is always safe.</p> +<p>The optional !nontemporal metadata must reference a single metatadata + name <index> corresponding to a metadata node with one i32 entry of + value 1. The existence of the !nontemporal metatadata on the + instruction tells the optimizer and code generator that this load is + not expected to be reused in the cache. The code generator may + select special instructions to save cache bandwidth, such as the + MOVNT instruction on x86.</p> + + <h5>Semantics:</h5> <p>The contents of memory are updated to contain '<tt><value></tt>' at the location specified by the '<tt><pointer></tt>' operand. If @@ -4943,7 +4964,7 @@ entry: <tt>op1</tt> is equal to <tt>op2</tt>.</li> <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and - <tt>op1</tt> is greather than <tt>op2</tt>.</li> + <tt>op1</tt> is greater than <tt>op2</tt>.</li> <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li> @@ -5119,7 +5140,7 @@ Loop: ; Infinite loop that counts from 0 on up... <li>The call is in tail position (ret immediately follows call and ret uses value of call or is void).</li> <li>Option <tt>-tailcallopt</tt> is enabled, - or <code>llvm::PerformTailCallOpt</code> is <code>true</code>.</li> + or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li> <li><a href="CodeGenerator.html#tailcallopt">Platform specific constraints are met.</a></li> </ul> @@ -5150,10 +5171,10 @@ Loop: ; Infinite loop that counts from 0 on up... to function value.</li> <li>'<tt>function args</tt>': argument list whose types match the function - signature argument types. All arguments must be of - <a href="#t_firstclass">first class</a> type. If the function signature - indicates the function accepts a variable number of arguments, the extra - arguments can be specified.</li> + signature argument types and parameter attributes. All arguments must be + of <a href="#t_firstclass">first class</a> type. If the function + signature indicates the function accepts a variable number of arguments, + the extra arguments can be specified.</li> <li>The optional <a href="#fnattrs">function attributes</a> list. Only '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and @@ -5188,7 +5209,7 @@ Loop: ; Infinite loop that counts from 0 on up... standard C99 library as being the C99 library functions, and may perform optimizations or generate code for them under that assumption. This is something we'd like to change in the future to provide better support for -freestanding environments and non-C-based langauges.</p> +freestanding environments and non-C-based languages.</p> </div> @@ -5744,7 +5765,7 @@ LLVM</a>.</p> <h5>Semantics:</h5> <p>This intrinsic does not modify the behavior of the program. Backends that do - not support this intrinisic may ignore it.</p> + not support this intrinsic may ignore it.</p> </div> @@ -5824,7 +5845,7 @@ LLVM</a>.</p> number of bytes to copy, and the fourth argument is the alignment of the source and destination locations.</p> -<p>If the call to this intrinisic has an alignment value that is not 0 or 1, +<p>If the call to this intrinsic has an alignment value that is not 0 or 1, then the caller guarantees that both the source and destination pointers are aligned to that boundary.</p> @@ -5874,7 +5895,7 @@ LLVM</a>.</p> number of bytes to copy, and the fourth argument is the alignment of the source and destination locations.</p> -<p>If the call to this intrinisic has an alignment value that is not 0 or 1, +<p>If the call to this intrinsic has an alignment value that is not 0 or 1, then the caller guarantees that the source and destination pointers are aligned to that boundary.</p> @@ -5922,7 +5943,7 @@ LLVM</a>.</p> specifying the number of bytes to fill, and the fourth argument is the known alignment of destination location.</p> -<p>If the call to this intrinisic has an alignment value that is not 0 or 1, +<p>If the call to this intrinsic has an alignment value that is not 0 or 1, then the caller guarantees that the destination pointer is aligned to that boundary.</p> @@ -6693,7 +6714,7 @@ LLVM</a>.</p> <h5>Arguments:</h5> <p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments. The first four arguments enables a specific barrier as listed below. The - fith argument specifies that the barrier applies to io or device or uncached + fifth argument specifies that the barrier applies to io or device or uncached memory.</p> <ul> @@ -7432,7 +7453,7 @@ LLVM</a>.</p> <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-12 21:49:41 +0100 (Fri, 12 Feb 2010) $ + Last modified: $Date: 2010-03-02 07:36:51 +0100 (Tue, 02 Mar 2010) $ </address> </body> diff --git a/docs/MakefileGuide.html b/docs/MakefileGuide.html index 36a4725..eb91a21 100644 --- a/docs/MakefileGuide.html +++ b/docs/MakefileGuide.html @@ -818,6 +818,10 @@ <tt>mklib</tt> by the <tt>configure</tt> script and always located in the <dt><a name="LLVMAS"><tt>LLVMAS</tt></a><small>(defaulted)</small></dt> <dd>Specifies the path to the <tt>llvm-as</tt> tool.</dd> + <dt><a name="LLVMCC"><tt>LLVMCC</tt></a></dt> + <dd>Specifies the path to the LLVM capable compiler.</dd> + <dt><a name="LLVMCXX"><tt>LLVMCXX</tt></a></dt> + <dd>Specifies the path to the LLVM C++ capable compiler.</dd> <dt><a name="LLVMGCC"><tt>LLVMGCC</tt></a><small>(defaulted)</small></dt> <dd>Specifies the path to the LLVM version of the GCC 'C' Compiler</dd> <dt><a name="LLVMGXX"><tt>LLVMGXX</tt></a><small>(defaulted)</small></dt> @@ -1021,7 +1025,7 @@ <a href="mailto:rspencer@x10sys.com">Reid Spencer</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-10-12 16:46:08 +0200 (Mon, 12 Oct 2009) $ + Last modified: $Date: 2010-02-23 11:00:53 +0100 (Tue, 23 Feb 2010) $ </address> </body> </html> diff --git a/docs/Packaging.html b/docs/Packaging.html new file mode 100644 index 0000000..678edb0 --- /dev/null +++ b/docs/Packaging.html @@ -0,0 +1,118 @@ +<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" + "http://www.w3.org/TR/html4/strict.dtd"> +<html> +<head> + <title>Advice on Packaging LLVM</title> + <link rel="stylesheet" href="llvm.css" type="text/css"> +</head> +<body> + +<div class="doc_title">Advice on Packaging LLVM</div> +<ol> + <li><a href="#overview">Overview</a></li> + <li><a href="#compilation">Compile Flags</a></li> + <li><a href="#cxx-features">C++ Features</a></li> + <li><a href="#shared-library">Shared Library</a></li> + <li><a href="#deps">Dependencies</a></li> +</ol> + +<!--=========================================================================--> +<div class="doc_section"><a name="overview">Overview</a></div> +<!--=========================================================================--> +<div class="doc_text"> + +<p>LLVM sets certain default configure options to make sure our developers don't +break things for constrained platforms. These settings are not optimal for most +desktop systems, and we hope that packagers (e.g., Redhat, Debian, MacPorts, +etc.) will tweak them. This document lists settings we suggest you tweak. +</p> + +<p>LLVM's API changes with each release, so users are likely to want, for +example, both LLVM-2.6 and LLVM-2.7 installed at the same time to support apps +developed against each. +</p> +</div> + +<!--=========================================================================--> +<div class="doc_section"><a name="compilation">Compile Flags</a></div> +<!--=========================================================================--> +<div class="doc_text"> + +<p>LLVM runs much more quickly when it's optimized and assertions are removed. +However, such a build is currently incompatible with users who build without +defining NDEBUG, and the lack of assertions makes it hard to debug problems in +user code. We recommend allowing users to install both optimized and debug +versions of LLVM in parallel. The following configure flags are relevant: +</p> + +<dl> + <dt><tt>--disable-assertions</tt></dt><dd>Builds LLVM with <tt>NDEBUG</tt> + defined. Changes the LLVM ABI. Also available by setting + <tt>DISABLE_ASSERTIONS=0|1</tt> in <tt>make</tt>'s environment. This defaults + to enabled regardless of the optimization setting, but it slows things + down.</dd> + + <dt><tt>--enable-debug-symbols</tt></dt><dd>Builds LLVM with <tt>-g</tt>. + Also available by setting <tt>DEBUG_SYMBOLS=0|1</tt> in <tt>make</tt>'s + environment. This defaults to disabled when optimizing, so you should turn it + back on to let users debug their programs.</dd> + + <dt><tt>--enable-optimized</tt></dt><dd>(For svn checkouts) Builds LLVM with + <tt>-O2</tt> and, by default, turns off debug symbols. Also available by + setting <tt>ENABLE_OPTIMIZED=0|1</tt> in <tt>make</tt>'s environment. This + defaults to enabled when not in a checkout.</dd> +</dl> +</div> + +<!--=========================================================================--> +<div class="doc_section"><a name="cxx-features">C++ Features</a></div> +<!--=========================================================================--> +<div class="doc_text"> + +<dl> + <dt>RTTI</dt><dd>LLVM disables RTTI by default. Add <tt>REQUIRES_RTTI=1</tt> + to your environment while running <tt>make</tt> to re-enable it. This will + allow users to build with RTTI enabled and still inherit from LLVM + classes.</dd> +</dl> +</div> + +<!--=========================================================================--> +<div class="doc_section"><a name="shared-library">Shared Library</a></div> +<!--=========================================================================--> +<div class="doc_text"> + +<p>Configure with <tt>--enable-shared</tt> to build +<tt>libLLVM-<var>major</var>.<var>minor</var>.(so|dylib)</tt> and link the tools +against it. This saves lots of binary size at the cost of some startup time. +</p> +</div> + +<!--=========================================================================--> +<div class="doc_section"><a name="deps">Dependencies</a></div> +<!--=========================================================================--> +<div class="doc_text"> + +<dl> +<dt><tt>--enable-libffi</tt></dt><dd>Depend on <a +href="http://sources.redhat.com/libffi/">libffi</a> to allow the LLVM +interpreter to call external functions.</dd> +<dt><tt>--with-oprofile</tt></dt><dd>Depend on <a +href="http://oprofile.sourceforge.net/doc/devel/index.html">libopagent</a> +(>=version 0.9.4) to let the LLVM JIT tell oprofile about function addresses and +line numbers.</dd> +</dl> +</div> + +<!-- *********************************************************************** --> +<hr> +<address> + <a href="http://jigsaw.w3.org/css-validator/check/referer"><img + src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a> + <a href="http://validator.w3.org/check/referer"><img + src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a> + <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> + Last modified: $Date: 2010-02-26 23:25:06 +0100 (Fri, 26 Feb 2010) $ +</address> +</body> +</html> diff --git a/docs/Passes.html b/docs/Passes.html index 2107e13..e3403b4 100644 --- a/docs/Passes.html +++ b/docs/Passes.html @@ -75,7 +75,6 @@ perl -e '$/ = undef; for (split(/\n/, <>)) { s:^ *///? ?::; print " <p>\n" if ! <tr><th colspan="2"><b>ANALYSIS PASSES</b></th></tr> <tr><th>Option</th><th>Name</th></tr> <tr><td><a href="#aa-eval">-aa-eval</a></td><td>Exhaustive Alias Analysis Precision Evaluator</td></tr> -<tr><td><a href="#anders-aa">-anders-aa</a></td><td>Andersen's Interprocedural Alias Analysis</td></tr> <tr><td><a href="#basicaa">-basicaa</a></td><td>Basic Alias Analysis (default AA impl)</td></tr> <tr><td><a href="#basiccg">-basiccg</a></td><td>Basic CallGraph Construction</td></tr> <tr><td><a href="#codegenprepare">-codegenprepare</a></td><td>Optimize for code generation</td></tr> @@ -204,80 +203,6 @@ perl -e '$/ = undef; for (split(/\n/, <>)) { s:^ *///? ?::; print " <p>\n" if ! <!-------------------------------------------------------------------------- --> <div class="doc_subsection"> - <a name="anders-aa">Andersen's Interprocedural Alias Analysis</a> -</div> -<div class="doc_text"> - <p> - This is an implementation of Andersen's interprocedural alias - analysis - </p> - - <p> - In pointer analysis terms, this is a subset-based, flow-insensitive, - field-sensitive, and context-insensitive algorithm pointer algorithm. - </p> - - <p> - This algorithm is implemented as three stages: - </p> - - <ol> - <li>Object identification.</li> - <li>Inclusion constraint identification.</li> - <li>Offline constraint graph optimization.</li> - <li>Inclusion constraint solving.</li> - </ol> - - <p> - The object identification stage identifies all of the memory objects in the - program, which includes globals, heap allocated objects, and stack allocated - objects. - </p> - - <p> - The inclusion constraint identification stage finds all inclusion constraints - in the program by scanning the program, looking for pointer assignments and - other statements that effect the points-to graph. For a statement like - <code><var>A</var> = <var>B</var></code>, this statement is processed to - indicate that <var>A</var> can point to anything that <var>B</var> can point - to. Constraints can handle copies, loads, and stores, and address taking. - </p> - - <p> - The offline constraint graph optimization portion includes offline variable - substitution algorithms intended to computer pointer and location - equivalences. Pointer equivalences are those pointers that will have the - same points-to sets, and location equivalences are those variables that - always appear together in points-to sets. - </p> - - <p> - The inclusion constraint solving phase iteratively propagates the inclusion - constraints until a fixed point is reached. This is an O(<var>n</var>³) - algorithm. - </p> - - <p> - Function constraints are handled as if they were structs with <var>X</var> - fields. Thus, an access to argument <var>X</var> of function <var>Y</var> is - an access to node index <code>getNode(<var>Y</var>) + <var>X</var></code>. - This representation allows handling of indirect calls without any issues. To - wit, an indirect call <code><var>Y</var>(<var>a</var>,<var>b</var>)</code> is - equivalent to <code>*(<var>Y</var> + 1) = <var>a</var>, *(<var>Y</var> + 2) = - <var>b</var></code>. The return node for a function <var>F</var> is always - located at <code>getNode(<var>F</var>) + CallReturnPos</code>. The arguments - start at <code>getNode(<var>F</var>) + CallArgPos</code>. - </p> - - <p> - Please keep in mind that the current andersen's pass has many known - problems and bugs. It should be considered "research quality". - </p> - -</div> - -<!-------------------------------------------------------------------------- --> -<div class="doc_subsection"> <a name="basicaa">Basic Alias Analysis (default AA impl)</a> </div> <div class="doc_text"> @@ -1848,7 +1773,7 @@ if (X < 3) {</pre> <a href="mailto:rspencer@x10sys.com">Reid Spencer</a><br> <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-10-28 05:47:06 +0100 (Wed, 28 Oct 2009) $ + Last modified: $Date: 2010-03-01 20:24:17 +0100 (Mon, 01 Mar 2010) $ </address> </body> diff --git a/docs/ProgrammersManual.html b/docs/ProgrammersManual.html index df4366a..ace44e7 100644 --- a/docs/ProgrammersManual.html +++ b/docs/ProgrammersManual.html @@ -525,7 +525,7 @@ lightweight <a href="http://en.wikipedia.org/wiki/Rope_(computer_science)">rope< which points to temporary (stack allocated) objects. Twines can be implicitly constructed as the result of the plus operator applied to strings (i.e., a C strings, an <tt>std::string</tt>, or a <tt>StringRef</tt>). The twine delays the -actual concatentation of strings until it is actually required, at which point +actual concatenation of strings until it is actually required, at which point it can be efficiently rendered directly into a character array. This avoids unnecessary heap allocation involved in constructing the temporary results of string concatenation. See @@ -1098,7 +1098,7 @@ in the default manner.</p> </div> <div class="doc_text"> -<p><tt>ilist</tt>s have another speciality that must be considered. To be a good +<p><tt>ilist</tt>s have another specialty that must be considered. To be a good citizen in the C++ ecosystem, it needs to support the standard container operations, such as <tt>begin</tt> and <tt>end</tt> iterators, etc. Also, the <tt>operator--</tt> must work correctly on the <tt>end</tt> iterator in the @@ -3921,7 +3921,7 @@ arguments. An argument has a pointer to the parent Function.</p> <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-15 17:12:20 +0100 (Mon, 15 Feb 2010) $ + Last modified: $Date: 2010-02-26 00:51:27 +0100 (Fri, 26 Feb 2010) $ </address> </body> diff --git a/docs/ReleaseNotes.html b/docs/ReleaseNotes.html index 21f5082..5bdbceb 100644 --- a/docs/ReleaseNotes.html +++ b/docs/ReleaseNotes.html @@ -451,6 +451,8 @@ release includes a few major enhancements and additions to the optimizers:</p> </ul> +<p>Also, -anders-aa was removed</p> + </div> @@ -644,6 +646,10 @@ Clients must replace calls to <li>The <tt>llvm/Support/DataTypes.h</tt> header has moved to <tt>llvm/System/DataTypes.h</tt>.</li> +<li>The <tt>isInteger</tt>, <tt>isIntOrIntVector</tt>, <tt>isFloatingPoint</tt>, +<tt>isFPOrFPVector</tt> and <tt>isFPOrFPVector</tt> methods have been renamed +<tt>isIntegerTy</tt>, <tt>isIntOrIntVectorTy</tt>, <tt>isFloatingPointTy</tt>, +<tt>isFPOrFPVectorTy</tt> and <tt>isFPOrFPVectorTy</tt> respectively.</li> </ul> </div> @@ -729,7 +735,6 @@ href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev list</a>.</p> experimental.</li> <li>The <tt>llc</tt> "<tt>-filetype=asm</tt>" (the default) is the only supported value for this option. The ELF writer is experimental.</li> -<li>The implementation of Andersen's Alias Analysis has many known bugs.</li> </ul> </div> @@ -992,7 +997,7 @@ lists</a>.</p> src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a> <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-14 02:47:19 +0100 (Sun, 14 Feb 2010) $ + Last modified: $Date: 2010-03-01 20:29:17 +0100 (Mon, 01 Mar 2010) $ </address> </body> diff --git a/docs/TableGenFundamentals.html b/docs/TableGenFundamentals.html index b96aeab..701d11f 100644 --- a/docs/TableGenFundamentals.html +++ b/docs/TableGenFundamentals.html @@ -155,7 +155,6 @@ file prints this (at the time of this writing):</p> <b>bit</b> hasCtrlDep = 0; <b>bit</b> isNotDuplicable = 0; <b>bit</b> hasSideEffects = 0; - <b>bit</b> mayHaveSideEffects = 0; <b>bit</b> neverHasSideEffects = 0; InstrItinClass Itinerary = NoItinerary; <b>string</b> Constraints = ""; @@ -189,7 +188,7 @@ backend, and is only shown as an example.</p> <p>As you can see, a lot of information is needed for every instruction supported by the code generator, and specifying it all manually would be -unmaintainble, prone to bugs, and tiring to do in the first place. Because we +unmaintainable, prone to bugs, and tiring to do in the first place. Because we are using TableGen, all of the information was derived from the following definition:</p> @@ -798,7 +797,7 @@ This should highlight the APIs in <tt>TableGen/Record.h</tt>.</p> <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-01-05 20:11:42 +0100 (Tue, 05 Jan 2010) $ + Last modified: $Date: 2010-02-28 00:47:46 +0100 (Sun, 28 Feb 2010) $ </address> </body> diff --git a/docs/TestingGuide.html b/docs/TestingGuide.html index 4f05d77..76d6de0 100644 --- a/docs/TestingGuide.html +++ b/docs/TestingGuide.html @@ -693,8 +693,6 @@ that FileCheck is not actually line-oriented when it matches, this allows you to define two separate CHECK lines that match on the same line. </p> - - </div> <!-- _______________________________________________________________________ --> @@ -761,12 +759,6 @@ substitutions</a></div> <dd>The full path to the <tt>llvm-gxx</tt> executable as specified in the configured LLVM environment</dd> - <dt><b>llvmgcc_version</b> (%llvmgcc_version)</dt> - <dd>The full version number of the <tt>llvm-gcc</tt> executable.</dd> - - <dt><b>llvmgccmajvers</b> (%llvmgccmajvers)</dt> - <dd>The major version number of the <tt>llvm-gcc</tt> executable.</dd> - <dt><b>gccpath</b></dt> <dd>The full path to the C compiler used to <i>build </i> LLVM. Note that this might not be gcc.</dd> @@ -824,22 +816,20 @@ substitutions</a></div> </dl> <p>Sometimes it is necessary to mark a test case as "expected fail" or XFAIL. - You can easily mark a test as XFAIL just by including <tt>XFAIL: </tt> on a + You can easily mark a test as XFAIL just by including <tt>XFAIL: </tt> on a line near the top of the file. This signals that the test case should succeed if the test fails. Such test cases are counted separately by DejaGnu. To specify an expected fail, use the XFAIL keyword in the comments of the test program followed by a colon and one or more regular expressions (separated by - a comma). The regular expressions allow you to XFAIL the test conditionally - by host platform. The regular expressions following the : are matched against - the target triplet or llvmgcc version number for the host machine. If there is - a match, the test is expected to fail. If not, the test is expected to - succeed. To XFAIL everywhere just specify <tt>XFAIL: *</tt>. When matching - the llvm-gcc version, you can specify the major (e.g. 3) or full version - (i.e. 3.4) number. Here is an example of an <tt>XFAIL</tt> line:</p> + a comma). The regular expressions allow you to XFAIL the test conditionally by + host platform. The regular expressions following the : are matched against the + target triplet for the host machine. If there is a match, the test is expected + to fail. If not, the test is expected to succeed. To XFAIL everywhere just + specify <tt>XFAIL: *</tt>. Here is an example of an <tt>XFAIL</tt> line:</p> <div class="doc_code"> <pre> -; XFAIL: darwin,sun,llvmgcc4 +; XFAIL: darwin,sun </pre> </div> @@ -1145,7 +1135,6 @@ example reports that can do fancy stuff.</p> </div> - <!--=========================================================================--> <div class="doc_section"><a name="nightly">Running the nightly tester</a></div> <!--=========================================================================--> @@ -1217,7 +1206,7 @@ know. Thanks!</p> John T. Criswell, Reid Spencer, and Tanya Lattner<br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-09-27 10:01:44 +0200 (Sun, 27 Sep 2009) $ + Last modified: $Date: 2010-02-26 22:23:59 +0100 (Fri, 26 Feb 2010) $ </address> </body> </html> diff --git a/docs/WritingAnLLVMPass.html b/docs/WritingAnLLVMPass.html index 218f8ef..d18181e 100644 --- a/docs/WritingAnLLVMPass.html +++ b/docs/WritingAnLLVMPass.html @@ -377,10 +377,10 @@ interesting way, we just throw away the result of <tt>opt</tt> (sending it to <tt>/dev/null</tt>).</p> <p>To see what happened to the other string you registered, try running -<tt>opt</tt> with the <tt>--help</tt> option:</p> +<tt>opt</tt> with the <tt>-help</tt> option:</p> <div class="doc_code"><pre> -$ opt -load ../../../Debug/lib/Hello.so --help +$ opt -load ../../../Debug/lib/Hello.so -help OVERVIEW: llvm .bc -> .bc modular optimizer USAGE: opt [options] <input bitcode> @@ -970,7 +970,7 @@ template, which requires you to pass at least two parameters. The first parameter is the name of the pass that is to be used on the command line to specify that the pass should be added to a program (for example, with <tt>opt</tt> or <tt>bugpoint</tt>). The second argument is the -name of the pass, which is to be used for the <tt>--help</tt> output of +name of the pass, which is to be used for the <tt>-help</tt> output of programs, as well as for debug output generated by the <tt>--debug-pass</tt> option.</p> @@ -1410,7 +1410,7 @@ allowing any analysis results to live across the execution of your pass.</p> options that is useful for debugging pass execution, seeing how things work, and diagnosing when you should be preserving more analyses than you currently are (To get information about all of the variants of the <tt>--debug-pass</tt> -option, just type '<tt>opt --help-hidden</tt>').</p> +option, just type '<tt>opt -help-hidden</tt>').</p> <p>By using the <tt>--debug-pass=Structure</tt> option, for example, we can see how our <a href="#basiccode">Hello World</a> pass interacts with other passes. @@ -1625,12 +1625,12 @@ form; </p> </pre></div> <p>Note the two spaces prior to the help string produces a tidy result on the ---help query.</p> +-help query.</p> <div class="doc_code"><pre> -$ llc --help +$ llc -help ... - -regalloc - Register allocator to use: (default = linearscan) + -regalloc - Register allocator to use (default=linearscan) =linearscan - linear scan register allocator =local - local register allocator =simple - simple register allocator @@ -1829,7 +1829,7 @@ Despite that, we have kept the LLVM passes SMP ready, and you should too.</p> <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2009-10-12 16:46:08 +0200 (Mon, 12 Oct 2009) $ + Last modified: $Date: 2010-02-18 15:37:52 +0100 (Thu, 18 Feb 2010) $ </address> </body> diff --git a/docs/index.html b/docs/index.html index 1ada7de..cd43f88 100644 --- a/docs/index.html +++ b/docs/index.html @@ -93,7 +93,6 @@ Current tools: <a href="/cmds/llvmc.html">llvmc</a> <a href="/cmds/llvmgcc.html">llvm-gcc</a>, <a href="/cmds/llvmgxx.html">llvm-g++</a>, - <a href="/cmds/stkrc.html">stkrc</a>, <a href="/cmds/bugpoint.html">bugpoint</a>, <a href="/cmds/llvm-bcanalyzer.html">llvm-bcanalyzer</a>, </li> @@ -117,6 +116,9 @@ manual for using the LLVM testing infrastructure.</li> <li><a href="GCCFEBuildInstrs.html">How to build the Ada/C/C++/Fortran front-ends</a> - Instructions for building gcc front-ends from source.</li> +<li><a href="Packaging.html">Packaging guide</a> - Advice on packaging +LLVM into a distribution.</li> + <li><a href="Lexicon.html">The LLVM Lexicon</a> - Definition of acronyms, terms and concepts used in LLVM.</li> @@ -285,7 +287,7 @@ times each day, making it a high volume list.</li> src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a> <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-03 19:49:55 +0100 (Wed, 03 Feb 2010) $ + Last modified: $Date: 2010-02-26 01:54:42 +0100 (Fri, 26 Feb 2010) $ </address> </body></html> diff --git a/docs/tutorial/LangImpl3.html b/docs/tutorial/LangImpl3.html index a628145..80c9fc2 100644 --- a/docs/tutorial/LangImpl3.html +++ b/docs/tutorial/LangImpl3.html @@ -170,7 +170,7 @@ internally (<tt>APFloat</tt> has the capability of holding floating point constants of <em>A</em>rbitrary <em>P</em>recision). This code basically just creates and returns a <tt>ConstantFP</tt>. Note that in the LLVM IR that constants are all uniqued together and shared. For this reason, the API -uses "the Context.get..." idiom instead of "new foo(..)" or "foo::Create(..)".</p> +uses the "foo::get(...)" idiom instead of "new foo(..)" or "foo::Create(..)".</p> <div class="doc_code"> <pre> @@ -323,10 +323,10 @@ really talks about the external interface for a function (not the value computed by an expression), it makes sense for it to return the LLVM Function it corresponds to when codegen'd.</p> -<p>The call to <tt>Context.get</tt> creates +<p>The call to <tt>FunctionType::get</tt> creates the <tt>FunctionType</tt> that should be used for a given Prototype. Since all function arguments in Kaleidoscope are of type double, the first line creates -a vector of "N" LLVM double types. It then uses the <tt>Context.get</tt> +a vector of "N" LLVM double types. It then uses the <tt>Functiontype::get</tt> method to create a function type that takes "N" doubles as arguments, returns one double as a result, and that is not vararg (the false parameter indicates this). Note that Types in LLVM are uniqued just like Constants are, so you @@ -535,8 +535,7 @@ ready> <b>4+5</b>; Read top-level expression: define double @""() { entry: - %addtmp = add double 4.000000e+00, 5.000000e+00 - ret double %addtmp + ret double 9.000000e+00 } </pre> </div> @@ -544,7 +543,8 @@ entry: <p>Note how the parser turns the top-level expression into anonymous functions for us. This will be handy when we add <a href="LangImpl4.html#jit">JIT support</a> in the next chapter. Also note that the code is very literally -transcribed, no optimizations are being performed. We will +transcribed, no optimizations are being performed except simple constant +folding done by IRBuilder. We will <a href="LangImpl4.html#trivialconstfold">add optimizations</a> explicitly in the next chapter.</p> @@ -554,12 +554,12 @@ ready> <b>def foo(a b) a*a + 2*a*b + b*b;</b> Read function definition: define double @foo(double %a, double %b) { entry: - %multmp = mul double %a, %a - %multmp1 = mul double 2.000000e+00, %a - %multmp2 = mul double %multmp1, %b - %addtmp = add double %multmp, %multmp2 - %multmp3 = mul double %b, %b - %addtmp4 = add double %addtmp, %multmp3 + %multmp = fmul double %a, %a + %multmp1 = fmul double 2.000000e+00, %a + %multmp2 = fmul double %multmp1, %b + %addtmp = fadd double %multmp, %multmp2 + %multmp3 = fmul double %b, %b + %addtmp4 = fadd double %addtmp, %multmp3 ret double %addtmp4 } </pre> @@ -576,7 +576,7 @@ define double @bar(double %a) { entry: %calltmp = call double @foo( double %a, double 4.000000e+00 ) %calltmp1 = call double @bar( double 3.133700e+04 ) - %addtmp = add double %calltmp, %calltmp1 + %addtmp = fadd double %calltmp, %calltmp1 ret double %addtmp } </pre> @@ -612,18 +612,18 @@ ready> <b>^D</b> define double @""() { entry: - %addtmp = add double 4.000000e+00, 5.000000e+00 + %addtmp = fadd double 4.000000e+00, 5.000000e+00 ret double %addtmp } define double @foo(double %a, double %b) { entry: - %multmp = mul double %a, %a - %multmp1 = mul double 2.000000e+00, %a - %multmp2 = mul double %multmp1, %b - %addtmp = add double %multmp, %multmp2 - %multmp3 = mul double %b, %b - %addtmp4 = add double %addtmp, %multmp3 + %multmp = fmul double %a, %a + %multmp1 = fmul double 2.000000e+00, %a + %multmp2 = fmul double %multmp1, %b + %addtmp = fadd double %multmp, %multmp2 + %multmp3 = fmul double %b, %b + %addtmp4 = fadd double %addtmp, %multmp3 ret double %addtmp4 } @@ -631,7 +631,7 @@ define double @bar(double %a) { entry: %calltmp = call double @foo( double %a, double 4.000000e+00 ) %calltmp1 = call double @bar( double 3.133700e+04 ) - %addtmp = add double %calltmp, %calltmp1 + %addtmp = fadd double %calltmp, %calltmp1 ret double %addtmp } @@ -1263,7 +1263,7 @@ int main() { <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-03 18:27:31 +0100 (Wed, 03 Feb 2010) $ + Last modified: $Date: 2010-03-02 02:11:08 +0100 (Tue, 02 Mar 2010) $ </address> </body> </html> diff --git a/docs/tutorial/LangImpl4.html b/docs/tutorial/LangImpl4.html index 85e48c5..4520c46 100644 --- a/docs/tutorial/LangImpl4.html +++ b/docs/tutorial/LangImpl4.html @@ -65,7 +65,7 @@ ready> <b>def test(x) 1+2+x;</b> Read function definition: define double @test(double %x) { entry: - %addtmp = add double 3.000000e+00, %x + %addtmp = fadd double 3.000000e+00, %x ret double %addtmp } </pre> @@ -80,8 +80,8 @@ ready> <b>def test(x) 1+2+x;</b> Read function definition: define double @test(double %x) { entry: - %addtmp = add double 2.000000e+00, 1.000000e+00 - %addtmp1 = add double %addtmp, %x + %addtmp = fadd double 2.000000e+00, 1.000000e+00 + %addtmp1 = fadd double %addtmp, %x ret double %addtmp1 } </pre> @@ -113,9 +113,9 @@ ready> <b>def test(x) (1+2+x)*(x+(1+2));</b> ready> Read function definition: define double @test(double %x) { entry: - %addtmp = add double 3.000000e+00, %x - %addtmp1 = add double %x, 3.000000e+00 - %multmp = mul double %addtmp, %addtmp1 + %addtmp = fadd double 3.000000e+00, %x + %addtmp1 = fadd double %x, 3.000000e+00 + %multmp = fmul double %addtmp, %addtmp1 ret double %multmp } </pre> @@ -240,8 +240,8 @@ ready> <b>def test(x) (1+2+x)*(x+(1+2));</b> ready> Read function definition: define double @test(double %x) { entry: - %addtmp = add double %x, 3.000000e+00 - %multmp = mul double %addtmp, %addtmp + %addtmp = fadd double %x, 3.000000e+00 + %multmp = fmul double %addtmp, %addtmp ret double %multmp } </pre> @@ -363,8 +363,8 @@ ready> <b>def testfunc(x y) x + y*2; </b> Read function definition: define double @testfunc(double %x, double %y) { entry: - %multmp = mul double %y, 2.000000e+00 - %addtmp = add double %multmp, %x + %multmp = fmul double %y, 2.000000e+00 + %addtmp = fadd double %multmp, %x ret double %addtmp } @@ -411,10 +411,10 @@ Read function definition: define double @foo(double %x) { entry: %calltmp = call double @sin( double %x ) - %multmp = mul double %calltmp, %calltmp + %multmp = fmul double %calltmp, %calltmp %calltmp2 = call double @cos( double %x ) - %multmp4 = mul double %calltmp2, %calltmp2 - %addtmp = add double %multmp, %multmp4 + %multmp4 = fmul double %calltmp2, %calltmp2 + %addtmp = fadd double %multmp, %multmp4 ret double %addtmp } @@ -1126,7 +1126,7 @@ int main() { <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-11 20:15:20 +0100 (Thu, 11 Feb 2010) $ + Last modified: $Date: 2010-03-02 02:11:08 +0100 (Tue, 02 Mar 2010) $ </address> </body> </html> diff --git a/docs/tutorial/LangImpl5.html b/docs/tutorial/LangImpl5.html index f80f3f3..3256c40 100644 --- a/docs/tutorial/LangImpl5.html +++ b/docs/tutorial/LangImpl5.html @@ -678,7 +678,7 @@ loop: ; preds = %loop, %entry ; body %calltmp = call double @putchard( double 4.200000e+01 ) ; increment - %nextvar = add double %i, 1.000000e+00 + %nextvar = fadd double %i, 1.000000e+00 ; termination test %cmptmp = fcmp ult double %i, %n @@ -1771,7 +1771,7 @@ int main() { <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-11 20:15:20 +0100 (Thu, 11 Feb 2010) $ + Last modified: $Date: 2010-03-02 02:11:08 +0100 (Tue, 02 Mar 2010) $ </address> </body> </html> diff --git a/docs/tutorial/LangImpl7.html b/docs/tutorial/LangImpl7.html index 1a779ba..3d24739 100644 --- a/docs/tutorial/LangImpl7.html +++ b/docs/tutorial/LangImpl7.html @@ -557,12 +557,12 @@ then: ; preds = %entry else: ; preds = %entry <b>%x3 = load double* %x1</b> - %subtmp = sub double %x3, 1.000000e+00 + %subtmp = fsub double %x3, 1.000000e+00 %calltmp = call double @fib( double %subtmp ) <b>%x4 = load double* %x1</b> - %subtmp5 = sub double %x4, 2.000000e+00 + %subtmp5 = fsub double %x4, 2.000000e+00 %calltmp6 = call double @fib( double %subtmp5 ) - %addtmp = add double %calltmp, %calltmp6 + %addtmp = fadd double %calltmp, %calltmp6 br label %ifcont ifcont: ; preds = %else, %then @@ -595,11 +595,11 @@ then: br label %ifcont else: - %subtmp = sub double <b>%x</b>, 1.000000e+00 + %subtmp = fsub double <b>%x</b>, 1.000000e+00 %calltmp = call double @fib( double %subtmp ) - %subtmp5 = sub double <b>%x</b>, 2.000000e+00 + %subtmp5 = fsub double <b>%x</b>, 2.000000e+00 %calltmp6 = call double @fib( double %subtmp5 ) - %addtmp = add double %calltmp, %calltmp6 + %addtmp = fadd double %calltmp, %calltmp6 br label %ifcont ifcont: ; preds = %else, %then @@ -625,11 +625,11 @@ entry: br i1 %ifcond, label %else, label %ifcont else: - %subtmp = sub double %x, 1.000000e+00 + %subtmp = fsub double %x, 1.000000e+00 %calltmp = call double @fib( double %subtmp ) - %subtmp5 = sub double %x, 2.000000e+00 + %subtmp5 = fsub double %x, 2.000000e+00 %calltmp6 = call double @fib( double %subtmp5 ) - %addtmp = add double %calltmp, %calltmp6 + %addtmp = fadd double %calltmp, %calltmp6 ret double %addtmp ifcont: @@ -2158,7 +2158,7 @@ int main() { <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-11 20:15:20 +0100 (Thu, 11 Feb 2010) $ + Last modified: $Date: 2010-03-02 02:11:08 +0100 (Tue, 02 Mar 2010) $ </address> </body> </html> diff --git a/docs/tutorial/OCamlLangImpl3.html b/docs/tutorial/OCamlLangImpl3.html index b335147..c69f644 100644 --- a/docs/tutorial/OCamlLangImpl3.html +++ b/docs/tutorial/OCamlLangImpl3.html @@ -484,7 +484,7 @@ ready> <b>4+5</b>; Read top-level expression: define double @""() { entry: - %addtmp = add double 4.000000e+00, 5.000000e+00 + %addtmp = fadd double 4.000000e+00, 5.000000e+00 ret double %addtmp } </pre> @@ -503,12 +503,12 @@ ready> <b>def foo(a b) a*a + 2*a*b + b*b;</b> Read function definition: define double @foo(double %a, double %b) { entry: - %multmp = mul double %a, %a - %multmp1 = mul double 2.000000e+00, %a - %multmp2 = mul double %multmp1, %b - %addtmp = add double %multmp, %multmp2 - %multmp3 = mul double %b, %b - %addtmp4 = add double %addtmp, %multmp3 + %multmp = fmul double %a, %a + %multmp1 = fmul double 2.000000e+00, %a + %multmp2 = fmul double %multmp1, %b + %addtmp = fadd double %multmp, %multmp2 + %multmp3 = fmul double %b, %b + %addtmp4 = fadd double %addtmp, %multmp3 ret double %addtmp4 } </pre> @@ -525,7 +525,7 @@ define double @bar(double %a) { entry: %calltmp = call double @foo( double %a, double 4.000000e+00 ) %calltmp1 = call double @bar( double 3.133700e+04 ) - %addtmp = add double %calltmp, %calltmp1 + %addtmp = fadd double %calltmp, %calltmp1 ret double %addtmp } </pre> @@ -561,18 +561,18 @@ ready> <b>^D</b> define double @""() { entry: - %addtmp = add double 4.000000e+00, 5.000000e+00 + %addtmp = fadd double 4.000000e+00, 5.000000e+00 ret double %addtmp } define double @foo(double %a, double %b) { entry: - %multmp = mul double %a, %a - %multmp1 = mul double 2.000000e+00, %a - %multmp2 = mul double %multmp1, %b - %addtmp = add double %multmp, %multmp2 - %multmp3 = mul double %b, %b - %addtmp4 = add double %addtmp, %multmp3 + %multmp = fmul double %a, %a + %multmp1 = fmul double 2.000000e+00, %a + %multmp2 = fmul double %multmp1, %b + %addtmp = fadd double %multmp, %multmp2 + %multmp3 = fmul double %b, %b + %addtmp4 = fadd double %addtmp, %multmp3 ret double %addtmp4 } @@ -580,7 +580,7 @@ define double @bar(double %a) { entry: %calltmp = call double @foo( double %a, double 4.000000e+00 ) %calltmp1 = call double @bar( double 3.133700e+04 ) - %addtmp = add double %calltmp, %calltmp1 + %addtmp = fadd double %calltmp, %calltmp1 ret double %addtmp } @@ -1085,7 +1085,7 @@ main () <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="mailto:idadesub@users.sourceforge.net">Erick Tryzelaar</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-03 18:27:31 +0100 (Wed, 03 Feb 2010) $ + Last modified: $Date: 2010-03-02 02:11:08 +0100 (Tue, 02 Mar 2010) $ </address> </body> </html> diff --git a/docs/tutorial/OCamlLangImpl4.html b/docs/tutorial/OCamlLangImpl4.html index 451ab11..4a72f65 100644 --- a/docs/tutorial/OCamlLangImpl4.html +++ b/docs/tutorial/OCamlLangImpl4.html @@ -72,8 +72,8 @@ ready> <b>def test(x) 1+2+x;</b> Read function definition: define double @test(double %x) { entry: - %addtmp = add double 1.000000e+00, 2.000000e+00 - %addtmp1 = add double %addtmp, %x + %addtmp = fadd double 1.000000e+00, 2.000000e+00 + %addtmp1 = fadd double %addtmp, %x ret double %addtmp1 } </pre> @@ -104,7 +104,7 @@ ready> <b>def test(x) 1+2+x;</b> Read function definition: define double @test(double %x) { entry: - %addtmp = add double 3.000000e+00, %x + %addtmp = fadd double 3.000000e+00, %x ret double %addtmp } </pre> @@ -127,9 +127,9 @@ ready> <b>def test(x) (1+2+x)*(x+(1+2));</b> ready> Read function definition: define double @test(double %x) { entry: - %addtmp = add double 3.000000e+00, %x - %addtmp1 = add double %x, 3.000000e+00 - %multmp = mul double %addtmp, %addtmp1 + %addtmp = fadd double 3.000000e+00, %x + %addtmp1 = fadd double %x, 3.000000e+00 + %multmp = fmul double %addtmp, %addtmp1 ret double %multmp } </pre> @@ -267,8 +267,8 @@ ready> <b>def test(x) (1+2+x)*(x+(1+2));</b> ready> Read function definition: define double @test(double %x) { entry: - %addtmp = add double %x, 3.000000e+00 - %multmp = mul double %addtmp, %addtmp + %addtmp = fadd double %x, 3.000000e+00 + %multmp = fmul double %addtmp, %addtmp ret double %multmp } </pre> @@ -388,8 +388,8 @@ ready> <b>def testfunc(x y) x + y*2; </b> Read function definition: define double @testfunc(double %x, double %y) { entry: - %multmp = mul double %y, 2.000000e+00 - %addtmp = add double %multmp, %x + %multmp = fmul double %y, 2.000000e+00 + %addtmp = fadd double %multmp, %x ret double %addtmp } @@ -436,10 +436,10 @@ Read function definition: define double @foo(double %x) { entry: %calltmp = call double @sin( double %x ) - %multmp = mul double %calltmp, %calltmp + %multmp = fmul double %calltmp, %calltmp %calltmp2 = call double @cos( double %x ) - %multmp4 = mul double %calltmp2, %calltmp2 - %addtmp = add double %multmp, %multmp4 + %multmp4 = fmul double %calltmp2, %calltmp2 + %addtmp = fadd double %multmp, %multmp4 ret double %addtmp } @@ -1032,7 +1032,7 @@ extern double putchard(double X) { <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="mailto:idadesub@users.sourceforge.net">Erick Tryzelaar</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-03 18:27:31 +0100 (Wed, 03 Feb 2010) $ + Last modified: $Date: 2010-03-02 02:11:08 +0100 (Tue, 02 Mar 2010) $ </address> </body> </html> diff --git a/docs/tutorial/OCamlLangImpl5.html b/docs/tutorial/OCamlLangImpl5.html index ae075ec..14f0efd 100644 --- a/docs/tutorial/OCamlLangImpl5.html +++ b/docs/tutorial/OCamlLangImpl5.html @@ -653,7 +653,7 @@ loop: ; preds = %loop, %entry ; body %calltmp = call double @putchard( double 4.200000e+01 ) ; increment - %nextvar = add double %i, 1.000000e+00 + %nextvar = fadd double %i, 1.000000e+00 ; termination test %cmptmp = fcmp ult double %i, %n @@ -1563,7 +1563,7 @@ operators</a> <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="mailto:idadesub@users.sourceforge.net">Erick Tryzelaar</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> - Last modified: $Date: 2010-02-03 18:27:31 +0100 (Wed, 03 Feb 2010) $ + Last modified: $Date: 2010-03-02 02:11:08 +0100 (Tue, 02 Mar 2010) $ </address> </body> </html> diff --git a/docs/tutorial/OCamlLangImpl7.html b/docs/tutorial/OCamlLangImpl7.html index ed1a558..823b4c0 100644 --- a/docs/tutorial/OCamlLangImpl7.html +++ b/docs/tutorial/OCamlLangImpl7.html @@ -581,12 +581,12 @@ then: ; preds = %entry else: ; preds = %entry <b>%x3 = load double* %x1</b> - %subtmp = sub double %x3, 1.000000e+00 + %subtmp = fsub double %x3, 1.000000e+00 %calltmp = call double @fib( double %subtmp ) <b>%x4 = load double* %x1</b> - %subtmp5 = sub double %x4, 2.000000e+00 + %subtmp5 = fsub double %x4, 2.000000e+00 %calltmp6 = call double @fib( double %subtmp5 ) - %addtmp = add double %calltmp, %calltmp6 + %addtmp = fadd double %calltmp, %calltmp6 br label %ifcont ifcont: ; preds = %else, %then @@ -619,11 +619,11 @@ then: br label %ifcont else: - %subtmp = sub double <b>%x</b>, 1.000000e+00 + %subtmp = fsub double <b>%x</b>, 1.000000e+00 %calltmp = call double @fib( double %subtmp ) - %subtmp5 = sub double <b>%x</b>, 2.000000e+00 + %subtmp5 = fsub double <b>%x</b>, 2.000000e+00 %calltmp6 = call double @fib( double %subtmp5 ) - %addtmp = add double %calltmp, %calltmp6 + %addtmp = fadd double %calltmp, %calltmp6 br label %ifcont ifcont: ; preds = %else, %then @@ -649,11 +649,11 @@ entry: br i1 %ifcond, label %else, label %ifcont else: - %subtmp = sub double %x, 1.000000e+00 + %subtmp = fsub double %x, 1.000000e+00 %calltmp = call double @fib( double %subtmp ) - %subtmp5 = sub double %x, 2.000000e+00 + %subtmp5 = fsub double %x, 2.000000e+00 %calltmp6 = call double @fib( double %subtmp5 ) - %addtmp = add double %calltmp, %calltmp6 + %addtmp = fadd double %calltmp, %calltmp6 ret double %addtmp ifcont: @@ -1901,7 +1901,7 @@ extern double printd(double X) { <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br> <a href="mailto:idadesub@users.sourceforge.net">Erick Tryzelaar</a><br> - Last modified: $Date: 2010-02-03 18:27:31 +0100 (Wed, 03 Feb 2010) $ + Last modified: $Date: 2010-03-02 02:11:08 +0100 (Tue, 02 Mar 2010) $ </address> </body> </html> diff --git a/examples/Makefile b/examples/Makefile index fc3a7d4..0737eaf 100644 --- a/examples/Makefile +++ b/examples/Makefile @@ -17,7 +17,12 @@ PARALLEL_DIRS += ParallelJIT endif ifeq ($(LLVM_ON_UNIX),1) -PARALLEL_DIRS += ExceptionDemo + ifeq ($(ARCH),x86) + PARALLEL_DIRS += ExceptionDemo + endif + ifeq ($(ARCH),x86_64) + PARALLEL_DIRS += ExceptionDemo + endif endif include $(LEVEL)/Makefile.common diff --git a/include/llvm-c/BitReader.h b/include/llvm-c/BitReader.h index 59269ce..6db6607 100644 --- a/include/llvm-c/BitReader.h +++ b/include/llvm-c/BitReader.h @@ -36,18 +36,28 @@ LLVMBool LLVMParseBitcodeInContext(LLVMContextRef ContextRef, LLVMMemoryBufferRef MemBuf, LLVMModuleRef *OutModule, char **OutMessage); -/* Reads a module from the specified path, returning via the OutMP parameter - a module provider which performs lazy deserialization. Returns 0 on success. - Optionally returns a human-readable error message via OutMessage. */ -LLVMBool LLVMGetBitcodeModuleProvider(LLVMMemoryBufferRef MemBuf, - LLVMModuleProviderRef *OutMP, - char **OutMessage); +/** Reads a module from the specified path, returning via the OutMP parameter + a module provider which performs lazy deserialization. Returns 0 on success. + Optionally returns a human-readable error message via OutMessage. */ +LLVMBool LLVMGetBitcodeModuleInContext(LLVMContextRef ContextRef, + LLVMMemoryBufferRef MemBuf, + LLVMModuleRef *OutM, + char **OutMessage); + +LLVMBool LLVMGetBitcodeModule(LLVMMemoryBufferRef MemBuf, LLVMModuleRef *OutM, + char **OutMessage); + +/** Deprecated: Use LLVMGetBitcodeModuleInContext instead. */ LLVMBool LLVMGetBitcodeModuleProviderInContext(LLVMContextRef ContextRef, LLVMMemoryBufferRef MemBuf, LLVMModuleProviderRef *OutMP, char **OutMessage); +/** Deprecated: Use LLVMGetBitcodeModule instead. */ +LLVMBool LLVMGetBitcodeModuleProvider(LLVMMemoryBufferRef MemBuf, + LLVMModuleProviderRef *OutMP, + char **OutMessage); #ifdef __cplusplus } diff --git a/include/llvm-c/Core.h b/include/llvm-c/Core.h index 063ab91..457436d 100644 --- a/include/llvm-c/Core.h +++ b/include/llvm-c/Core.h @@ -92,11 +92,8 @@ typedef struct LLVMOpaqueMemoryBuffer *LLVMMemoryBufferRef; /** See the llvm::PassManagerBase class. */ typedef struct LLVMOpaquePassManager *LLVMPassManagerRef; -/** - * Used to iterate through the uses of a Value, allowing access to all Values - * that use this Value. See the llvm::Use and llvm::value_use_iterator classes. - */ -typedef struct LLVMOpaqueUseIterator *LLVMUseIteratorRef; +/** Used to get the users and usees of a Value. See the llvm::Use class. */ +typedef struct LLVMOpaqueUse *LLVMUseRef; typedef enum { LLVMZExtAttribute = 1<<0, @@ -282,13 +279,19 @@ typedef enum { void LLVMDisposeMessage(char *Message); -/*===-- Modules -----------------------------------------------------------===*/ +/*===-- Contexts ----------------------------------------------------------===*/ /* Create and destroy contexts. */ LLVMContextRef LLVMContextCreate(void); LLVMContextRef LLVMGetGlobalContext(void); void LLVMContextDispose(LLVMContextRef C); +unsigned LLVMGetMDKindIDInContext(LLVMContextRef C, const char* Name, + unsigned SLen); +unsigned LLVMGetMDKindID(const char* Name, unsigned SLen); + +/*===-- Modules -----------------------------------------------------------===*/ + /* Create and destroy modules. */ /** See llvm::Module::Module. */ LLVMModuleRef LLVMModuleCreateWithName(const char *ModuleID); @@ -497,6 +500,9 @@ const char *LLVMGetValueName(LLVMValueRef Val); void LLVMSetValueName(LLVMValueRef Val, const char *Name); void LLVMDumpValue(LLVMValueRef Val); void LLVMReplaceAllUsesWith(LLVMValueRef OldVal, LLVMValueRef NewVal); +int LLVMHasMetadata(LLVMValueRef Val); +LLVMValueRef LLVMGetMetadata(LLVMValueRef Val, unsigned KindID); +void LLVMSetMetadata(LLVMValueRef Val, unsigned KindID, LLVMValueRef Node); /* Conversion functions. Return the input value if it is an instance of the specified class, otherwise NULL. See llvm::dyn_cast_or_null<>. */ @@ -505,10 +511,10 @@ void LLVMReplaceAllUsesWith(LLVMValueRef OldVal, LLVMValueRef NewVal); LLVM_FOR_EACH_VALUE_SUBCLASS(LLVM_DECLARE_VALUE_CAST) /* Operations on Uses */ -LLVMUseIteratorRef LLVMGetFirstUse(LLVMValueRef Val); -LLVMUseIteratorRef LLVMGetNextUse(LLVMUseIteratorRef U); -LLVMValueRef LLVMGetUser(LLVMUseIteratorRef U); -LLVMValueRef LLVMGetUsedValue(LLVMUseIteratorRef U); +LLVMUseRef LLVMGetFirstUse(LLVMValueRef Val); +LLVMUseRef LLVMGetNextUse(LLVMUseRef U); +LLVMValueRef LLVMGetUser(LLVMUseRef U); +LLVMValueRef LLVMGetUsedValue(LLVMUseRef U); /* Operations on Users */ LLVMValueRef LLVMGetOperand(LLVMValueRef Val, unsigned Index); @@ -522,6 +528,14 @@ LLVMBool LLVMIsNull(LLVMValueRef Val); LLVMBool LLVMIsUndef(LLVMValueRef Val); LLVMValueRef LLVMConstPointerNull(LLVMTypeRef Ty); +/* Operations on metadata */ +LLVMValueRef LLVMMDStringInContext(LLVMContextRef C, const char *Str, + unsigned SLen); +LLVMValueRef LLVMMDString(const char *Str, unsigned SLen); +LLVMValueRef LLVMMDNodeInContext(LLVMContextRef C, LLVMValueRef *Vals, + unsigned Count); +LLVMValueRef LLVMMDNode(LLVMValueRef *Vals, unsigned Count); + /* Operations on scalar constants */ LLVMValueRef LLVMConstInt(LLVMTypeRef IntTy, unsigned long long N, LLVMBool SignExtend); @@ -551,20 +565,28 @@ LLVMValueRef LLVMConstArray(LLVMTypeRef ElementTy, LLVMValueRef LLVMConstStruct(LLVMValueRef *ConstantVals, unsigned Count, LLVMBool Packed); LLVMValueRef LLVMConstVector(LLVMValueRef *ScalarConstantVals, unsigned Size); +LLVMValueRef LLVMConstUnion(LLVMTypeRef Ty, LLVMValueRef Val); /* Constant expressions */ LLVMOpcode LLVMGetConstOpcode(LLVMValueRef ConstantVal); LLVMValueRef LLVMAlignOf(LLVMTypeRef Ty); LLVMValueRef LLVMSizeOf(LLVMTypeRef Ty); LLVMValueRef LLVMConstNeg(LLVMValueRef ConstantVal); +LLVMValueRef LLVMConstNSWNeg(LLVMValueRef ConstantVal); +LLVMValueRef LLVMConstNUWNeg(LLVMValueRef ConstantVal); LLVMValueRef LLVMConstFNeg(LLVMValueRef ConstantVal); LLVMValueRef LLVMConstNot(LLVMValueRef ConstantVal); LLVMValueRef LLVMConstAdd(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); LLVMValueRef LLVMConstNSWAdd(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstNUWAdd(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); LLVMValueRef LLVMConstFAdd(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); LLVMValueRef LLVMConstSub(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstNSWSub(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstNUWSub(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); LLVMValueRef LLVMConstFSub(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); LLVMValueRef LLVMConstMul(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstNSWMul(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); +LLVMValueRef LLVMConstNUWMul(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); LLVMValueRef LLVMConstFMul(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); LLVMValueRef LLVMConstUDiv(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); LLVMValueRef LLVMConstSDiv(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant); @@ -630,6 +652,7 @@ LLVMValueRef LLVMConstInsertValue(LLVMValueRef AggConstant, LLVMValueRef LLVMConstInlineAsm(LLVMTypeRef Ty, const char *AsmString, const char *Constraints, LLVMBool HasSideEffects, LLVMBool IsAlignStack); +LLVMValueRef LLVMBlockAddress(LLVMValueRef F, LLVMBasicBlockRef BB); /* Operations on global variables, functions, and aliases (globals) */ LLVMModuleRef LLVMGetGlobalParent(LLVMValueRef Global); @@ -645,6 +668,9 @@ void LLVMSetAlignment(LLVMValueRef Global, unsigned Bytes); /* Operations on global variables */ LLVMValueRef LLVMAddGlobal(LLVMModuleRef M, LLVMTypeRef Ty, const char *Name); +LLVMValueRef LLVMAddGlobalInAddressSpace(LLVMModuleRef M, LLVMTypeRef Ty, + const char *Name, + unsigned AddressSpace); LLVMValueRef LLVMGetNamedGlobal(LLVMModuleRef M, const char *Name); LLVMValueRef LLVMGetFirstGlobal(LLVMModuleRef M); LLVMValueRef LLVMGetLastGlobal(LLVMModuleRef M); @@ -765,6 +791,11 @@ void LLVMInsertIntoBuilderWithName(LLVMBuilderRef Builder, LLVMValueRef Instr, const char *Name); void LLVMDisposeBuilder(LLVMBuilderRef Builder); +/* Metadata */ +void LLVMSetCurrentDebugLocation(LLVMBuilderRef Builder, LLVMValueRef L); +LLVMValueRef LLVMGetCurrentDebugLocation(LLVMBuilderRef Builder); +void LLVMSetInstDebugLocation(LLVMBuilderRef Builder, LLVMValueRef Inst); + /* Terminators */ LLVMValueRef LLVMBuildRetVoid(LLVMBuilderRef); LLVMValueRef LLVMBuildRet(LLVMBuilderRef, LLVMValueRef V); @@ -775,6 +806,8 @@ LLVMValueRef LLVMBuildCondBr(LLVMBuilderRef, LLVMValueRef If, LLVMBasicBlockRef Then, LLVMBasicBlockRef Else); LLVMValueRef LLVMBuildSwitch(LLVMBuilderRef, LLVMValueRef V, LLVMBasicBlockRef Else, unsigned NumCases); +LLVMValueRef LLVMBuildIndirectBr(LLVMBuilderRef B, LLVMValueRef Addr, + unsigned NumDests); LLVMValueRef LLVMBuildInvoke(LLVMBuilderRef, LLVMValueRef Fn, LLVMValueRef *Args, unsigned NumArgs, LLVMBasicBlockRef Then, LLVMBasicBlockRef Catch, @@ -786,19 +819,32 @@ LLVMValueRef LLVMBuildUnreachable(LLVMBuilderRef); void LLVMAddCase(LLVMValueRef Switch, LLVMValueRef OnVal, LLVMBasicBlockRef Dest); +/* Add a destination to the indirectbr instruction */ +void LLVMAddDestination(LLVMValueRef IndirectBr, LLVMBasicBlockRef Dest); + /* Arithmetic */ LLVMValueRef LLVMBuildAdd(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name); LLVMValueRef LLVMBuildNSWAdd(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name); +LLVMValueRef LLVMBuildNUWAdd(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name); LLVMValueRef LLVMBuildFAdd(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name); LLVMValueRef LLVMBuildSub(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name); +LLVMValueRef LLVMBuildNSWSub(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name); +LLVMValueRef LLVMBuildNUWSub(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name); LLVMValueRef LLVMBuildFSub(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name); LLVMValueRef LLVMBuildMul(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name); +LLVMValueRef LLVMBuildNSWMul(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name); +LLVMValueRef LLVMBuildNUWMul(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name); LLVMValueRef LLVMBuildFMul(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name); LLVMValueRef LLVMBuildUDiv(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, @@ -827,7 +873,14 @@ LLVMValueRef LLVMBuildOr(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name); LLVMValueRef LLVMBuildXor(LLVMBuilderRef, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name); +LLVMValueRef LLVMBuildBinOp(LLVMBuilderRef B, LLVMOpcode Op, + LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name); LLVMValueRef LLVMBuildNeg(LLVMBuilderRef, LLVMValueRef V, const char *Name); +LLVMValueRef LLVMBuildNSWNeg(LLVMBuilderRef B, LLVMValueRef V, + const char *Name); +LLVMValueRef LLVMBuildNUWNeg(LLVMBuilderRef B, LLVMValueRef V, + const char *Name); LLVMValueRef LLVMBuildFNeg(LLVMBuilderRef, LLVMValueRef V, const char *Name); LLVMValueRef LLVMBuildNot(LLVMBuilderRef, LLVMValueRef V, const char *Name); @@ -886,6 +939,8 @@ LLVMValueRef LLVMBuildSExtOrBitCast(LLVMBuilderRef, LLVMValueRef Val, LLVMTypeRef DestTy, const char *Name); LLVMValueRef LLVMBuildTruncOrBitCast(LLVMBuilderRef, LLVMValueRef Val, LLVMTypeRef DestTy, const char *Name); +LLVMValueRef LLVMBuildCast(LLVMBuilderRef B, LLVMOpcode Op, LLVMValueRef Val, + LLVMTypeRef DestTy, const char *Name); LLVMValueRef LLVMBuildPointerCast(LLVMBuilderRef, LLVMValueRef Val, LLVMTypeRef DestTy, const char *Name); LLVMValueRef LLVMBuildIntCast(LLVMBuilderRef, LLVMValueRef Val, /*Signed cast!*/ @@ -967,6 +1022,9 @@ LLVMPassManagerRef LLVMCreatePassManager(void); provider. It does not take ownership of the module provider. This type of pipeline is suitable for code generation and JIT compilation tasks. See llvm::FunctionPassManager::FunctionPassManager. */ +LLVMPassManagerRef LLVMCreateFunctionPassManagerForModule(LLVMModuleRef M); + +/** Deprecated: Use LLVMCreateFunctionPassManagerForModule instead. */ LLVMPassManagerRef LLVMCreateFunctionPassManager(LLVMModuleProviderRef MP); /** Initializes, executes on the provided module, and finalizes all of the @@ -1038,7 +1096,7 @@ namespace llvm { DEFINE_SIMPLE_CONVERSION_FUNCTIONS(PATypeHolder, LLVMTypeHandleRef ) DEFINE_SIMPLE_CONVERSION_FUNCTIONS(MemoryBuffer, LLVMMemoryBufferRef ) DEFINE_SIMPLE_CONVERSION_FUNCTIONS(LLVMContext, LLVMContextRef ) - DEFINE_SIMPLE_CONVERSION_FUNCTIONS(Use, LLVMUseIteratorRef ) + DEFINE_SIMPLE_CONVERSION_FUNCTIONS(Use, LLVMUseRef ) DEFINE_STDCXX_CONVERSION_FUNCTIONS(PassManagerBase, LLVMPassManagerRef ) /* LLVMModuleProviderRef exists for historical reasons, but now just holds a * Module. diff --git a/include/llvm-c/ExecutionEngine.h b/include/llvm-c/ExecutionEngine.h index 151c935..5a98a77 100644 --- a/include/llvm-c/ExecutionEngine.h +++ b/include/llvm-c/ExecutionEngine.h @@ -55,14 +55,30 @@ void LLVMDisposeGenericValue(LLVMGenericValueRef GenVal); /*===-- Operations on execution engines -----------------------------------===*/ +LLVMBool LLVMCreateExecutionEngineForModule(LLVMExecutionEngineRef *OutEE, + LLVMModuleRef M, + char **OutError); + +LLVMBool LLVMCreateInterpreterForModule(LLVMExecutionEngineRef *OutInterp, + LLVMModuleRef M, + char **OutError); + +LLVMBool LLVMCreateJITCompilerForModule(LLVMExecutionEngineRef *OutJIT, + LLVMModuleRef M, + unsigned OptLevel, + char **OutError); + +/** Deprecated: Use LLVMCreateExecutionEngineForModule instead. */ LLVMBool LLVMCreateExecutionEngine(LLVMExecutionEngineRef *OutEE, LLVMModuleProviderRef MP, char **OutError); +/** Deprecated: Use LLVMCreateInterpreterForModule instead. */ LLVMBool LLVMCreateInterpreter(LLVMExecutionEngineRef *OutInterp, LLVMModuleProviderRef MP, char **OutError); +/** Deprecated: Use LLVMCreateJITCompilerForModule instead. */ LLVMBool LLVMCreateJITCompiler(LLVMExecutionEngineRef *OutJIT, LLVMModuleProviderRef MP, unsigned OptLevel, @@ -84,8 +100,15 @@ LLVMGenericValueRef LLVMRunFunction(LLVMExecutionEngineRef EE, LLVMValueRef F, void LLVMFreeMachineCodeForFunction(LLVMExecutionEngineRef EE, LLVMValueRef F); +void LLVMAddModule(LLVMExecutionEngineRef EE, LLVMModuleRef M); + +/** Deprecated: Use LLVMAddModule instead. */ void LLVMAddModuleProvider(LLVMExecutionEngineRef EE, LLVMModuleProviderRef MP); +LLVMBool LLVMRemoveModule(LLVMExecutionEngineRef EE, LLVMModuleRef M, + LLVMModuleRef *OutMod, char **OutError); + +/** Deprecated: Use LLVMRemoveModule instead. */ LLVMBool LLVMRemoveModuleProvider(LLVMExecutionEngineRef EE, LLVMModuleProviderRef MP, LLVMModuleRef *OutMod, char **OutError); diff --git a/include/llvm/ADT/APFloat.h b/include/llvm/ADT/APFloat.h index f81109a..861b7b9 100644 --- a/include/llvm/ADT/APFloat.h +++ b/include/llvm/ADT/APFloat.h @@ -173,11 +173,16 @@ namespace llvm { fcZero }; + enum uninitializedTag { + uninitialized + }; + // Constructors. APFloat(const fltSemantics &); // Default construct to 0.0 APFloat(const fltSemantics &, const StringRef &); APFloat(const fltSemantics &, integerPart); - APFloat(const fltSemantics &, fltCategory, bool negative, unsigned type=0); + APFloat(const fltSemantics &, fltCategory, bool negative); + APFloat(const fltSemantics &, uninitializedTag); explicit APFloat(double d); explicit APFloat(float f); explicit APFloat(const APInt &, bool isIEEE = false); @@ -199,7 +204,26 @@ namespace llvm { /// default. The value is truncated as necessary. static APFloat getNaN(const fltSemantics &Sem, bool Negative = false, unsigned type = 0) { - return APFloat(Sem, fcNaN, Negative, type); + if (type) { + APInt fill(64, type); + return getQNaN(Sem, Negative, &fill); + } else { + return getQNaN(Sem, Negative, 0); + } + } + + /// getQNan - Factory for QNaN values. + static APFloat getQNaN(const fltSemantics &Sem, + bool Negative = false, + const APInt *payload = 0) { + return makeNaN(Sem, false, Negative, payload); + } + + /// getSNan - Factory for SNaN values. + static APFloat getSNaN(const fltSemantics &Sem, + bool Negative = false, + const APInt *payload = 0) { + return makeNaN(Sem, true, Negative, payload); } /// getLargest - Returns the largest finite number in the given @@ -350,7 +374,9 @@ namespace llvm { opStatus modSpecials(const APFloat &); /* Miscellany. */ - void makeNaN(unsigned = 0); + static APFloat makeNaN(const fltSemantics &Sem, bool SNaN, bool Negative, + const APInt *fill); + void makeNaN(bool SNaN = false, bool Neg = false, const APInt *fill = 0); opStatus normalize(roundingMode, lostFraction); opStatus addOrSubtract(const APFloat &, roundingMode, bool subtract); cmpResult compareAbsoluteValue(const APFloat &) const; diff --git a/include/llvm/ADT/APInt.h b/include/llvm/ADT/APInt.h index 88aa995..3f67ffb 100644 --- a/include/llvm/ADT/APInt.h +++ b/include/llvm/ADT/APInt.h @@ -150,7 +150,17 @@ class APInt { return isSingleWord() ? VAL : pVal[whichWord(bitPosition)]; } + /// Converts a string into a number. The string must be non-empty + /// and well-formed as a number of the given base. The bit-width + /// must be sufficient to hold the result. + /// /// This is used by the constructors that take string arguments. + /// + /// StringRef::getAsInteger is superficially similar but (1) does + /// not assume that the string is well-formed and (2) grows the + /// result to hold the input. + /// + /// @param radix 2, 8, 10, or 16 /// @brief Convert a char array into an APInt void fromString(unsigned numBits, const StringRef &str, uint8_t radix); @@ -571,6 +581,21 @@ public: /// @brief Bitwise OR assignment operator. APInt& operator|=(const APInt& RHS); + /// Performs a bitwise OR operation on this APInt and RHS. RHS is + /// logically zero-extended or truncated to match the bit-width of + /// the LHS. + /// + /// @brief Bitwise OR assignment operator. + APInt& operator|=(uint64_t RHS) { + if (isSingleWord()) { + VAL |= RHS; + clearUnusedBits(); + } else { + pVal[0] |= RHS; + } + return *this; + } + /// Performs a bitwise XOR operation on this APInt and RHS. The result is /// assigned to *this. /// @returns *this after XORing with RHS. @@ -1308,6 +1333,9 @@ public: /// Set the given bit of a bignum. Zero-based. static void tcSetBit(integerPart *, unsigned int bit); + /// Clear the given bit of a bignum. Zero-based. + static void tcClearBit(integerPart *, unsigned int bit); + /// Returns the bit number of the least or most significant set bit /// of a number. If the input number has no bits set -1U is /// returned. diff --git a/include/llvm/ADT/DeltaAlgorithm.h b/include/llvm/ADT/DeltaAlgorithm.h index 1facfa0..45ba198 100644 --- a/include/llvm/ADT/DeltaAlgorithm.h +++ b/include/llvm/ADT/DeltaAlgorithm.h @@ -52,7 +52,7 @@ private: /// \return - The test result. bool GetTestResult(const changeset_ty &Changes); - /// Split - Partition a set of changes \arg Sinto one or two subsets. + /// Split - Partition a set of changes \arg S into one or two subsets. void Split(const changeset_ty &S, changesetlist_ty &Res); /// Delta - Minimize a set of \arg Changes which has been partioned into diff --git a/include/llvm/ADT/ScopedHashTable.h b/include/llvm/ADT/ScopedHashTable.h index d538295..f325e2b 100644 --- a/include/llvm/ADT/ScopedHashTable.h +++ b/include/llvm/ADT/ScopedHashTable.h @@ -36,10 +36,10 @@ namespace llvm { -template <typename K, typename V> +template <typename K, typename V, typename KInfo = DenseMapInfo<K> > class ScopedHashTable; -template <typename K, typename V> +template <typename K, typename V, typename KInfo = DenseMapInfo<K> > class ScopedHashTableVal { ScopedHashTableVal *NextInScope; ScopedHashTableVal *NextForKey; @@ -61,35 +61,39 @@ public: ScopedHashTableVal *getNextInScope() { return NextInScope; } }; -template <typename K, typename V> +template <typename K, typename V, typename KInfo = DenseMapInfo<K> > class ScopedHashTableScope { /// HT - The hashtable that we are active for. - ScopedHashTable<K, V> &HT; + ScopedHashTable<K, V, KInfo> &HT; /// PrevScope - This is the scope that we are shadowing in HT. ScopedHashTableScope *PrevScope; /// LastValInScope - This is the last value that was inserted for this scope /// or null if none have been inserted yet. - ScopedHashTableVal<K,V> *LastValInScope; + ScopedHashTableVal<K, V, KInfo> *LastValInScope; void operator=(ScopedHashTableScope&); // DO NOT IMPLEMENT ScopedHashTableScope(ScopedHashTableScope&); // DO NOT IMPLEMENT public: - ScopedHashTableScope(ScopedHashTable<K, V> &HT); + ScopedHashTableScope(ScopedHashTable<K, V, KInfo> &HT); ~ScopedHashTableScope(); private: - friend class ScopedHashTable<K, V>; - ScopedHashTableVal<K, V> *getLastValInScope() { return LastValInScope; } - void setLastValInScope(ScopedHashTableVal<K,V> *Val) { LastValInScope = Val; } + friend class ScopedHashTable<K, V, KInfo>; + ScopedHashTableVal<K, V, KInfo> *getLastValInScope() { + return LastValInScope; + } + void setLastValInScope(ScopedHashTableVal<K, V, KInfo> *Val) { + LastValInScope = Val; + } }; -template <typename K, typename V> +template <typename K, typename V, typename KInfo = DenseMapInfo<K> > class ScopedHashTableIterator { - ScopedHashTableVal<K,V> *Node; + ScopedHashTableVal<K, V, KInfo> *Node; public: - ScopedHashTableIterator(ScopedHashTableVal<K,V> *node) : Node(node){} + ScopedHashTableIterator(ScopedHashTableVal<K, V, KInfo> *node) : Node(node) {} V &operator*() const { assert(Node && "Dereference end()"); @@ -117,35 +121,43 @@ public: }; -template <typename K, typename V> +template <typename K, typename V, typename KInfo> class ScopedHashTable { - DenseMap<K, ScopedHashTableVal<K,V>*> TopLevelMap; - ScopedHashTableScope<K, V> *CurScope; + DenseMap<K, ScopedHashTableVal<K, V, KInfo>*, KInfo> TopLevelMap; + ScopedHashTableScope<K, V, KInfo> *CurScope; ScopedHashTable(const ScopedHashTable&); // NOT YET IMPLEMENTED void operator=(const ScopedHashTable&); // NOT YET IMPLEMENTED - friend class ScopedHashTableScope<K, V>; + friend class ScopedHashTableScope<K, V, KInfo>; public: ScopedHashTable() : CurScope(0) {} ~ScopedHashTable() { assert(CurScope == 0 && TopLevelMap.empty() && "Scope imbalance!"); } + bool count(const K &Key) const { + return TopLevelMap.count(Key); + } + + V lookup(const K &Key) { + return TopLevelMap[Key].getValue(); + } + void insert(const K &Key, const V &Val) { assert(CurScope && "No scope active!"); - ScopedHashTableVal<K,V> *&KeyEntry = TopLevelMap[Key]; + ScopedHashTableVal<K, V, KInfo> *&KeyEntry = TopLevelMap[Key]; - KeyEntry = new ScopedHashTableVal<K,V>(CurScope->getLastValInScope(), - KeyEntry, Key, Val); + KeyEntry= new ScopedHashTableVal<K, V, KInfo>(CurScope->getLastValInScope(), + KeyEntry, Key, Val); CurScope->setLastValInScope(KeyEntry); } - typedef ScopedHashTableIterator<K, V> iterator; + typedef ScopedHashTableIterator<K, V, KInfo> iterator; iterator end() { return iterator(0); } iterator begin(const K &Key) { - typename DenseMap<K, ScopedHashTableVal<K,V>*>::iterator I = + typename DenseMap<K, ScopedHashTableVal<K, V, KInfo>*, KInfo>::iterator I = TopLevelMap.find(Key); if (I == TopLevelMap.end()) return end(); return iterator(I->second); @@ -154,28 +166,29 @@ public: /// ScopedHashTableScope ctor - Install this as the current scope for the hash /// table. -template <typename K, typename V> -ScopedHashTableScope<K, V>::ScopedHashTableScope(ScopedHashTable<K, V> &ht) - : HT(ht) { +template <typename K, typename V, typename KInfo> +ScopedHashTableScope<K, V, KInfo>:: + ScopedHashTableScope(ScopedHashTable<K, V, KInfo> &ht) : HT(ht) { PrevScope = HT.CurScope; HT.CurScope = this; LastValInScope = 0; } -template <typename K, typename V> -ScopedHashTableScope<K, V>::~ScopedHashTableScope() { +template <typename K, typename V, typename KInfo> +ScopedHashTableScope<K, V, KInfo>::~ScopedHashTableScope() { assert(HT.CurScope == this && "Scope imbalance!"); HT.CurScope = PrevScope; // Pop and delete all values corresponding to this scope. - while (ScopedHashTableVal<K, V> *ThisEntry = LastValInScope) { + while (ScopedHashTableVal<K, V, KInfo> *ThisEntry = LastValInScope) { // Pop this value out of the TopLevelMap. if (ThisEntry->getNextForKey() == 0) { assert(HT.TopLevelMap[ThisEntry->getKey()] == ThisEntry && "Scope imbalance!"); HT.TopLevelMap.erase(ThisEntry->getKey()); } else { - ScopedHashTableVal<K,V> *&KeyEntry = HT.TopLevelMap[ThisEntry->getKey()]; + ScopedHashTableVal<K, V, KInfo> *&KeyEntry = + HT.TopLevelMap[ThisEntry->getKey()]; assert(KeyEntry == ThisEntry && "Scope imbalance!"); KeyEntry = ThisEntry->getNextForKey(); } diff --git a/include/llvm/ADT/StringRef.h b/include/llvm/ADT/StringRef.h index 3064af3..9257770 100644 --- a/include/llvm/ADT/StringRef.h +++ b/include/llvm/ADT/StringRef.h @@ -18,6 +18,7 @@ namespace llvm { template<typename T> class SmallVectorImpl; + class APInt; /// StringRef - Represent a constant reference to a string, i.e. a character /// array and a length, which need not be null terminated. @@ -273,6 +274,19 @@ namespace llvm { // TODO: Provide overloads for int/unsigned that check for overflow. + /// getAsInteger - Parse the current string as an integer of the + /// specified radix, or of an autosensed radix if the radix given + /// is 0. The current value in Result is discarded, and the + /// storage is changed to be wide enough to store the parsed + /// integer. + /// + /// Returns true if the string does not solely consist of a valid + /// non-empty number in the appropriate base. + /// + /// APInt::fromString is superficially similar but assumes the + /// string is well-formed in the given radix. + bool getAsInteger(unsigned Radix, APInt &Result) const; + /// @} /// @name Substring Operations /// @{ @@ -355,7 +369,7 @@ namespace llvm { /// \param A - Where to put the substrings. /// \param Separator - The string to split on. /// \param MaxSplit - The maximum number of times the string is split. - /// \parm KeepEmpty - True if empty substring should be added. + /// \param KeepEmpty - True if empty substring should be added. void split(SmallVectorImpl<StringRef> &A, StringRef Separator, int MaxSplit = -1, bool KeepEmpty = true) const; diff --git a/include/llvm/ADT/Triple.h b/include/llvm/ADT/Triple.h index 8798b0e..be31ea0 100644 --- a/include/llvm/ADT/Triple.h +++ b/include/llvm/ADT/Triple.h @@ -73,6 +73,7 @@ public: x86, // X86: i[3-9]86 x86_64, // X86-64: amd64, x86_64 xcore, // XCore: xcore + mblaze, // MBlaze: mblaze InvalidArch }; diff --git a/include/llvm/Analysis/DebugInfo.h b/include/llvm/Analysis/DebugInfo.h index ccf0105..f105c20 100644 --- a/include/llvm/Analysis/DebugInfo.h +++ b/include/llvm/Analysis/DebugInfo.h @@ -459,9 +459,11 @@ namespace llvm { if (DbgNode && !isLexicalBlock()) DbgNode = 0; } - DIScope getContext() const { return getFieldAs<DIScope>(1); } - StringRef getDirectory() const { return getContext().getDirectory(); } - StringRef getFilename() const { return getContext().getFilename(); } + DIScope getContext() const { return getFieldAs<DIScope>(1); } + StringRef getDirectory() const { return getContext().getDirectory(); } + StringRef getFilename() const { return getContext().getFilename(); } + unsigned getLineNumber() const { return getUnsignedField(2); } + unsigned getColumnNumber() const { return getUnsignedField(3); } }; /// DINameSpace - A wrapper for a C++ style name space. @@ -636,7 +638,8 @@ namespace llvm { /// CreateLexicalBlock - This creates a descriptor for a lexical block /// with the specified parent context. - DILexicalBlock CreateLexicalBlock(DIDescriptor Context); + DILexicalBlock CreateLexicalBlock(DIDescriptor Context, unsigned Line = 0, + unsigned Col = 0); /// CreateNameSpace - This creates new descriptor for a namespace /// with the specified parent context. diff --git a/include/llvm/Analysis/Dominators.h b/include/llvm/Analysis/Dominators.h index 31c19c4..1e94f30 100644 --- a/include/llvm/Analysis/Dominators.h +++ b/include/llvm/Analysis/Dominators.h @@ -52,7 +52,7 @@ protected: Roots(), IsPostDominators(isPostDom) {} public: - /// getRoots - Return the root blocks of the current CFG. This may include + /// getRoots - Return the root blocks of the current CFG. This may include /// multiple blocks if we are computing post dominators. For forward /// dominators, this will always be a single block (the entry node). /// @@ -225,7 +225,7 @@ protected: DenseMap<NodeT*, InfoRec> Info; void reset() { - for (typename DomTreeNodeMapType::iterator I = this->DomTreeNodes.begin(), + for (typename DomTreeNodeMapType::iterator I = this->DomTreeNodes.begin(), E = DomTreeNodes.end(); I != E; ++I) delete I->second; DomTreeNodes.clear(); @@ -248,7 +248,7 @@ protected: for (typename GraphTraits<Inverse<N> >::ChildIteratorType PI = GraphTraits<Inverse<N> >::child_begin(NewBB), PE = GraphTraits<Inverse<N> >::child_end(NewBB); PI != PE; ++PI) - PredBlocks.push_back(*PI); + PredBlocks.push_back(*PI); assert(!PredBlocks.empty() && "No predblocks??"); @@ -310,7 +310,7 @@ public: if (DomTreeNodes.size() != OtherDomTreeNodes.size()) return true; - for (typename DomTreeNodeMapType::const_iterator + for (typename DomTreeNodeMapType::const_iterator I = this->DomTreeNodes.begin(), E = this->DomTreeNodes.end(); I != E; ++I) { NodeT *BB = I->first; @@ -361,7 +361,7 @@ public: return properlyDominates(getNode(A), getNode(B)); } - bool dominatedBySlowTreeWalk(const DomTreeNodeBase<NodeT> *A, + bool dominatedBySlowTreeWalk(const DomTreeNodeBase<NodeT> *A, const DomTreeNodeBase<NodeT> *B) const { const DomTreeNodeBase<NodeT> *IDom; if (A == 0 || B == 0) return false; @@ -374,7 +374,7 @@ public: /// isReachableFromEntry - Return true if A is dominated by the entry /// block of the function containing it. bool isReachableFromEntry(NodeT* A) { - assert (!this->isPostDominator() + assert (!this->isPostDominator() && "This is not implemented for post dominators"); return dominates(&A->getParent()->front(), A); } @@ -384,7 +384,7 @@ public: /// inline bool dominates(const DomTreeNodeBase<NodeT> *A, const DomTreeNodeBase<NodeT> *B) { - if (B == A) + if (B == A) return true; // A node trivially dominates itself. if (A == 0 || B == 0) @@ -412,7 +412,7 @@ public: } inline bool dominates(const NodeT *A, const NodeT *B) { - if (A == B) + if (A == B) return true; // Cast away the const qualifiers here. This is ok since @@ -431,9 +431,9 @@ public: /// for basic block A and B. If there is no such block then return NULL. NodeT *findNearestCommonDominator(NodeT *A, NodeT *B) { - assert (!this->isPostDominator() + assert (!this->isPostDominator() && "This is not implemented for post dominators"); - assert (A->getParent() == B->getParent() + assert (A->getParent() == B->getParent() && "Two blocks are not in same function"); // If either A or B is a entry block then it is nearest common dominator. @@ -478,14 +478,14 @@ public: // the CFG... /// addNewBlock - Add a new node to the dominator tree information. This - /// creates a new node as a child of DomBB dominator node,linking it into + /// creates a new node as a child of DomBB dominator node,linking it into /// the children list of the immediate dominator. DomTreeNodeBase<NodeT> *addNewBlock(NodeT *BB, NodeT *DomBB) { assert(getNode(BB) == 0 && "Block already in dominator tree!"); DomTreeNodeBase<NodeT> *IDomNode = getNode(DomBB); assert(IDomNode && "Not immediate dominator specified for block!"); DFSInfoValid = false; - return DomTreeNodes[BB] = + return DomTreeNodes[BB] = IDomNode->addChild(new DomTreeNodeBase<NodeT>(BB, IDomNode)); } @@ -503,7 +503,7 @@ public: changeImmediateDominator(getNode(BB), getNode(NewBB)); } - /// eraseNode - Removes a node from the dominator tree. Block must not + /// eraseNode - Removes a node from the dominator tree. Block must not /// domiante any other blocks. Removes node from its immediate dominator's /// children list. Deletes dominator node associated with basic block BB. void eraseNode(NodeT *BB) { @@ -708,7 +708,7 @@ public: DominatorTreeBase<BasicBlock>& getBase() { return *DT; } - /// getRoots - Return the root blocks of the current CFG. This may include + /// getRoots - Return the root blocks of the current CFG. This may include /// multiple blocks if we are computing post dominators. For forward /// dominators, this will always be a single block (the entry node). /// @@ -785,7 +785,7 @@ public: } /// addNewBlock - Add a new node to the dominator tree information. This - /// creates a new node as a child of DomBB dominator node,linking it into + /// creates a new node as a child of DomBB dominator node,linking it into /// the children list of the immediate dominator. inline DomTreeNode *addNewBlock(BasicBlock *BB, BasicBlock *DomBB) { return DT->addNewBlock(BB, DomBB); @@ -802,7 +802,7 @@ public: DT->changeImmediateDominator(N, NewIDom); } - /// eraseNode - Removes a node from the dominator tree. Block must not + /// eraseNode - Removes a node from the dominator tree. Block must not /// domiante any other blocks. Removes node from its immediate dominator's /// children list. Deletes dominator node associated with basic block BB. inline void eraseNode(BasicBlock *BB) { @@ -820,7 +820,7 @@ public: } - virtual void releaseMemory() { + virtual void releaseMemory() { DT->releaseMemory(); } @@ -886,10 +886,10 @@ protected: const bool IsPostDominators; public: - DominanceFrontierBase(void *ID, bool isPostDom) + DominanceFrontierBase(void *ID, bool isPostDom) : FunctionPass(ID), IsPostDominators(isPostDom) {} - /// getRoots - Return the root blocks of the current CFG. This may include + /// getRoots - Return the root blocks of the current CFG. This may include /// multiple blocks if we are computing post dominators. For forward /// dominators, this will always be a single block (the entry node). /// @@ -940,7 +940,7 @@ public: bool compareDomSet(DomSetType &DS1, const DomSetType &DS2) const { std::set<BasicBlock *> tmpSet; for (DomSetType::const_iterator I = DS2.begin(), - E = DS2.end(); I != E; ++I) + E = DS2.end(); I != E; ++I) tmpSet.insert(*I); for (DomSetType::const_iterator I = DS1.begin(), @@ -965,14 +965,14 @@ public: bool compare(DominanceFrontierBase &Other) const { DomSetMapType tmpFrontiers; for (DomSetMapType::const_iterator I = Other.begin(), - E = Other.end(); I != E; ++I) + E = Other.end(); I != E; ++I) tmpFrontiers.insert(std::make_pair(I->first, I->second)); for (DomSetMapType::iterator I = tmpFrontiers.begin(), E = tmpFrontiers.end(); I != E; ) { BasicBlock *Node = I->first; const_iterator DFI = find(Node); - if (DFI == end()) + if (DFI == end()) return true; if (compareDomSet(I->second, DFI->second)) @@ -1001,7 +1001,7 @@ public: class DominanceFrontier : public DominanceFrontierBase { public: static char ID; // Pass ID, replacement for typeid - DominanceFrontier() : + DominanceFrontier() : DominanceFrontierBase(&ID, false) {} BasicBlock *getRoot() const { @@ -1033,7 +1033,7 @@ public: /// to reflect this change. void changeImmediateDominator(BasicBlock *BB, BasicBlock *NewBB, DominatorTree *DT) { - // NewBB is now dominating BB. Which means BB's dominance + // NewBB is now dominating BB. Which means BB's dominance // frontier is now part of NewBB's dominance frontier. However, BB // itself is not member of NewBB's dominance frontier. DominanceFrontier::iterator NewDFI = find(NewBB); diff --git a/include/llvm/Analysis/IVUsers.h b/include/llvm/Analysis/IVUsers.h index e6e9c71..dc616ca 100644 --- a/include/llvm/Analysis/IVUsers.h +++ b/include/llvm/Analysis/IVUsers.h @@ -61,7 +61,7 @@ public: Stride = Val; } - /// getOffset - Return the offset to add to a theoeretical induction + /// getOffset - Return the offset to add to a theoretical induction /// variable that starts at zero and counts up by the stride to compute /// the value for the use. This always has the same type as the stride. const SCEV *getOffset() const { return Offset; } @@ -116,7 +116,7 @@ private: bool IsUseOfPostIncrementedValue; /// Deleted - Implementation of CallbackVH virtual function to - /// recieve notification when the User is deleted. + /// receive notification when the User is deleted. virtual void deleted(); }; diff --git a/include/llvm/Analysis/LoopInfo.h b/include/llvm/Analysis/LoopInfo.h index d5e4d51..f792a7f 100644 --- a/include/llvm/Analysis/LoopInfo.h +++ b/include/llvm/Analysis/LoopInfo.h @@ -835,7 +835,7 @@ public: } else if (BlockLoop != Child) { LoopT *SubLoop = BlockLoop; // Reparent all of the blocks which used to belong to BlockLoops - for (unsigned j = 0, e = SubLoop->Blocks.size(); j != e; ++j) + for (unsigned j = 0, f = SubLoop->Blocks.size(); j != f; ++j) ContainingLoops[SubLoop->Blocks[j]] = Child; // There is already a loop which contains this block, that means diff --git a/include/llvm/Analysis/MemoryDependenceAnalysis.h b/include/llvm/Analysis/MemoryDependenceAnalysis.h index f83cc4f..f6aab03 100644 --- a/include/llvm/Analysis/MemoryDependenceAnalysis.h +++ b/include/llvm/Analysis/MemoryDependenceAnalysis.h @@ -249,7 +249,7 @@ namespace llvm { SmallPtrSet<Instruction*, 4> > ReverseDepMapType; ReverseDepMapType ReverseLocalDeps; - // A reverse mapping form dependencies to the non-local dependees. + // A reverse mapping from dependencies to the non-local dependees. ReverseDepMapType ReverseNonLocalDeps; /// Current AA implementation, just a cache. @@ -312,6 +312,11 @@ namespace llvm { /// value and replaces the other value with ptr. This can make Ptr available /// in more places that cached info does not necessarily keep. void invalidateCachedPointerInfo(Value *Ptr); + + /// invalidateCachedPredecessors - Clear the PredIteratorCache info. + /// This needs to be done when the CFG changes, e.g., due to splitting + /// critical edges. + void invalidateCachedPredecessors(); private: MemDepResult getPointerDependencyFrom(Value *Pointer, uint64_t MemSize, diff --git a/include/llvm/Analysis/PHITransAddr.h b/include/llvm/Analysis/PHITransAddr.h index b612316..033efba 100644 --- a/include/llvm/Analysis/PHITransAddr.h +++ b/include/llvm/Analysis/PHITransAddr.h @@ -66,9 +66,11 @@ public: bool IsPotentiallyPHITranslatable() const; /// PHITranslateValue - PHI translate the current address up the CFG from - /// CurBB to Pred, updating our state the reflect any needed changes. This - /// returns true on failure and sets Addr to null. - bool PHITranslateValue(BasicBlock *CurBB, BasicBlock *PredBB); + /// CurBB to Pred, updating our state to reflect any needed changes. If the + /// dominator tree DT is non-null, the translated value must dominate + /// PredBB. This returns true on failure and sets Addr to null. + bool PHITranslateValue(BasicBlock *CurBB, BasicBlock *PredBB, + const DominatorTree *DT); /// PHITranslateWithInsertion - PHI translate this value into the specified /// predecessor block, inserting a computation of the value if it is @@ -88,14 +90,8 @@ public: /// returns false. bool Verify() const; private: - Value *PHITranslateSubExpr(Value *V, BasicBlock *CurBB, BasicBlock *PredBB); - - - /// GetAvailablePHITranslatedSubExpr - Return the value computed by - /// PHITranslateSubExpr if it dominates PredBB, otherwise return null. - Value *GetAvailablePHITranslatedSubExpr(Value *V, - BasicBlock *CurBB, BasicBlock *PredBB, - const DominatorTree &DT) const; + Value *PHITranslateSubExpr(Value *V, BasicBlock *CurBB, BasicBlock *PredBB, + const DominatorTree *DT); /// InsertPHITranslatedSubExpr - Insert a computation of the PHI translated /// version of 'V' for the edge PredBB->CurBB into the end of the PredBB diff --git a/include/llvm/Analysis/Passes.h b/include/llvm/Analysis/Passes.h index 2f39c6a..1a5cbb2 100644 --- a/include/llvm/Analysis/Passes.h +++ b/include/llvm/Analysis/Passes.h @@ -81,13 +81,6 @@ namespace llvm { //===--------------------------------------------------------------------===// // - // createAndersensPass - This pass implements Andersen's interprocedural alias - // analysis. - // - ModulePass *createAndersensPass(); - - //===--------------------------------------------------------------------===// - // // createProfileLoaderPass - This pass loads information from a profile dump // file. // diff --git a/include/llvm/Analysis/ScalarEvolution.h b/include/llvm/Analysis/ScalarEvolution.h index 383ee88..96d29ba 100644 --- a/include/llvm/Analysis/ScalarEvolution.h +++ b/include/llvm/Analysis/ScalarEvolution.h @@ -8,7 +8,7 @@ //===----------------------------------------------------------------------===// // // The ScalarEvolution class is an LLVM pass which can be used to analyze and -// catagorize scalar expressions in loops. It specializes in recognizing +// categorize scalar expressions in loops. It specializes in recognizing // general induction variables, representing them with the abstract and opaque // SCEV class. Given this analysis, trip counts of loops and other important // properties can be obtained. @@ -55,7 +55,7 @@ namespace llvm { protected: /// SubclassData - This field is initialized to zero and may be used in - /// subclasses to store miscelaneous information. + /// subclasses to store miscellaneous information. unsigned short SubclassData; private: @@ -177,7 +177,7 @@ namespace llvm { /// LoopInfo *LI; - /// TD - The target data information for the target we are targetting. + /// TD - The target data information for the target we are targeting. /// TargetData *TD; @@ -194,7 +194,7 @@ namespace llvm { std::map<SCEVCallbackVH, const SCEV *> Scalars; /// BackedgeTakenInfo - Information about the backedge-taken count - /// of a loop. This currently inclues an exact count and a maximum count. + /// of a loop. This currently includes an exact count and a maximum count. /// struct BackedgeTakenInfo { /// Exact - An expression indicating the exact backedge-taken count of @@ -305,7 +305,7 @@ namespace llvm { /// ComputeLoadConstantCompareBackedgeTakenCount - Given an exit condition /// of 'icmp op load X, cst', try to see if we can compute the /// backedge-taken count. - const SCEV * + BackedgeTakenInfo ComputeLoadConstantCompareBackedgeTakenCount(LoadInst *LI, Constant *RHS, const Loop *L, @@ -323,12 +323,12 @@ namespace llvm { /// HowFarToZero - Return the number of times a backedge comparing the /// specified value to zero will execute. If not computable, return /// CouldNotCompute. - const SCEV *HowFarToZero(const SCEV *V, const Loop *L); + BackedgeTakenInfo HowFarToZero(const SCEV *V, const Loop *L); /// HowFarToNonZero - Return the number of times a backedge checking the /// specified value for nonzero will execute. If not computable, return /// CouldNotCompute. - const SCEV *HowFarToNonZero(const SCEV *V, const Loop *L); + BackedgeTakenInfo HowFarToNonZero(const SCEV *V, const Loop *L); /// HowManyLessThans - Return the number of times a backedge containing the /// specified less-than comparison will execute. If not computable, return @@ -353,14 +353,14 @@ namespace llvm { bool Inverse); /// isImpliedCondOperands - Test whether the condition described by Pred, - /// LHS, and RHS is true whenever the condition desribed by Pred, FoundLHS, + /// LHS, and RHS is true whenever the condition described by Pred, FoundLHS, /// and FoundRHS is true. bool isImpliedCondOperands(ICmpInst::Predicate Pred, const SCEV *LHS, const SCEV *RHS, const SCEV *FoundLHS, const SCEV *FoundRHS); /// isImpliedCondOperandsHelper - Test whether the condition described by - /// Pred, LHS, and RHS is true whenever the condition desribed by Pred, + /// Pred, LHS, and RHS is true whenever the condition described by Pred, /// FoundLHS, and FoundRHS is true. bool isImpliedCondOperandsHelper(ICmpInst::Predicate Pred, const SCEV *LHS, const SCEV *RHS, @@ -586,6 +586,11 @@ namespace llvm { /// compute a trip count, or if the loop is deleted. void forgetLoop(const Loop *L); + /// forgetValue - This method should be called by the client when it has + /// changed a value in a way that may effect its value, or which may + /// disconnect it from a def-use chain linking it to a loop. + void forgetValue(Value *V); + /// GetMinTrailingZeros - Determine the minimum number of zero bits that S /// is guaranteed to end in (at every loop iteration). It is, at the same /// time, the minimum number of times S is divisible by 2. For example, diff --git a/include/llvm/Attributes.h b/include/llvm/Attributes.h index 126c290..1296d67 100644 --- a/include/llvm/Attributes.h +++ b/include/llvm/Attributes.h @@ -60,11 +60,11 @@ const Attributes NoImplicitFloat = 1<<23; /// disable implicit floating point const Attributes Naked = 1<<24; ///< Naked function const Attributes InlineHint = 1<<25; ///< source said inlining was ///desirable -const Attributes StackAlignment = 31<<26; ///< Alignment of stack for - ///function (5 bits) stored as log2 - ///of alignment with +1 bias - ///0 means unaligned (different from - ///alignstack(1)) +const Attributes StackAlignment = 7<<26; ///< Alignment of stack for + ///function (3 bits) stored as log2 + ///of alignment with +1 bias + ///0 means unaligned (different from + ///alignstack(1)) /// @brief Attributes that only apply to function parameters. const Attributes ParameterOnly = ByVal | Nest | StructRet | NoCapture; @@ -118,7 +118,7 @@ inline Attributes constructStackAlignmentFromInt(unsigned i) { return 0; assert(isPowerOf2_32(i) && "Alignment must be a power of two."); - assert(i <= 0x40000000 && "Alignment too large."); + assert(i <= 0x100 && "Alignment too large."); return (Log2_32(i)+1) << 26; } diff --git a/include/llvm/CodeGen/AsmPrinter.h b/include/llvm/CodeGen/AsmPrinter.h index 28a1a3e..8ade1bd 100644 --- a/include/llvm/CodeGen/AsmPrinter.h +++ b/include/llvm/CodeGen/AsmPrinter.h @@ -223,7 +223,7 @@ namespace llvm { void EmitFunctionBody(); /// EmitInstruction - Targets should implement this to emit instructions. - virtual void EmitInstruction(const MachineInstr *MI) { + virtual void EmitInstruction(const MachineInstr *) { assert(0 && "EmitInstruction not implemented"); } @@ -356,6 +356,11 @@ namespace llvm { /// printOffset - This is just convenient handler for printing offsets. void printOffset(int64_t Offset) const; + /// isBlockOnlyReachableByFallthough - Return true if the basic block has + /// exactly one predecessor and the control transfer mechanism between + /// the predecessor and this block is a fall-through. + virtual bool isBlockOnlyReachableByFallthrough(const MachineBasicBlock *MBB) const; + private: /// processDebugLoc - Processes the debug information of each machine diff --git a/include/llvm/CodeGen/DAGISelHeader.h b/include/llvm/CodeGen/DAGISelHeader.h deleted file mode 100644 index 7fda6f7..0000000 --- a/include/llvm/CodeGen/DAGISelHeader.h +++ /dev/null @@ -1,414 +0,0 @@ -//==-llvm/CodeGen/DAGISelHeader.h - Common DAG ISel definitions -*- C++ -*-==// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file provides definitions of the common, target-independent methods and -// data, which is used by SelectionDAG-based instruction selectors. -// -// *** NOTE: This file is #included into the middle of the target -// instruction selector class. These functions are really methods. -// This is a little awkward, but it allows this code to be shared -// by all the targets while still being able to call into -// target-specific code without using a virtual function call. -// -//===----------------------------------------------------------------------===// - -#ifndef LLVM_CODEGEN_DAGISEL_HEADER_H -#define LLVM_CODEGEN_DAGISEL_HEADER_H - -/// ISelPosition - Node iterator marking the current position of -/// instruction selection as it procedes through the topologically-sorted -/// node list. -SelectionDAG::allnodes_iterator ISelPosition; - -/// IsChainCompatible - Returns true if Chain is Op or Chain does -/// not reach Op. -static bool IsChainCompatible(SDNode *Chain, SDNode *Op) { - if (Chain->getOpcode() == ISD::EntryToken) - return true; - if (Chain->getOpcode() == ISD::TokenFactor) - return false; - if (Chain->getNumOperands() > 0) { - SDValue C0 = Chain->getOperand(0); - if (C0.getValueType() == MVT::Other) - return C0.getNode() != Op && IsChainCompatible(C0.getNode(), Op); - } - return true; -} - -/// ISelUpdater - helper class to handle updates of the -/// instruciton selection graph. -class VISIBILITY_HIDDEN ISelUpdater : public SelectionDAG::DAGUpdateListener { - SelectionDAG::allnodes_iterator &ISelPosition; -public: - explicit ISelUpdater(SelectionDAG::allnodes_iterator &isp) - : ISelPosition(isp) {} - - /// NodeDeleted - Handle nodes deleted from the graph. If the - /// node being deleted is the current ISelPosition node, update - /// ISelPosition. - /// - virtual void NodeDeleted(SDNode *N, SDNode *E) { - if (ISelPosition == SelectionDAG::allnodes_iterator(N)) - ++ISelPosition; - } - - /// NodeUpdated - Ignore updates for now. - virtual void NodeUpdated(SDNode *N) {} -}; - -/// ReplaceUses - replace all uses of the old node F with the use -/// of the new node T. -DISABLE_INLINE void ReplaceUses(SDValue F, SDValue T) { - ISelUpdater ISU(ISelPosition); - CurDAG->ReplaceAllUsesOfValueWith(F, T, &ISU); -} - -/// ReplaceUses - replace all uses of the old nodes F with the use -/// of the new nodes T. -DISABLE_INLINE void ReplaceUses(const SDValue *F, const SDValue *T, - unsigned Num) { - ISelUpdater ISU(ISelPosition); - CurDAG->ReplaceAllUsesOfValuesWith(F, T, Num, &ISU); -} - -/// ReplaceUses - replace all uses of the old node F with the use -/// of the new node T. -DISABLE_INLINE void ReplaceUses(SDNode *F, SDNode *T) { - ISelUpdater ISU(ISelPosition); - CurDAG->ReplaceAllUsesWith(F, T, &ISU); -} - -/// SelectRoot - Top level entry to DAG instruction selector. -/// Selects instructions starting at the root of the current DAG. -void SelectRoot(SelectionDAG &DAG) { - SelectRootInit(); - - // Create a dummy node (which is not added to allnodes), that adds - // a reference to the root node, preventing it from being deleted, - // and tracking any changes of the root. - HandleSDNode Dummy(CurDAG->getRoot()); - ISelPosition = llvm::next(SelectionDAG::allnodes_iterator(CurDAG->getRoot().getNode())); - - // The AllNodes list is now topological-sorted. Visit the - // nodes by starting at the end of the list (the root of the - // graph) and preceding back toward the beginning (the entry - // node). - while (ISelPosition != CurDAG->allnodes_begin()) { - SDNode *Node = --ISelPosition; - // Skip dead nodes. DAGCombiner is expected to eliminate all dead nodes, - // but there are currently some corner cases that it misses. Also, this - // makes it theoretically possible to disable the DAGCombiner. - if (Node->use_empty()) - continue; -#if 0 - DAG.setSubgraphColor(Node, "red"); -#endif - SDNode *ResNode = Select(Node); - // If node should not be replaced, continue with the next one. - if (ResNode == Node) - continue; - // Replace node. - if (ResNode) { -#if 0 - DAG.setSubgraphColor(ResNode, "yellow"); - DAG.setSubgraphColor(ResNode, "black"); -#endif - ReplaceUses(Node, ResNode); - } - // If after the replacement this node is not used any more, - // remove this dead node. - if (Node->use_empty()) { // Don't delete EntryToken, etc. - ISelUpdater ISU(ISelPosition); - CurDAG->RemoveDeadNode(Node, &ISU); - } - } - - CurDAG->setRoot(Dummy.getValue()); -} - - -/// CheckInteger - Return true if the specified node is not a ConstantSDNode or -/// if it doesn't have the specified value. -static bool CheckInteger(SDValue V, int64_t Val) { - ConstantSDNode *C = dyn_cast<ConstantSDNode>(V); - return C == 0 || C->getSExtValue() != Val; -} - -/// CheckAndImmediate - Check to see if the specified node is an and with an -/// immediate returning true on failure. -/// -/// FIXME: Inline this gunk into CheckAndMask. -bool CheckAndImmediate(SDValue V, int64_t Val) { - if (V->getOpcode() == ISD::AND) - if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(V->getOperand(1))) - if (CheckAndMask(V.getOperand(0), C, Val)) - return false; - return true; -} - -/// CheckOrImmediate - Check to see if the specified node is an or with an -/// immediate returning true on failure. -/// -/// FIXME: Inline this gunk into CheckOrMask. -bool CheckOrImmediate(SDValue V, int64_t Val) { - if (V->getOpcode() == ISD::OR) - if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(V->getOperand(1))) - if (CheckOrMask(V.getOperand(0), C, Val)) - return false; - return true; -} - -// These functions are marked always inline so that Idx doesn't get pinned to -// the stack. -ALWAYS_INLINE static int8_t -GetInt1(const unsigned char *MatcherTable, unsigned &Idx) { - return MatcherTable[Idx++]; -} - -ALWAYS_INLINE static int16_t -GetInt2(const unsigned char *MatcherTable, unsigned &Idx) { - int16_t Val = GetInt1(MatcherTable, Idx); - Val |= int16_t(GetInt1(MatcherTable, Idx)) << 8; - return Val; -} - -ALWAYS_INLINE static int32_t -GetInt4(const unsigned char *MatcherTable, unsigned &Idx) { - int32_t Val = GetInt2(MatcherTable, Idx); - Val |= int32_t(GetInt2(MatcherTable, Idx)) << 16; - return Val; -} - -ALWAYS_INLINE static int64_t -GetInt8(const unsigned char *MatcherTable, unsigned &Idx) { - int64_t Val = GetInt4(MatcherTable, Idx); - Val |= int64_t(GetInt4(MatcherTable, Idx)) << 32; - return Val; -} - -enum BuiltinOpcodes { - OPC_Emit, - OPC_Push, - OPC_Record, - OPC_MoveChild, - OPC_MoveParent, - OPC_CheckSame, - OPC_CheckPatternPredicate, - OPC_CheckPredicate, - OPC_CheckOpcode, - OPC_CheckType, - OPC_CheckInteger1, OPC_CheckInteger2, OPC_CheckInteger4, OPC_CheckInteger8, - OPC_CheckCondCode, - OPC_CheckValueType, - OPC_CheckComplexPat, - OPC_CheckAndImm1, OPC_CheckAndImm2, OPC_CheckAndImm4, OPC_CheckAndImm8, - OPC_CheckOrImm1, OPC_CheckOrImm2, OPC_CheckOrImm4, OPC_CheckOrImm8, - OPC_IsProfitableToFold, - OPC_IsLegalToFold -}; - -struct MatchScope { - /// FailIndex - If this match fails, this is the index to continue with. - unsigned FailIndex; - - /// NodeStackSize - The size of the node stack when the scope was formed. - unsigned NodeStackSize; - - /// NumRecordedNodes - The number of recorded nodes when the scope was formed. - unsigned NumRecordedNodes; -}; - -SDNode *SelectCodeCommon(SDNode *NodeToMatch, const unsigned char *MatcherTable, - unsigned TableSize) { - switch (NodeToMatch->getOpcode()) { - default: - break; - case ISD::EntryToken: // These nodes remain the same. - case ISD::BasicBlock: - case ISD::Register: - case ISD::HANDLENODE: - case ISD::TargetConstant: - case ISD::TargetConstantFP: - case ISD::TargetConstantPool: - case ISD::TargetFrameIndex: - case ISD::TargetExternalSymbol: - case ISD::TargetBlockAddress: - case ISD::TargetJumpTable: - case ISD::TargetGlobalTLSAddress: - case ISD::TargetGlobalAddress: - case ISD::TokenFactor: - case ISD::CopyFromReg: - case ISD::CopyToReg: - return 0; - case ISD::AssertSext: - case ISD::AssertZext: - ReplaceUses(SDValue(NodeToMatch, 0), NodeToMatch->getOperand(0)); - return 0; - case ISD::INLINEASM: return Select_INLINEASM(NodeToMatch); - case ISD::EH_LABEL: return Select_EH_LABEL(NodeToMatch); - case ISD::UNDEF: return Select_UNDEF(NodeToMatch); - } - - assert(!NodeToMatch->isMachineOpcode() && "Node already selected!"); - - SmallVector<MatchScope, 8> MatchScopes; - - // RecordedNodes - This is the set of nodes that have been recorded by the - // state machine. - SmallVector<SDValue, 8> RecordedNodes; - - // Set up the node stack with NodeToMatch as the only node on the stack. - SmallVector<SDValue, 8> NodeStack; - SDValue N = SDValue(NodeToMatch, 0); - NodeStack.push_back(N); - - // Interpreter starts at opcode #0. - unsigned MatcherIndex = 0; - while (1) { - assert(MatcherIndex < TableSize && "Invalid index"); - switch ((BuiltinOpcodes)MatcherTable[MatcherIndex++]) { - case OPC_Emit: { - errs() << "EMIT NODE\n"; - return 0; - } - case OPC_Push: { - unsigned NumToSkip = MatcherTable[MatcherIndex++]; - MatchScope NewEntry; - NewEntry.FailIndex = MatcherIndex+NumToSkip; - NewEntry.NodeStackSize = NodeStack.size(); - NewEntry.NumRecordedNodes = RecordedNodes.size(); - MatchScopes.push_back(NewEntry); - continue; - } - case OPC_Record: - // Remember this node, it may end up being an operand in the pattern. - RecordedNodes.push_back(N); - continue; - - case OPC_MoveChild: { - unsigned Child = MatcherTable[MatcherIndex++]; - if (Child >= N.getNumOperands()) - break; // Match fails if out of range child #. - N = N.getOperand(Child); - NodeStack.push_back(N); - continue; - } - - case OPC_MoveParent: - // Pop the current node off the NodeStack. - NodeStack.pop_back(); - assert(!NodeStack.empty() && "Node stack imbalance!"); - N = NodeStack.back(); - continue; - - case OPC_CheckSame: { - // Accept if it is exactly the same as a previously recorded node. - unsigned RecNo = MatcherTable[MatcherIndex++]; - assert(RecNo < RecordedNodes.size() && "Invalid CheckSame"); - if (N != RecordedNodes[RecNo]) break; - continue; - } - case OPC_CheckPatternPredicate: - if (!CheckPatternPredicate(MatcherTable[MatcherIndex++])) break; - continue; - case OPC_CheckPredicate: - if (!CheckNodePredicate(N.getNode(), MatcherTable[MatcherIndex++])) break; - continue; - case OPC_CheckComplexPat: { - unsigned PatNo = MatcherTable[MatcherIndex++]; - (void)PatNo; - // FIXME: CHECK IT. - continue; - } - - case OPC_CheckOpcode: - if (N->getOpcode() != MatcherTable[MatcherIndex++]) break; - continue; - case OPC_CheckType: - if (N.getValueType() != - (MVT::SimpleValueType)MatcherTable[MatcherIndex++]) break; - continue; - case OPC_CheckCondCode: - if (cast<CondCodeSDNode>(N)->get() != - (ISD::CondCode)MatcherTable[MatcherIndex++]) break; - continue; - case OPC_CheckValueType: - if (cast<VTSDNode>(N)->getVT() != - (MVT::SimpleValueType)MatcherTable[MatcherIndex++]) break; - continue; - - case OPC_CheckInteger1: - if (CheckInteger(N, GetInt1(MatcherTable, MatcherIndex))) break; - continue; - case OPC_CheckInteger2: - if (CheckInteger(N, GetInt2(MatcherTable, MatcherIndex))) break; - continue; - case OPC_CheckInteger4: - if (CheckInteger(N, GetInt4(MatcherTable, MatcherIndex))) break; - continue; - case OPC_CheckInteger8: - if (CheckInteger(N, GetInt8(MatcherTable, MatcherIndex))) break; - continue; - - case OPC_CheckAndImm1: - if (CheckAndImmediate(N, GetInt1(MatcherTable, MatcherIndex))) break; - continue; - case OPC_CheckAndImm2: - if (CheckAndImmediate(N, GetInt2(MatcherTable, MatcherIndex))) break; - continue; - case OPC_CheckAndImm4: - if (CheckAndImmediate(N, GetInt4(MatcherTable, MatcherIndex))) break; - continue; - case OPC_CheckAndImm8: - if (CheckAndImmediate(N, GetInt8(MatcherTable, MatcherIndex))) break; - continue; - - case OPC_CheckOrImm1: - if (CheckOrImmediate(N, GetInt1(MatcherTable, MatcherIndex))) break; - continue; - case OPC_CheckOrImm2: - if (CheckOrImmediate(N, GetInt2(MatcherTable, MatcherIndex))) break; - continue; - case OPC_CheckOrImm4: - if (CheckOrImmediate(N, GetInt4(MatcherTable, MatcherIndex))) break; - continue; - case OPC_CheckOrImm8: - if (CheckOrImmediate(N, GetInt8(MatcherTable, MatcherIndex))) break; - continue; - - case OPC_IsProfitableToFold: - assert(!NodeStack.size() == 1 && "No parent node"); - if (!IsProfitableToFold(N, NodeStack[NodeStack.size()-2].getNode(), - NodeToMatch)) - break; - continue; - case OPC_IsLegalToFold: - assert(!NodeStack.size() == 1 && "No parent node"); - if (!IsLegalToFold(N, NodeStack[NodeStack.size()-2].getNode(), - NodeToMatch)) - break; - continue; - } - - // If the code reached this point, then the match failed pop out to the next - // match scope. - if (MatchScopes.empty()) { - CannotYetSelect(NodeToMatch); - return 0; - } - - RecordedNodes.resize(MatchScopes.back().NumRecordedNodes); - NodeStack.resize(MatchScopes.back().NodeStackSize); - MatcherIndex = MatchScopes.back().FailIndex; - MatchScopes.pop_back(); - } -} - - -#endif /* LLVM_CODEGEN_DAGISEL_HEADER_H */ diff --git a/include/llvm/CodeGen/LiveInterval.h b/include/llvm/CodeGen/LiveInterval.h index 512c94d..eb5901c 100644 --- a/include/llvm/CodeGen/LiveInterval.h +++ b/include/llvm/CodeGen/LiveInterval.h @@ -569,6 +569,16 @@ namespace llvm { /// unsigned getSize() const; + /// isSpillable - Can this interval be spilled? + bool isSpillable() const { + return weight != HUGE_VALF; + } + + /// markNotSpillable - Mark interval as not spillable + void markNotSpillable() { + weight = HUGE_VALF; + } + /// ComputeJoinedWeight - Set the weight of a live interval after /// Other has been merged into it. void ComputeJoinedWeight(const LiveInterval &Other); diff --git a/include/llvm/CodeGen/LiveIntervalAnalysis.h b/include/llvm/CodeGen/LiveIntervalAnalysis.h index d7ff8da..e8856ac 100644 --- a/include/llvm/CodeGen/LiveIntervalAnalysis.h +++ b/include/llvm/CodeGen/LiveIntervalAnalysis.h @@ -70,8 +70,15 @@ namespace llvm { static char ID; // Pass identification, replacement for typeid LiveIntervals() : MachineFunctionPass(&ID) {} - static float getSpillWeight(bool isDef, bool isUse, unsigned loopDepth) { - return (isDef + isUse) * powf(10.0F, (float)loopDepth); + // Calculate the spill weight to assign to a single instruction. + static float getSpillWeight(bool isDef, bool isUse, unsigned loopDepth); + + // After summing the spill weights of all defs and uses, the final weight + // should be normalized, dividing the weight of the interval by its size. + // This encourages spilling of intervals that are large and have few uses, + // and discourages spilling of small intervals with many uses. + void normalizeSpillWeight(LiveInterval &li) { + li.weight /= getApproximateInstructionCount(li) + 25; } typedef Reg2IntervalMap::iterator iterator; @@ -409,6 +416,9 @@ namespace llvm { DenseMap<unsigned,unsigned> &MBBVRegsMap, std::vector<LiveInterval*> &NewLIs); + // Normalize the spill weight of all the intervals in NewLIs. + void normalizeSpillWeights(std::vector<LiveInterval*> &NewLIs); + static LiveInterval* createInterval(unsigned Reg); void printInstrs(raw_ostream &O) const; diff --git a/include/llvm/CodeGen/LiveVariables.h b/include/llvm/CodeGen/LiveVariables.h index a7bf600..fc5ea6f 100644 --- a/include/llvm/CodeGen/LiveVariables.h +++ b/include/llvm/CodeGen/LiveVariables.h @@ -124,6 +124,11 @@ private: /// std::vector<VarInfo> VirtRegInfo; + /// PHIJoins - list of virtual registers that are PHI joins. These registers + /// may have multiple definitions, and they require special handling when + /// building live intervals. + SparseBitVector<> PHIJoins; + /// ReservedRegisters - This vector keeps track of which registers /// are reserved register which are not allocatable by the target machine. /// We can not track liveness for values that are in this set. @@ -295,6 +300,12 @@ public: void addNewBlock(MachineBasicBlock *BB, MachineBasicBlock *DomBB, MachineBasicBlock *SuccBB); + + /// isPHIJoin - Return true if Reg is a phi join register. + bool isPHIJoin(unsigned Reg) { return PHIJoins.test(Reg); } + + /// setPHIJoin - Mark Reg as a phi join register. + void setPHIJoin(unsigned Reg) { PHIJoins.set(Reg); } }; } // End llvm namespace diff --git a/include/llvm/CodeGen/MachineBasicBlock.h b/include/llvm/CodeGen/MachineBasicBlock.h index db82ba5..d92650b4 100644 --- a/include/llvm/CodeGen/MachineBasicBlock.h +++ b/include/llvm/CodeGen/MachineBasicBlock.h @@ -285,11 +285,6 @@ public: /// it returns end() iterator getFirstTerminator(); - /// isOnlyReachableViaFallthough - Return true if this basic block has - /// exactly one predecessor and the control transfer mechanism between - /// the predecessor and this block is a fall-through. - bool isOnlyReachableByFallthrough() const; - void pop_front() { Insts.pop_front(); } void pop_back() { Insts.pop_back(); } void push_back(MachineInstr *MI) { Insts.push_back(MI); } diff --git a/include/llvm/CodeGen/MachineInstr.h b/include/llvm/CodeGen/MachineInstr.h index 6e33fb3..66be26c 100644 --- a/include/llvm/CodeGen/MachineInstr.h +++ b/include/llvm/CodeGen/MachineInstr.h @@ -179,17 +179,16 @@ public: return MemRefsEnd - MemRefs == 1; } + enum MICheckType { + CheckDefs, // Check all operands for equality + IgnoreDefs, // Ignore all definitions + IgnoreVRegDefs // Ignore virtual register definitions + }; + /// isIdenticalTo - Return true if this instruction is identical to (same /// opcode and same operands as) the specified instruction. - bool isIdenticalTo(const MachineInstr *Other) const { - if (Other->getOpcode() != getOpcode() || - Other->getNumOperands() != getNumOperands()) - return false; - for (unsigned i = 0, e = getNumOperands(); i != e; ++i) - if (!getOperand(i).isIdenticalTo(Other->getOperand(i))) - return false; - return true; - } + bool isIdenticalTo(const MachineInstr *Other, + MICheckType Check = CheckDefs) const; /// removeFromParent - This method unlinks 'this' from the containing basic /// block, and returns it, but does not delete it. @@ -331,13 +330,13 @@ public: /// isSafeToMove - Return true if it is safe to move this instruction. If /// SawStore is set to true, it means that there is a store (or call) between /// the instruction's location and its intended destination. - bool isSafeToMove(const TargetInstrInfo *TII, bool &SawStore, - AliasAnalysis *AA) const; + bool isSafeToMove(const TargetInstrInfo *TII, AliasAnalysis *AA, + bool &SawStore) const; /// isSafeToReMat - Return true if it's safe to rematerialize the specified /// instruction which defined the specified register instead of copying it. - bool isSafeToReMat(const TargetInstrInfo *TII, unsigned DstReg, - AliasAnalysis *AA) const; + bool isSafeToReMat(const TargetInstrInfo *TII, AliasAnalysis *AA, + unsigned DstReg) const; /// hasVolatileMemoryRef - Return true if this instruction may have a /// volatile memory reference, or if the information describing the diff --git a/include/llvm/CodeGen/MachineInstrBuilder.h b/include/llvm/CodeGen/MachineInstrBuilder.h index a263a97..47f7cf7 100644 --- a/include/llvm/CodeGen/MachineInstrBuilder.h +++ b/include/llvm/CodeGen/MachineInstrBuilder.h @@ -127,7 +127,7 @@ public: return *this; } - const MachineInstrBuilder &addMetadata(MDNode *MD) const { + const MachineInstrBuilder &addMetadata(const MDNode *MD) const { MI->addOperand(MachineOperand::CreateMetadata(MD)); return *this; } diff --git a/include/llvm/CodeGen/MachineOperand.h b/include/llvm/CodeGen/MachineOperand.h index dac0092..0978057 100644 --- a/include/llvm/CodeGen/MachineOperand.h +++ b/include/llvm/CodeGen/MachineOperand.h @@ -19,14 +19,14 @@ namespace llvm { -class ConstantFP; class BlockAddress; -class MachineBasicBlock; +class ConstantFP; class GlobalValue; +class MachineBasicBlock; class MachineInstr; -class TargetMachine; class MachineRegisterInfo; class MDNode; +class TargetMachine; class raw_ostream; /// MachineOperand class - Representation of each machine instruction operand. @@ -100,7 +100,7 @@ private: MachineBasicBlock *MBB; // For MO_MachineBasicBlock. const ConstantFP *CFP; // For MO_FPImmediate. int64_t ImmVal; // For MO_Immediate. - MDNode *MD; // For MO_Metadata. + const MDNode *MD; // For MO_Metadata. struct { // For MO_Register. unsigned RegNo; @@ -467,7 +467,7 @@ public: Op.setTargetFlags(TargetFlags); return Op; } - static MachineOperand CreateMetadata(MDNode *Meta) { + static MachineOperand CreateMetadata(const MDNode *Meta) { MachineOperand Op(MachineOperand::MO_Metadata); Op.Contents.MD = Meta; return Op; diff --git a/include/llvm/CodeGen/Passes.h b/include/llvm/CodeGen/Passes.h index dbc73cb..911be22 100644 --- a/include/llvm/CodeGen/Passes.h +++ b/include/llvm/CodeGen/Passes.h @@ -162,6 +162,10 @@ namespace llvm { /// FunctionPass *createGCInfoPrinter(raw_ostream &OS); + /// createMachineCSEPass - This pass performs global CSE on machine + /// instructions. + FunctionPass *createMachineCSEPass(); + /// createMachineLICMPass - This pass performs LICM on machine instructions. /// FunctionPass *createMachineLICMPass(); @@ -187,7 +191,7 @@ namespace llvm { /// createMachineVerifierPass - This pass verifies cenerated machine code /// instructions for correctness. /// - /// @param allowPhysDoubleDefs ignore double definitions of + /// @param allowDoubleDefs ignore double definitions of /// registers. Useful before LiveVariables has run. FunctionPass *createMachineVerifierPass(bool allowDoubleDefs); diff --git a/include/llvm/CodeGen/SelectionDAG.h b/include/llvm/CodeGen/SelectionDAG.h index e628603..ad01e89 100644 --- a/include/llvm/CodeGen/SelectionDAG.h +++ b/include/llvm/CodeGen/SelectionDAG.h @@ -668,27 +668,8 @@ public: SDNode *SelectNodeTo(SDNode *N, unsigned TargetOpc, SDVTList VTs, const SDValue *Ops, unsigned NumOps); - /// MorphNodeTo - These *mutate* the specified node to have the specified + /// MorphNodeTo - This *mutates* the specified node to have the specified /// return type, opcode, and operands. - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT, SDValue Op1); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT, - SDValue Op1, SDValue Op2); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT, - SDValue Op1, SDValue Op2, SDValue Op3); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT, - const SDValue *Ops, unsigned NumOps); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT1, EVT VT2); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT1, - EVT VT2, const SDValue *Ops, unsigned NumOps); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT1, - EVT VT2, EVT VT3, const SDValue *Ops, unsigned NumOps); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT1, - EVT VT2, SDValue Op1); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT1, - EVT VT2, SDValue Op1, SDValue Op2); - SDNode *MorphNodeTo(SDNode *N, unsigned Opc, EVT VT1, - EVT VT2, SDValue Op1, SDValue Op2, SDValue Op3); SDNode *MorphNodeTo(SDNode *N, unsigned Opc, SDVTList VTs, const SDValue *Ops, unsigned NumOps); @@ -898,6 +879,15 @@ public: /// isKnownNeverNan - Test whether the given SDValue is known to never be NaN. bool isKnownNeverNaN(SDValue Op) const; + /// isKnownNeverZero - Test whether the given SDValue is known to never be + /// positive or negative Zero. + bool isKnownNeverZero(SDValue Op) const; + + /// isEqualTo - Test whether two SDValues are known to compare equal. This + /// is true if they are the same value, or if one is negative zero and the + /// other positive zero. + bool isEqualTo(SDValue A, SDValue B) const; + /// isVerifiedDebugInfoDesc - Returns true if the specified SDValue has /// been verified as a debug information descriptor. bool isVerifiedDebugInfoDesc(SDValue Op) const; diff --git a/include/llvm/CodeGen/SelectionDAGISel.h b/include/llvm/CodeGen/SelectionDAGISel.h index 0be91b4..23ff001 100644 --- a/include/llvm/CodeGen/SelectionDAGISel.h +++ b/include/llvm/CodeGen/SelectionDAGISel.h @@ -68,12 +68,18 @@ public: unsigned MakeReg(EVT VT); virtual void EmitFunctionEntryCode(Function &Fn, MachineFunction &MF) {} - virtual void InstructionSelect() = 0; - void SelectRootInit() { - DAGSize = CurDAG->AssignTopologicalOrder(); - } - + /// PreprocessISelDAG - This hook allows targets to hack on the graph before + /// instruction selection starts. + virtual void PreprocessISelDAG() {} + + /// PostprocessISelDAG() - This hook allows the target to hack on the graph + /// right after selection. + virtual void PostprocessISelDAG() {} + + /// Select - Main hook targets implement to select a node. + virtual SDNode *Select(SDNode *N) = 0; + /// SelectInlineAsmMemoryOperand - Select the specified address as a target /// addressing mode, according to the specified constraint code. If this does /// not match or is not implemented, return true. The resultant operands @@ -91,21 +97,138 @@ public: /// IsLegalToFold - Returns true if the specific operand node N of /// U can be folded during instruction selection that starts at Root. - virtual bool IsLegalToFold(SDValue N, SDNode *U, SDNode *Root) const; + bool IsLegalToFold(SDValue N, SDNode *U, SDNode *Root, + bool IgnoreChains = false) const; /// CreateTargetHazardRecognizer - Return a newly allocated hazard recognizer /// to use for this target when scheduling the DAG. virtual ScheduleHazardRecognizer *CreateTargetHazardRecognizer(); + + // Opcodes used by the DAG state machine: + enum BuiltinOpcodes { + OPC_Scope, + OPC_RecordNode, + OPC_RecordChild0, OPC_RecordChild1, OPC_RecordChild2, OPC_RecordChild3, + OPC_RecordChild4, OPC_RecordChild5, OPC_RecordChild6, OPC_RecordChild7, + OPC_RecordMemRef, + OPC_CaptureFlagInput, + OPC_MoveChild, + OPC_MoveParent, + OPC_CheckSame, + OPC_CheckPatternPredicate, + OPC_CheckPredicate, + OPC_CheckOpcode, + OPC_SwitchOpcode, + OPC_CheckType, + OPC_SwitchType, + OPC_CheckChild0Type, OPC_CheckChild1Type, OPC_CheckChild2Type, + OPC_CheckChild3Type, OPC_CheckChild4Type, OPC_CheckChild5Type, + OPC_CheckChild6Type, OPC_CheckChild7Type, + OPC_CheckInteger, + OPC_CheckCondCode, + OPC_CheckValueType, + OPC_CheckComplexPat, + OPC_CheckAndImm, OPC_CheckOrImm, + OPC_CheckFoldableChainNode, + + OPC_EmitInteger, + OPC_EmitRegister, + OPC_EmitConvertToTarget, + OPC_EmitMergeInputChains, + OPC_EmitCopyToReg, + OPC_EmitNodeXForm, + OPC_EmitNode, + OPC_MorphNodeTo, + OPC_MarkFlagResults, + OPC_CompleteMatch + }; + + enum { + OPFL_None = 0, // Node has no chain or flag input and isn't variadic. + OPFL_Chain = 1, // Node has a chain input. + OPFL_FlagInput = 2, // Node has a flag input. + OPFL_FlagOutput = 4, // Node has a flag output. + OPFL_MemRefs = 8, // Node gets accumulated MemRefs. + OPFL_Variadic0 = 1<<4, // Node is variadic, root has 0 fixed inputs. + OPFL_Variadic1 = 2<<4, // Node is variadic, root has 1 fixed inputs. + OPFL_Variadic2 = 3<<4, // Node is variadic, root has 2 fixed inputs. + OPFL_Variadic3 = 4<<4, // Node is variadic, root has 3 fixed inputs. + OPFL_Variadic4 = 5<<4, // Node is variadic, root has 4 fixed inputs. + OPFL_Variadic5 = 6<<4, // Node is variadic, root has 5 fixed inputs. + OPFL_Variadic6 = 7<<4, // Node is variadic, root has 6 fixed inputs. + + OPFL_VariadicInfo = OPFL_Variadic6 + }; + + /// getNumFixedFromVariadicInfo - Transform an EmitNode flags word into the + /// number of fixed arity values that should be skipped when copying from the + /// root. + static inline int getNumFixedFromVariadicInfo(unsigned Flags) { + return ((Flags&OPFL_VariadicInfo) >> 4)-1; + } + + protected: /// DAGSize - Size of DAG being instruction selected. /// unsigned DAGSize; + + /// ISelPosition - Node iterator marking the current position of + /// instruction selection as it procedes through the topologically-sorted + /// node list. + SelectionDAG::allnodes_iterator ISelPosition; + + + /// ISelUpdater - helper class to handle updates of the + /// instruction selection graph. + class ISelUpdater : public SelectionDAG::DAGUpdateListener { + SelectionDAG::allnodes_iterator &ISelPosition; + public: + explicit ISelUpdater(SelectionDAG::allnodes_iterator &isp) + : ISelPosition(isp) {} + + /// NodeDeleted - Handle nodes deleted from the graph. If the + /// node being deleted is the current ISelPosition node, update + /// ISelPosition. + /// + virtual void NodeDeleted(SDNode *N, SDNode *E) { + if (ISelPosition == SelectionDAG::allnodes_iterator(N)) + ++ISelPosition; + } + + /// NodeUpdated - Ignore updates for now. + virtual void NodeUpdated(SDNode *N) {} + }; + + /// ReplaceUses - replace all uses of the old node F with the use + /// of the new node T. + void ReplaceUses(SDValue F, SDValue T) { + ISelUpdater ISU(ISelPosition); + CurDAG->ReplaceAllUsesOfValueWith(F, T, &ISU); + } + + /// ReplaceUses - replace all uses of the old nodes F with the use + /// of the new nodes T. + void ReplaceUses(const SDValue *F, const SDValue *T, unsigned Num) { + ISelUpdater ISU(ISelPosition); + CurDAG->ReplaceAllUsesOfValuesWith(F, T, Num, &ISU); + } + + /// ReplaceUses - replace all uses of the old node F with the use + /// of the new node T. + void ReplaceUses(SDNode *F, SDNode *T) { + ISelUpdater ISU(ISelPosition); + CurDAG->ReplaceAllUsesWith(F, T, &ISU); + } + /// SelectInlineAsmMemoryOperands - Calls to this are automatically generated /// by tblgen. Others should not call it. void SelectInlineAsmMemoryOperands(std::vector<SDValue> &Ops); + +public: // Calls to these predicates are generated by tblgen. bool CheckAndMask(SDValue LHS, ConstantSDNode *RHS, int64_t DesiredMaskS) const; @@ -122,8 +245,8 @@ protected: return 0; } - /// CheckNodePredicate - This function is generated by tblgen in the - /// target. It runs node predicate #PredNo and returns true if it succeeds or + /// CheckNodePredicate - This function is generated by tblgen in the target. + /// It runs node predicate number PredNo and returns true if it succeeds or /// false if it fails. The number is a private implementation /// detail to the code tblgen produces. virtual bool CheckNodePredicate(SDNode *N, unsigned PredNo) const { @@ -131,6 +254,23 @@ protected: return 0; } + virtual bool CheckComplexPattern(SDNode *Root, SDValue N, unsigned PatternNo, + SmallVectorImpl<SDValue> &Result) { + assert(0 && "Tblgen should generate the implementation of this!"); + return false; + } + + virtual SDValue RunSDNodeXForm(SDValue V, unsigned XFormNo) { + assert(0 && "Tblgen shoudl generate this!"); + return SDValue(); + } + + SDNode *SelectCodeCommon(SDNode *NodeToMatch, + const unsigned char *MatcherTable, + unsigned TableSize); + +private: + // Calls to these functions are generated by tblgen. SDNode *Select_INLINEASM(SDNode *N); SDNode *Select_UNDEF(SDNode *N); @@ -139,6 +279,10 @@ protected: void CannotYetSelectIntrinsic(SDNode *N); private: + void DoInstructionSelection(); + SDNode *MorphNode(SDNode *Node, unsigned TargetOpc, SDVTList VTs, + const SDValue *Ops, unsigned NumOps, unsigned EmitNodeInfo); + void SelectAllBasicBlocks(Function &Fn, MachineFunction &MF, MachineModuleInfo *MMI, DwarfWriter *DW, @@ -164,6 +308,16 @@ private: /// one preferred by the target. /// ScheduleDAGSDNodes *CreateScheduler(); + + /// OpcodeOffset - This is a cache used to dispatch efficiently into isel + /// state machines that start with a OPC_SwitchOpcode node. + std::vector<unsigned> OpcodeOffset; + + void UpdateChainsAndFlags(SDNode *NodeToMatch, SDValue InputChain, + const SmallVectorImpl<SDNode*> &ChainNodesMatched, + SDValue InputFlag,const SmallVectorImpl<SDNode*> &F, + bool isMorphNodeTo); + }; } diff --git a/include/llvm/CodeGen/SelectionDAGNodes.h b/include/llvm/CodeGen/SelectionDAGNodes.h index df1f91b..950963e 100644 --- a/include/llvm/CodeGen/SelectionDAGNodes.h +++ b/include/llvm/CodeGen/SelectionDAGNodes.h @@ -1595,8 +1595,10 @@ public: return SubclassData; } + // We access subclass data here so that we can check consistency + // with MachineMemOperand information. bool isVolatile() const { return (SubclassData >> 5) & 1; } - bool isNonTemporal() const { return MMO->isNonTemporal(); } + bool isNonTemporal() const { return (SubclassData >> 6) & 1; } /// Returns the SrcValue and offset that describes the location of the access const Value *getSrcValue() const { return MMO->getValue(); } @@ -1762,7 +1764,12 @@ public: bool isSplat() const { return isSplatMask(Mask, getValueType(0)); } int getSplatIndex() const { assert(isSplat() && "Cannot get splat index for non-splat!"); - return Mask[0]; + EVT VT = getValueType(0); + for (unsigned i = 0, e = VT.getVectorNumElements(); i != e; ++i) { + if (Mask[i] != -1) + return Mask[i]; + } + return -1; } static bool isSplatMask(const int *Mask, EVT VT); @@ -1808,6 +1815,12 @@ public: const APFloat& getValueAPF() const { return Value->getValueAPF(); } const ConstantFP *getConstantFPValue() const { return Value; } + /// isZero - Return true if the value is positive or negative zero. + bool isZero() const { return Value->isZero(); } + + /// isNaN - Return true if the value is a NaN. + bool isNaN() const { return Value->isNaN(); } + /// isExactlyValue - We don't rely on operator== working on double values, as /// it returns true for things that are clearly not equal, like -0.0 and 0.0. /// As such, this method can be used to do an exact bit-for-bit comparison of diff --git a/include/llvm/CodeGen/TargetLoweringObjectFileImpl.h b/include/llvm/CodeGen/TargetLoweringObjectFileImpl.h index c5aa626..3d99fa7 100644 --- a/include/llvm/CodeGen/TargetLoweringObjectFileImpl.h +++ b/include/llvm/CodeGen/TargetLoweringObjectFileImpl.h @@ -171,6 +171,11 @@ public: virtual const MCExpr * getSymbolForDwarfGlobalReference(const GlobalValue *GV, Mangler *Mang, MachineModuleInfo *MMI, unsigned Encoding) const; + + virtual unsigned getPersonalityEncoding() const; + virtual unsigned getLSDAEncoding() const; + virtual unsigned getFDEEncoding() const; + virtual unsigned getTTypeEncoding() const; }; diff --git a/include/llvm/CompilerDriver/Common.td b/include/llvm/CompilerDriver/Common.td index 9c3e861..479bd6e 100644 --- a/include/llvm/CompilerDriver/Common.td +++ b/include/llvm/CompilerDriver/Common.td @@ -20,9 +20,12 @@ class Tool<list<dag> l> { def in_language; def out_language; def output_suffix; -def cmd_line; +def command; +def out_file_option; +def in_file_option; def join; def sink; +def works_on_empty; def actions; // Possible option types. @@ -46,6 +49,7 @@ def optional; def really_hidden; def required; def comma_separated; +def forward_not_split; // The 'case' construct. def case; @@ -81,6 +85,7 @@ def forward_as; def forward_value; def forward_transformed_value; def stop_compilation; +def no_out_file; def unpack_values; def warning; def error; diff --git a/include/llvm/CompilerDriver/Main.inc b/include/llvm/CompilerDriver/Main.inc index fc8b503..71bb8cb 100644 --- a/include/llvm/CompilerDriver/Main.inc +++ b/include/llvm/CompilerDriver/Main.inc @@ -10,7 +10,7 @@ // This tool provides a single point of access to the LLVM // compilation tools. It has many options. To discover the options // supported please refer to the tools' manual page or run the tool -// with the --help option. +// with the -help option. // // This file provides the default entry point for the driver executable. // diff --git a/include/llvm/CompilerDriver/Tool.h b/include/llvm/CompilerDriver/Tool.h index a982e2d..85d1690 100644 --- a/include/llvm/CompilerDriver/Tool.h +++ b/include/llvm/CompilerDriver/Tool.h @@ -20,15 +20,19 @@ #include "llvm/ADT/StringSet.h" #include "llvm/System/Path.h" +#include <string> #include <vector> +#include <utility> namespace llvmc { class LanguageMap; + typedef std::vector<std::pair<unsigned, std::string> > ArgsVector; typedef std::vector<llvm::sys::Path> PathVector; + typedef std::vector<std::string> StrVector; typedef llvm::StringSet<> InputLanguagesSet; - /// Tool - A class + /// Tool - Represents a single tool. class Tool : public llvm::RefCountedBaseVPTR<Tool> { public: @@ -51,6 +55,7 @@ namespace llvmc { virtual const char* OutputLanguage() const = 0; virtual bool IsJoin() const = 0; + virtual bool WorksOnEmpty() const = 0; protected: /// OutFileName - Generate the output file name. @@ -58,6 +63,8 @@ namespace llvmc { const llvm::sys::Path& TempDir, bool StopCompilation, const char* OutputSuffix) const; + + StrVector SortArgs(ArgsVector& Args) const; }; /// JoinTool - A Tool that has an associated input file list. diff --git a/include/llvm/Constants.h b/include/llvm/Constants.h index bd14303d..1cebb20 100644 --- a/include/llvm/Constants.h +++ b/include/llvm/Constants.h @@ -276,6 +276,12 @@ public: return Val.isZero() && Val.isNegative(); } + /// isZero - Return true if the value is positive or negative zero. + bool isZero() const { return Val.isZero(); } + + /// isNaN - Return true if the value is a NaN. + bool isNaN() const { return Val.isNaN(); } + /// isExactlyValue - We don't rely on operator== working on double values, as /// it returns true for things that are clearly not equal, like -0.0 and 0.0. /// As such, this method can be used to do an exact bit-for-bit comparison of @@ -692,8 +698,9 @@ public: /// independent way (Note: the return type is an i64). static Constant *getAlignOf(const Type* Ty); - /// getSizeOf constant expr - computes the size of a type in a target - /// independent way (Note: the return type is an i64). + /// getSizeOf constant expr - computes the (alloc) size of a type (in + /// address-units, not bits) in a target independent way (Note: the return + /// type is an i64). /// static Constant *getSizeOf(const Type* Ty); diff --git a/include/llvm/Instructions.h b/include/llvm/Instructions.h index 1ae495f..f07291c 100644 --- a/include/llvm/Instructions.h +++ b/include/llvm/Instructions.h @@ -591,7 +591,7 @@ public: "Both operands to ICmp instruction are not of the same type!"); // Check that the operands are the right type assert((getOperand(0)->getType()->isIntOrIntVectorTy() || - isa<PointerType>(getOperand(0)->getType())) && + getOperand(0)->getType()->isPointerTy()) && "Invalid operand types for ICmp instruction"); } @@ -612,7 +612,7 @@ public: "Both operands to ICmp instruction are not of the same type!"); // Check that the operands are the right type assert((getOperand(0)->getType()->isIntOrIntVectorTy() || - isa<PointerType>(getOperand(0)->getType())) && + getOperand(0)->getType()->isPointerTy()) && "Invalid operand types for ICmp instruction"); } @@ -631,7 +631,7 @@ public: "Both operands to ICmp instruction are not of the same type!"); // Check that the operands are the right type assert((getOperand(0)->getType()->isIntOrIntVectorTy() || - isa<PointerType>(getOperand(0)->getType())) && + getOperand(0)->getType()->isPointerTy()) && "Invalid operand types for ICmp instruction"); } @@ -1805,7 +1805,7 @@ public: return i/2; } - /// getIncomingBlock - Return incoming basic block #i. + /// getIncomingBlock - Return incoming basic block number @p i. /// BasicBlock *getIncomingBlock(unsigned i) const { return cast<BasicBlock>(getOperand(i*2+1)); diff --git a/include/llvm/LinkAllPasses.h b/include/llvm/LinkAllPasses.h index a7e2e05..ae53851 100644 --- a/include/llvm/LinkAllPasses.h +++ b/include/llvm/LinkAllPasses.h @@ -46,7 +46,6 @@ namespace { (void) llvm::createAggressiveDCEPass(); (void) llvm::createAliasAnalysisCounterPass(); (void) llvm::createAliasDebugger(); - (void) llvm::createAndersensPass(); (void) llvm::createArgumentPromotionPass(); (void) llvm::createStructRetPromotionPass(); (void) llvm::createBasicAliasAnalysisPass(); diff --git a/include/llvm/MC/MCAssembler.h b/include/llvm/MC/MCAssembler.h index 4527f3c..882929f 100644 --- a/include/llvm/MC/MCAssembler.h +++ b/include/llvm/MC/MCAssembler.h @@ -195,12 +195,16 @@ class MCAlignFragment : public MCFragment { /// cannot be satisfied in this width then this fragment is ignored. unsigned MaxBytesToEmit; + /// EmitNops - true when aligning code and optimal nops to be used for filling + bool EmitNops; + public: MCAlignFragment(unsigned _Alignment, int64_t _Value, unsigned _ValueSize, - unsigned _MaxBytesToEmit, MCSectionData *SD = 0) + unsigned _MaxBytesToEmit, bool _EmitNops, + MCSectionData *SD = 0) : MCFragment(FT_Align, SD), Alignment(_Alignment), Value(_Value),ValueSize(_ValueSize), - MaxBytesToEmit(_MaxBytesToEmit) {} + MaxBytesToEmit(_MaxBytesToEmit), EmitNops(_EmitNops) {} /// @name Accessors /// @{ @@ -217,6 +221,8 @@ public: unsigned getMaxBytesToEmit() const { return MaxBytesToEmit; } + unsigned getEmitNops() const { return EmitNops; } + /// @} static bool classof(const MCFragment *F) { diff --git a/include/llvm/MC/MCContext.h b/include/llvm/MC/MCContext.h index 95c6bd4..74415e2 100644 --- a/include/llvm/MC/MCContext.h +++ b/include/llvm/MC/MCContext.h @@ -46,30 +46,28 @@ namespace llvm { /// @name Symbol Managment /// @{ - /// CreateSymbol - Create a new symbol with the specified @param Name. + /// CreateSymbol - Create a new symbol with the specified @p Name. /// /// @param Name - The symbol name, which must be unique across all symbols. MCSymbol *CreateSymbol(StringRef Name); /// GetOrCreateSymbol - Lookup the symbol inside with the specified - /// @param Name. If it exists, return it. If not, create a forward + /// @p Name. If it exists, return it. If not, create a forward /// reference and return it. /// /// @param Name - The symbol name, which must be unique across all symbols. - /// @param IsTemporary - Whether this symbol is an assembler temporary, - /// which should not survive into the symbol table for the translation unit. MCSymbol *GetOrCreateSymbol(StringRef Name); MCSymbol *GetOrCreateSymbol(const Twine &Name); /// CreateTemporarySymbol - Create a new temporary symbol with the specified - /// @param Name. + /// @p Name. /// /// @param Name - The symbol name, for debugging purposes only, temporary /// symbols do not surive assembly. If non-empty the name must be unique /// across all symbols. MCSymbol *CreateTemporarySymbol(StringRef Name = ""); - /// LookupSymbol - Get the symbol for @param Name, or null. + /// LookupSymbol - Get the symbol for \p Name, or null. MCSymbol *LookupSymbol(StringRef Name) const; /// @} diff --git a/include/llvm/MC/MCStreamer.h b/include/llvm/MC/MCStreamer.h index 624d9a6..696d024 100644 --- a/include/llvm/MC/MCStreamer.h +++ b/include/llvm/MC/MCStreamer.h @@ -92,12 +92,12 @@ namespace llvm { const MCSection *getCurrentSection() const { return CurSection; } /// SwitchSection - Set the current section where code is being emitted to - /// @param Section. This is required to update CurSection. + /// @p Section. This is required to update CurSection. /// /// This corresponds to assembler directives like .section, .text, etc. virtual void SwitchSection(const MCSection *Section) = 0; - /// EmitLabel - Emit a label for @param Symbol into the current section. + /// EmitLabel - Emit a label for @p Symbol into the current section. /// /// This corresponds to an assembler statement such as: /// foo: @@ -107,10 +107,10 @@ namespace llvm { /// used in an assignment. virtual void EmitLabel(MCSymbol *Symbol) = 0; - /// EmitAssemblerFlag - Note in the output the specified @param Flag + /// EmitAssemblerFlag - Note in the output the specified @p Flag virtual void EmitAssemblerFlag(MCAssemblerFlag Flag) = 0; - /// EmitAssignment - Emit an assignment of @param Value to @param Symbol. + /// EmitAssignment - Emit an assignment of @p Value to @p Symbol. /// /// This corresponds to an assembler statement such as: /// symbol = value @@ -123,11 +123,11 @@ namespace llvm { /// @param Value - The value for the symbol. virtual void EmitAssignment(MCSymbol *Symbol, const MCExpr *Value) = 0; - /// EmitSymbolAttribute - Add the given @param Attribute to @param Symbol. + /// EmitSymbolAttribute - Add the given @p Attribute to @p Symbol. virtual void EmitSymbolAttribute(MCSymbol *Symbol, MCSymbolAttr Attribute) = 0; - /// EmitSymbolDesc - Set the @param DescValue for the @param Symbol. + /// EmitSymbolDesc - Set the @p DescValue for the @p Symbol. /// /// @param Symbol - The symbol to have its n_desc field set. /// @param DescValue - The value to set into the n_desc field. @@ -176,8 +176,8 @@ namespace llvm { /// etc. virtual void EmitBytes(StringRef Data, unsigned AddrSpace) = 0; - /// EmitValue - Emit the expression @param Value into the output as a native - /// integer of the given @param Size bytes. + /// EmitValue - Emit the expression @p Value into the output as a native + /// integer of the given @p Size bytes. /// /// This is used to implement assembler directives such as .word, .quad, /// etc. @@ -192,7 +192,7 @@ namespace llvm { /// to pass in a MCExpr for constant integers. virtual void EmitIntValue(uint64_t Value, unsigned Size,unsigned AddrSpace); - /// EmitGPRel32Value - Emit the expression @param Value into the output as a + /// EmitGPRel32Value - Emit the expression @p Value into the output as a /// gprel32 (32-bit GP relative) value. /// /// This is used to implement assembler directives such as .gprel32 on @@ -211,11 +211,11 @@ namespace llvm { } - /// EmitValueToAlignment - Emit some number of copies of @param Value until - /// the byte alignment @param ByteAlignment is reached. + /// EmitValueToAlignment - Emit some number of copies of @p Value until + /// the byte alignment @p ByteAlignment is reached. /// /// If the number of bytes need to emit for the alignment is not a multiple - /// of @param ValueSize, then the contents of the emitted fill bytes is + /// of @p ValueSize, then the contents of the emitted fill bytes is /// undefined. /// /// This used to implement the .align assembler directive. @@ -223,8 +223,8 @@ namespace llvm { /// @param ByteAlignment - The alignment to reach. This must be a power of /// two on some targets. /// @param Value - The value to use when filling bytes. - /// @param Size - The size of the integer (in bytes) to emit for @param - /// Value. This must match a native machine width. + /// @param ValueSize - The size of the integer (in bytes) to emit for + /// @p Value. This must match a native machine width. /// @param MaxBytesToEmit - The maximum numbers of bytes to emit, or 0. If /// the alignment cannot be reached in this many bytes, no bytes are /// emitted. @@ -232,8 +232,22 @@ namespace llvm { unsigned ValueSize = 1, unsigned MaxBytesToEmit = 0) = 0; - /// EmitValueToOffset - Emit some number of copies of @param Value until the - /// byte offset @param Offset is reached. + /// EmitCodeAlignment - Emit nops until the byte alignment @p ByteAlignment + /// is reached. + /// + /// This used to align code where the alignment bytes may be executed. This + /// can emit different bytes for different sizes to optimize execution. + /// + /// @param ByteAlignment - The alignment to reach. This must be a power of + /// two on some targets. + /// @param MaxBytesToEmit - The maximum numbers of bytes to emit, or 0. If + /// the alignment cannot be reached in this many bytes, no bytes are + /// emitted. + virtual void EmitCodeAlignment(unsigned ByteAlignment, + unsigned MaxBytesToEmit = 0) = 0; + + /// EmitValueToOffset - Emit some number of copies of @p Value until the + /// byte offset @p Offset is reached. /// /// This is used to implement assembler directives such as .org. /// @@ -254,7 +268,7 @@ namespace llvm { /// directive. virtual void EmitDwarfFileDirective(unsigned FileNo,StringRef Filename) = 0; - /// EmitInstruction - Emit the given @param Instruction into the current + /// EmitInstruction - Emit the given @p Instruction into the current /// section. virtual void EmitInstruction(const MCInst &Inst) = 0; diff --git a/include/llvm/Metadata.h b/include/llvm/Metadata.h index 4e459bf..e536322 100644 --- a/include/llvm/Metadata.h +++ b/include/llvm/Metadata.h @@ -75,6 +75,7 @@ class MDNode : public Value, public FoldingSetNode { MDNode(const MDNode &); // DO NOT IMPLEMENT void operator=(const MDNode &); // DO NOT IMPLEMENT friend class MDNodeOperand; + friend class LLVMContextImpl; /// NumOperands - This many 'MDNodeOperand' items are co-allocated onto the /// end of this MDNode. @@ -103,9 +104,13 @@ class MDNode : public Value, public FoldingSetNode { FL_Yes = 1 }; - // Replace each instance of F from the operand list of this node with T. + /// replaceOperand - Replace each instance of F from the operand list of this + /// node with T. void replaceOperand(MDNodeOperand *Op, Value *NewVal); ~MDNode(); + /// replaceAllOperandsWithNull - This is used while destroying llvm context to + /// gracefully delete all nodes. This method replaces all operands with null. + void replaceAllOperandsWithNull(); protected: explicit MDNode(LLVMContext &C, Value *const *Vals, unsigned NumVals, diff --git a/include/llvm/Pass.h b/include/llvm/Pass.h index e822a0f..8fc3a53 100644 --- a/include/llvm/Pass.h +++ b/include/llvm/Pass.h @@ -163,7 +163,7 @@ public: /// an analysis interface through multiple inheritance. If needed, it should /// override this to adjust the this pointer as needed for the specified pass /// info. - virtual void *getAdjustedAnalysisPointer(const PassInfo *PI) { + virtual void *getAdjustedAnalysisPointer(const PassInfo *) { return this; } virtual ImmutablePass *getAsImmutablePass() { return 0; } diff --git a/include/llvm/Support/CommandLine.h b/include/llvm/Support/CommandLine.h index 3ee2313..61c3256 100644 --- a/include/llvm/Support/CommandLine.h +++ b/include/llvm/Support/CommandLine.h @@ -93,9 +93,9 @@ enum ValueExpected { // Is a value required for the option? }; enum OptionHidden { // Control whether -help shows this option - NotHidden = 0x20, // Option included in --help & --help-hidden - Hidden = 0x40, // -help doesn't, but --help-hidden does - ReallyHidden = 0x60, // Neither --help nor --help-hidden show this arg + NotHidden = 0x20, // Option included in -help & -help-hidden + Hidden = 0x40, // -help doesn't, but -help-hidden does + ReallyHidden = 0x60, // Neither -help nor -help-hidden show this arg HiddenMask = 0x60 }; @@ -159,7 +159,7 @@ class Option { Option *NextRegistered; // Singly linked list of registered options. public: const char *ArgStr; // The argument string itself (ex: "help", "o") - const char *HelpStr; // The descriptive text message for --help + const char *HelpStr; // The descriptive text message for -help const char *ValueStr; // String describing what the value of this option is inline enum NumOccurrencesFlag getNumOccurrencesFlag() const { @@ -251,14 +251,14 @@ public: // command line option parsers... // -// desc - Modifier to set the description shown in the --help output... +// desc - Modifier to set the description shown in the -help output... struct desc { const char *Desc; desc(const char *Str) : Desc(Str) {} void apply(Option &O) const { O.setDescription(Desc); } }; -// value_desc - Modifier to set the value description shown in the --help +// value_desc - Modifier to set the value description shown in the -help // output... struct value_desc { const char *Desc; @@ -437,7 +437,7 @@ protected: // Default parser implementation - This implementation depends on having a // mapping of recognized options to values of some sort. In addition to this, // each entry in the mapping also tracks a help message that is printed with the -// command line option for --help. Because this is a simple mapping parser, the +// command line option for -help. Because this is a simple mapping parser, the // data type can be any unsupported type. // template <class DataType> @@ -1373,7 +1373,7 @@ struct extrahelp { void PrintVersionMessage(); // This function just prints the help message, exactly the same way as if the -// --help option had been given on the command line. +// -help option had been given on the command line. // NOTE: THIS FUNCTION TERMINATES THE PROGRAM! void PrintHelpMessage(); diff --git a/include/llvm/Support/Regex.h b/include/llvm/Support/Regex.h index c954c0d..591af00 100644 --- a/include/llvm/Support/Regex.h +++ b/include/llvm/Support/Regex.h @@ -56,6 +56,19 @@ namespace llvm { /// /// This returns true on a successful match. bool match(const StringRef &String, SmallVectorImpl<StringRef> *Matches=0); + + /// sub - Return the result of replacing the first match of the regex in + /// \arg String with the \arg Repl string. Backreferences like "\0" in the + /// replacement string are replaced with the appropriate match substring. + /// + /// Note that the replacement string has backslash escaping performed on + /// it. Invalid backreferences are ignored (replaced by empty strings). + /// + /// \param Error If non-null, any errors in the substitution (invalid + /// backreferences, trailing backslashes) will be recorded as a non-empty + /// string. + std::string sub(StringRef Repl, StringRef String, std::string *Error = 0); + private: struct llvm_regex *preg; int error; diff --git a/include/llvm/Support/TargetFolder.h b/include/llvm/Support/TargetFolder.h index 384c493..d34f35f 100644 --- a/include/llvm/Support/TargetFolder.h +++ b/include/llvm/Support/TargetFolder.h @@ -185,7 +185,9 @@ public: return C; // avoid calling Fold return Fold(ConstantExpr::getIntegerCast(C, DestTy, isSigned)); } - + Constant *CreatePointerCast(Constant *C, const Type *DestTy) const { + return ConstantExpr::getPointerCast(C, DestTy); + } Constant *CreateBitCast(Constant *C, const Type *DestTy) const { return CreateCast(Instruction::BitCast, C, DestTy); } diff --git a/include/llvm/System/Path.h b/include/llvm/System/Path.h index 1be27b2..d4af478 100644 --- a/include/llvm/System/Path.h +++ b/include/llvm/System/Path.h @@ -203,7 +203,7 @@ namespace sys { } /// Makes a copy of \p that to \p this. - /// @param \p that A StringRef denoting the path + /// @param that A StringRef denoting the path /// @returns \p this /// @brief Assignment Operator Path &operator=(StringRef that); diff --git a/include/llvm/Target/Target.td b/include/llvm/Target/Target.td index 9a117df..0cffffb 100644 --- a/include/llvm/Target/Target.td +++ b/include/llvm/Target/Target.td @@ -211,16 +211,9 @@ class Instruction { // hasSideEffects - The instruction has side effects that are not // captured by any operands of the instruction or other flags. // - // mayHaveSideEffects - Some instances of the instruction can have side - // effects. The virtual method "isReallySideEffectFree" is called to - // determine this. Load instructions are an example of where this is - // useful. In general, loads always have side effects. However, loads from - // constant pools don't. Individual back ends make this determination. - // // neverHasSideEffects - Set on an instruction with no pattern if it has no // side effects. bit hasSideEffects = 0; - bit mayHaveSideEffects = 0; bit neverHasSideEffects = 0; // Is this instruction a "real" instruction (with a distinct machine diff --git a/include/llvm/Target/TargetAsmBackend.h b/include/llvm/Target/TargetAsmBackend.h new file mode 100644 index 0000000..dfdabdb --- /dev/null +++ b/include/llvm/Target/TargetAsmBackend.h @@ -0,0 +1,35 @@ +//===-- llvm/Target/TargetAsmBackend.h - Target Asm Backend -----*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_TARGET_TARGETASMBACKEND_H +#define LLVM_TARGET_TARGETASMBACKEND_H + +namespace llvm { +class Target; + +/// TargetAsmBackend - Generic interface to target specific assembler backends. +class TargetAsmBackend { + TargetAsmBackend(const TargetAsmBackend &); // DO NOT IMPLEMENT + void operator=(const TargetAsmBackend &); // DO NOT IMPLEMENT +protected: // Can only create subclasses. + TargetAsmBackend(const Target &); + + /// TheTarget - The Target that this machine was created for. + const Target &TheTarget; + +public: + virtual ~TargetAsmBackend(); + + const Target &getTarget() const { return TheTarget; } + +}; + +} // End llvm namespace + +#endif diff --git a/include/llvm/Target/TargetAsmParser.h b/include/llvm/Target/TargetAsmParser.h index da9ba2b..85315c1 100644 --- a/include/llvm/Target/TargetAsmParser.h +++ b/include/llvm/Target/TargetAsmParser.h @@ -42,8 +42,8 @@ public: /// line should be parsed up to and including the end-of-statement token. On /// failure, the parser is not required to read to the end of the line. // - /// \param AP - The current parser object. /// \param Name - The instruction name. + /// \param NameLoc - The source location of the name. /// \param Operands [out] - The list of parsed operands, this returns /// ownership of them to the caller. /// \return True on failure. @@ -59,7 +59,7 @@ public: /// the target, the entire line is parsed up to and including the /// end-of-statement token and false is returned. /// - /// \param ID - the identifier token of the directive. + /// \param DirectiveID - the identifier token of the directive. virtual bool ParseDirective(AsmToken DirectiveID) = 0; /// MatchInstruction - Recognize a series of operands of a parsed instruction diff --git a/include/llvm/Target/TargetInstrInfo.h b/include/llvm/Target/TargetInstrInfo.h index d95e4e8..4b26beb 100644 --- a/include/llvm/Target/TargetInstrInfo.h +++ b/include/llvm/Target/TargetInstrInfo.h @@ -19,14 +19,14 @@ namespace llvm { -class MCAsmInfo; -class TargetRegisterClass; -class TargetRegisterInfo; -class LiveVariables; class CalleeSavedInfo; +class LiveVariables; +class MCAsmInfo; +class MachineMemOperand; class SDNode; class SelectionDAG; -class MachineMemOperand; +class TargetRegisterClass; +class TargetRegisterInfo; template<class T> class SmallVectorImpl; @@ -243,13 +243,11 @@ public: virtual bool findCommutedOpIndices(MachineInstr *MI, unsigned &SrcOpIdx1, unsigned &SrcOpIdx2) const = 0; - /// isIdentical - Return true if two instructions are identical. This differs - /// from MachineInstr::isIdenticalTo() in that it does not require the - /// virtual destination registers to be the same. This is used by MachineLICM - /// and other MI passes to perform CSE. - virtual bool isIdentical(const MachineInstr *MI, - const MachineInstr *Other, - const MachineRegisterInfo *MRI) const = 0; + /// produceSameValue - Return true if two machine instructions would produce + /// identical values. By default, this is only true when the two instructions + /// are deemed identical except for defs. + virtual bool produceSameValue(const MachineInstr *MI0, + const MachineInstr *MI1) const = 0; /// AnalyzeBranch - Analyze the branching code at the end of MBB, returning /// true if it cannot be understood (e.g. it's a switch dispatch or isn't @@ -560,10 +558,8 @@ public: const TargetRegisterInfo *TRI) const; virtual MachineInstr *duplicate(MachineInstr *Orig, MachineFunction &MF) const; - virtual bool isIdentical(const MachineInstr *MI, - const MachineInstr *Other, - const MachineRegisterInfo *MRI) const; - + virtual bool produceSameValue(const MachineInstr *MI0, + const MachineInstr *MI1) const; virtual unsigned GetFunctionSizeInBytes(const MachineFunction &MF) const; }; diff --git a/include/llvm/Target/TargetLowering.h b/include/llvm/Target/TargetLowering.h index c6ac89a..60b7ebd 100644 --- a/include/llvm/Target/TargetLowering.h +++ b/include/llvm/Target/TargetLowering.h @@ -1164,7 +1164,7 @@ public: bool isVarArg, bool isInreg, unsigned NumFixedArgs, CallingConv::ID CallConv, bool isTailCall, bool isReturnValueUsed, SDValue Callee, ArgListTy &Args, - SelectionDAG &DAG, DebugLoc dl, unsigned Order); + SelectionDAG &DAG, DebugLoc dl); /// LowerCall - This hook must be implemented to lower calls into the /// the specified DAG. The outgoing arguments to the call are described diff --git a/include/llvm/Target/TargetMachine.h b/include/llvm/Target/TargetMachine.h index c496e11..a7062ac 100644 --- a/include/llvm/Target/TargetMachine.h +++ b/include/llvm/Target/TargetMachine.h @@ -189,8 +189,9 @@ public: /// is not supported, or false on success. virtual bool addPassesToEmitFile(PassManagerBase &, formatted_raw_ostream &, - CodeGenFileType Filetype, - CodeGenOpt::Level) { + CodeGenFileType, + CodeGenOpt::Level, + bool DisableVerify = true) { return true; } @@ -202,7 +203,8 @@ public: /// virtual bool addPassesToEmitMachineCode(PassManagerBase &, JITCodeEmitter &, - CodeGenOpt::Level) { + CodeGenOpt::Level, + bool DisableVerify = true) { return true; } @@ -212,7 +214,8 @@ public: virtual bool WantsWholeFile() const { return false; } virtual bool addPassesToEmitWholeFile(PassManager &, formatted_raw_ostream &, CodeGenFileType, - CodeGenOpt::Level) { + CodeGenOpt::Level, + bool DisableVerify = true) { return true; } }; @@ -227,7 +230,8 @@ protected: // Can only create subclasses. /// addCommonCodeGenPasses - Add standard LLVM codegen passes used for /// both emitting to assembly files or machine code output. /// - bool addCommonCodeGenPasses(PassManagerBase &, CodeGenOpt::Level); + bool addCommonCodeGenPasses(PassManagerBase &, CodeGenOpt::Level, + bool DisableVerify); private: virtual void setCodeModelForJIT(); @@ -242,7 +246,8 @@ public: virtual bool addPassesToEmitFile(PassManagerBase &PM, formatted_raw_ostream &Out, CodeGenFileType FileType, - CodeGenOpt::Level); + CodeGenOpt::Level, + bool DisableVerify = true); /// addPassesToEmitMachineCode - Add passes to the specified pass manager to /// get machine code emitted. This uses a JITCodeEmitter object to handle @@ -252,7 +257,8 @@ public: /// virtual bool addPassesToEmitMachineCode(PassManagerBase &PM, JITCodeEmitter &MCE, - CodeGenOpt::Level); + CodeGenOpt::Level, + bool DisableVerify = true); /// Target-Independent Code Generator Pass Configuration Options. diff --git a/include/llvm/Target/TargetOpcodes.h b/include/llvm/Target/TargetOpcodes.h index 10cb45f..48665b7 100644 --- a/include/llvm/Target/TargetOpcodes.h +++ b/include/llvm/Target/TargetOpcodes.h @@ -16,7 +16,7 @@ namespace llvm { -// Invariant opcodes: All instruction sets have these as their low opcodes. +/// Invariant opcodes: All instruction sets have these as their low opcodes. namespace TargetOpcode { enum { PHI = 0, @@ -63,7 +63,7 @@ namespace TargetOpcode { /// the copy are emitted with the TargetInstrInfo::copyRegToReg hook. COPY_TO_REGCLASS = 10, - // DBG_VALUE - a mapping of the llvm.dbg.value intrinsic + /// DBG_VALUE - a mapping of the llvm.dbg.value intrinsic DBG_VALUE = 11 }; } // end namespace TargetOpcode diff --git a/include/llvm/Target/TargetRegisterInfo.h b/include/llvm/Target/TargetRegisterInfo.h index 65b60f7..212cc93 100644 --- a/include/llvm/Target/TargetRegisterInfo.h +++ b/include/llvm/Target/TargetRegisterInfo.h @@ -587,6 +587,17 @@ public: return !hasFP(MF); } + /// canSimplifyCallFramePseudos - When possible, it's best to simplify the + /// call frame pseudo ops before doing frame index elimination. This is + /// possible only when frame index references between the pseudos won't + /// need adjusted for the call frame adjustments. Normally, that's true + /// if the function has a reserved call frame or a frame pointer. Some + /// targets (Thumb2, for example) may have more complicated criteria, + /// however, and can override this behavior. + virtual bool canSimplifyCallFramePseudos(MachineFunction &MF) const { + return hasReservedCallFrame(MF) || hasFP(MF); + } + /// hasReservedSpillSlot - Return true if target has reserved a spill slot in /// the stack frame of the given function for the specified register. e.g. On /// x86, if the frame register is required, the first fixed stack object is diff --git a/include/llvm/Target/TargetRegistry.h b/include/llvm/Target/TargetRegistry.h index 37380ab..a409b62 100644 --- a/include/llvm/Target/TargetRegistry.h +++ b/include/llvm/Target/TargetRegistry.h @@ -26,6 +26,7 @@ namespace llvm { class AsmPrinter; class Module; + class MCAssembler; class MCAsmInfo; class MCAsmParser; class MCCodeEmitter; @@ -33,6 +34,7 @@ namespace llvm { class MCDisassembler; class MCInstPrinter; class MCStreamer; + class TargetAsmBackend; class TargetAsmLexer; class TargetAsmParser; class TargetMachine; @@ -63,6 +65,8 @@ namespace llvm { MCContext &Ctx, MCStreamer &Streamer, const MCAsmInfo *MAI); + typedef TargetAsmBackend *(*AsmBackendCtorTy)(const Target &T, + MCAssembler &A); typedef TargetAsmLexer *(*AsmLexerCtorTy)(const Target &T, const MCAsmInfo &MAI); typedef TargetAsmParser *(*AsmParserCtorTy)(const Target &T,MCAsmParser &P); @@ -94,32 +98,35 @@ namespace llvm { bool HasJIT; AsmInfoCtorFnTy AsmInfoCtorFn; - + /// TargetMachineCtorFn - Construction function for this target's /// TargetMachine, if registered. TargetMachineCtorTy TargetMachineCtorFn; - /// AsmPrinterCtorFn - Construction function for this target's AsmPrinter, - /// if registered. - AsmPrinterCtorTy AsmPrinterCtorFn; + /// AsmBackendCtorFn - Construction function for this target's + /// TargetAsmBackend, if registered. + AsmBackendCtorTy AsmBackendCtorFn; /// AsmLexerCtorFn - Construction function for this target's TargetAsmLexer, /// if registered. AsmLexerCtorTy AsmLexerCtorFn; - + /// AsmParserCtorFn - Construction function for this target's /// TargetAsmParser, if registered. AsmParserCtorTy AsmParserCtorFn; - + + /// AsmPrinterCtorFn - Construction function for this target's AsmPrinter, + /// if registered. + AsmPrinterCtorTy AsmPrinterCtorFn; + /// MCDisassemblerCtorFn - Construction function for this target's /// MCDisassembler, if registered. MCDisassemblerCtorTy MCDisassemblerCtorFn; - - /// MCInstPrinterCtorFn - Construction function for this target's + /// MCInstPrinterCtorFn - Construction function for this target's /// MCInstPrinter, if registered. MCInstPrinterCtorTy MCInstPrinterCtorFn; - + /// CodeEmitterCtorFn - Construction function for this target's CodeEmitter, /// if registered. CodeEmitterCtorTy CodeEmitterCtorFn; @@ -147,12 +154,18 @@ namespace llvm { /// hasTargetMachine - Check if this target supports code generation. bool hasTargetMachine() const { return TargetMachineCtorFn != 0; } - /// hasAsmPrinter - Check if this target supports .s printing. - bool hasAsmPrinter() const { return AsmPrinterCtorFn != 0; } + /// hasAsmBackend - Check if this target supports .o generation. + bool hasAsmBackend() const { return AsmBackendCtorFn != 0; } + + /// hasAsmLexer - Check if this target supports .s lexing. + bool hasAsmLexer() const { return AsmLexerCtorFn != 0; } /// hasAsmParser - Check if this target supports .s parsing. bool hasAsmParser() const { return AsmParserCtorFn != 0; } - + + /// hasAsmPrinter - Check if this target supports .s printing. + bool hasAsmPrinter() const { return AsmPrinterCtorFn != 0; } + /// hasMCDisassembler - Check if this target has a disassembler. bool hasMCDisassembler() const { return MCDisassemblerCtorFn != 0; } @@ -165,7 +178,7 @@ namespace llvm { /// @} /// @name Feature Constructors /// @{ - + /// createAsmInfo - Create a MCAsmInfo implementation for the specified /// target triple. /// @@ -178,7 +191,7 @@ namespace llvm { return 0; return AsmInfoCtorFn(*this, Triple); } - + /// createTargetMachine - Create a target specific machine implementation /// for the specified \arg Triple. /// @@ -193,14 +206,13 @@ namespace llvm { return TargetMachineCtorFn(*this, Triple, Features); } - /// createAsmPrinter - Create a target specific assembly printer pass. This - /// takes ownership of the MCContext and MCStreamer objects but not the MAI. - AsmPrinter *createAsmPrinter(formatted_raw_ostream &OS, TargetMachine &TM, - MCContext &Ctx, MCStreamer &Streamer, - const MCAsmInfo *MAI) const { - if (!AsmPrinterCtorFn) + /// createAsmBackend - Create a target specific assembly parser. + /// + /// \arg Backend - The target independent assembler object. + TargetAsmBackend *createAsmBackend(MCAssembler &Backend) const { + if (!AsmBackendCtorFn) return 0; - return AsmPrinterCtorFn(OS, TM, Ctx, Streamer, MAI); + return AsmBackendCtorFn(*this, Backend); } /// createAsmLexer - Create a target specific assembly lexer. @@ -210,7 +222,7 @@ namespace llvm { return 0; return AsmLexerCtorFn(*this, MAI); } - + /// createAsmParser - Create a target specific assembly parser. /// /// \arg Parser - The target independent parser implementation to use for @@ -220,7 +232,17 @@ namespace llvm { return 0; return AsmParserCtorFn(*this, Parser); } - + + /// createAsmPrinter - Create a target specific assembly printer pass. This + /// takes ownership of the MCContext and MCStreamer objects but not the MAI. + AsmPrinter *createAsmPrinter(formatted_raw_ostream &OS, TargetMachine &TM, + MCContext &Ctx, MCStreamer &Streamer, + const MCAsmInfo *MAI) const { + if (!AsmPrinterCtorFn) + return 0; + return AsmPrinterCtorFn(OS, TM, Ctx, Streamer, MAI); + } + const MCDisassembler *createMCDisassembler() const { if (!MCDisassemblerCtorFn) return 0; @@ -234,8 +256,8 @@ namespace llvm { return 0; return MCInstPrinterCtorFn(*this, SyntaxVariant, MAI, O); } - - + + /// createCodeEmitter - Create a target specific code emitter. MCCodeEmitter *createCodeEmitter(TargetMachine &TM, MCContext &Ctx) const { if (!CodeEmitterCtorFn) @@ -270,8 +292,8 @@ namespace llvm { return *this; } iterator operator++(int) { // Postincrement - iterator tmp = *this; - ++*this; + iterator tmp = *this; + ++*this; return tmp; } @@ -313,7 +335,7 @@ namespace llvm { /// RegisterTarget - Register the given target. Attempts to register a /// target which has already been registered will be ignored. - /// + /// /// Clients are responsible for ensuring that registration doesn't occur /// while another thread is attempting to access the registry. Typically /// this is done by initializing all targets at program startup. @@ -321,7 +343,7 @@ namespace llvm { /// @param T - The target being registered. /// @param Name - The target name. This should be a static string. /// @param ShortDesc - A short target description. This should be a static - /// string. + /// string. /// @param TQualityFn - The triple match quality computation function for /// this target. /// @param HasJIT - Whether the target supports JIT code @@ -334,11 +356,11 @@ namespace llvm { /// RegisterAsmInfo - Register a MCAsmInfo implementation for the /// given target. - /// + /// /// Clients are responsible for ensuring that registration doesn't occur /// while another thread is attempting to access the registry. Typically /// this is done by initializing all targets at program startup. - /// + /// /// @param T - The target being registered. /// @param Fn - A function to construct a MCAsmInfo for the target. static void RegisterAsmInfo(Target &T, Target::AsmInfoCtorFnTy Fn) { @@ -346,76 +368,90 @@ namespace llvm { if (!T.AsmInfoCtorFn) T.AsmInfoCtorFn = Fn; } - + /// RegisterTargetMachine - Register a TargetMachine implementation for the /// given target. - /// + /// /// Clients are responsible for ensuring that registration doesn't occur /// while another thread is attempting to access the registry. Typically /// this is done by initializing all targets at program startup. - /// + /// /// @param T - The target being registered. /// @param Fn - A function to construct a TargetMachine for the target. - static void RegisterTargetMachine(Target &T, + static void RegisterTargetMachine(Target &T, Target::TargetMachineCtorTy Fn) { // Ignore duplicate registration. if (!T.TargetMachineCtorFn) T.TargetMachineCtorFn = Fn; } - /// RegisterAsmPrinter - Register an AsmPrinter implementation for the given - /// target. - /// + /// RegisterAsmBackend - Register a TargetAsmBackend implementation for the + /// given target. + /// /// Clients are responsible for ensuring that registration doesn't occur /// while another thread is attempting to access the registry. Typically /// this is done by initializing all targets at program startup. /// /// @param T - The target being registered. - /// @param Fn - A function to construct an AsmPrinter for the target. - static void RegisterAsmPrinter(Target &T, Target::AsmPrinterCtorTy Fn) { - // Ignore duplicate registration. - if (!T.AsmPrinterCtorFn) - T.AsmPrinterCtorFn = Fn; + /// @param Fn - A function to construct an AsmBackend for the target. + static void RegisterAsmBackend(Target &T, Target::AsmBackendCtorTy Fn) { + if (!T.AsmBackendCtorFn) + T.AsmBackendCtorFn = Fn; } /// RegisterAsmLexer - Register a TargetAsmLexer implementation for the /// given target. - /// + /// /// Clients are responsible for ensuring that registration doesn't occur /// while another thread is attempting to access the registry. Typically /// this is done by initializing all targets at program startup. /// /// @param T - The target being registered. - /// @param Fn - A function to construct an AsmPrinter for the target. + /// @param Fn - A function to construct an AsmLexer for the target. static void RegisterAsmLexer(Target &T, Target::AsmLexerCtorTy Fn) { if (!T.AsmLexerCtorFn) T.AsmLexerCtorFn = Fn; } - + /// RegisterAsmParser - Register a TargetAsmParser implementation for the /// given target. - /// + /// /// Clients are responsible for ensuring that registration doesn't occur /// while another thread is attempting to access the registry. Typically /// this is done by initializing all targets at program startup. /// /// @param T - The target being registered. - /// @param Fn - A function to construct an AsmPrinter for the target. + /// @param Fn - A function to construct an AsmParser for the target. static void RegisterAsmParser(Target &T, Target::AsmParserCtorTy Fn) { if (!T.AsmParserCtorFn) T.AsmParserCtorFn = Fn; } - + + /// RegisterAsmPrinter - Register an AsmPrinter implementation for the given + /// target. + /// + /// Clients are responsible for ensuring that registration doesn't occur + /// while another thread is attempting to access the registry. Typically + /// this is done by initializing all targets at program startup. + /// + /// @param T - The target being registered. + /// @param Fn - A function to construct an AsmPrinter for the target. + static void RegisterAsmPrinter(Target &T, Target::AsmPrinterCtorTy Fn) { + // Ignore duplicate registration. + if (!T.AsmPrinterCtorFn) + T.AsmPrinterCtorFn = Fn; + } + /// RegisterMCDisassembler - Register a MCDisassembler implementation for /// the given target. - /// + /// /// Clients are responsible for ensuring that registration doesn't occur /// while another thread is attempting to access the registry. Typically /// this is done by initializing all targets at program startup. /// /// @param T - The target being registered. /// @param Fn - A function to construct an MCDisassembler for the target. - static void RegisterMCDisassembler(Target &T, + static void RegisterMCDisassembler(Target &T, Target::MCDisassemblerCtorTy Fn) { if (!T.MCDisassemblerCtorFn) T.MCDisassemblerCtorFn = Fn; @@ -423,7 +459,7 @@ namespace llvm { /// RegisterMCInstPrinter - Register a MCInstPrinter implementation for the /// given target. - /// + /// /// Clients are responsible for ensuring that registration doesn't occur /// while another thread is attempting to access the registry. Typically /// this is done by initializing all targets at program startup. @@ -435,7 +471,7 @@ namespace llvm { if (!T.MCInstPrinterCtorFn) T.MCInstPrinterCtorFn = Fn; } - + /// RegisterCodeEmitter - Register a MCCodeEmitter implementation for the /// given target. /// @@ -444,7 +480,7 @@ namespace llvm { /// this is done by initializing all targets at program startup. /// /// @param T - The target being registered. - /// @param Fn - A function to construct an AsmPrinter for the target. + /// @param Fn - A function to construct an MCCodeEmitter for the target. static void RegisterCodeEmitter(Target &T, Target::CodeEmitterCtorTy Fn) { if (!T.CodeEmitterCtorFn) T.CodeEmitterCtorFn = Fn; @@ -498,7 +534,7 @@ namespace llvm { static const MCAsmInfo *Allocator(const Target &T, StringRef TT) { return new MCAsmInfoImpl(T, TT); } - + }; /// RegisterAsmInfoFn - Helper template for registering a target assembly info @@ -537,25 +573,22 @@ namespace llvm { } }; - /// RegisterAsmPrinter - Helper template for registering a target specific - /// assembly printer, for use in the target machine initialization - /// function. Usage: + /// RegisterAsmBackend - Helper template for registering a target specific + /// assembler backend. Usage: /// - /// extern "C" void LLVMInitializeFooAsmPrinter() { + /// extern "C" void LLVMInitializeFooAsmBackend() { /// extern Target TheFooTarget; - /// RegisterAsmPrinter<FooAsmPrinter> X(TheFooTarget); + /// RegisterAsmBackend<FooAsmLexer> X(TheFooTarget); /// } - template<class AsmPrinterImpl> - struct RegisterAsmPrinter { - RegisterAsmPrinter(Target &T) { - TargetRegistry::RegisterAsmPrinter(T, &Allocator); + template<class AsmBackendImpl> + struct RegisterAsmBackend { + RegisterAsmBackend(Target &T) { + TargetRegistry::RegisterAsmBackend(T, &Allocator); } private: - static AsmPrinter *Allocator(formatted_raw_ostream &OS, TargetMachine &TM, - MCContext &Ctx, MCStreamer &Streamer, - const MCAsmInfo *MAI) { - return new AsmPrinterImpl(OS, TM, Ctx, Streamer, MAI); + static TargetAsmBackend *Allocator(const Target &T, MCAssembler &Backend) { + return new AsmBackendImpl(T, Backend); } }; @@ -572,7 +605,7 @@ namespace llvm { RegisterAsmLexer(Target &T) { TargetRegistry::RegisterAsmLexer(T, &Allocator); } - + private: static TargetAsmLexer *Allocator(const Target &T, const MCAsmInfo &MAI) { return new AsmLexerImpl(T, MAI); @@ -599,6 +632,28 @@ namespace llvm { } }; + /// RegisterAsmPrinter - Helper template for registering a target specific + /// assembly printer, for use in the target machine initialization + /// function. Usage: + /// + /// extern "C" void LLVMInitializeFooAsmPrinter() { + /// extern Target TheFooTarget; + /// RegisterAsmPrinter<FooAsmPrinter> X(TheFooTarget); + /// } + template<class AsmPrinterImpl> + struct RegisterAsmPrinter { + RegisterAsmPrinter(Target &T) { + TargetRegistry::RegisterAsmPrinter(T, &Allocator); + } + + private: + static AsmPrinter *Allocator(formatted_raw_ostream &OS, TargetMachine &TM, + MCContext &Ctx, MCStreamer &Streamer, + const MCAsmInfo *MAI) { + return new AsmPrinterImpl(OS, TM, Ctx, Streamer, MAI); + } + }; + /// RegisterCodeEmitter - Helper template for registering a target specific /// machine code emitter, for use in the target initialization /// function. Usage: diff --git a/include/llvm/Target/TargetSelectionDAG.td b/include/llvm/Target/TargetSelectionDAG.td index 4b72f81..4365d33 100644 --- a/include/llvm/Target/TargetSelectionDAG.td +++ b/include/llvm/Target/TargetSelectionDAG.td @@ -479,7 +479,6 @@ class PatLeaf<dag frag, code pred = [{}], SDNodeXForm xform = NOOP_SDNodeXForm> def vtInt : PatLeaf<(vt), [{ return N->getVT().isInteger(); }]>; def vtFP : PatLeaf<(vt), [{ return N->getVT().isFloatingPoint(); }]>; -def immAllOnes : PatLeaf<(imm), [{ return N->isAllOnesValue(); }]>; def immAllOnesV: PatLeaf<(build_vector), [{ return ISD::isBuildVectorAllOnes(N); }]>; @@ -496,7 +495,7 @@ def immAllZerosV_bc: PatLeaf<(bitconvert), [{ // Other helper fragments. -def not : PatFrag<(ops node:$in), (xor node:$in, immAllOnes)>; +def not : PatFrag<(ops node:$in), (xor node:$in, -1)>; def vnot : PatFrag<(ops node:$in), (xor node:$in, immAllOnesV)>; def vnot_conv : PatFrag<(ops node:$in), (xor node:$in, immAllOnesV_bc)>; def ineg : PatFrag<(ops node:$in), (sub 0, node:$in)>; diff --git a/include/llvm/Transforms/Scalar.h b/include/llvm/Transforms/Scalar.h index 7159f86..6893bad 100644 --- a/include/llvm/Transforms/Scalar.h +++ b/include/llvm/Transforms/Scalar.h @@ -263,7 +263,7 @@ extern const PassInfo *const LCSSAID; // GVN - This pass performs global value numbering and redundant load // elimination cotemporaneously. // -FunctionPass *createGVNPass(bool NoPRE = false, bool NoLoads = false); +FunctionPass *createGVNPass(bool NoLoads = false); //===----------------------------------------------------------------------===// // diff --git a/include/llvm/Transforms/Utils/BasicBlockUtils.h b/include/llvm/Transforms/Utils/BasicBlockUtils.h index 3f4571e..5279e96 100644 --- a/include/llvm/Transforms/Utils/BasicBlockUtils.h +++ b/include/llvm/Transforms/Utils/BasicBlockUtils.h @@ -102,6 +102,12 @@ void FindFunctionBackedges(const Function &F, // void RemoveSuccessor(TerminatorInst *TI, unsigned SuccNum); +/// GetSuccessorNumber - Search for the specified successor of basic block BB +/// and return its position in the terminator instruction's list of +/// successors. It is an error to call this with a block that is not a +/// successor. +unsigned GetSuccessorNumber(BasicBlock *BB, BasicBlock *Succ); + /// isCriticalEdge - Return true if the specified edge is a critical edge. /// Critical edges are edges from a block with multiple successors to a block /// with multiple predecessors. diff --git a/include/llvm/Type.h b/include/llvm/Type.h index c52419c..d09913a 100644 --- a/include/llvm/Type.h +++ b/include/llvm/Type.h @@ -89,7 +89,7 @@ public: VectorTyID, ///< 15: SIMD 'packed' format, or other vector type NumTypeIDs, // Must remain as last defined ID - LastPrimitiveTyID = LabelTyID, + LastPrimitiveTyID = MetadataTyID, FirstDerivedTyID = IntegerTyID }; @@ -182,6 +182,9 @@ public: // are defined in private classes defined in Type.cpp for primitive types. // + /// getDescription - Return the string representation of the type. + std::string getDescription() const; + /// getTypeID - Return the type id for the type. This will return one /// of the TypeID enum elements defined above. /// @@ -205,15 +208,21 @@ public: /// isPPC_FP128Ty - Return true if this is powerpc long double. bool isPPC_FP128Ty() const { return ID == PPC_FP128TyID; } + /// isFloatingPointTy - Return true if this is one of the five floating point + /// types + bool isFloatingPointTy() const { return ID == FloatTyID || ID == DoubleTyID || + ID == X86_FP80TyID || ID == FP128TyID || ID == PPC_FP128TyID; } + + /// isFPOrFPVectorTy - Return true if this is a FP type or a vector of FP. + /// + bool isFPOrFPVectorTy() const; + /// isLabelTy - Return true if this is 'label'. bool isLabelTy() const { return ID == LabelTyID; } /// isMetadataTy - Return true if this is 'metadata'. bool isMetadataTy() const { return ID == MetadataTyID; } - /// getDescription - Return the string representation of the type. - std::string getDescription() const; - /// isIntegerTy - True if this is an instance of IntegerType. /// bool isIntegerTy() const { return ID == IntegerTyID; } @@ -226,15 +235,6 @@ public: /// bool isIntOrIntVectorTy() const; - /// isFloatingPointTy - Return true if this is one of the five floating point - /// types - bool isFloatingPointTy() const { return ID == FloatTyID || ID == DoubleTyID || - ID == X86_FP80TyID || ID == FP128TyID || ID == PPC_FP128TyID; } - - /// isFPOrFPVectorTy - Return true if this is a FP type or a vector of FP. - /// - bool isFPOrFPVectorTy() const; - /// isFunctionTy - True if this is an instance of FunctionType. /// bool isFunctionTy() const { return ID == FunctionTyID; } @@ -243,6 +243,10 @@ public: /// bool isStructTy() const { return ID == StructTyID; } + /// isUnionTy - True if this is an instance of UnionType. + /// + bool isUnionTy() const { return ID == UnionTyID; } + /// isArrayTy - True if this is an instance of ArrayType. /// bool isArrayTy() const { return ID == ArrayTyID; } @@ -251,6 +255,10 @@ public: /// bool isPointerTy() const { return ID == PointerTyID; } + /// isOpaqueTy - True if this is an instance of OpaqueType. + /// + bool isOpaqueTy() const { return ID == OpaqueTyID; } + /// isVectorTy - True if this is an instance of VectorType. /// bool isVectorTy() const { return ID == VectorTyID; } diff --git a/lib/Analysis/AliasAnalysisCounter.cpp b/lib/Analysis/AliasAnalysisCounter.cpp index 761cd46..1053955 100644 --- a/lib/Analysis/AliasAnalysisCounter.cpp +++ b/lib/Analysis/AliasAnalysisCounter.cpp @@ -162,7 +162,7 @@ AliasAnalysisCounter::getModRefInfo(CallSite CS, Value *P, unsigned Size) { errs() << MRString << ": Ptr: "; errs() << "[" << Size << "B] "; WriteAsOperand(errs(), P, true, M); - errs() << "\t<->" << *CS.getInstruction(); + errs() << "\t<->" << *CS.getInstruction() << '\n'; } return R; } diff --git a/lib/Analysis/AliasAnalysisEvaluator.cpp b/lib/Analysis/AliasAnalysisEvaluator.cpp index 6b0a956..308b9e3 100644 --- a/lib/Analysis/AliasAnalysisEvaluator.cpp +++ b/lib/Analysis/AliasAnalysisEvaluator.cpp @@ -115,11 +115,11 @@ bool AAEval::runOnFunction(Function &F) { SetVector<CallSite> CallSites; for (Function::arg_iterator I = F.arg_begin(), E = F.arg_end(); I != E; ++I) - if (isa<PointerType>(I->getType())) // Add all pointer arguments + if (I->getType()->isPointerTy()) // Add all pointer arguments Pointers.insert(I); for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I) { - if (isa<PointerType>(I->getType())) // Add all pointer instructions + if (I->getType()->isPointerTy()) // Add all pointer instructions Pointers.insert(&*I); Instruction &Inst = *I; User::op_iterator OI = Inst.op_begin(); @@ -128,7 +128,7 @@ bool AAEval::runOnFunction(Function &F) { isa<Function>(CS.getCalledValue())) ++OI; // Skip actual functions for direct function calls. for (; OI != Inst.op_end(); ++OI) - if (isa<PointerType>((*OI)->getType()) && !isa<ConstantPointerNull>(*OI)) + if ((*OI)->getType()->isPointerTy() && !isa<ConstantPointerNull>(*OI)) Pointers.insert(*OI); if (CS.getInstruction()) CallSites.insert(CS); diff --git a/lib/Analysis/BasicAliasAnalysis.cpp b/lib/Analysis/BasicAliasAnalysis.cpp index 36b831c..31a649d 100644 --- a/lib/Analysis/BasicAliasAnalysis.cpp +++ b/lib/Analysis/BasicAliasAnalysis.cpp @@ -290,7 +290,7 @@ BasicAliasAnalysis::getModRefInfo(CallSite CS, Value *P, unsigned Size) { for (CallSite::arg_iterator CI = CS.arg_begin(), CE = CS.arg_end(); CI != CE; ++CI, ++ArgNo) { // Only look at the no-capture pointer arguments. - if (!isa<PointerType>((*CI)->getType()) || + if (!(*CI)->getType()->isPointerTy() || !CS.paramHasAttr(ArgNo+1, Attribute::NoCapture)) continue; @@ -662,7 +662,7 @@ BasicAliasAnalysis::aliasCheck(const Value *V1, unsigned V1Size, // Are we checking for alias of the same value? if (V1 == V2) return MustAlias; - if (!isa<PointerType>(V1->getType()) || !isa<PointerType>(V2->getType())) + if (!V1->getType()->isPointerTy() || !V2->getType()->isPointerTy()) return NoAlias; // Scalars cannot alias each other // Figure out what objects these things are pointing to if we can. diff --git a/lib/Analysis/CaptureTracking.cpp b/lib/Analysis/CaptureTracking.cpp index 10a8b11..8767c18 100644 --- a/lib/Analysis/CaptureTracking.cpp +++ b/lib/Analysis/CaptureTracking.cpp @@ -44,7 +44,7 @@ static int const Threshold = 20; /// counts as capturing it or not. bool llvm::PointerMayBeCaptured(const Value *V, bool ReturnCaptures, bool StoreCaptures) { - assert(isa<PointerType>(V->getType()) && "Capture is for pointers only!"); + assert(V->getType()->isPointerTy() && "Capture is for pointers only!"); SmallVector<Use*, Threshold> Worklist; SmallSet<Use*, Threshold> Visited; int Count = 0; diff --git a/lib/Analysis/ConstantFolding.cpp b/lib/Analysis/ConstantFolding.cpp index 808e6fa..114db2d 100644 --- a/lib/Analysis/ConstantFolding.cpp +++ b/lib/Analysis/ConstantFolding.cpp @@ -359,7 +359,7 @@ static Constant *FoldReinterpretLoadFromConstPtr(Constant *C, MapTy = Type::getInt32PtrTy(C->getContext()); else if (LoadTy->isDoubleTy()) MapTy = Type::getInt64PtrTy(C->getContext()); - else if (isa<VectorType>(LoadTy)) { + else if (LoadTy->isVectorTy()) { MapTy = IntegerType::get(C->getContext(), TD.getTypeAllocSizeInBits(LoadTy)); MapTy = PointerType::getUnqual(MapTy); @@ -605,7 +605,7 @@ static Constant *SymbolicallyEvaluateGEP(Constant *const *Ops, unsigned NumOps, SmallVector<Constant*, 32> NewIdxs; do { if (const SequentialType *ATy = dyn_cast<SequentialType>(Ty)) { - if (isa<PointerType>(ATy)) { + if (ATy->isPointerTy()) { // The only pointer indexing we'll do is on the first index of the GEP. if (!NewIdxs.empty()) break; @@ -783,45 +783,12 @@ Constant *llvm::ConstantFoldInstOperands(unsigned Opcode, const Type *DestTy, // If the input is a ptrtoint, turn the pair into a ptr to ptr bitcast if // the int size is >= the ptr size. This requires knowing the width of a // pointer, so it can't be done in ConstantExpr::getCast. - if (ConstantExpr *CE = dyn_cast<ConstantExpr>(Ops[0])) { + if (ConstantExpr *CE = dyn_cast<ConstantExpr>(Ops[0])) if (TD && - TD->getPointerSizeInBits() <= - CE->getType()->getScalarSizeInBits()) { - if (CE->getOpcode() == Instruction::PtrToInt) - return FoldBitCast(CE->getOperand(0), DestTy, *TD); - - // If there's a constant offset added to the integer value before - // it is casted back to a pointer, see if the expression can be - // converted into a GEP. - if (CE->getOpcode() == Instruction::Add) - if (ConstantInt *L = dyn_cast<ConstantInt>(CE->getOperand(0))) - if (ConstantExpr *R = dyn_cast<ConstantExpr>(CE->getOperand(1))) - if (R->getOpcode() == Instruction::PtrToInt) - if (GlobalVariable *GV = - dyn_cast<GlobalVariable>(R->getOperand(0))) { - const PointerType *GVTy = cast<PointerType>(GV->getType()); - if (const ArrayType *AT = - dyn_cast<ArrayType>(GVTy->getElementType())) { - const Type *ElTy = AT->getElementType(); - uint64_t AllocSize = TD->getTypeAllocSize(ElTy); - APInt PSA(L->getValue().getBitWidth(), AllocSize); - if (ElTy == cast<PointerType>(DestTy)->getElementType() && - L->getValue().urem(PSA) == 0) { - APInt ElemIdx = L->getValue().udiv(PSA); - if (ElemIdx.ult(APInt(ElemIdx.getBitWidth(), - AT->getNumElements()))) { - Constant *Index[] = { - Constant::getNullValue(CE->getType()), - ConstantInt::get(ElTy->getContext(), ElemIdx) - }; - return - ConstantExpr::getGetElementPtr(GV, &Index[0], 2); - } - } - } - } - } - } + TD->getPointerSizeInBits() <= CE->getType()->getScalarSizeInBits() && + CE->getOpcode() == Instruction::PtrToInt) + return FoldBitCast(CE->getOperand(0), DestTy, *TD); + return ConstantExpr::getCast(Opcode, Ops[0], DestTy); case Instruction::Trunc: case Instruction::ZExt: @@ -1179,6 +1146,12 @@ llvm::ConstantFoldCall(Function *F, return 0; } + if (isa<UndefValue>(Operands[0])) { + if (Name.startswith("llvm.bswap")) + return Operands[0]; + return 0; + } + return 0; } diff --git a/lib/Analysis/DebugInfo.cpp b/lib/Analysis/DebugInfo.cpp index 258f1db..5cfe666 100644 --- a/lib/Analysis/DebugInfo.cpp +++ b/lib/Analysis/DebugInfo.cpp @@ -1007,12 +1007,15 @@ DIVariable DIFactory::CreateComplexVariable(unsigned Tag, DIDescriptor Context, /// CreateBlock - This creates a descriptor for a lexical block with the /// specified parent VMContext. -DILexicalBlock DIFactory::CreateLexicalBlock(DIDescriptor Context) { +DILexicalBlock DIFactory::CreateLexicalBlock(DIDescriptor Context, + unsigned LineNo, unsigned Col) { Value *Elts[] = { GetTagConstant(dwarf::DW_TAG_lexical_block), - Context.getNode() + Context.getNode(), + ConstantInt::get(Type::getInt32Ty(VMContext), LineNo), + ConstantInt::get(Type::getInt32Ty(VMContext), Col) }; - return DILexicalBlock(MDNode::get(VMContext, &Elts[0], 2)); + return DILexicalBlock(MDNode::get(VMContext, &Elts[0], 4)); } /// CreateNameSpace - This creates new descriptor for a namespace diff --git a/lib/Analysis/IPA/Andersens.cpp b/lib/Analysis/IPA/Andersens.cpp deleted file mode 100644 index 4180206..0000000 --- a/lib/Analysis/IPA/Andersens.cpp +++ /dev/null @@ -1,2868 +0,0 @@ -//===- Andersens.cpp - Andersen's Interprocedural Alias Analysis ----------===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file defines an implementation of Andersen's interprocedural alias -// analysis -// -// In pointer analysis terms, this is a subset-based, flow-insensitive, -// field-sensitive, and context-insensitive algorithm pointer algorithm. -// -// This algorithm is implemented as three stages: -// 1. Object identification. -// 2. Inclusion constraint identification. -// 3. Offline constraint graph optimization -// 4. Inclusion constraint solving. -// -// The object identification stage identifies all of the memory objects in the -// program, which includes globals, heap allocated objects, and stack allocated -// objects. -// -// The inclusion constraint identification stage finds all inclusion constraints -// in the program by scanning the program, looking for pointer assignments and -// other statements that effect the points-to graph. For a statement like "A = -// B", this statement is processed to indicate that A can point to anything that -// B can point to. Constraints can handle copies, loads, and stores, and -// address taking. -// -// The offline constraint graph optimization portion includes offline variable -// substitution algorithms intended to compute pointer and location -// equivalences. Pointer equivalences are those pointers that will have the -// same points-to sets, and location equivalences are those variables that -// always appear together in points-to sets. It also includes an offline -// cycle detection algorithm that allows cycles to be collapsed sooner -// during solving. -// -// The inclusion constraint solving phase iteratively propagates the inclusion -// constraints until a fixed point is reached. This is an O(N^3) algorithm. -// -// Function constraints are handled as if they were structs with X fields. -// Thus, an access to argument X of function Y is an access to node index -// getNode(Y) + X. This representation allows handling of indirect calls -// without any issues. To wit, an indirect call Y(a,b) is equivalent to -// *(Y + 1) = a, *(Y + 2) = b. -// The return node for a function is always located at getNode(F) + -// CallReturnPos. The arguments start at getNode(F) + CallArgPos. -// -// Future Improvements: -// Use of BDD's. -//===----------------------------------------------------------------------===// - -#define DEBUG_TYPE "anders-aa" -#include "llvm/Constants.h" -#include "llvm/DerivedTypes.h" -#include "llvm/Instructions.h" -#include "llvm/Module.h" -#include "llvm/Pass.h" -#include "llvm/Support/ErrorHandling.h" -#include "llvm/Support/InstIterator.h" -#include "llvm/Support/InstVisitor.h" -#include "llvm/Analysis/AliasAnalysis.h" -#include "llvm/Analysis/MemoryBuiltins.h" -#include "llvm/Analysis/Passes.h" -#include "llvm/Support/Debug.h" -#include "llvm/System/Atomic.h" -#include "llvm/ADT/Statistic.h" -#include "llvm/ADT/SparseBitVector.h" -#include "llvm/ADT/DenseSet.h" -#include <algorithm> -#include <set> -#include <list> -#include <map> -#include <stack> -#include <vector> -#include <queue> - -// Determining the actual set of nodes the universal set can consist of is very -// expensive because it means propagating around very large sets. We rely on -// other analysis being able to determine which nodes can never be pointed to in -// order to disambiguate further than "points-to anything". -#define FULL_UNIVERSAL 0 - -using namespace llvm; -#ifndef NDEBUG -STATISTIC(NumIters , "Number of iterations to reach convergence"); -#endif -STATISTIC(NumConstraints, "Number of constraints"); -STATISTIC(NumNodes , "Number of nodes"); -STATISTIC(NumUnified , "Number of variables unified"); -STATISTIC(NumErased , "Number of redundant constraints erased"); - -static const unsigned SelfRep = (unsigned)-1; -static const unsigned Unvisited = (unsigned)-1; -// Position of the function return node relative to the function node. -static const unsigned CallReturnPos = 1; -// Position of the function call node relative to the function node. -static const unsigned CallFirstArgPos = 2; - -namespace { - struct BitmapKeyInfo { - static inline SparseBitVector<> *getEmptyKey() { - return reinterpret_cast<SparseBitVector<> *>(-1); - } - static inline SparseBitVector<> *getTombstoneKey() { - return reinterpret_cast<SparseBitVector<> *>(-2); - } - static unsigned getHashValue(const SparseBitVector<> *bitmap) { - return bitmap->getHashValue(); - } - static bool isEqual(const SparseBitVector<> *LHS, - const SparseBitVector<> *RHS) { - if (LHS == RHS) - return true; - else if (LHS == getEmptyKey() || RHS == getEmptyKey() - || LHS == getTombstoneKey() || RHS == getTombstoneKey()) - return false; - - return *LHS == *RHS; - } - }; - - class Andersens : public ModulePass, public AliasAnalysis, - private InstVisitor<Andersens> { - struct Node; - - /// Constraint - Objects of this structure are used to represent the various - /// constraints identified by the algorithm. The constraints are 'copy', - /// for statements like "A = B", 'load' for statements like "A = *B", - /// 'store' for statements like "*A = B", and AddressOf for statements like - /// A = alloca; The Offset is applied as *(A + K) = B for stores, - /// A = *(B + K) for loads, and A = B + K for copies. It is - /// illegal on addressof constraints (because it is statically - /// resolvable to A = &C where C = B + K) - - struct Constraint { - enum ConstraintType { Copy, Load, Store, AddressOf } Type; - unsigned Dest; - unsigned Src; - unsigned Offset; - - Constraint(ConstraintType Ty, unsigned D, unsigned S, unsigned O = 0) - : Type(Ty), Dest(D), Src(S), Offset(O) { - assert((Offset == 0 || Ty != AddressOf) && - "Offset is illegal on addressof constraints"); - } - - bool operator==(const Constraint &RHS) const { - return RHS.Type == Type - && RHS.Dest == Dest - && RHS.Src == Src - && RHS.Offset == Offset; - } - - bool operator!=(const Constraint &RHS) const { - return !(*this == RHS); - } - - bool operator<(const Constraint &RHS) const { - if (RHS.Type != Type) - return RHS.Type < Type; - else if (RHS.Dest != Dest) - return RHS.Dest < Dest; - else if (RHS.Src != Src) - return RHS.Src < Src; - return RHS.Offset < Offset; - } - }; - - // Information DenseSet requires implemented in order to be able to do - // it's thing - struct PairKeyInfo { - static inline std::pair<unsigned, unsigned> getEmptyKey() { - return std::make_pair(~0U, ~0U); - } - static inline std::pair<unsigned, unsigned> getTombstoneKey() { - return std::make_pair(~0U - 1, ~0U - 1); - } - static unsigned getHashValue(const std::pair<unsigned, unsigned> &P) { - return P.first ^ P.second; - } - static unsigned isEqual(const std::pair<unsigned, unsigned> &LHS, - const std::pair<unsigned, unsigned> &RHS) { - return LHS == RHS; - } - }; - - struct ConstraintKeyInfo { - static inline Constraint getEmptyKey() { - return Constraint(Constraint::Copy, ~0U, ~0U, ~0U); - } - static inline Constraint getTombstoneKey() { - return Constraint(Constraint::Copy, ~0U - 1, ~0U - 1, ~0U - 1); - } - static unsigned getHashValue(const Constraint &C) { - return C.Src ^ C.Dest ^ C.Type ^ C.Offset; - } - static bool isEqual(const Constraint &LHS, - const Constraint &RHS) { - return LHS.Type == RHS.Type && LHS.Dest == RHS.Dest - && LHS.Src == RHS.Src && LHS.Offset == RHS.Offset; - } - }; - - // Node class - This class is used to represent a node in the constraint - // graph. Due to various optimizations, it is not always the case that - // there is a mapping from a Node to a Value. In particular, we add - // artificial Node's that represent the set of pointed-to variables shared - // for each location equivalent Node. - struct Node { - private: - static volatile sys::cas_flag Counter; - - public: - Value *Val; - SparseBitVector<> *Edges; - SparseBitVector<> *PointsTo; - SparseBitVector<> *OldPointsTo; - std::list<Constraint> Constraints; - - // Pointer and location equivalence labels - unsigned PointerEquivLabel; - unsigned LocationEquivLabel; - // Predecessor edges, both real and implicit - SparseBitVector<> *PredEdges; - SparseBitVector<> *ImplicitPredEdges; - // Set of nodes that point to us, only use for location equivalence. - SparseBitVector<> *PointedToBy; - // Number of incoming edges, used during variable substitution to early - // free the points-to sets - unsigned NumInEdges; - // True if our points-to set is in the Set2PEClass map - bool StoredInHash; - // True if our node has no indirect constraints (complex or otherwise) - bool Direct; - // True if the node is address taken, *or* it is part of a group of nodes - // that must be kept together. This is set to true for functions and - // their arg nodes, which must be kept at the same position relative to - // their base function node. - bool AddressTaken; - - // Nodes in cycles (or in equivalence classes) are united together using a - // standard union-find representation with path compression. NodeRep - // gives the index into GraphNodes for the representative Node. - unsigned NodeRep; - - // Modification timestamp. Assigned from Counter. - // Used for work list prioritization. - unsigned Timestamp; - - explicit Node(bool direct = true) : - Val(0), Edges(0), PointsTo(0), OldPointsTo(0), - PointerEquivLabel(0), LocationEquivLabel(0), PredEdges(0), - ImplicitPredEdges(0), PointedToBy(0), NumInEdges(0), - StoredInHash(false), Direct(direct), AddressTaken(false), - NodeRep(SelfRep), Timestamp(0) { } - - Node *setValue(Value *V) { - assert(Val == 0 && "Value already set for this node!"); - Val = V; - return this; - } - - /// getValue - Return the LLVM value corresponding to this node. - /// - Value *getValue() const { return Val; } - - /// addPointerTo - Add a pointer to the list of pointees of this node, - /// returning true if this caused a new pointer to be added, or false if - /// we already knew about the points-to relation. - bool addPointerTo(unsigned Node) { - return PointsTo->test_and_set(Node); - } - - /// intersects - Return true if the points-to set of this node intersects - /// with the points-to set of the specified node. - bool intersects(Node *N) const; - - /// intersectsIgnoring - Return true if the points-to set of this node - /// intersects with the points-to set of the specified node on any nodes - /// except for the specified node to ignore. - bool intersectsIgnoring(Node *N, unsigned) const; - - // Timestamp a node (used for work list prioritization) - void Stamp() { - Timestamp = sys::AtomicIncrement(&Counter); - --Timestamp; - } - - bool isRep() const { - return( (int) NodeRep < 0 ); - } - }; - - struct WorkListElement { - Node* node; - unsigned Timestamp; - WorkListElement(Node* n, unsigned t) : node(n), Timestamp(t) {} - - // Note that we reverse the sense of the comparison because we - // actually want to give low timestamps the priority over high, - // whereas priority is typically interpreted as a greater value is - // given high priority. - bool operator<(const WorkListElement& that) const { - return( this->Timestamp > that.Timestamp ); - } - }; - - // Priority-queue based work list specialized for Nodes. - class WorkList { - std::priority_queue<WorkListElement> Q; - - public: - void insert(Node* n) { - Q.push( WorkListElement(n, n->Timestamp) ); - } - - // We automatically discard non-representative nodes and nodes - // that were in the work list twice (we keep a copy of the - // timestamp in the work list so we can detect this situation by - // comparing against the node's current timestamp). - Node* pop() { - while( !Q.empty() ) { - WorkListElement x = Q.top(); Q.pop(); - Node* INode = x.node; - - if( INode->isRep() && - INode->Timestamp == x.Timestamp ) { - return(x.node); - } - } - return(0); - } - - bool empty() { - return Q.empty(); - } - }; - - /// GraphNodes - This vector is populated as part of the object - /// identification stage of the analysis, which populates this vector with a - /// node for each memory object and fills in the ValueNodes map. - std::vector<Node> GraphNodes; - - /// ValueNodes - This map indicates the Node that a particular Value* is - /// represented by. This contains entries for all pointers. - DenseMap<Value*, unsigned> ValueNodes; - - /// ObjectNodes - This map contains entries for each memory object in the - /// program: globals, alloca's and mallocs. - DenseMap<Value*, unsigned> ObjectNodes; - - /// ReturnNodes - This map contains an entry for each function in the - /// program that returns a value. - DenseMap<Function*, unsigned> ReturnNodes; - - /// VarargNodes - This map contains the entry used to represent all pointers - /// passed through the varargs portion of a function call for a particular - /// function. An entry is not present in this map for functions that do not - /// take variable arguments. - DenseMap<Function*, unsigned> VarargNodes; - - - /// Constraints - This vector contains a list of all of the constraints - /// identified by the program. - std::vector<Constraint> Constraints; - - // Map from graph node to maximum K value that is allowed (for functions, - // this is equivalent to the number of arguments + CallFirstArgPos) - std::map<unsigned, unsigned> MaxK; - - /// This enum defines the GraphNodes indices that correspond to important - /// fixed sets. - enum { - UniversalSet = 0, - NullPtr = 1, - NullObject = 2, - NumberSpecialNodes - }; - // Stack for Tarjan's - std::stack<unsigned> SCCStack; - // Map from Graph Node to DFS number - std::vector<unsigned> Node2DFS; - // Map from Graph Node to Deleted from graph. - std::vector<bool> Node2Deleted; - // Same as Node Maps, but implemented as std::map because it is faster to - // clear - std::map<unsigned, unsigned> Tarjan2DFS; - std::map<unsigned, bool> Tarjan2Deleted; - // Current DFS number - unsigned DFSNumber; - - // Work lists. - WorkList w1, w2; - WorkList *CurrWL, *NextWL; // "current" and "next" work lists - - // Offline variable substitution related things - - // Temporary rep storage, used because we can't collapse SCC's in the - // predecessor graph by uniting the variables permanently, we can only do so - // for the successor graph. - std::vector<unsigned> VSSCCRep; - // Mapping from node to whether we have visited it during SCC finding yet. - std::vector<bool> Node2Visited; - // During variable substitution, we create unknowns to represent the unknown - // value that is a dereference of a variable. These nodes are known as - // "ref" nodes (since they represent the value of dereferences). - unsigned FirstRefNode; - // During HVN, we create represent address taken nodes as if they were - // unknown (since HVN, unlike HU, does not evaluate unions). - unsigned FirstAdrNode; - // Current pointer equivalence class number - unsigned PEClass; - // Mapping from points-to sets to equivalence classes - typedef DenseMap<SparseBitVector<> *, unsigned, BitmapKeyInfo> BitVectorMap; - BitVectorMap Set2PEClass; - // Mapping from pointer equivalences to the representative node. -1 if we - // have no representative node for this pointer equivalence class yet. - std::vector<int> PEClass2Node; - // Mapping from pointer equivalences to representative node. This includes - // pointer equivalent but not location equivalent variables. -1 if we have - // no representative node for this pointer equivalence class yet. - std::vector<int> PENLEClass2Node; - // Union/Find for HCD - std::vector<unsigned> HCDSCCRep; - // HCD's offline-detected cycles; "Statically DeTected" - // -1 if not part of such a cycle, otherwise a representative node. - std::vector<int> SDT; - // Whether to use SDT (UniteNodes can use it during solving, but not before) - bool SDTActive; - - public: - static char ID; - Andersens() : ModulePass(&ID) {} - - bool runOnModule(Module &M) { - InitializeAliasAnalysis(this); - IdentifyObjects(M); - CollectConstraints(M); -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa-constraints" - DEBUG(PrintConstraints()); -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa" - SolveConstraints(); - DEBUG(PrintPointsToGraph()); - - // Free the constraints list, as we don't need it to respond to alias - // requests. - std::vector<Constraint>().swap(Constraints); - //These are needed for Print() (-analyze in opt) - //ObjectNodes.clear(); - //ReturnNodes.clear(); - //VarargNodes.clear(); - return false; - } - - void releaseMemory() { - // FIXME: Until we have transitively required passes working correctly, - // this cannot be enabled! Otherwise, using -count-aa with the pass - // causes memory to be freed too early. :( -#if 0 - // The memory objects and ValueNodes data structures at the only ones that - // are still live after construction. - std::vector<Node>().swap(GraphNodes); - ValueNodes.clear(); -#endif - } - - virtual void getAnalysisUsage(AnalysisUsage &AU) const { - AliasAnalysis::getAnalysisUsage(AU); - AU.setPreservesAll(); // Does not transform code - } - - /// getAdjustedAnalysisPointer - This method is used when a pass implements - /// an analysis interface through multiple inheritance. If needed, it - /// should override this to adjust the this pointer as needed for the - /// specified pass info. - virtual void *getAdjustedAnalysisPointer(const PassInfo *PI) { - if (PI->isPassID(&AliasAnalysis::ID)) - return (AliasAnalysis*)this; - return this; - } - - //------------------------------------------------ - // Implement the AliasAnalysis API - // - AliasResult alias(const Value *V1, unsigned V1Size, - const Value *V2, unsigned V2Size); - virtual ModRefResult getModRefInfo(CallSite CS, Value *P, unsigned Size); - virtual ModRefResult getModRefInfo(CallSite CS1, CallSite CS2); - bool pointsToConstantMemory(const Value *P); - - virtual void deleteValue(Value *V) { - ValueNodes.erase(V); - getAnalysis<AliasAnalysis>().deleteValue(V); - } - - virtual void copyValue(Value *From, Value *To) { - ValueNodes[To] = ValueNodes[From]; - getAnalysis<AliasAnalysis>().copyValue(From, To); - } - - private: - /// getNode - Return the node corresponding to the specified pointer scalar. - /// - unsigned getNode(Value *V) { - if (Constant *C = dyn_cast<Constant>(V)) - if (!isa<GlobalValue>(C)) - return getNodeForConstantPointer(C); - - DenseMap<Value*, unsigned>::iterator I = ValueNodes.find(V); - if (I == ValueNodes.end()) { -#ifndef NDEBUG - V->dump(); -#endif - llvm_unreachable("Value does not have a node in the points-to graph!"); - } - return I->second; - } - - /// getObject - Return the node corresponding to the memory object for the - /// specified global or allocation instruction. - unsigned getObject(Value *V) const { - DenseMap<Value*, unsigned>::const_iterator I = ObjectNodes.find(V); - assert(I != ObjectNodes.end() && - "Value does not have an object in the points-to graph!"); - return I->second; - } - - /// getReturnNode - Return the node representing the return value for the - /// specified function. - unsigned getReturnNode(Function *F) const { - DenseMap<Function*, unsigned>::const_iterator I = ReturnNodes.find(F); - assert(I != ReturnNodes.end() && "Function does not return a value!"); - return I->second; - } - - /// getVarargNode - Return the node representing the variable arguments - /// formal for the specified function. - unsigned getVarargNode(Function *F) const { - DenseMap<Function*, unsigned>::const_iterator I = VarargNodes.find(F); - assert(I != VarargNodes.end() && "Function does not take var args!"); - return I->second; - } - - /// getNodeValue - Get the node for the specified LLVM value and set the - /// value for it to be the specified value. - unsigned getNodeValue(Value &V) { - unsigned Index = getNode(&V); - GraphNodes[Index].setValue(&V); - return Index; - } - - unsigned UniteNodes(unsigned First, unsigned Second, - bool UnionByRank = true); - unsigned FindNode(unsigned Node); - unsigned FindNode(unsigned Node) const; - - void IdentifyObjects(Module &M); - void CollectConstraints(Module &M); - bool AnalyzeUsesOfFunction(Value *); - void CreateConstraintGraph(); - void OptimizeConstraints(); - unsigned FindEquivalentNode(unsigned, unsigned); - void ClumpAddressTaken(); - void RewriteConstraints(); - void HU(); - void HVN(); - void HCD(); - void Search(unsigned Node); - void UnitePointerEquivalences(); - void SolveConstraints(); - bool QueryNode(unsigned Node); - void Condense(unsigned Node); - void HUValNum(unsigned Node); - void HVNValNum(unsigned Node); - unsigned getNodeForConstantPointer(Constant *C); - unsigned getNodeForConstantPointerTarget(Constant *C); - void AddGlobalInitializerConstraints(unsigned, Constant *C); - - void AddConstraintsForNonInternalLinkage(Function *F); - void AddConstraintsForCall(CallSite CS, Function *F); - bool AddConstraintsForExternalCall(CallSite CS, Function *F); - - - void PrintNode(const Node *N) const; - void PrintConstraints() const ; - void PrintConstraint(const Constraint &) const; - void PrintLabels() const; - void PrintPointsToGraph() const; - - //===------------------------------------------------------------------===// - // Instruction visitation methods for adding constraints - // - friend class InstVisitor<Andersens>; - void visitReturnInst(ReturnInst &RI); - void visitInvokeInst(InvokeInst &II) { visitCallSite(CallSite(&II)); } - void visitCallInst(CallInst &CI) { - if (isMalloc(&CI)) visitAlloc(CI); - else visitCallSite(CallSite(&CI)); - } - void visitCallSite(CallSite CS); - void visitAllocaInst(AllocaInst &I); - void visitAlloc(Instruction &I); - void visitLoadInst(LoadInst &LI); - void visitStoreInst(StoreInst &SI); - void visitGetElementPtrInst(GetElementPtrInst &GEP); - void visitPHINode(PHINode &PN); - void visitCastInst(CastInst &CI); - void visitICmpInst(ICmpInst &ICI) {} // NOOP! - void visitFCmpInst(FCmpInst &ICI) {} // NOOP! - void visitSelectInst(SelectInst &SI); - void visitVAArg(VAArgInst &I); - void visitInstruction(Instruction &I); - - //===------------------------------------------------------------------===// - // Implement Analyize interface - // - void print(raw_ostream &O, const Module*) const { - PrintPointsToGraph(); - } - }; -} - -char Andersens::ID = 0; -static RegisterPass<Andersens> -X("anders-aa", "Andersen's Interprocedural Alias Analysis (experimental)", - false, true); -static RegisterAnalysisGroup<AliasAnalysis> Y(X); - -// Initialize Timestamp Counter (static). -volatile llvm::sys::cas_flag Andersens::Node::Counter = 0; - -ModulePass *llvm::createAndersensPass() { return new Andersens(); } - -//===----------------------------------------------------------------------===// -// AliasAnalysis Interface Implementation -//===----------------------------------------------------------------------===// - -AliasAnalysis::AliasResult Andersens::alias(const Value *V1, unsigned V1Size, - const Value *V2, unsigned V2Size) { - Node *N1 = &GraphNodes[FindNode(getNode(const_cast<Value*>(V1)))]; - Node *N2 = &GraphNodes[FindNode(getNode(const_cast<Value*>(V2)))]; - - // Check to see if the two pointers are known to not alias. They don't alias - // if their points-to sets do not intersect. - if (!N1->intersectsIgnoring(N2, NullObject)) - return NoAlias; - - return AliasAnalysis::alias(V1, V1Size, V2, V2Size); -} - -AliasAnalysis::ModRefResult -Andersens::getModRefInfo(CallSite CS, Value *P, unsigned Size) { - // The only thing useful that we can contribute for mod/ref information is - // when calling external function calls: if we know that memory never escapes - // from the program, it cannot be modified by an external call. - // - // NOTE: This is not really safe, at least not when the entire program is not - // available. The deal is that the external function could call back into the - // program and modify stuff. We ignore this technical niggle for now. This - // is, after all, a "research quality" implementation of Andersen's analysis. - if (Function *F = CS.getCalledFunction()) - if (F->isDeclaration()) { - Node *N1 = &GraphNodes[FindNode(getNode(P))]; - - if (N1->PointsTo->empty()) - return NoModRef; -#if FULL_UNIVERSAL - if (!UniversalSet->PointsTo->test(FindNode(getNode(P)))) - return NoModRef; // Universal set does not contain P -#else - if (!N1->PointsTo->test(UniversalSet)) - return NoModRef; // P doesn't point to the universal set. -#endif - } - - return AliasAnalysis::getModRefInfo(CS, P, Size); -} - -AliasAnalysis::ModRefResult -Andersens::getModRefInfo(CallSite CS1, CallSite CS2) { - return AliasAnalysis::getModRefInfo(CS1,CS2); -} - -/// pointsToConstantMemory - If we can determine that this pointer only points -/// to constant memory, return true. In practice, this means that if the -/// pointer can only point to constant globals, functions, or the null pointer, -/// return true. -/// -bool Andersens::pointsToConstantMemory(const Value *P) { - Node *N = &GraphNodes[FindNode(getNode(const_cast<Value*>(P)))]; - unsigned i; - - for (SparseBitVector<>::iterator bi = N->PointsTo->begin(); - bi != N->PointsTo->end(); - ++bi) { - i = *bi; - Node *Pointee = &GraphNodes[i]; - if (Value *V = Pointee->getValue()) { - if (!isa<GlobalValue>(V) || (isa<GlobalVariable>(V) && - !cast<GlobalVariable>(V)->isConstant())) - return AliasAnalysis::pointsToConstantMemory(P); - } else { - if (i != NullObject) - return AliasAnalysis::pointsToConstantMemory(P); - } - } - - return true; -} - -//===----------------------------------------------------------------------===// -// Object Identification Phase -//===----------------------------------------------------------------------===// - -/// IdentifyObjects - This stage scans the program, adding an entry to the -/// GraphNodes list for each memory object in the program (global stack or -/// heap), and populates the ValueNodes and ObjectNodes maps for these objects. -/// -void Andersens::IdentifyObjects(Module &M) { - unsigned NumObjects = 0; - - // Object #0 is always the universal set: the object that we don't know - // anything about. - assert(NumObjects == UniversalSet && "Something changed!"); - ++NumObjects; - - // Object #1 always represents the null pointer. - assert(NumObjects == NullPtr && "Something changed!"); - ++NumObjects; - - // Object #2 always represents the null object (the object pointed to by null) - assert(NumObjects == NullObject && "Something changed!"); - ++NumObjects; - - // Add all the globals first. - for (Module::global_iterator I = M.global_begin(), E = M.global_end(); - I != E; ++I) { - ObjectNodes[I] = NumObjects++; - ValueNodes[I] = NumObjects++; - } - - // Add nodes for all of the functions and the instructions inside of them. - for (Module::iterator F = M.begin(), E = M.end(); F != E; ++F) { - // The function itself is a memory object. - unsigned First = NumObjects; - ValueNodes[F] = NumObjects++; - if (isa<PointerType>(F->getFunctionType()->getReturnType())) - ReturnNodes[F] = NumObjects++; - if (F->getFunctionType()->isVarArg()) - VarargNodes[F] = NumObjects++; - - - // Add nodes for all of the incoming pointer arguments. - for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end(); - I != E; ++I) - { - if (isa<PointerType>(I->getType())) - ValueNodes[I] = NumObjects++; - } - MaxK[First] = NumObjects - First; - - // Scan the function body, creating a memory object for each heap/stack - // allocation in the body of the function and a node to represent all - // pointer values defined by instructions and used as operands. - for (inst_iterator II = inst_begin(F), E = inst_end(F); II != E; ++II) { - // If this is an heap or stack allocation, create a node for the memory - // object. - if (isa<PointerType>(II->getType())) { - ValueNodes[&*II] = NumObjects++; - if (AllocaInst *AI = dyn_cast<AllocaInst>(&*II)) - ObjectNodes[AI] = NumObjects++; - else if (isMalloc(&*II)) - ObjectNodes[&*II] = NumObjects++; - } - - // Calls to inline asm need to be added as well because the callee isn't - // referenced anywhere else. - if (CallInst *CI = dyn_cast<CallInst>(&*II)) { - Value *Callee = CI->getCalledValue(); - if (isa<InlineAsm>(Callee)) - ValueNodes[Callee] = NumObjects++; - } - } - } - - // Now that we know how many objects to create, make them all now! - GraphNodes.resize(NumObjects); - NumNodes += NumObjects; -} - -//===----------------------------------------------------------------------===// -// Constraint Identification Phase -//===----------------------------------------------------------------------===// - -/// getNodeForConstantPointer - Return the node corresponding to the constant -/// pointer itself. -unsigned Andersens::getNodeForConstantPointer(Constant *C) { - assert(isa<PointerType>(C->getType()) && "Not a constant pointer!"); - - if (isa<ConstantPointerNull>(C) || isa<UndefValue>(C)) - return NullPtr; - else if (GlobalValue *GV = dyn_cast<GlobalValue>(C)) - return getNode(GV); - else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(C)) { - switch (CE->getOpcode()) { - case Instruction::GetElementPtr: - return getNodeForConstantPointer(CE->getOperand(0)); - case Instruction::IntToPtr: - return UniversalSet; - case Instruction::BitCast: - return getNodeForConstantPointer(CE->getOperand(0)); - default: - errs() << "Constant Expr not yet handled: " << *CE << "\n"; - llvm_unreachable(0); - } - } else { - llvm_unreachable("Unknown constant pointer!"); - } - return 0; -} - -/// getNodeForConstantPointerTarget - Return the node POINTED TO by the -/// specified constant pointer. -unsigned Andersens::getNodeForConstantPointerTarget(Constant *C) { - assert(isa<PointerType>(C->getType()) && "Not a constant pointer!"); - - if (isa<ConstantPointerNull>(C)) - return NullObject; - else if (GlobalValue *GV = dyn_cast<GlobalValue>(C)) - return getObject(GV); - else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(C)) { - switch (CE->getOpcode()) { - case Instruction::GetElementPtr: - return getNodeForConstantPointerTarget(CE->getOperand(0)); - case Instruction::IntToPtr: - return UniversalSet; - case Instruction::BitCast: - return getNodeForConstantPointerTarget(CE->getOperand(0)); - default: - errs() << "Constant Expr not yet handled: " << *CE << "\n"; - llvm_unreachable(0); - } - } else { - llvm_unreachable("Unknown constant pointer!"); - } - return 0; -} - -/// AddGlobalInitializerConstraints - Add inclusion constraints for the memory -/// object N, which contains values indicated by C. -void Andersens::AddGlobalInitializerConstraints(unsigned NodeIndex, - Constant *C) { - if (C->getType()->isSingleValueType()) { - if (isa<PointerType>(C->getType())) - Constraints.push_back(Constraint(Constraint::Copy, NodeIndex, - getNodeForConstantPointer(C))); - } else if (C->isNullValue()) { - Constraints.push_back(Constraint(Constraint::Copy, NodeIndex, - NullObject)); - return; - } else if (!isa<UndefValue>(C)) { - // If this is an array or struct, include constraints for each element. - assert(isa<ConstantArray>(C) || isa<ConstantStruct>(C)); - for (unsigned i = 0, e = C->getNumOperands(); i != e; ++i) - AddGlobalInitializerConstraints(NodeIndex, - cast<Constant>(C->getOperand(i))); - } -} - -/// AddConstraintsForNonInternalLinkage - If this function does not have -/// internal linkage, realize that we can't trust anything passed into or -/// returned by this function. -void Andersens::AddConstraintsForNonInternalLinkage(Function *F) { - for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end(); I != E; ++I) - if (isa<PointerType>(I->getType())) - // If this is an argument of an externally accessible function, the - // incoming pointer might point to anything. - Constraints.push_back(Constraint(Constraint::Copy, getNode(I), - UniversalSet)); -} - -/// AddConstraintsForCall - If this is a call to a "known" function, add the -/// constraints and return true. If this is a call to an unknown function, -/// return false. -bool Andersens::AddConstraintsForExternalCall(CallSite CS, Function *F) { - assert(F->isDeclaration() && "Not an external function!"); - - // These functions don't induce any points-to constraints. - if (F->getName() == "atoi" || F->getName() == "atof" || - F->getName() == "atol" || F->getName() == "atoll" || - F->getName() == "remove" || F->getName() == "unlink" || - F->getName() == "rename" || F->getName() == "memcmp" || - F->getName() == "llvm.memset" || - F->getName() == "strcmp" || F->getName() == "strncmp" || - F->getName() == "execl" || F->getName() == "execlp" || - F->getName() == "execle" || F->getName() == "execv" || - F->getName() == "execvp" || F->getName() == "chmod" || - F->getName() == "puts" || F->getName() == "write" || - F->getName() == "open" || F->getName() == "create" || - F->getName() == "truncate" || F->getName() == "chdir" || - F->getName() == "mkdir" || F->getName() == "rmdir" || - F->getName() == "read" || F->getName() == "pipe" || - F->getName() == "wait" || F->getName() == "time" || - F->getName() == "stat" || F->getName() == "fstat" || - F->getName() == "lstat" || F->getName() == "strtod" || - F->getName() == "strtof" || F->getName() == "strtold" || - F->getName() == "fopen" || F->getName() == "fdopen" || - F->getName() == "freopen" || - F->getName() == "fflush" || F->getName() == "feof" || - F->getName() == "fileno" || F->getName() == "clearerr" || - F->getName() == "rewind" || F->getName() == "ftell" || - F->getName() == "ferror" || F->getName() == "fgetc" || - F->getName() == "fgetc" || F->getName() == "_IO_getc" || - F->getName() == "fwrite" || F->getName() == "fread" || - F->getName() == "fgets" || F->getName() == "ungetc" || - F->getName() == "fputc" || - F->getName() == "fputs" || F->getName() == "putc" || - F->getName() == "ftell" || F->getName() == "rewind" || - F->getName() == "_IO_putc" || F->getName() == "fseek" || - F->getName() == "fgetpos" || F->getName() == "fsetpos" || - F->getName() == "printf" || F->getName() == "fprintf" || - F->getName() == "sprintf" || F->getName() == "vprintf" || - F->getName() == "vfprintf" || F->getName() == "vsprintf" || - F->getName() == "scanf" || F->getName() == "fscanf" || - F->getName() == "sscanf" || F->getName() == "__assert_fail" || - F->getName() == "modf") - return true; - - - // These functions do induce points-to edges. - if (F->getName() == "llvm.memcpy" || - F->getName() == "llvm.memmove" || - F->getName() == "memmove") { - - const FunctionType *FTy = F->getFunctionType(); - if (FTy->getNumParams() > 1 && - isa<PointerType>(FTy->getParamType(0)) && - isa<PointerType>(FTy->getParamType(1))) { - - // *Dest = *Src, which requires an artificial graph node to represent the - // constraint. It is broken up into *Dest = temp, temp = *Src - unsigned FirstArg = getNode(CS.getArgument(0)); - unsigned SecondArg = getNode(CS.getArgument(1)); - unsigned TempArg = GraphNodes.size(); - GraphNodes.push_back(Node()); - Constraints.push_back(Constraint(Constraint::Store, - FirstArg, TempArg)); - Constraints.push_back(Constraint(Constraint::Load, - TempArg, SecondArg)); - // In addition, Dest = Src - Constraints.push_back(Constraint(Constraint::Copy, - FirstArg, SecondArg)); - return true; - } - } - - // Result = Arg0 - if (F->getName() == "realloc" || F->getName() == "strchr" || - F->getName() == "strrchr" || F->getName() == "strstr" || - F->getName() == "strtok") { - const FunctionType *FTy = F->getFunctionType(); - if (FTy->getNumParams() > 0 && - isa<PointerType>(FTy->getParamType(0))) { - Constraints.push_back(Constraint(Constraint::Copy, - getNode(CS.getInstruction()), - getNode(CS.getArgument(0)))); - return true; - } - } - - return false; -} - - - -/// AnalyzeUsesOfFunction - Look at all of the users of the specified function. -/// If this is used by anything complex (i.e., the address escapes), return -/// true. -bool Andersens::AnalyzeUsesOfFunction(Value *V) { - - if (!isa<PointerType>(V->getType())) return true; - - for (Value::use_iterator UI = V->use_begin(), E = V->use_end(); UI != E; ++UI) - if (isa<LoadInst>(*UI)) { - return false; - } else if (StoreInst *SI = dyn_cast<StoreInst>(*UI)) { - if (V == SI->getOperand(1)) { - return false; - } else if (SI->getOperand(1)) { - return true; // Storing the pointer - } - } else if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(*UI)) { - if (AnalyzeUsesOfFunction(GEP)) return true; - } else if (isFreeCall(*UI)) { - return false; - } else if (CallInst *CI = dyn_cast<CallInst>(*UI)) { - // Make sure that this is just the function being called, not that it is - // passing into the function. - for (unsigned i = 1, e = CI->getNumOperands(); i != e; ++i) - if (CI->getOperand(i) == V) return true; - } else if (InvokeInst *II = dyn_cast<InvokeInst>(*UI)) { - // Make sure that this is just the function being called, not that it is - // passing into the function. - for (unsigned i = 3, e = II->getNumOperands(); i != e; ++i) - if (II->getOperand(i) == V) return true; - } else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(*UI)) { - if (CE->getOpcode() == Instruction::GetElementPtr || - CE->getOpcode() == Instruction::BitCast) { - if (AnalyzeUsesOfFunction(CE)) - return true; - } else { - return true; - } - } else if (ICmpInst *ICI = dyn_cast<ICmpInst>(*UI)) { - if (!isa<ConstantPointerNull>(ICI->getOperand(1))) - return true; // Allow comparison against null. - } else { - return true; - } - return false; -} - -/// CollectConstraints - This stage scans the program, adding a constraint to -/// the Constraints list for each instruction in the program that induces a -/// constraint, and setting up the initial points-to graph. -/// -void Andersens::CollectConstraints(Module &M) { - // First, the universal set points to itself. - Constraints.push_back(Constraint(Constraint::AddressOf, UniversalSet, - UniversalSet)); - Constraints.push_back(Constraint(Constraint::Store, UniversalSet, - UniversalSet)); - - // Next, the null pointer points to the null object. - Constraints.push_back(Constraint(Constraint::AddressOf, NullPtr, NullObject)); - - // Next, add any constraints on global variables and their initializers. - for (Module::global_iterator I = M.global_begin(), E = M.global_end(); - I != E; ++I) { - // Associate the address of the global object as pointing to the memory for - // the global: &G = <G memory> - unsigned ObjectIndex = getObject(I); - Node *Object = &GraphNodes[ObjectIndex]; - Object->setValue(I); - Constraints.push_back(Constraint(Constraint::AddressOf, getNodeValue(*I), - ObjectIndex)); - - if (I->hasDefinitiveInitializer()) { - AddGlobalInitializerConstraints(ObjectIndex, I->getInitializer()); - } else { - // If it doesn't have an initializer (i.e. it's defined in another - // translation unit), it points to the universal set. - Constraints.push_back(Constraint(Constraint::Copy, ObjectIndex, - UniversalSet)); - } - } - - for (Module::iterator F = M.begin(), E = M.end(); F != E; ++F) { - // Set up the return value node. - if (isa<PointerType>(F->getFunctionType()->getReturnType())) - GraphNodes[getReturnNode(F)].setValue(F); - if (F->getFunctionType()->isVarArg()) - GraphNodes[getVarargNode(F)].setValue(F); - - // Set up incoming argument nodes. - for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end(); - I != E; ++I) - if (isa<PointerType>(I->getType())) - getNodeValue(*I); - - // At some point we should just add constraints for the escaping functions - // at solve time, but this slows down solving. For now, we simply mark - // address taken functions as escaping and treat them as external. - if (!F->hasLocalLinkage() || AnalyzeUsesOfFunction(F)) - AddConstraintsForNonInternalLinkage(F); - - if (!F->isDeclaration()) { - // Scan the function body, creating a memory object for each heap/stack - // allocation in the body of the function and a node to represent all - // pointer values defined by instructions and used as operands. - visit(F); - } else { - // External functions that return pointers return the universal set. - if (isa<PointerType>(F->getFunctionType()->getReturnType())) - Constraints.push_back(Constraint(Constraint::Copy, - getReturnNode(F), - UniversalSet)); - - // Any pointers that are passed into the function have the universal set - // stored into them. - for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end(); - I != E; ++I) - if (isa<PointerType>(I->getType())) { - // Pointers passed into external functions could have anything stored - // through them. - Constraints.push_back(Constraint(Constraint::Store, getNode(I), - UniversalSet)); - // Memory objects passed into external function calls can have the - // universal set point to them. -#if FULL_UNIVERSAL - Constraints.push_back(Constraint(Constraint::Copy, - UniversalSet, - getNode(I))); -#else - Constraints.push_back(Constraint(Constraint::Copy, - getNode(I), - UniversalSet)); -#endif - } - - // If this is an external varargs function, it can also store pointers - // into any pointers passed through the varargs section. - if (F->getFunctionType()->isVarArg()) - Constraints.push_back(Constraint(Constraint::Store, getVarargNode(F), - UniversalSet)); - } - } - NumConstraints += Constraints.size(); -} - - -void Andersens::visitInstruction(Instruction &I) { -#ifdef NDEBUG - return; // This function is just a big assert. -#endif - if (isa<BinaryOperator>(I)) - return; - // Most instructions don't have any effect on pointer values. - switch (I.getOpcode()) { - case Instruction::Br: - case Instruction::Switch: - case Instruction::Unwind: - case Instruction::Unreachable: - case Instruction::ICmp: - case Instruction::FCmp: - return; - default: - // Is this something we aren't handling yet? - errs() << "Unknown instruction: " << I; - llvm_unreachable(0); - } -} - -void Andersens::visitAllocaInst(AllocaInst &I) { - visitAlloc(I); -} - -void Andersens::visitAlloc(Instruction &I) { - unsigned ObjectIndex = getObject(&I); - GraphNodes[ObjectIndex].setValue(&I); - Constraints.push_back(Constraint(Constraint::AddressOf, getNodeValue(I), - ObjectIndex)); -} - -void Andersens::visitReturnInst(ReturnInst &RI) { - if (RI.getNumOperands() && isa<PointerType>(RI.getOperand(0)->getType())) - // return V --> <Copy/retval{F}/v> - Constraints.push_back(Constraint(Constraint::Copy, - getReturnNode(RI.getParent()->getParent()), - getNode(RI.getOperand(0)))); -} - -void Andersens::visitLoadInst(LoadInst &LI) { - if (isa<PointerType>(LI.getType())) - // P1 = load P2 --> <Load/P1/P2> - Constraints.push_back(Constraint(Constraint::Load, getNodeValue(LI), - getNode(LI.getOperand(0)))); -} - -void Andersens::visitStoreInst(StoreInst &SI) { - if (isa<PointerType>(SI.getOperand(0)->getType())) - // store P1, P2 --> <Store/P2/P1> - Constraints.push_back(Constraint(Constraint::Store, - getNode(SI.getOperand(1)), - getNode(SI.getOperand(0)))); -} - -void Andersens::visitGetElementPtrInst(GetElementPtrInst &GEP) { - // P1 = getelementptr P2, ... --> <Copy/P1/P2> - Constraints.push_back(Constraint(Constraint::Copy, getNodeValue(GEP), - getNode(GEP.getOperand(0)))); -} - -void Andersens::visitPHINode(PHINode &PN) { - if (isa<PointerType>(PN.getType())) { - unsigned PNN = getNodeValue(PN); - for (unsigned i = 0, e = PN.getNumIncomingValues(); i != e; ++i) - // P1 = phi P2, P3 --> <Copy/P1/P2>, <Copy/P1/P3>, ... - Constraints.push_back(Constraint(Constraint::Copy, PNN, - getNode(PN.getIncomingValue(i)))); - } -} - -void Andersens::visitCastInst(CastInst &CI) { - Value *Op = CI.getOperand(0); - if (isa<PointerType>(CI.getType())) { - if (isa<PointerType>(Op->getType())) { - // P1 = cast P2 --> <Copy/P1/P2> - Constraints.push_back(Constraint(Constraint::Copy, getNodeValue(CI), - getNode(CI.getOperand(0)))); - } else { - // P1 = cast int --> <Copy/P1/Univ> -#if 0 - Constraints.push_back(Constraint(Constraint::Copy, getNodeValue(CI), - UniversalSet)); -#else - getNodeValue(CI); -#endif - } - } else if (isa<PointerType>(Op->getType())) { - // int = cast P1 --> <Copy/Univ/P1> -#if 0 - Constraints.push_back(Constraint(Constraint::Copy, - UniversalSet, - getNode(CI.getOperand(0)))); -#else - getNode(CI.getOperand(0)); -#endif - } -} - -void Andersens::visitSelectInst(SelectInst &SI) { - if (isa<PointerType>(SI.getType())) { - unsigned SIN = getNodeValue(SI); - // P1 = select C, P2, P3 ---> <Copy/P1/P2>, <Copy/P1/P3> - Constraints.push_back(Constraint(Constraint::Copy, SIN, - getNode(SI.getOperand(1)))); - Constraints.push_back(Constraint(Constraint::Copy, SIN, - getNode(SI.getOperand(2)))); - } -} - -void Andersens::visitVAArg(VAArgInst &I) { - llvm_unreachable("vaarg not handled yet!"); -} - -/// AddConstraintsForCall - Add constraints for a call with actual arguments -/// specified by CS to the function specified by F. Note that the types of -/// arguments might not match up in the case where this is an indirect call and -/// the function pointer has been casted. If this is the case, do something -/// reasonable. -void Andersens::AddConstraintsForCall(CallSite CS, Function *F) { - Value *CallValue = CS.getCalledValue(); - bool IsDeref = F == NULL; - - // If this is a call to an external function, try to handle it directly to get - // some taste of context sensitivity. - if (F && F->isDeclaration() && AddConstraintsForExternalCall(CS, F)) - return; - - if (isa<PointerType>(CS.getType())) { - unsigned CSN = getNode(CS.getInstruction()); - if (!F || isa<PointerType>(F->getFunctionType()->getReturnType())) { - if (IsDeref) - Constraints.push_back(Constraint(Constraint::Load, CSN, - getNode(CallValue), CallReturnPos)); - else - Constraints.push_back(Constraint(Constraint::Copy, CSN, - getNode(CallValue) + CallReturnPos)); - } else { - // If the function returns a non-pointer value, handle this just like we - // treat a nonpointer cast to pointer. - Constraints.push_back(Constraint(Constraint::Copy, CSN, - UniversalSet)); - } - } else if (F && isa<PointerType>(F->getFunctionType()->getReturnType())) { -#if FULL_UNIVERSAL - Constraints.push_back(Constraint(Constraint::Copy, - UniversalSet, - getNode(CallValue) + CallReturnPos)); -#else - Constraints.push_back(Constraint(Constraint::Copy, - getNode(CallValue) + CallReturnPos, - UniversalSet)); -#endif - - - } - - CallSite::arg_iterator ArgI = CS.arg_begin(), ArgE = CS.arg_end(); - bool external = !F || F->isDeclaration(); - if (F) { - // Direct Call - Function::arg_iterator AI = F->arg_begin(), AE = F->arg_end(); - for (; AI != AE && ArgI != ArgE; ++AI, ++ArgI) - { -#if !FULL_UNIVERSAL - if (external && isa<PointerType>((*ArgI)->getType())) - { - // Add constraint that ArgI can now point to anything due to - // escaping, as can everything it points to. The second portion of - // this should be taken care of by universal = *universal - Constraints.push_back(Constraint(Constraint::Copy, - getNode(*ArgI), - UniversalSet)); - } -#endif - if (isa<PointerType>(AI->getType())) { - if (isa<PointerType>((*ArgI)->getType())) { - // Copy the actual argument into the formal argument. - Constraints.push_back(Constraint(Constraint::Copy, getNode(AI), - getNode(*ArgI))); - } else { - Constraints.push_back(Constraint(Constraint::Copy, getNode(AI), - UniversalSet)); - } - } else if (isa<PointerType>((*ArgI)->getType())) { -#if FULL_UNIVERSAL - Constraints.push_back(Constraint(Constraint::Copy, - UniversalSet, - getNode(*ArgI))); -#else - Constraints.push_back(Constraint(Constraint::Copy, - getNode(*ArgI), - UniversalSet)); -#endif - } - } - } else { - //Indirect Call - unsigned ArgPos = CallFirstArgPos; - for (; ArgI != ArgE; ++ArgI) { - if (isa<PointerType>((*ArgI)->getType())) { - // Copy the actual argument into the formal argument. - Constraints.push_back(Constraint(Constraint::Store, - getNode(CallValue), - getNode(*ArgI), ArgPos++)); - } else { - Constraints.push_back(Constraint(Constraint::Store, - getNode (CallValue), - UniversalSet, ArgPos++)); - } - } - } - // Copy all pointers passed through the varargs section to the varargs node. - if (F && F->getFunctionType()->isVarArg()) - for (; ArgI != ArgE; ++ArgI) - if (isa<PointerType>((*ArgI)->getType())) - Constraints.push_back(Constraint(Constraint::Copy, getVarargNode(F), - getNode(*ArgI))); - // If more arguments are passed in than we track, just drop them on the floor. -} - -void Andersens::visitCallSite(CallSite CS) { - if (isa<PointerType>(CS.getType())) - getNodeValue(*CS.getInstruction()); - - if (Function *F = CS.getCalledFunction()) { - AddConstraintsForCall(CS, F); - } else { - AddConstraintsForCall(CS, NULL); - } -} - -//===----------------------------------------------------------------------===// -// Constraint Solving Phase -//===----------------------------------------------------------------------===// - -/// intersects - Return true if the points-to set of this node intersects -/// with the points-to set of the specified node. -bool Andersens::Node::intersects(Node *N) const { - return PointsTo->intersects(N->PointsTo); -} - -/// intersectsIgnoring - Return true if the points-to set of this node -/// intersects with the points-to set of the specified node on any nodes -/// except for the specified node to ignore. -bool Andersens::Node::intersectsIgnoring(Node *N, unsigned Ignoring) const { - // TODO: If we are only going to call this with the same value for Ignoring, - // we should move the special values out of the points-to bitmap. - bool WeHadIt = PointsTo->test(Ignoring); - bool NHadIt = N->PointsTo->test(Ignoring); - bool Result = false; - if (WeHadIt) - PointsTo->reset(Ignoring); - if (NHadIt) - N->PointsTo->reset(Ignoring); - Result = PointsTo->intersects(N->PointsTo); - if (WeHadIt) - PointsTo->set(Ignoring); - if (NHadIt) - N->PointsTo->set(Ignoring); - return Result; -} - - -/// Clump together address taken variables so that the points-to sets use up -/// less space and can be operated on faster. - -void Andersens::ClumpAddressTaken() { -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa-renumber" - std::vector<unsigned> Translate; - std::vector<Node> NewGraphNodes; - - Translate.resize(GraphNodes.size()); - unsigned NewPos = 0; - - for (unsigned i = 0; i < Constraints.size(); ++i) { - Constraint &C = Constraints[i]; - if (C.Type == Constraint::AddressOf) { - GraphNodes[C.Src].AddressTaken = true; - } - } - for (unsigned i = 0; i < NumberSpecialNodes; ++i) { - unsigned Pos = NewPos++; - Translate[i] = Pos; - NewGraphNodes.push_back(GraphNodes[i]); - DEBUG(dbgs() << "Renumbering node " << i << " to node " << Pos << "\n"); - } - - // I believe this ends up being faster than making two vectors and splicing - // them. - for (unsigned i = NumberSpecialNodes; i < GraphNodes.size(); ++i) { - if (GraphNodes[i].AddressTaken) { - unsigned Pos = NewPos++; - Translate[i] = Pos; - NewGraphNodes.push_back(GraphNodes[i]); - DEBUG(dbgs() << "Renumbering node " << i << " to node " << Pos << "\n"); - } - } - - for (unsigned i = NumberSpecialNodes; i < GraphNodes.size(); ++i) { - if (!GraphNodes[i].AddressTaken) { - unsigned Pos = NewPos++; - Translate[i] = Pos; - NewGraphNodes.push_back(GraphNodes[i]); - DEBUG(dbgs() << "Renumbering node " << i << " to node " << Pos << "\n"); - } - } - - for (DenseMap<Value*, unsigned>::iterator Iter = ValueNodes.begin(); - Iter != ValueNodes.end(); - ++Iter) - Iter->second = Translate[Iter->second]; - - for (DenseMap<Value*, unsigned>::iterator Iter = ObjectNodes.begin(); - Iter != ObjectNodes.end(); - ++Iter) - Iter->second = Translate[Iter->second]; - - for (DenseMap<Function*, unsigned>::iterator Iter = ReturnNodes.begin(); - Iter != ReturnNodes.end(); - ++Iter) - Iter->second = Translate[Iter->second]; - - for (DenseMap<Function*, unsigned>::iterator Iter = VarargNodes.begin(); - Iter != VarargNodes.end(); - ++Iter) - Iter->second = Translate[Iter->second]; - - for (unsigned i = 0; i < Constraints.size(); ++i) { - Constraint &C = Constraints[i]; - C.Src = Translate[C.Src]; - C.Dest = Translate[C.Dest]; - } - - GraphNodes.swap(NewGraphNodes); -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa" -} - -/// The technique used here is described in "Exploiting Pointer and Location -/// Equivalence to Optimize Pointer Analysis. In the 14th International Static -/// Analysis Symposium (SAS), August 2007." It is known as the "HVN" algorithm, -/// and is equivalent to value numbering the collapsed constraint graph without -/// evaluating unions. This is used as a pre-pass to HU in order to resolve -/// first order pointer dereferences and speed up/reduce memory usage of HU. -/// Running both is equivalent to HRU without the iteration -/// HVN in more detail: -/// Imagine the set of constraints was simply straight line code with no loops -/// (we eliminate cycles, so there are no loops), such as: -/// E = &D -/// E = &C -/// E = F -/// F = G -/// G = F -/// Applying value numbering to this code tells us: -/// G == F == E -/// -/// For HVN, this is as far as it goes. We assign new value numbers to every -/// "address node", and every "reference node". -/// To get the optimal result for this, we use a DFS + SCC (since all nodes in a -/// cycle must have the same value number since the = operation is really -/// inclusion, not overwrite), and value number nodes we receive points-to sets -/// before we value our own node. -/// The advantage of HU over HVN is that HU considers the inclusion property, so -/// that if you have -/// E = &D -/// E = &C -/// E = F -/// F = G -/// F = &D -/// G = F -/// HU will determine that G == F == E. HVN will not, because it cannot prove -/// that the points to information ends up being the same because they all -/// receive &D from E anyway. - -void Andersens::HVN() { - DEBUG(dbgs() << "Beginning HVN\n"); - // Build a predecessor graph. This is like our constraint graph with the - // edges going in the opposite direction, and there are edges for all the - // constraints, instead of just copy constraints. We also build implicit - // edges for constraints are implied but not explicit. I.E for the constraint - // a = &b, we add implicit edges *a = b. This helps us capture more cycles - for (unsigned i = 0, e = Constraints.size(); i != e; ++i) { - Constraint &C = Constraints[i]; - if (C.Type == Constraint::AddressOf) { - GraphNodes[C.Src].AddressTaken = true; - GraphNodes[C.Src].Direct = false; - - // Dest = &src edge - unsigned AdrNode = C.Src + FirstAdrNode; - if (!GraphNodes[C.Dest].PredEdges) - GraphNodes[C.Dest].PredEdges = new SparseBitVector<>; - GraphNodes[C.Dest].PredEdges->set(AdrNode); - - // *Dest = src edge - unsigned RefNode = C.Dest + FirstRefNode; - if (!GraphNodes[RefNode].ImplicitPredEdges) - GraphNodes[RefNode].ImplicitPredEdges = new SparseBitVector<>; - GraphNodes[RefNode].ImplicitPredEdges->set(C.Src); - } else if (C.Type == Constraint::Load) { - if (C.Offset == 0) { - // dest = *src edge - if (!GraphNodes[C.Dest].PredEdges) - GraphNodes[C.Dest].PredEdges = new SparseBitVector<>; - GraphNodes[C.Dest].PredEdges->set(C.Src + FirstRefNode); - } else { - GraphNodes[C.Dest].Direct = false; - } - } else if (C.Type == Constraint::Store) { - if (C.Offset == 0) { - // *dest = src edge - unsigned RefNode = C.Dest + FirstRefNode; - if (!GraphNodes[RefNode].PredEdges) - GraphNodes[RefNode].PredEdges = new SparseBitVector<>; - GraphNodes[RefNode].PredEdges->set(C.Src); - } - } else { - // Dest = Src edge and *Dest = *Src edge - if (!GraphNodes[C.Dest].PredEdges) - GraphNodes[C.Dest].PredEdges = new SparseBitVector<>; - GraphNodes[C.Dest].PredEdges->set(C.Src); - unsigned RefNode = C.Dest + FirstRefNode; - if (!GraphNodes[RefNode].ImplicitPredEdges) - GraphNodes[RefNode].ImplicitPredEdges = new SparseBitVector<>; - GraphNodes[RefNode].ImplicitPredEdges->set(C.Src + FirstRefNode); - } - } - PEClass = 1; - // Do SCC finding first to condense our predecessor graph - DFSNumber = 0; - Node2DFS.insert(Node2DFS.begin(), GraphNodes.size(), 0); - Node2Deleted.insert(Node2Deleted.begin(), GraphNodes.size(), false); - Node2Visited.insert(Node2Visited.begin(), GraphNodes.size(), false); - - for (unsigned i = 0; i < FirstRefNode; ++i) { - unsigned Node = VSSCCRep[i]; - if (!Node2Visited[Node]) - HVNValNum(Node); - } - for (BitVectorMap::iterator Iter = Set2PEClass.begin(); - Iter != Set2PEClass.end(); - ++Iter) - delete Iter->first; - Set2PEClass.clear(); - Node2DFS.clear(); - Node2Deleted.clear(); - Node2Visited.clear(); - DEBUG(dbgs() << "Finished HVN\n"); - -} - -/// This is the workhorse of HVN value numbering. We combine SCC finding at the -/// same time because it's easy. -void Andersens::HVNValNum(unsigned NodeIndex) { - unsigned MyDFS = DFSNumber++; - Node *N = &GraphNodes[NodeIndex]; - Node2Visited[NodeIndex] = true; - Node2DFS[NodeIndex] = MyDFS; - - // First process all our explicit edges - if (N->PredEdges) - for (SparseBitVector<>::iterator Iter = N->PredEdges->begin(); - Iter != N->PredEdges->end(); - ++Iter) { - unsigned j = VSSCCRep[*Iter]; - if (!Node2Deleted[j]) { - if (!Node2Visited[j]) - HVNValNum(j); - if (Node2DFS[NodeIndex] > Node2DFS[j]) - Node2DFS[NodeIndex] = Node2DFS[j]; - } - } - - // Now process all the implicit edges - if (N->ImplicitPredEdges) - for (SparseBitVector<>::iterator Iter = N->ImplicitPredEdges->begin(); - Iter != N->ImplicitPredEdges->end(); - ++Iter) { - unsigned j = VSSCCRep[*Iter]; - if (!Node2Deleted[j]) { - if (!Node2Visited[j]) - HVNValNum(j); - if (Node2DFS[NodeIndex] > Node2DFS[j]) - Node2DFS[NodeIndex] = Node2DFS[j]; - } - } - - // See if we found any cycles - if (MyDFS == Node2DFS[NodeIndex]) { - while (!SCCStack.empty() && Node2DFS[SCCStack.top()] >= MyDFS) { - unsigned CycleNodeIndex = SCCStack.top(); - Node *CycleNode = &GraphNodes[CycleNodeIndex]; - VSSCCRep[CycleNodeIndex] = NodeIndex; - // Unify the nodes - N->Direct &= CycleNode->Direct; - - if (CycleNode->PredEdges) { - if (!N->PredEdges) - N->PredEdges = new SparseBitVector<>; - *(N->PredEdges) |= CycleNode->PredEdges; - delete CycleNode->PredEdges; - CycleNode->PredEdges = NULL; - } - if (CycleNode->ImplicitPredEdges) { - if (!N->ImplicitPredEdges) - N->ImplicitPredEdges = new SparseBitVector<>; - *(N->ImplicitPredEdges) |= CycleNode->ImplicitPredEdges; - delete CycleNode->ImplicitPredEdges; - CycleNode->ImplicitPredEdges = NULL; - } - - SCCStack.pop(); - } - - Node2Deleted[NodeIndex] = true; - - if (!N->Direct) { - GraphNodes[NodeIndex].PointerEquivLabel = PEClass++; - return; - } - - // Collect labels of successor nodes - bool AllSame = true; - unsigned First = ~0; - SparseBitVector<> *Labels = new SparseBitVector<>; - bool Used = false; - - if (N->PredEdges) - for (SparseBitVector<>::iterator Iter = N->PredEdges->begin(); - Iter != N->PredEdges->end(); - ++Iter) { - unsigned j = VSSCCRep[*Iter]; - unsigned Label = GraphNodes[j].PointerEquivLabel; - // Ignore labels that are equal to us or non-pointers - if (j == NodeIndex || Label == 0) - continue; - if (First == (unsigned)~0) - First = Label; - else if (First != Label) - AllSame = false; - Labels->set(Label); - } - - // We either have a non-pointer, a copy of an existing node, or a new node. - // Assign the appropriate pointer equivalence label. - if (Labels->empty()) { - GraphNodes[NodeIndex].PointerEquivLabel = 0; - } else if (AllSame) { - GraphNodes[NodeIndex].PointerEquivLabel = First; - } else { - GraphNodes[NodeIndex].PointerEquivLabel = Set2PEClass[Labels]; - if (GraphNodes[NodeIndex].PointerEquivLabel == 0) { - unsigned EquivClass = PEClass++; - Set2PEClass[Labels] = EquivClass; - GraphNodes[NodeIndex].PointerEquivLabel = EquivClass; - Used = true; - } - } - if (!Used) - delete Labels; - } else { - SCCStack.push(NodeIndex); - } -} - -/// The technique used here is described in "Exploiting Pointer and Location -/// Equivalence to Optimize Pointer Analysis. In the 14th International Static -/// Analysis Symposium (SAS), August 2007." It is known as the "HU" algorithm, -/// and is equivalent to value numbering the collapsed constraint graph -/// including evaluating unions. -void Andersens::HU() { - DEBUG(dbgs() << "Beginning HU\n"); - // Build a predecessor graph. This is like our constraint graph with the - // edges going in the opposite direction, and there are edges for all the - // constraints, instead of just copy constraints. We also build implicit - // edges for constraints are implied but not explicit. I.E for the constraint - // a = &b, we add implicit edges *a = b. This helps us capture more cycles - for (unsigned i = 0, e = Constraints.size(); i != e; ++i) { - Constraint &C = Constraints[i]; - if (C.Type == Constraint::AddressOf) { - GraphNodes[C.Src].AddressTaken = true; - GraphNodes[C.Src].Direct = false; - - GraphNodes[C.Dest].PointsTo->set(C.Src); - // *Dest = src edge - unsigned RefNode = C.Dest + FirstRefNode; - if (!GraphNodes[RefNode].ImplicitPredEdges) - GraphNodes[RefNode].ImplicitPredEdges = new SparseBitVector<>; - GraphNodes[RefNode].ImplicitPredEdges->set(C.Src); - GraphNodes[C.Src].PointedToBy->set(C.Dest); - } else if (C.Type == Constraint::Load) { - if (C.Offset == 0) { - // dest = *src edge - if (!GraphNodes[C.Dest].PredEdges) - GraphNodes[C.Dest].PredEdges = new SparseBitVector<>; - GraphNodes[C.Dest].PredEdges->set(C.Src + FirstRefNode); - } else { - GraphNodes[C.Dest].Direct = false; - } - } else if (C.Type == Constraint::Store) { - if (C.Offset == 0) { - // *dest = src edge - unsigned RefNode = C.Dest + FirstRefNode; - if (!GraphNodes[RefNode].PredEdges) - GraphNodes[RefNode].PredEdges = new SparseBitVector<>; - GraphNodes[RefNode].PredEdges->set(C.Src); - } - } else { - // Dest = Src edge and *Dest = *Src edg - if (!GraphNodes[C.Dest].PredEdges) - GraphNodes[C.Dest].PredEdges = new SparseBitVector<>; - GraphNodes[C.Dest].PredEdges->set(C.Src); - unsigned RefNode = C.Dest + FirstRefNode; - if (!GraphNodes[RefNode].ImplicitPredEdges) - GraphNodes[RefNode].ImplicitPredEdges = new SparseBitVector<>; - GraphNodes[RefNode].ImplicitPredEdges->set(C.Src + FirstRefNode); - } - } - PEClass = 1; - // Do SCC finding first to condense our predecessor graph - DFSNumber = 0; - Node2DFS.insert(Node2DFS.begin(), GraphNodes.size(), 0); - Node2Deleted.insert(Node2Deleted.begin(), GraphNodes.size(), false); - Node2Visited.insert(Node2Visited.begin(), GraphNodes.size(), false); - - for (unsigned i = 0; i < FirstRefNode; ++i) { - if (FindNode(i) == i) { - unsigned Node = VSSCCRep[i]; - if (!Node2Visited[Node]) - Condense(Node); - } - } - - // Reset tables for actual labeling - Node2DFS.clear(); - Node2Visited.clear(); - Node2Deleted.clear(); - // Pre-grow our densemap so that we don't get really bad behavior - Set2PEClass.resize(GraphNodes.size()); - - // Visit the condensed graph and generate pointer equivalence labels. - Node2Visited.insert(Node2Visited.begin(), GraphNodes.size(), false); - for (unsigned i = 0; i < FirstRefNode; ++i) { - if (FindNode(i) == i) { - unsigned Node = VSSCCRep[i]; - if (!Node2Visited[Node]) - HUValNum(Node); - } - } - // PEClass nodes will be deleted by the deleting of N->PointsTo in our caller. - Set2PEClass.clear(); - DEBUG(dbgs() << "Finished HU\n"); -} - - -/// Implementation of standard Tarjan SCC algorithm as modified by Nuutilla. -void Andersens::Condense(unsigned NodeIndex) { - unsigned MyDFS = DFSNumber++; - Node *N = &GraphNodes[NodeIndex]; - Node2Visited[NodeIndex] = true; - Node2DFS[NodeIndex] = MyDFS; - - // First process all our explicit edges - if (N->PredEdges) - for (SparseBitVector<>::iterator Iter = N->PredEdges->begin(); - Iter != N->PredEdges->end(); - ++Iter) { - unsigned j = VSSCCRep[*Iter]; - if (!Node2Deleted[j]) { - if (!Node2Visited[j]) - Condense(j); - if (Node2DFS[NodeIndex] > Node2DFS[j]) - Node2DFS[NodeIndex] = Node2DFS[j]; - } - } - - // Now process all the implicit edges - if (N->ImplicitPredEdges) - for (SparseBitVector<>::iterator Iter = N->ImplicitPredEdges->begin(); - Iter != N->ImplicitPredEdges->end(); - ++Iter) { - unsigned j = VSSCCRep[*Iter]; - if (!Node2Deleted[j]) { - if (!Node2Visited[j]) - Condense(j); - if (Node2DFS[NodeIndex] > Node2DFS[j]) - Node2DFS[NodeIndex] = Node2DFS[j]; - } - } - - // See if we found any cycles - if (MyDFS == Node2DFS[NodeIndex]) { - while (!SCCStack.empty() && Node2DFS[SCCStack.top()] >= MyDFS) { - unsigned CycleNodeIndex = SCCStack.top(); - Node *CycleNode = &GraphNodes[CycleNodeIndex]; - VSSCCRep[CycleNodeIndex] = NodeIndex; - // Unify the nodes - N->Direct &= CycleNode->Direct; - - *(N->PointsTo) |= CycleNode->PointsTo; - delete CycleNode->PointsTo; - CycleNode->PointsTo = NULL; - if (CycleNode->PredEdges) { - if (!N->PredEdges) - N->PredEdges = new SparseBitVector<>; - *(N->PredEdges) |= CycleNode->PredEdges; - delete CycleNode->PredEdges; - CycleNode->PredEdges = NULL; - } - if (CycleNode->ImplicitPredEdges) { - if (!N->ImplicitPredEdges) - N->ImplicitPredEdges = new SparseBitVector<>; - *(N->ImplicitPredEdges) |= CycleNode->ImplicitPredEdges; - delete CycleNode->ImplicitPredEdges; - CycleNode->ImplicitPredEdges = NULL; - } - SCCStack.pop(); - } - - Node2Deleted[NodeIndex] = true; - - // Set up number of incoming edges for other nodes - if (N->PredEdges) - for (SparseBitVector<>::iterator Iter = N->PredEdges->begin(); - Iter != N->PredEdges->end(); - ++Iter) - ++GraphNodes[VSSCCRep[*Iter]].NumInEdges; - } else { - SCCStack.push(NodeIndex); - } -} - -void Andersens::HUValNum(unsigned NodeIndex) { - Node *N = &GraphNodes[NodeIndex]; - Node2Visited[NodeIndex] = true; - - // Eliminate dereferences of non-pointers for those non-pointers we have - // already identified. These are ref nodes whose non-ref node: - // 1. Has already been visited determined to point to nothing (and thus, a - // dereference of it must point to nothing) - // 2. Any direct node with no predecessor edges in our graph and with no - // points-to set (since it can't point to anything either, being that it - // receives no points-to sets and has none). - if (NodeIndex >= FirstRefNode) { - unsigned j = VSSCCRep[FindNode(NodeIndex - FirstRefNode)]; - if ((Node2Visited[j] && !GraphNodes[j].PointerEquivLabel) - || (GraphNodes[j].Direct && !GraphNodes[j].PredEdges - && GraphNodes[j].PointsTo->empty())){ - return; - } - } - // Process all our explicit edges - if (N->PredEdges) - for (SparseBitVector<>::iterator Iter = N->PredEdges->begin(); - Iter != N->PredEdges->end(); - ++Iter) { - unsigned j = VSSCCRep[*Iter]; - if (!Node2Visited[j]) - HUValNum(j); - - // If this edge turned out to be the same as us, or got no pointer - // equivalence label (and thus points to nothing) , just decrement our - // incoming edges and continue. - if (j == NodeIndex || GraphNodes[j].PointerEquivLabel == 0) { - --GraphNodes[j].NumInEdges; - continue; - } - - *(N->PointsTo) |= GraphNodes[j].PointsTo; - - // If we didn't end up storing this in the hash, and we're done with all - // the edges, we don't need the points-to set anymore. - --GraphNodes[j].NumInEdges; - if (!GraphNodes[j].NumInEdges && !GraphNodes[j].StoredInHash) { - delete GraphNodes[j].PointsTo; - GraphNodes[j].PointsTo = NULL; - } - } - // If this isn't a direct node, generate a fresh variable. - if (!N->Direct) { - N->PointsTo->set(FirstRefNode + NodeIndex); - } - - // See If we have something equivalent to us, if not, generate a new - // equivalence class. - if (N->PointsTo->empty()) { - delete N->PointsTo; - N->PointsTo = NULL; - } else { - if (N->Direct) { - N->PointerEquivLabel = Set2PEClass[N->PointsTo]; - if (N->PointerEquivLabel == 0) { - unsigned EquivClass = PEClass++; - N->StoredInHash = true; - Set2PEClass[N->PointsTo] = EquivClass; - N->PointerEquivLabel = EquivClass; - } - } else { - N->PointerEquivLabel = PEClass++; - } - } -} - -/// Rewrite our list of constraints so that pointer equivalent nodes are -/// replaced by their the pointer equivalence class representative. -void Andersens::RewriteConstraints() { - std::vector<Constraint> NewConstraints; - DenseSet<Constraint, ConstraintKeyInfo> Seen; - - PEClass2Node.clear(); - PENLEClass2Node.clear(); - - // We may have from 1 to Graphnodes + 1 equivalence classes. - PEClass2Node.insert(PEClass2Node.begin(), GraphNodes.size() + 1, -1); - PENLEClass2Node.insert(PENLEClass2Node.begin(), GraphNodes.size() + 1, -1); - - // Rewrite constraints, ignoring non-pointer constraints, uniting equivalent - // nodes, and rewriting constraints to use the representative nodes. - for (unsigned i = 0, e = Constraints.size(); i != e; ++i) { - Constraint &C = Constraints[i]; - unsigned RHSNode = FindNode(C.Src); - unsigned LHSNode = FindNode(C.Dest); - unsigned RHSLabel = GraphNodes[VSSCCRep[RHSNode]].PointerEquivLabel; - unsigned LHSLabel = GraphNodes[VSSCCRep[LHSNode]].PointerEquivLabel; - - // First we try to eliminate constraints for things we can prove don't point - // to anything. - if (LHSLabel == 0) { - DEBUG(PrintNode(&GraphNodes[LHSNode])); - DEBUG(dbgs() << " is a non-pointer, ignoring constraint.\n"); - continue; - } - if (RHSLabel == 0) { - DEBUG(PrintNode(&GraphNodes[RHSNode])); - DEBUG(dbgs() << " is a non-pointer, ignoring constraint.\n"); - continue; - } - // This constraint may be useless, and it may become useless as we translate - // it. - if (C.Src == C.Dest && C.Type == Constraint::Copy) - continue; - - C.Src = FindEquivalentNode(RHSNode, RHSLabel); - C.Dest = FindEquivalentNode(FindNode(LHSNode), LHSLabel); - if ((C.Src == C.Dest && C.Type == Constraint::Copy) - || Seen.count(C)) - continue; - - Seen.insert(C); - NewConstraints.push_back(C); - } - Constraints.swap(NewConstraints); - PEClass2Node.clear(); -} - -/// See if we have a node that is pointer equivalent to the one being asked -/// about, and if so, unite them and return the equivalent node. Otherwise, -/// return the original node. -unsigned Andersens::FindEquivalentNode(unsigned NodeIndex, - unsigned NodeLabel) { - if (!GraphNodes[NodeIndex].AddressTaken) { - if (PEClass2Node[NodeLabel] != -1) { - // We found an existing node with the same pointer label, so unify them. - // We specifically request that Union-By-Rank not be used so that - // PEClass2Node[NodeLabel] U= NodeIndex and not the other way around. - return UniteNodes(PEClass2Node[NodeLabel], NodeIndex, false); - } else { - PEClass2Node[NodeLabel] = NodeIndex; - PENLEClass2Node[NodeLabel] = NodeIndex; - } - } else if (PENLEClass2Node[NodeLabel] == -1) { - PENLEClass2Node[NodeLabel] = NodeIndex; - } - - return NodeIndex; -} - -void Andersens::PrintLabels() const { - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - if (i < FirstRefNode) { - PrintNode(&GraphNodes[i]); - } else if (i < FirstAdrNode) { - DEBUG(dbgs() << "REF("); - PrintNode(&GraphNodes[i-FirstRefNode]); - DEBUG(dbgs() <<")"); - } else { - DEBUG(dbgs() << "ADR("); - PrintNode(&GraphNodes[i-FirstAdrNode]); - DEBUG(dbgs() <<")"); - } - - DEBUG(dbgs() << " has pointer label " << GraphNodes[i].PointerEquivLabel - << " and SCC rep " << VSSCCRep[i] - << " and is " << (GraphNodes[i].Direct ? "Direct" : "Not direct") - << "\n"); - } -} - -/// The technique used here is described in "The Ant and the -/// Grasshopper: Fast and Accurate Pointer Analysis for Millions of -/// Lines of Code. In Programming Language Design and Implementation -/// (PLDI), June 2007." It is known as the "HCD" (Hybrid Cycle -/// Detection) algorithm. It is called a hybrid because it performs an -/// offline analysis and uses its results during the solving (online) -/// phase. This is just the offline portion; the results of this -/// operation are stored in SDT and are later used in SolveContraints() -/// and UniteNodes(). -void Andersens::HCD() { - DEBUG(dbgs() << "Starting HCD.\n"); - HCDSCCRep.resize(GraphNodes.size()); - - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - GraphNodes[i].Edges = new SparseBitVector<>; - HCDSCCRep[i] = i; - } - - for (unsigned i = 0, e = Constraints.size(); i != e; ++i) { - Constraint &C = Constraints[i]; - assert (C.Src < GraphNodes.size() && C.Dest < GraphNodes.size()); - if (C.Type == Constraint::AddressOf) { - continue; - } else if (C.Type == Constraint::Load) { - if( C.Offset == 0 ) - GraphNodes[C.Dest].Edges->set(C.Src + FirstRefNode); - } else if (C.Type == Constraint::Store) { - if( C.Offset == 0 ) - GraphNodes[C.Dest + FirstRefNode].Edges->set(C.Src); - } else { - GraphNodes[C.Dest].Edges->set(C.Src); - } - } - - Node2DFS.insert(Node2DFS.begin(), GraphNodes.size(), 0); - Node2Deleted.insert(Node2Deleted.begin(), GraphNodes.size(), false); - Node2Visited.insert(Node2Visited.begin(), GraphNodes.size(), false); - SDT.insert(SDT.begin(), GraphNodes.size() / 2, -1); - - DFSNumber = 0; - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - unsigned Node = HCDSCCRep[i]; - if (!Node2Deleted[Node]) - Search(Node); - } - - for (unsigned i = 0; i < GraphNodes.size(); ++i) - if (GraphNodes[i].Edges != NULL) { - delete GraphNodes[i].Edges; - GraphNodes[i].Edges = NULL; - } - - while( !SCCStack.empty() ) - SCCStack.pop(); - - Node2DFS.clear(); - Node2Visited.clear(); - Node2Deleted.clear(); - HCDSCCRep.clear(); - DEBUG(dbgs() << "HCD complete.\n"); -} - -// Component of HCD: -// Use Nuutila's variant of Tarjan's algorithm to detect -// Strongly-Connected Components (SCCs). For non-trivial SCCs -// containing ref nodes, insert the appropriate information in SDT. -void Andersens::Search(unsigned Node) { - unsigned MyDFS = DFSNumber++; - - Node2Visited[Node] = true; - Node2DFS[Node] = MyDFS; - - for (SparseBitVector<>::iterator Iter = GraphNodes[Node].Edges->begin(), - End = GraphNodes[Node].Edges->end(); - Iter != End; - ++Iter) { - unsigned J = HCDSCCRep[*Iter]; - assert(GraphNodes[J].isRep() && "Debug check; must be representative"); - if (!Node2Deleted[J]) { - if (!Node2Visited[J]) - Search(J); - if (Node2DFS[Node] > Node2DFS[J]) - Node2DFS[Node] = Node2DFS[J]; - } - } - - if( MyDFS != Node2DFS[Node] ) { - SCCStack.push(Node); - return; - } - - // This node is the root of a SCC, so process it. - // - // If the SCC is "non-trivial" (not a singleton) and contains a reference - // node, we place this SCC into SDT. We unite the nodes in any case. - if (!SCCStack.empty() && Node2DFS[SCCStack.top()] >= MyDFS) { - SparseBitVector<> SCC; - - SCC.set(Node); - - bool Ref = (Node >= FirstRefNode); - - Node2Deleted[Node] = true; - - do { - unsigned P = SCCStack.top(); SCCStack.pop(); - Ref |= (P >= FirstRefNode); - SCC.set(P); - HCDSCCRep[P] = Node; - } while (!SCCStack.empty() && Node2DFS[SCCStack.top()] >= MyDFS); - - if (Ref) { - unsigned Rep = SCC.find_first(); - assert(Rep < FirstRefNode && "The SCC didn't have a non-Ref node!"); - - SparseBitVector<>::iterator i = SCC.begin(); - - // Skip over the non-ref nodes - while( *i < FirstRefNode ) - ++i; - - while( i != SCC.end() ) - SDT[ (*i++) - FirstRefNode ] = Rep; - } - } -} - - -/// Optimize the constraints by performing offline variable substitution and -/// other optimizations. -void Andersens::OptimizeConstraints() { - DEBUG(dbgs() << "Beginning constraint optimization\n"); - - SDTActive = false; - - // Function related nodes need to stay in the same relative position and can't - // be location equivalent. - for (std::map<unsigned, unsigned>::iterator Iter = MaxK.begin(); - Iter != MaxK.end(); - ++Iter) { - for (unsigned i = Iter->first; - i != Iter->first + Iter->second; - ++i) { - GraphNodes[i].AddressTaken = true; - GraphNodes[i].Direct = false; - } - } - - ClumpAddressTaken(); - FirstRefNode = GraphNodes.size(); - FirstAdrNode = FirstRefNode + GraphNodes.size(); - GraphNodes.insert(GraphNodes.end(), 2 * GraphNodes.size(), - Node(false)); - VSSCCRep.resize(GraphNodes.size()); - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - VSSCCRep[i] = i; - } - HVN(); - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - Node *N = &GraphNodes[i]; - delete N->PredEdges; - N->PredEdges = NULL; - delete N->ImplicitPredEdges; - N->ImplicitPredEdges = NULL; - } -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa-labels" - DEBUG(PrintLabels()); -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa" - RewriteConstraints(); - // Delete the adr nodes. - GraphNodes.resize(FirstRefNode * 2); - - // Now perform HU - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - Node *N = &GraphNodes[i]; - if (FindNode(i) == i) { - N->PointsTo = new SparseBitVector<>; - N->PointedToBy = new SparseBitVector<>; - // Reset our labels - } - VSSCCRep[i] = i; - N->PointerEquivLabel = 0; - } - HU(); -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa-labels" - DEBUG(PrintLabels()); -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa" - RewriteConstraints(); - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - if (FindNode(i) == i) { - Node *N = &GraphNodes[i]; - delete N->PointsTo; - N->PointsTo = NULL; - delete N->PredEdges; - N->PredEdges = NULL; - delete N->ImplicitPredEdges; - N->ImplicitPredEdges = NULL; - delete N->PointedToBy; - N->PointedToBy = NULL; - } - } - - // perform Hybrid Cycle Detection (HCD) - HCD(); - SDTActive = true; - - // No longer any need for the upper half of GraphNodes (for ref nodes). - GraphNodes.erase(GraphNodes.begin() + FirstRefNode, GraphNodes.end()); - - // HCD complete. - - DEBUG(dbgs() << "Finished constraint optimization\n"); - FirstRefNode = 0; - FirstAdrNode = 0; -} - -/// Unite pointer but not location equivalent variables, now that the constraint -/// graph is built. -void Andersens::UnitePointerEquivalences() { - DEBUG(dbgs() << "Uniting remaining pointer equivalences\n"); - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - if (GraphNodes[i].AddressTaken && GraphNodes[i].isRep()) { - unsigned Label = GraphNodes[i].PointerEquivLabel; - - if (Label && PENLEClass2Node[Label] != -1) - UniteNodes(i, PENLEClass2Node[Label]); - } - } - DEBUG(dbgs() << "Finished remaining pointer equivalences\n"); - PENLEClass2Node.clear(); -} - -/// Create the constraint graph used for solving points-to analysis. -/// -void Andersens::CreateConstraintGraph() { - for (unsigned i = 0, e = Constraints.size(); i != e; ++i) { - Constraint &C = Constraints[i]; - assert (C.Src < GraphNodes.size() && C.Dest < GraphNodes.size()); - if (C.Type == Constraint::AddressOf) - GraphNodes[C.Dest].PointsTo->set(C.Src); - else if (C.Type == Constraint::Load) - GraphNodes[C.Src].Constraints.push_back(C); - else if (C.Type == Constraint::Store) - GraphNodes[C.Dest].Constraints.push_back(C); - else if (C.Offset != 0) - GraphNodes[C.Src].Constraints.push_back(C); - else - GraphNodes[C.Src].Edges->set(C.Dest); - } -} - -// Perform DFS and cycle detection. -bool Andersens::QueryNode(unsigned Node) { - assert(GraphNodes[Node].isRep() && "Querying a non-rep node"); - unsigned OurDFS = ++DFSNumber; - SparseBitVector<> ToErase; - SparseBitVector<> NewEdges; - Tarjan2DFS[Node] = OurDFS; - - // Changed denotes a change from a recursive call that we will bubble up. - // Merged is set if we actually merge a node ourselves. - bool Changed = false, Merged = false; - - for (SparseBitVector<>::iterator bi = GraphNodes[Node].Edges->begin(); - bi != GraphNodes[Node].Edges->end(); - ++bi) { - unsigned RepNode = FindNode(*bi); - // If this edge points to a non-representative node but we are - // already planning to add an edge to its representative, we have no - // need for this edge anymore. - if (RepNode != *bi && NewEdges.test(RepNode)){ - ToErase.set(*bi); - continue; - } - - // Continue about our DFS. - if (!Tarjan2Deleted[RepNode]){ - if (Tarjan2DFS[RepNode] == 0) { - Changed |= QueryNode(RepNode); - // May have been changed by QueryNode - RepNode = FindNode(RepNode); - } - if (Tarjan2DFS[RepNode] < Tarjan2DFS[Node]) - Tarjan2DFS[Node] = Tarjan2DFS[RepNode]; - } - - // We may have just discovered that this node is part of a cycle, in - // which case we can also erase it. - if (RepNode != *bi) { - ToErase.set(*bi); - NewEdges.set(RepNode); - } - } - - GraphNodes[Node].Edges->intersectWithComplement(ToErase); - GraphNodes[Node].Edges |= NewEdges; - - // If this node is a root of a non-trivial SCC, place it on our - // worklist to be processed. - if (OurDFS == Tarjan2DFS[Node]) { - while (!SCCStack.empty() && Tarjan2DFS[SCCStack.top()] >= OurDFS) { - Node = UniteNodes(Node, SCCStack.top()); - - SCCStack.pop(); - Merged = true; - } - Tarjan2Deleted[Node] = true; - - if (Merged) - NextWL->insert(&GraphNodes[Node]); - } else { - SCCStack.push(Node); - } - - return(Changed | Merged); -} - -/// SolveConstraints - This stage iteratively processes the constraints list -/// propagating constraints (adding edges to the Nodes in the points-to graph) -/// until a fixed point is reached. -/// -/// We use a variant of the technique called "Lazy Cycle Detection", which is -/// described in "The Ant and the Grasshopper: Fast and Accurate Pointer -/// Analysis for Millions of Lines of Code. In Programming Language Design and -/// Implementation (PLDI), June 2007." -/// The paper describes performing cycle detection one node at a time, which can -/// be expensive if there are no cycles, but there are long chains of nodes that -/// it heuristically believes are cycles (because it will DFS from each node -/// without state from previous nodes). -/// Instead, we use the heuristic to build a worklist of nodes to check, then -/// cycle detect them all at the same time to do this more cheaply. This -/// catches cycles slightly later than the original technique did, but does it -/// make significantly cheaper. - -void Andersens::SolveConstraints() { - CurrWL = &w1; - NextWL = &w2; - - OptimizeConstraints(); -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa-constraints" - DEBUG(PrintConstraints()); -#undef DEBUG_TYPE -#define DEBUG_TYPE "anders-aa" - - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - Node *N = &GraphNodes[i]; - N->PointsTo = new SparseBitVector<>; - N->OldPointsTo = new SparseBitVector<>; - N->Edges = new SparseBitVector<>; - } - CreateConstraintGraph(); - UnitePointerEquivalences(); - assert(SCCStack.empty() && "SCC Stack should be empty by now!"); - Node2DFS.clear(); - Node2Deleted.clear(); - Node2DFS.insert(Node2DFS.begin(), GraphNodes.size(), 0); - Node2Deleted.insert(Node2Deleted.begin(), GraphNodes.size(), false); - DFSNumber = 0; - DenseSet<Constraint, ConstraintKeyInfo> Seen; - DenseSet<std::pair<unsigned,unsigned>, PairKeyInfo> EdgesChecked; - - // Order graph and add initial nodes to work list. - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - Node *INode = &GraphNodes[i]; - - // Add to work list if it's a representative and can contribute to the - // calculation right now. - if (INode->isRep() && !INode->PointsTo->empty() - && (!INode->Edges->empty() || !INode->Constraints.empty())) { - INode->Stamp(); - CurrWL->insert(INode); - } - } - std::queue<unsigned int> TarjanWL; -#if !FULL_UNIVERSAL - // "Rep and special variables" - in order for HCD to maintain conservative - // results when !FULL_UNIVERSAL, we need to treat the special variables in - // the same way that the !FULL_UNIVERSAL tweak does throughout the rest of - // the analysis - it's ok to add edges from the special nodes, but never - // *to* the special nodes. - std::vector<unsigned int> RSV; -#endif - while( !CurrWL->empty() ) { - DEBUG(dbgs() << "Starting iteration #" << ++NumIters << "\n"); - - Node* CurrNode; - unsigned CurrNodeIndex; - - // Actual cycle checking code. We cycle check all of the lazy cycle - // candidates from the last iteration in one go. - if (!TarjanWL.empty()) { - DFSNumber = 0; - - Tarjan2DFS.clear(); - Tarjan2Deleted.clear(); - while (!TarjanWL.empty()) { - unsigned int ToTarjan = TarjanWL.front(); - TarjanWL.pop(); - if (!Tarjan2Deleted[ToTarjan] - && GraphNodes[ToTarjan].isRep() - && Tarjan2DFS[ToTarjan] == 0) - QueryNode(ToTarjan); - } - } - - // Add to work list if it's a representative and can contribute to the - // calculation right now. - while( (CurrNode = CurrWL->pop()) != NULL ) { - CurrNodeIndex = CurrNode - &GraphNodes[0]; - CurrNode->Stamp(); - - - // Figure out the changed points to bits - SparseBitVector<> CurrPointsTo; - CurrPointsTo.intersectWithComplement(CurrNode->PointsTo, - CurrNode->OldPointsTo); - if (CurrPointsTo.empty()) - continue; - - *(CurrNode->OldPointsTo) |= CurrPointsTo; - - // Check the offline-computed equivalencies from HCD. - bool SCC = false; - unsigned Rep; - - if (SDT[CurrNodeIndex] >= 0) { - SCC = true; - Rep = FindNode(SDT[CurrNodeIndex]); - -#if !FULL_UNIVERSAL - RSV.clear(); -#endif - for (SparseBitVector<>::iterator bi = CurrPointsTo.begin(); - bi != CurrPointsTo.end(); ++bi) { - unsigned Node = FindNode(*bi); -#if !FULL_UNIVERSAL - if (Node < NumberSpecialNodes) { - RSV.push_back(Node); - continue; - } -#endif - Rep = UniteNodes(Rep,Node); - } -#if !FULL_UNIVERSAL - RSV.push_back(Rep); -#endif - - NextWL->insert(&GraphNodes[Rep]); - - if ( ! CurrNode->isRep() ) - continue; - } - - Seen.clear(); - - /* Now process the constraints for this node. */ - for (std::list<Constraint>::iterator li = CurrNode->Constraints.begin(); - li != CurrNode->Constraints.end(); ) { - li->Src = FindNode(li->Src); - li->Dest = FindNode(li->Dest); - - // Delete redundant constraints - if( Seen.count(*li) ) { - std::list<Constraint>::iterator lk = li; li++; - - CurrNode->Constraints.erase(lk); - ++NumErased; - continue; - } - Seen.insert(*li); - - // Src and Dest will be the vars we are going to process. - // This may look a bit ugly, but what it does is allow us to process - // both store and load constraints with the same code. - // Load constraints say that every member of our RHS solution has K - // added to it, and that variable gets an edge to LHS. We also union - // RHS+K's solution into the LHS solution. - // Store constraints say that every member of our LHS solution has K - // added to it, and that variable gets an edge from RHS. We also union - // RHS's solution into the LHS+K solution. - unsigned *Src; - unsigned *Dest; - unsigned K = li->Offset; - unsigned CurrMember; - if (li->Type == Constraint::Load) { - Src = &CurrMember; - Dest = &li->Dest; - } else if (li->Type == Constraint::Store) { - Src = &li->Src; - Dest = &CurrMember; - } else { - // TODO Handle offseted copy constraint - li++; - continue; - } - - // See if we can use Hybrid Cycle Detection (that is, check - // if it was a statically detected offline equivalence that - // involves pointers; if so, remove the redundant constraints). - if( SCC && K == 0 ) { -#if FULL_UNIVERSAL - CurrMember = Rep; - - if (GraphNodes[*Src].Edges->test_and_set(*Dest)) - if (GraphNodes[*Dest].PointsTo |= *(GraphNodes[*Src].PointsTo)) - NextWL->insert(&GraphNodes[*Dest]); -#else - for (unsigned i=0; i < RSV.size(); ++i) { - CurrMember = RSV[i]; - - if (*Dest < NumberSpecialNodes) - continue; - if (GraphNodes[*Src].Edges->test_and_set(*Dest)) - if (GraphNodes[*Dest].PointsTo |= *(GraphNodes[*Src].PointsTo)) - NextWL->insert(&GraphNodes[*Dest]); - } -#endif - // since all future elements of the points-to set will be - // equivalent to the current ones, the complex constraints - // become redundant. - // - std::list<Constraint>::iterator lk = li; li++; -#if !FULL_UNIVERSAL - // In this case, we can still erase the constraints when the - // elements of the points-to sets are referenced by *Dest, - // but not when they are referenced by *Src (i.e. for a Load - // constraint). This is because if another special variable is - // put into the points-to set later, we still need to add the - // new edge from that special variable. - if( lk->Type != Constraint::Load) -#endif - GraphNodes[CurrNodeIndex].Constraints.erase(lk); - } else { - const SparseBitVector<> &Solution = CurrPointsTo; - - for (SparseBitVector<>::iterator bi = Solution.begin(); - bi != Solution.end(); - ++bi) { - CurrMember = *bi; - - // Need to increment the member by K since that is where we are - // supposed to copy to/from. Note that in positive weight cycles, - // which occur in address taking of fields, K can go past - // MaxK[CurrMember] elements, even though that is all it could point - // to. - if (K > 0 && K > MaxK[CurrMember]) - continue; - else - CurrMember = FindNode(CurrMember + K); - - // Add an edge to the graph, so we can just do regular - // bitmap ior next time. It may also let us notice a cycle. -#if !FULL_UNIVERSAL - if (*Dest < NumberSpecialNodes) - continue; -#endif - if (GraphNodes[*Src].Edges->test_and_set(*Dest)) - if (GraphNodes[*Dest].PointsTo |= *(GraphNodes[*Src].PointsTo)) - NextWL->insert(&GraphNodes[*Dest]); - - } - li++; - } - } - SparseBitVector<> NewEdges; - SparseBitVector<> ToErase; - - // Now all we have left to do is propagate points-to info along the - // edges, erasing the redundant edges. - for (SparseBitVector<>::iterator bi = CurrNode->Edges->begin(); - bi != CurrNode->Edges->end(); - ++bi) { - - unsigned DestVar = *bi; - unsigned Rep = FindNode(DestVar); - - // If we ended up with this node as our destination, or we've already - // got an edge for the representative, delete the current edge. - if (Rep == CurrNodeIndex || - (Rep != DestVar && NewEdges.test(Rep))) { - ToErase.set(DestVar); - continue; - } - - std::pair<unsigned,unsigned> edge(CurrNodeIndex,Rep); - - // This is where we do lazy cycle detection. - // If this is a cycle candidate (equal points-to sets and this - // particular edge has not been cycle-checked previously), add to the - // list to check for cycles on the next iteration. - if (!EdgesChecked.count(edge) && - *(GraphNodes[Rep].PointsTo) == *(CurrNode->PointsTo)) { - EdgesChecked.insert(edge); - TarjanWL.push(Rep); - } - // Union the points-to sets into the dest -#if !FULL_UNIVERSAL - if (Rep >= NumberSpecialNodes) -#endif - if (GraphNodes[Rep].PointsTo |= CurrPointsTo) { - NextWL->insert(&GraphNodes[Rep]); - } - // If this edge's destination was collapsed, rewrite the edge. - if (Rep != DestVar) { - ToErase.set(DestVar); - NewEdges.set(Rep); - } - } - CurrNode->Edges->intersectWithComplement(ToErase); - CurrNode->Edges |= NewEdges; - } - - // Switch to other work list. - WorkList* t = CurrWL; CurrWL = NextWL; NextWL = t; - } - - - Node2DFS.clear(); - Node2Deleted.clear(); - for (unsigned i = 0; i < GraphNodes.size(); ++i) { - Node *N = &GraphNodes[i]; - delete N->OldPointsTo; - delete N->Edges; - } - SDTActive = false; - SDT.clear(); -} - -//===----------------------------------------------------------------------===// -// Union-Find -//===----------------------------------------------------------------------===// - -// Unite nodes First and Second, returning the one which is now the -// representative node. First and Second are indexes into GraphNodes -unsigned Andersens::UniteNodes(unsigned First, unsigned Second, - bool UnionByRank) { - assert (First < GraphNodes.size() && Second < GraphNodes.size() && - "Attempting to merge nodes that don't exist"); - - Node *FirstNode = &GraphNodes[First]; - Node *SecondNode = &GraphNodes[Second]; - - assert (SecondNode->isRep() && FirstNode->isRep() && - "Trying to unite two non-representative nodes!"); - if (First == Second) - return First; - - if (UnionByRank) { - int RankFirst = (int) FirstNode ->NodeRep; - int RankSecond = (int) SecondNode->NodeRep; - - // Rank starts at -1 and gets decremented as it increases. - // Translation: higher rank, lower NodeRep value, which is always negative. - if (RankFirst > RankSecond) { - unsigned t = First; First = Second; Second = t; - Node* tp = FirstNode; FirstNode = SecondNode; SecondNode = tp; - } else if (RankFirst == RankSecond) { - FirstNode->NodeRep = (unsigned) (RankFirst - 1); - } - } - - SecondNode->NodeRep = First; -#if !FULL_UNIVERSAL - if (First >= NumberSpecialNodes) -#endif - if (FirstNode->PointsTo && SecondNode->PointsTo) - FirstNode->PointsTo |= *(SecondNode->PointsTo); - if (FirstNode->Edges && SecondNode->Edges) - FirstNode->Edges |= *(SecondNode->Edges); - if (!SecondNode->Constraints.empty()) - FirstNode->Constraints.splice(FirstNode->Constraints.begin(), - SecondNode->Constraints); - if (FirstNode->OldPointsTo) { - delete FirstNode->OldPointsTo; - FirstNode->OldPointsTo = new SparseBitVector<>; - } - - // Destroy interesting parts of the merged-from node. - delete SecondNode->OldPointsTo; - delete SecondNode->Edges; - delete SecondNode->PointsTo; - SecondNode->Edges = NULL; - SecondNode->PointsTo = NULL; - SecondNode->OldPointsTo = NULL; - - NumUnified++; - DEBUG(dbgs() << "Unified Node "); - DEBUG(PrintNode(FirstNode)); - DEBUG(dbgs() << " and Node "); - DEBUG(PrintNode(SecondNode)); - DEBUG(dbgs() << "\n"); - - if (SDTActive) - if (SDT[Second] >= 0) { - if (SDT[First] < 0) - SDT[First] = SDT[Second]; - else { - UniteNodes( FindNode(SDT[First]), FindNode(SDT[Second]) ); - First = FindNode(First); - } - } - - return First; -} - -// Find the index into GraphNodes of the node representing Node, performing -// path compression along the way -unsigned Andersens::FindNode(unsigned NodeIndex) { - assert (NodeIndex < GraphNodes.size() - && "Attempting to find a node that can't exist"); - Node *N = &GraphNodes[NodeIndex]; - if (N->isRep()) - return NodeIndex; - else - return (N->NodeRep = FindNode(N->NodeRep)); -} - -// Find the index into GraphNodes of the node representing Node, -// don't perform path compression along the way (for Print) -unsigned Andersens::FindNode(unsigned NodeIndex) const { - assert (NodeIndex < GraphNodes.size() - && "Attempting to find a node that can't exist"); - const Node *N = &GraphNodes[NodeIndex]; - if (N->isRep()) - return NodeIndex; - else - return FindNode(N->NodeRep); -} - -//===----------------------------------------------------------------------===// -// Debugging Output -//===----------------------------------------------------------------------===// - -void Andersens::PrintNode(const Node *N) const { - if (N == &GraphNodes[UniversalSet]) { - dbgs() << "<universal>"; - return; - } else if (N == &GraphNodes[NullPtr]) { - dbgs() << "<nullptr>"; - return; - } else if (N == &GraphNodes[NullObject]) { - dbgs() << "<null>"; - return; - } - if (!N->getValue()) { - dbgs() << "artificial" << (intptr_t) N; - return; - } - - assert(N->getValue() != 0 && "Never set node label!"); - Value *V = N->getValue(); - if (Function *F = dyn_cast<Function>(V)) { - if (isa<PointerType>(F->getFunctionType()->getReturnType()) && - N == &GraphNodes[getReturnNode(F)]) { - dbgs() << F->getName() << ":retval"; - return; - } else if (F->getFunctionType()->isVarArg() && - N == &GraphNodes[getVarargNode(F)]) { - dbgs() << F->getName() << ":vararg"; - return; - } - } - - if (Instruction *I = dyn_cast<Instruction>(V)) - dbgs() << I->getParent()->getParent()->getName() << ":"; - else if (Argument *Arg = dyn_cast<Argument>(V)) - dbgs() << Arg->getParent()->getName() << ":"; - - if (V->hasName()) - dbgs() << V->getName(); - else - dbgs() << "(unnamed)"; - - if (isa<GlobalValue>(V) || isa<AllocaInst>(V) || isMalloc(V)) - if (N == &GraphNodes[getObject(V)]) - dbgs() << "<mem>"; -} -void Andersens::PrintConstraint(const Constraint &C) const { - if (C.Type == Constraint::Store) { - dbgs() << "*"; - if (C.Offset != 0) - dbgs() << "("; - } - PrintNode(&GraphNodes[C.Dest]); - if (C.Type == Constraint::Store && C.Offset != 0) - dbgs() << " + " << C.Offset << ")"; - dbgs() << " = "; - if (C.Type == Constraint::Load) { - dbgs() << "*"; - if (C.Offset != 0) - dbgs() << "("; - } - else if (C.Type == Constraint::AddressOf) - dbgs() << "&"; - PrintNode(&GraphNodes[C.Src]); - if (C.Offset != 0 && C.Type != Constraint::Store) - dbgs() << " + " << C.Offset; - if (C.Type == Constraint::Load && C.Offset != 0) - dbgs() << ")"; - dbgs() << "\n"; -} - -void Andersens::PrintConstraints() const { - dbgs() << "Constraints:\n"; - - for (unsigned i = 0, e = Constraints.size(); i != e; ++i) - PrintConstraint(Constraints[i]); -} - -void Andersens::PrintPointsToGraph() const { - dbgs() << "Points-to graph:\n"; - for (unsigned i = 0, e = GraphNodes.size(); i != e; ++i) { - const Node *N = &GraphNodes[i]; - if (FindNode(i) != i) { - PrintNode(N); - dbgs() << "\t--> same as "; - PrintNode(&GraphNodes[FindNode(i)]); - dbgs() << "\n"; - } else { - dbgs() << "[" << (N->PointsTo->count()) << "] "; - PrintNode(N); - dbgs() << "\t--> "; - - bool first = true; - for (SparseBitVector<>::iterator bi = N->PointsTo->begin(); - bi != N->PointsTo->end(); - ++bi) { - if (!first) - dbgs() << ", "; - PrintNode(&GraphNodes[*bi]); - first = false; - } - dbgs() << "\n"; - } - } -} diff --git a/lib/Analysis/IPA/CMakeLists.txt b/lib/Analysis/IPA/CMakeLists.txt index 1ebb0be..007ad22 100644 --- a/lib/Analysis/IPA/CMakeLists.txt +++ b/lib/Analysis/IPA/CMakeLists.txt @@ -1,5 +1,4 @@ add_llvm_library(LLVMipa - Andersens.cpp CallGraph.cpp CallGraphSCCPass.cpp FindUsedTypes.cpp diff --git a/lib/Analysis/IPA/GlobalsModRef.cpp b/lib/Analysis/IPA/GlobalsModRef.cpp index ec94bc8..7b43089 100644 --- a/lib/Analysis/IPA/GlobalsModRef.cpp +++ b/lib/Analysis/IPA/GlobalsModRef.cpp @@ -213,7 +213,7 @@ void GlobalsModRef::AnalyzeGlobals(Module &M) { ++NumNonAddrTakenGlobalVars; // If this global holds a pointer type, see if it is an indirect global. - if (isa<PointerType>(I->getType()->getElementType()) && + if (I->getType()->getElementType()->isPointerTy() && AnalyzeIndirectGlobalMemory(I)) ++NumIndirectGlobalVars; } @@ -231,7 +231,7 @@ bool GlobalsModRef::AnalyzeUsesOfPointer(Value *V, std::vector<Function*> &Readers, std::vector<Function*> &Writers, GlobalValue *OkayStoreDest) { - if (!isa<PointerType>(V->getType())) return true; + if (!V->getType()->isPointerTy()) return true; for (Value::use_iterator UI = V->use_begin(), E = V->use_end(); UI != E; ++UI) if (LoadInst *LI = dyn_cast<LoadInst>(*UI)) { diff --git a/lib/Analysis/IVUsers.cpp b/lib/Analysis/IVUsers.cpp index 4ce6868..98a436f 100644 --- a/lib/Analysis/IVUsers.cpp +++ b/lib/Analysis/IVUsers.cpp @@ -142,8 +142,7 @@ static bool getSCEVStartAndStride(const SCEV *&SH, Loop *L, Loop *UseLoop, /// the loop, resulting in reg-reg copies (if we use the pre-inc value when we /// should use the post-inc value). static bool IVUseShouldUsePostIncValue(Instruction *User, Instruction *IV, - Loop *L, LoopInfo *LI, DominatorTree *DT, - Pass *P) { + Loop *L, DominatorTree *DT) { // If the user is in the loop, use the preinc value. if (L->contains(User)) return false; @@ -223,7 +222,7 @@ bool IVUsers::AddUsersIfInteresting(Instruction *I) { // Descend recursively, but not into PHI nodes outside the current loop. // It's important to see the entire expression outside the loop to get // choices that depend on addressing mode use right, although we won't - // consider references ouside the loop in all cases. + // consider references outside the loop in all cases. // If User is already in Processed, we don't want to recurse into it again, // but do want to record a second reference in the same instruction. bool AddUserToIVUsers = false; @@ -245,7 +244,7 @@ bool IVUsers::AddUsersIfInteresting(Instruction *I) { // Okay, we found a user that we cannot reduce. Analyze the instruction // and decide what to do with it. If we are a use inside of the loop, use // the value before incrementation, otherwise use it after incrementation. - if (IVUseShouldUsePostIncValue(User, I, L, LI, DT, this)) { + if (IVUseShouldUsePostIncValue(User, I, L, DT)) { // The value used will be incremented by the stride more than we are // expecting, so subtract this off. const SCEV *NewStart = SE->getMinusSCEV(Start, Stride); @@ -331,7 +330,7 @@ void IVUsers::print(raw_ostream &OS, const Module *M) const { } OS << ":\n"; - // Use a defualt AssemblyAnnotationWriter to suppress the default info + // Use a default AssemblyAnnotationWriter to suppress the default info // comments, which aren't relevant here. AssemblyAnnotationWriter Annotator; for (ilist<IVStrideUse>::const_iterator UI = IVUses.begin(), diff --git a/lib/Analysis/InlineCost.cpp b/lib/Analysis/InlineCost.cpp index 972d034..ca50a17 100644 --- a/lib/Analysis/InlineCost.cpp +++ b/lib/Analysis/InlineCost.cpp @@ -84,7 +84,7 @@ unsigned InlineCostAnalyzer::FunctionInfo:: // unsigned InlineCostAnalyzer::FunctionInfo:: CountCodeReductionForAlloca(Value *V) { - if (!isa<PointerType>(V->getType())) return 0; // Not a pointer + if (!V->getType()->isPointerTy()) return 0; // Not a pointer unsigned Reduction = 0; for (Value::use_iterator UI = V->use_begin(), E = V->use_end(); UI != E;++UI){ Instruction *I = cast<Instruction>(*UI); @@ -175,7 +175,7 @@ void CodeMetrics::analyzeBasicBlock(const BasicBlock *BB) { this->usesDynamicAlloca = true; } - if (isa<ExtractElementInst>(II) || isa<VectorType>(II->getType())) + if (isa<ExtractElementInst>(II) || II->getType()->isVectorTy()) ++NumVectorInsts; if (const CastInst *CI = dyn_cast<CastInst>(II)) { diff --git a/lib/Analysis/InstructionSimplify.cpp b/lib/Analysis/InstructionSimplify.cpp index b53ac13..1f8053a 100644 --- a/lib/Analysis/InstructionSimplify.cpp +++ b/lib/Analysis/InstructionSimplify.cpp @@ -283,6 +283,32 @@ Value *llvm::SimplifyFCmpInst(unsigned Predicate, Value *LHS, Value *RHS, // True if unordered. return ConstantInt::getTrue(CFP->getContext()); } + // Check whether the constant is an infinity. + if (CFP->getValueAPF().isInfinity()) { + if (CFP->getValueAPF().isNegative()) { + switch (Pred) { + case FCmpInst::FCMP_OLT: + // No value is ordered and less than negative infinity. + return ConstantInt::getFalse(CFP->getContext()); + case FCmpInst::FCMP_UGE: + // All values are unordered with or at least negative infinity. + return ConstantInt::getTrue(CFP->getContext()); + default: + break; + } + } else { + switch (Pred) { + case FCmpInst::FCMP_OGT: + // No value is ordered and greater than infinity. + return ConstantInt::getFalse(CFP->getContext()); + case FCmpInst::FCMP_ULE: + // All values are unordered with and at most infinity. + return ConstantInt::getTrue(CFP->getContext()); + default: + break; + } + } + } } } diff --git a/lib/Analysis/MemoryDependenceAnalysis.cpp b/lib/Analysis/MemoryDependenceAnalysis.cpp index 2d74709d..2aa2f17 100644 --- a/lib/Analysis/MemoryDependenceAnalysis.cpp +++ b/lib/Analysis/MemoryDependenceAnalysis.cpp @@ -580,7 +580,7 @@ MemoryDependenceAnalysis::getNonLocalCallDependency(CallSite QueryCS) { void MemoryDependenceAnalysis:: getNonLocalPointerDependency(Value *Pointer, bool isLoad, BasicBlock *FromBB, SmallVectorImpl<NonLocalDepResult> &Result) { - assert(isa<PointerType>(Pointer->getType()) && + assert(Pointer->getType()->isPointerTy() && "Can't get pointer deps of a non-pointer!"); Result.clear(); @@ -861,7 +861,7 @@ getNonLocalPointerDepFromBB(const PHITransAddr &Pointer, uint64_t PointeeSize, // Get the PHI translated pointer in this predecessor. This can fail if // not translatable, in which case the getAddr() returns null. PHITransAddr PredPointer(Pointer); - PredPointer.PHITranslateValue(BB, Pred); + PredPointer.PHITranslateValue(BB, Pred, 0); Value *PredPtrVal = PredPointer.getAddr(); @@ -1009,13 +1009,20 @@ RemoveCachedNonLocalPointerDependencies(ValueIsLoadPair P) { /// in more places that cached info does not necessarily keep. void MemoryDependenceAnalysis::invalidateCachedPointerInfo(Value *Ptr) { // If Ptr isn't really a pointer, just ignore it. - if (!isa<PointerType>(Ptr->getType())) return; + if (!Ptr->getType()->isPointerTy()) return; // Flush store info for the pointer. RemoveCachedNonLocalPointerDependencies(ValueIsLoadPair(Ptr, false)); // Flush load info for the pointer. RemoveCachedNonLocalPointerDependencies(ValueIsLoadPair(Ptr, true)); } +/// invalidateCachedPredecessors - Clear the PredIteratorCache info. +/// This needs to be done when the CFG changes, e.g., due to splitting +/// critical edges. +void MemoryDependenceAnalysis::invalidateCachedPredecessors() { + PredCache->clear(); +} + /// removeInstruction - Remove an instruction from the dependence analysis, /// updating the dependence of instructions that previously depended on it. /// This method attempts to keep the cache coherent using the reverse map. @@ -1050,7 +1057,7 @@ void MemoryDependenceAnalysis::removeInstruction(Instruction *RemInst) { // Remove it from both the load info and the store info. The instruction // can't be in either of these maps if it is non-pointer. - if (isa<PointerType>(RemInst->getType())) { + if (RemInst->getType()->isPointerTy()) { RemoveCachedNonLocalPointerDependencies(ValueIsLoadPair(RemInst, false)); RemoveCachedNonLocalPointerDependencies(ValueIsLoadPair(RemInst, true)); } diff --git a/lib/Analysis/PHITransAddr.cpp b/lib/Analysis/PHITransAddr.cpp index 334a188..8e4fa03 100644 --- a/lib/Analysis/PHITransAddr.cpp +++ b/lib/Analysis/PHITransAddr.cpp @@ -134,7 +134,8 @@ static void RemoveInstInputs(Value *V, } Value *PHITransAddr::PHITranslateSubExpr(Value *V, BasicBlock *CurBB, - BasicBlock *PredBB) { + BasicBlock *PredBB, + const DominatorTree *DT) { // If this is a non-instruction value, it can't require PHI translation. Instruction *Inst = dyn_cast<Instruction>(V); if (Inst == 0) return V; @@ -177,7 +178,7 @@ Value *PHITransAddr::PHITranslateSubExpr(Value *V, BasicBlock *CurBB, // operands need to be phi translated, and if so, reconstruct it. if (BitCastInst *BC = dyn_cast<BitCastInst>(Inst)) { - Value *PHIIn = PHITranslateSubExpr(BC->getOperand(0), CurBB, PredBB); + Value *PHIIn = PHITranslateSubExpr(BC->getOperand(0), CurBB, PredBB, DT); if (PHIIn == 0) return 0; if (PHIIn == BC->getOperand(0)) return BC; @@ -193,7 +194,8 @@ Value *PHITransAddr::PHITranslateSubExpr(Value *V, BasicBlock *CurBB, for (Value::use_iterator UI = PHIIn->use_begin(), E = PHIIn->use_end(); UI != E; ++UI) { if (BitCastInst *BCI = dyn_cast<BitCastInst>(*UI)) - if (BCI->getType() == BC->getType()) + if (BCI->getType() == BC->getType() && + (!DT || DT->dominates(BCI->getParent(), PredBB))) return BCI; } return 0; @@ -204,7 +206,7 @@ Value *PHITransAddr::PHITranslateSubExpr(Value *V, BasicBlock *CurBB, SmallVector<Value*, 8> GEPOps; bool AnyChanged = false; for (unsigned i = 0, e = GEP->getNumOperands(); i != e; ++i) { - Value *GEPOp = PHITranslateSubExpr(GEP->getOperand(i), CurBB, PredBB); + Value *GEPOp = PHITranslateSubExpr(GEP->getOperand(i), CurBB, PredBB, DT); if (GEPOp == 0) return 0; AnyChanged |= GEPOp != GEP->getOperand(i); @@ -229,7 +231,8 @@ Value *PHITransAddr::PHITranslateSubExpr(Value *V, BasicBlock *CurBB, if (GetElementPtrInst *GEPI = dyn_cast<GetElementPtrInst>(*UI)) if (GEPI->getType() == GEP->getType() && GEPI->getNumOperands() == GEPOps.size() && - GEPI->getParent()->getParent() == CurBB->getParent()) { + GEPI->getParent()->getParent() == CurBB->getParent() && + (!DT || DT->dominates(GEPI->getParent(), PredBB))) { bool Mismatch = false; for (unsigned i = 0, e = GEPOps.size(); i != e; ++i) if (GEPI->getOperand(i) != GEPOps[i]) { @@ -251,7 +254,7 @@ Value *PHITransAddr::PHITranslateSubExpr(Value *V, BasicBlock *CurBB, bool isNSW = cast<BinaryOperator>(Inst)->hasNoSignedWrap(); bool isNUW = cast<BinaryOperator>(Inst)->hasNoUnsignedWrap(); - Value *LHS = PHITranslateSubExpr(Inst->getOperand(0), CurBB, PredBB); + Value *LHS = PHITranslateSubExpr(Inst->getOperand(0), CurBB, PredBB, DT); if (LHS == 0) return 0; // If the PHI translated LHS is an add of a constant, fold the immediates. @@ -287,7 +290,8 @@ Value *PHITransAddr::PHITranslateSubExpr(Value *V, BasicBlock *CurBB, if (BinaryOperator *BO = dyn_cast<BinaryOperator>(*UI)) if (BO->getOpcode() == Instruction::Add && BO->getOperand(0) == LHS && BO->getOperand(1) == RHS && - BO->getParent()->getParent() == CurBB->getParent()) + BO->getParent()->getParent() == CurBB->getParent() && + (!DT || DT->dominates(BO->getParent(), PredBB))) return BO; } @@ -300,33 +304,24 @@ Value *PHITransAddr::PHITranslateSubExpr(Value *V, BasicBlock *CurBB, /// PHITranslateValue - PHI translate the current address up the CFG from -/// CurBB to Pred, updating our state the reflect any needed changes. This -/// returns true on failure and sets Addr to null. -bool PHITransAddr::PHITranslateValue(BasicBlock *CurBB, BasicBlock *PredBB) { +/// CurBB to Pred, updating our state to reflect any needed changes. If the +/// dominator tree DT is non-null, the translated value must dominate +/// PredBB. This returns true on failure and sets Addr to null. +bool PHITransAddr::PHITranslateValue(BasicBlock *CurBB, BasicBlock *PredBB, + const DominatorTree *DT) { assert(Verify() && "Invalid PHITransAddr!"); - Addr = PHITranslateSubExpr(Addr, CurBB, PredBB); + Addr = PHITranslateSubExpr(Addr, CurBB, PredBB, DT); assert(Verify() && "Invalid PHITransAddr!"); - return Addr == 0; -} -/// GetAvailablePHITranslatedSubExpr - Return the value computed by -/// PHITranslateSubExpr if it dominates PredBB, otherwise return null. -Value *PHITransAddr:: -GetAvailablePHITranslatedSubExpr(Value *V, BasicBlock *CurBB,BasicBlock *PredBB, - const DominatorTree &DT) const { - PHITransAddr Tmp(V, TD); - Tmp.PHITranslateValue(CurBB, PredBB); - - // See if PHI translation succeeds. - V = Tmp.getAddr(); - - // Make sure the value is live in the predecessor. - if (Instruction *Inst = dyn_cast_or_null<Instruction>(V)) - if (!DT.dominates(Inst->getParent(), PredBB)) - return 0; - return V; -} + if (DT) { + // Make sure the value is live in the predecessor. + if (Instruction *Inst = dyn_cast_or_null<Instruction>(Addr)) + if (!DT->dominates(Inst->getParent(), PredBB)) + Addr = 0; + } + return Addr == 0; +} /// PHITranslateWithInsertion - PHI translate this value into the specified /// predecessor block, inserting a computation of the value if it is @@ -365,8 +360,9 @@ InsertPHITranslatedSubExpr(Value *InVal, BasicBlock *CurBB, SmallVectorImpl<Instruction*> &NewInsts) { // See if we have a version of this value already available and dominating // PredBB. If so, there is no need to insert a new instance of it. - if (Value *Res = GetAvailablePHITranslatedSubExpr(InVal, CurBB, PredBB, DT)) - return Res; + PHITransAddr Tmp(InVal, TD); + if (!Tmp.PHITranslateValue(CurBB, PredBB, &DT)) + return Tmp.getAddr(); // If we don't have an available version of this value, it must be an // instruction. diff --git a/lib/Analysis/PointerTracking.cpp b/lib/Analysis/PointerTracking.cpp index 8da07e7..ce7ac89 100644 --- a/lib/Analysis/PointerTracking.cpp +++ b/lib/Analysis/PointerTracking.cpp @@ -231,7 +231,7 @@ void PointerTracking::print(raw_ostream &OS, const Module* M) const { // this should be safe for the same reason its safe for SCEV. PointerTracking &PT = *const_cast<PointerTracking*>(this); for (inst_iterator I=inst_begin(*FF), E=inst_end(*FF); I != E; ++I) { - if (!isa<PointerType>(I->getType())) + if (!I->getType()->isPointerTy()) continue; Value *Base; const SCEV *Limit, *Offset; diff --git a/lib/Analysis/ScalarEvolution.cpp b/lib/Analysis/ScalarEvolution.cpp index 9ee7d3a..b979f33 100644 --- a/lib/Analysis/ScalarEvolution.cpp +++ b/lib/Analysis/ScalarEvolution.cpp @@ -214,8 +214,8 @@ bool SCEVCastExpr::properlyDominates(BasicBlock *BB, DominatorTree *DT) const { SCEVTruncateExpr::SCEVTruncateExpr(const FoldingSetNodeID &ID, const SCEV *op, const Type *ty) : SCEVCastExpr(ID, scTruncate, op, ty) { - assert((Op->getType()->isIntegerTy() || isa<PointerType>(Op->getType())) && - (Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert((Op->getType()->isIntegerTy() || Op->getType()->isPointerTy()) && + (Ty->isIntegerTy() || Ty->isPointerTy()) && "Cannot truncate non-integer value!"); } @@ -226,8 +226,8 @@ void SCEVTruncateExpr::print(raw_ostream &OS) const { SCEVZeroExtendExpr::SCEVZeroExtendExpr(const FoldingSetNodeID &ID, const SCEV *op, const Type *ty) : SCEVCastExpr(ID, scZeroExtend, op, ty) { - assert((Op->getType()->isIntegerTy() || isa<PointerType>(Op->getType())) && - (Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert((Op->getType()->isIntegerTy() || Op->getType()->isPointerTy()) && + (Ty->isIntegerTy() || Ty->isPointerTy()) && "Cannot zero extend non-integer value!"); } @@ -238,8 +238,8 @@ void SCEVZeroExtendExpr::print(raw_ostream &OS) const { SCEVSignExtendExpr::SCEVSignExtendExpr(const FoldingSetNodeID &ID, const SCEV *op, const Type *ty) : SCEVCastExpr(ID, scSignExtend, op, ty) { - assert((Op->getType()->isIntegerTy() || isa<PointerType>(Op->getType())) && - (Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert((Op->getType()->isIntegerTy() || Op->getType()->isPointerTy()) && + (Ty->isIntegerTy() || Ty->isPointerTy()) && "Cannot sign extend non-integer value!"); } @@ -416,7 +416,7 @@ bool SCEVUnknown::isOffsetOf(const Type *&CTy, Constant *&FieldNo) const { cast<PointerType>(CE->getOperand(0)->getType())->getElementType(); // Ignore vector types here so that ScalarEvolutionExpander doesn't // emit getelementptrs that index into vectors. - if (isa<StructType>(Ty) || isa<ArrayType>(Ty)) { + if (Ty->isStructTy() || Ty->isArrayTy()) { CTy = Ty; FieldNo = CE->getOperand(2); return true; @@ -518,9 +518,9 @@ namespace { // Order pointer values after integer values. This helps SCEVExpander // form GEPs. - if (isa<PointerType>(LU->getType()) && !isa<PointerType>(RU->getType())) + if (LU->getType()->isPointerTy() && !RU->getType()->isPointerTy()) return false; - if (isa<PointerType>(RU->getType()) && !isa<PointerType>(LU->getType())) + if (RU->getType()->isPointerTy() && !LU->getType()->isPointerTy()) return true; // Compare getValueID values. @@ -616,7 +616,7 @@ namespace { /// When this routine is finished, we know that any duplicates in the vector are /// consecutive and that complexity is monotonically increasing. /// -/// Note that we go take special precautions to ensure that we get determinstic +/// Note that we go take special precautions to ensure that we get deterministic /// results from this routine. In other words, we don't want the results of /// this to depend on where the addresses of various SCEV objects happened to /// land in memory. @@ -744,7 +744,7 @@ static const SCEV *BinomialCoefficient(const SCEV *It, unsigned K, // We need at least W + T bits for the multiplication step unsigned CalculationBits = W + T; - // Calcuate 2^T, at width T+W. + // Calculate 2^T, at width T+W. APInt DivFactor = APInt(CalculationBits, 1).shl(T); // Calculate the multiplicative inverse of K! / 2^T; @@ -921,9 +921,7 @@ const SCEV *ScalarEvolution::getZeroExtendExpr(const SCEV *Op, if (MaxBECount == RecastedMaxBECount) { const Type *WideTy = IntegerType::get(getContext(), BitWidth * 2); // Check whether Start+Step*MaxBECount has no unsigned overflow. - const SCEV *ZMul = - getMulExpr(CastedMaxBECount, - getTruncateOrZeroExtend(Step, Start->getType())); + const SCEV *ZMul = getMulExpr(CastedMaxBECount, Step); const SCEV *Add = getAddExpr(Start, ZMul); const SCEV *OperandExtendedAdd = getAddExpr(getZeroExtendExpr(Start, WideTy), @@ -937,9 +935,7 @@ const SCEV *ScalarEvolution::getZeroExtendExpr(const SCEV *Op, // Similar to above, only this time treat the step value as signed. // This covers loops that count down. - const SCEV *SMul = - getMulExpr(CastedMaxBECount, - getTruncateOrSignExtend(Step, Start->getType())); + const SCEV *SMul = getMulExpr(CastedMaxBECount, Step); Add = getAddExpr(Start, SMul); OperandExtendedAdd = getAddExpr(getZeroExtendExpr(Start, WideTy), @@ -1060,9 +1056,7 @@ const SCEV *ScalarEvolution::getSignExtendExpr(const SCEV *Op, if (MaxBECount == RecastedMaxBECount) { const Type *WideTy = IntegerType::get(getContext(), BitWidth * 2); // Check whether Start+Step*MaxBECount has no signed overflow. - const SCEV *SMul = - getMulExpr(CastedMaxBECount, - getTruncateOrSignExtend(Step, Start->getType())); + const SCEV *SMul = getMulExpr(CastedMaxBECount, Step); const SCEV *Add = getAddExpr(Start, SMul); const SCEV *OperandExtendedAdd = getAddExpr(getSignExtendExpr(Start, WideTy), @@ -1076,9 +1070,7 @@ const SCEV *ScalarEvolution::getSignExtendExpr(const SCEV *Op, // Similar to above, only this time treat the step value as unsigned. // This covers loops that count up with an unsigned step. - const SCEV *UMul = - getMulExpr(CastedMaxBECount, - getTruncateOrZeroExtend(Step, Start->getType())); + const SCEV *UMul = getMulExpr(CastedMaxBECount, Step); Add = getAddExpr(Start, UMul); OperandExtendedAdd = getAddExpr(getSignExtendExpr(Start, WideTy), @@ -1418,7 +1410,7 @@ const SCEV *ScalarEvolution::getAddExpr(SmallVectorImpl<const SCEV *> &Ops, // If we deleted at least one add, we added operands to the end of the list, // and they are not necessarily sorted. Recurse to resort and resimplify - // any operands we just aquired. + // any operands we just acquired. if (DeletedAdd) return getAddExpr(Ops); } @@ -1725,7 +1717,7 @@ const SCEV *ScalarEvolution::getMulExpr(SmallVectorImpl<const SCEV *> &Ops, // If we deleted at least one mul, we added operands to the end of the list, // and they are not necessarily sorted. Recurse to resort and resimplify - // any operands we just aquired. + // any operands we just acquired. if (DeletedMul) return getMulExpr(Ops); } @@ -1973,6 +1965,12 @@ ScalarEvolution::getAddRecExpr(SmallVectorImpl<const SCEV *> &Operands, return getAddRecExpr(Operands, L, HasNUW, HasNSW); // {X,+,0} --> X } + // It's tempting to want to call getMaxBackedgeTakenCount count here and + // use that information to infer NUW and NSW flags. However, computing a + // BE count requires calling getAddRecExpr, so we may not yet have a + // meaningful BE count at this point (and if we don't, we'd be stuck + // with a SCEVCouldNotCompute as the cached BE count). + // If HasNSW is true and all the operands are non-negative, infer HasNUW. if (!HasNUW && HasNSW) { bool All = true; @@ -2308,7 +2306,7 @@ const SCEV *ScalarEvolution::getUnknown(Value *V) { /// has access to target-specific information. bool ScalarEvolution::isSCEVable(const Type *Ty) const { // Integers and pointers are always SCEVable. - return Ty->isIntegerTy() || isa<PointerType>(Ty); + return Ty->isIntegerTy() || Ty->isPointerTy(); } /// getTypeSizeInBits - Return the size in bits of the specified type, @@ -2326,7 +2324,7 @@ uint64_t ScalarEvolution::getTypeSizeInBits(const Type *Ty) const { // The only other support type is pointer. Without TargetData, conservatively // assume pointers are 64-bit. - assert(isa<PointerType>(Ty) && "isSCEVable permitted a non-SCEVable type!"); + assert(Ty->isPointerTy() && "isSCEVable permitted a non-SCEVable type!"); return 64; } @@ -2341,7 +2339,7 @@ const Type *ScalarEvolution::getEffectiveSCEVType(const Type *Ty) const { return Ty; // The only other support type is pointer. - assert(isa<PointerType>(Ty) && "Unexpected non-pointer non-integer type!"); + assert(Ty->isPointerTy() && "Unexpected non-pointer non-integer type!"); if (TD) return TD->getIntPtrType(getContext()); // Without TargetData, conservatively assume pointers are 64-bit. @@ -2412,8 +2410,8 @@ const SCEV * ScalarEvolution::getTruncateOrZeroExtend(const SCEV *V, const Type *Ty) { const Type *SrcTy = V->getType(); - assert((SrcTy->isIntegerTy() || isa<PointerType>(SrcTy)) && - (Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert((SrcTy->isIntegerTy() || SrcTy->isPointerTy()) && + (Ty->isIntegerTy() || Ty->isPointerTy()) && "Cannot truncate or zero extend with non-integer arguments!"); if (getTypeSizeInBits(SrcTy) == getTypeSizeInBits(Ty)) return V; // No conversion @@ -2429,8 +2427,8 @@ const SCEV * ScalarEvolution::getTruncateOrSignExtend(const SCEV *V, const Type *Ty) { const Type *SrcTy = V->getType(); - assert((SrcTy->isIntegerTy() || isa<PointerType>(SrcTy)) && - (Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert((SrcTy->isIntegerTy() || SrcTy->isPointerTy()) && + (Ty->isIntegerTy() || Ty->isPointerTy()) && "Cannot truncate or zero extend with non-integer arguments!"); if (getTypeSizeInBits(SrcTy) == getTypeSizeInBits(Ty)) return V; // No conversion @@ -2445,8 +2443,8 @@ ScalarEvolution::getTruncateOrSignExtend(const SCEV *V, const SCEV * ScalarEvolution::getNoopOrZeroExtend(const SCEV *V, const Type *Ty) { const Type *SrcTy = V->getType(); - assert((SrcTy->isIntegerTy() || isa<PointerType>(SrcTy)) && - (Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert((SrcTy->isIntegerTy() || SrcTy->isPointerTy()) && + (Ty->isIntegerTy() || Ty->isPointerTy()) && "Cannot noop or zero extend with non-integer arguments!"); assert(getTypeSizeInBits(SrcTy) <= getTypeSizeInBits(Ty) && "getNoopOrZeroExtend cannot truncate!"); @@ -2461,8 +2459,8 @@ ScalarEvolution::getNoopOrZeroExtend(const SCEV *V, const Type *Ty) { const SCEV * ScalarEvolution::getNoopOrSignExtend(const SCEV *V, const Type *Ty) { const Type *SrcTy = V->getType(); - assert((SrcTy->isIntegerTy() || isa<PointerType>(SrcTy)) && - (Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert((SrcTy->isIntegerTy() || SrcTy->isPointerTy()) && + (Ty->isIntegerTy() || Ty->isPointerTy()) && "Cannot noop or sign extend with non-integer arguments!"); assert(getTypeSizeInBits(SrcTy) <= getTypeSizeInBits(Ty) && "getNoopOrSignExtend cannot truncate!"); @@ -2478,8 +2476,8 @@ ScalarEvolution::getNoopOrSignExtend(const SCEV *V, const Type *Ty) { const SCEV * ScalarEvolution::getNoopOrAnyExtend(const SCEV *V, const Type *Ty) { const Type *SrcTy = V->getType(); - assert((SrcTy->isIntegerTy() || isa<PointerType>(SrcTy)) && - (Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert((SrcTy->isIntegerTy() || SrcTy->isPointerTy()) && + (Ty->isIntegerTy() || Ty->isPointerTy()) && "Cannot noop or any extend with non-integer arguments!"); assert(getTypeSizeInBits(SrcTy) <= getTypeSizeInBits(Ty) && "getNoopOrAnyExtend cannot truncate!"); @@ -2493,8 +2491,8 @@ ScalarEvolution::getNoopOrAnyExtend(const SCEV *V, const Type *Ty) { const SCEV * ScalarEvolution::getTruncateOrNoop(const SCEV *V, const Type *Ty) { const Type *SrcTy = V->getType(); - assert((SrcTy->isIntegerTy() || isa<PointerType>(SrcTy)) && - (Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert((SrcTy->isIntegerTy() || SrcTy->isPointerTy()) && + (Ty->isIntegerTy() || Ty->isPointerTy()) && "Cannot truncate or noop with non-integer arguments!"); assert(getTypeSizeInBits(SrcTy) >= getTypeSizeInBits(Ty) && "getTruncateOrNoop cannot extend!"); @@ -2551,12 +2549,12 @@ PushDefUseChildren(Instruction *I, /// the Scalars map if they reference SymName. This is used during PHI /// resolution. void -ScalarEvolution::ForgetSymbolicName(Instruction *I, const SCEV *SymName) { +ScalarEvolution::ForgetSymbolicName(Instruction *PN, const SCEV *SymName) { SmallVector<Instruction *, 16> Worklist; - PushDefUseChildren(I, Worklist); + PushDefUseChildren(PN, Worklist); SmallPtrSet<Instruction *, 8> Visited; - Visited.insert(I); + Visited.insert(PN); while (!Worklist.empty()) { Instruction *I = Worklist.pop_back_val(); if (!Visited.insert(I)) continue; @@ -2570,12 +2568,15 @@ ScalarEvolution::ForgetSymbolicName(Instruction *I, const SCEV *SymName) { continue; // SCEVUnknown for a PHI either means that it has an unrecognized - // structure, or it's a PHI that's in the progress of being computed - // by createNodeForPHI. In the former case, additional loop trip - // count information isn't going to change anything. In the later - // case, createNodeForPHI will perform the necessary updates on its - // own when it gets to that point. - if (!isa<PHINode>(I) || !isa<SCEVUnknown>(It->second)) { + // structure, it's a PHI that's in the progress of being computed + // by createNodeForPHI, or it's a single-value PHI. In the first case, + // additional loop trip count information isn't going to change anything. + // In the second case, createNodeForPHI will perform the necessary + // updates on its own when it gets to that point. In the third, we do + // want to forget the SCEVUnknown. + if (!isa<PHINode>(I) || + !isa<SCEVUnknown>(It->second) || + (I != PN && It->second == SymName)) { ValuesAtScopes.erase(It->second); Scalars.erase(It); } @@ -2698,9 +2699,21 @@ const SCEV *ScalarEvolution::createNodeForPHI(PHINode *PN) { return SymbolicName; } - // It's tempting to recognize PHIs with a unique incoming value, however - // this leads passes like indvars to break LCSSA form. Fortunately, such - // PHIs are rare, as instcombine zaps them. + // If the PHI has a single incoming value, follow that value, unless the + // PHI's incoming blocks are in a different loop, in which case doing so + // risks breaking LCSSA form. Instcombine would normally zap these, but + // it doesn't have DominatorTree information, so it may miss cases. + if (Value *V = PN->hasConstantValue(DT)) { + bool AllSameLoop = true; + Loop *PNLoop = LI->getLoopFor(PN->getParent()); + for (size_t i = 0, e = PN->getNumIncomingValues(); i != e; ++i) + if (LI->getLoopFor(PN->getIncomingBlock(i)) != PNLoop) { + AllSameLoop = false; + break; + } + if (AllSameLoop) + return getSCEV(V); + } // If it's not a loop phi, we can't handle it yet. return getUnknown(PN); @@ -2733,7 +2746,7 @@ const SCEV *ScalarEvolution::createNodeForGEP(GEPOperator *GEP) { } else { // For an array, add the element offset, explicitly scaled. const SCEV *LocalOffset = getSCEV(Index); - // Getelementptr indicies are signed. + // Getelementptr indices are signed. LocalOffset = getTruncateOrSignExtend(LocalOffset, IntPtrTy); // Lower "inbounds" GEPs to NSW arithmetic. LocalOffset = getMulExpr(LocalOffset, getSizeOfExpr(*GTI), @@ -2936,7 +2949,6 @@ ScalarEvolution::getUnsignedRange(const SCEV *S) { if (const SCEVUnknown *U = dyn_cast<SCEVUnknown>(S)) { // For a SCEVUnknown, ask ValueTracking. - unsigned BitWidth = getTypeSizeInBits(U->getType()); APInt Mask = APInt::getAllOnesValue(BitWidth); APInt Zeros(BitWidth, 0), Ones(BitWidth, 0); ComputeMaskedBits(U->getValue(), Mask, Zeros, Ones, TD); @@ -3208,7 +3220,7 @@ const SCEV *ScalarEvolution::createSCEV(Value *V) { const Type *Z0Ty = Z0->getType(); unsigned Z0TySize = getTypeSizeInBits(Z0Ty); - // If C is a low-bits mask, the zero extend is zerving to + // If C is a low-bits mask, the zero extend is serving to // mask off the high bits. Complement the operand and // re-apply the zext. if (APIntOps::isMask(Z0TySize, CI->getValue())) @@ -3393,7 +3405,7 @@ PushLoopPHIs(const Loop *L, SmallVectorImpl<Instruction *> &Worklist) { const ScalarEvolution::BackedgeTakenInfo & ScalarEvolution::getBackedgeTakenInfo(const Loop *L) { // Initially insert a CouldNotCompute for this loop. If the insertion - // succeeds, procede to actually compute a backedge-taken count and + // succeeds, proceed to actually compute a backedge-taken count and // update the value. The temporary CouldNotCompute value tells SCEV // code elsewhere that it shouldn't attempt to request a new // backedge-taken count, which could result in infinite recursion. @@ -3485,6 +3497,35 @@ void ScalarEvolution::forgetLoop(const Loop *L) { } } +/// forgetValue - This method should be called by the client when it has +/// changed a value in a way that may effect its value, or which may +/// disconnect it from a def-use chain linking it to a loop. +void ScalarEvolution::forgetValue(Value *V) { + Instruction *I = dyn_cast<Instruction>(V); + if (!I) return; + + // Drop information about expressions based on loop-header PHIs. + SmallVector<Instruction *, 16> Worklist; + Worklist.push_back(I); + + SmallPtrSet<Instruction *, 8> Visited; + while (!Worklist.empty()) { + I = Worklist.pop_back_val(); + if (!Visited.insert(I)) continue; + + std::map<SCEVCallbackVH, const SCEV *>::iterator It = + Scalars.find(static_cast<Value *>(I)); + if (It != Scalars.end()) { + ValuesAtScopes.erase(It->second); + Scalars.erase(It); + if (PHINode *PN = dyn_cast<PHINode>(I)) + ConstantEvolutionLoopExitValue.erase(PN); + } + + PushDefUseChildren(I, Worklist); + } +} + /// ComputeBackedgeTakenCount - Compute the number of times the backedge /// of the specified loop will execute. ScalarEvolution::BackedgeTakenInfo @@ -3581,7 +3622,7 @@ ScalarEvolution::ComputeBackedgeTakenCountFromExit(const Loop *L, return getCouldNotCompute(); } - // Procede to the next level to examine the exit condition expression. + // Proceed to the next level to examine the exit condition expression. return ComputeBackedgeTakenCountFromExitCond(L, ExitBr->getCondition(), ExitBr->getSuccessor(0), ExitBr->getSuccessor(1)); @@ -3670,10 +3711,23 @@ ScalarEvolution::ComputeBackedgeTakenCountFromExitCond(const Loop *L, } // With an icmp, it may be feasible to compute an exact backedge-taken count. - // Procede to the next level to examine the icmp. + // Proceed to the next level to examine the icmp. if (ICmpInst *ExitCondICmp = dyn_cast<ICmpInst>(ExitCond)) return ComputeBackedgeTakenCountFromExitCondICmp(L, ExitCondICmp, TBB, FBB); + // Check for a constant condition. These are normally stripped out by + // SimplifyCFG, but ScalarEvolution may be used by a pass which wishes to + // preserve the CFG and is temporarily leaving constant conditions + // in place. + if (ConstantInt *CI = dyn_cast<ConstantInt>(ExitCond)) { + if (L->contains(FBB) == !CI->getZExtValue()) + // The backedge is always taken. + return getCouldNotCompute(); + else + // The backedge is never taken. + return getIntegerSCEV(0, CI->getType()); + } + // If it's not an integer or pointer comparison then compute it the hard way. return ComputeBackedgeTakenCountExhaustively(L, ExitCond, !L->contains(TBB)); } @@ -3697,14 +3751,10 @@ ScalarEvolution::ComputeBackedgeTakenCountFromExitCondICmp(const Loop *L, // Handle common loops like: for (X = "string"; *X; ++X) if (LoadInst *LI = dyn_cast<LoadInst>(ExitCond->getOperand(0))) if (Constant *RHS = dyn_cast<Constant>(ExitCond->getOperand(1))) { - const SCEV *ItCnt = + BackedgeTakenInfo ItCnt = ComputeLoadConstantCompareBackedgeTakenCount(LI, RHS, L, Cond); - if (!isa<SCEVCouldNotCompute>(ItCnt)) { - unsigned BitWidth = getTypeSizeInBits(ItCnt->getType()); - return BackedgeTakenInfo(ItCnt, - isa<SCEVConstant>(ItCnt) ? ItCnt : - getConstant(APInt::getMaxValue(BitWidth)-1)); - } + if (ItCnt.hasAnyInfo()) + return ItCnt; } const SCEV *LHS = getSCEV(ExitCond->getOperand(0)); @@ -3738,14 +3788,14 @@ ScalarEvolution::ComputeBackedgeTakenCountFromExitCondICmp(const Loop *L, switch (Cond) { case ICmpInst::ICMP_NE: { // while (X != Y) // Convert to: while (X-Y != 0) - const SCEV *TC = HowFarToZero(getMinusSCEV(LHS, RHS), L); - if (!isa<SCEVCouldNotCompute>(TC)) return TC; + BackedgeTakenInfo BTI = HowFarToZero(getMinusSCEV(LHS, RHS), L); + if (BTI.hasAnyInfo()) return BTI; break; } case ICmpInst::ICMP_EQ: { // while (X == Y) // Convert to: while (X-Y == 0) - const SCEV *TC = HowFarToNonZero(getMinusSCEV(LHS, RHS), L); - if (!isa<SCEVCouldNotCompute>(TC)) return TC; + BackedgeTakenInfo BTI = HowFarToNonZero(getMinusSCEV(LHS, RHS), L); + if (BTI.hasAnyInfo()) return BTI; break; } case ICmpInst::ICMP_SLT: { @@ -3832,7 +3882,7 @@ GetAddressedElementFromGlobal(GlobalVariable *GV, /// ComputeLoadConstantCompareBackedgeTakenCount - Given an exit condition of /// 'icmp op load X, cst', try to see if we can compute the backedge /// execution count. -const SCEV * +ScalarEvolution::BackedgeTakenInfo ScalarEvolution::ComputeLoadConstantCompareBackedgeTakenCount( LoadInst *LI, Constant *RHS, @@ -3841,6 +3891,7 @@ ScalarEvolution::ComputeLoadConstantCompareBackedgeTakenCount( if (LI->isVolatile()) return getCouldNotCompute(); // Check to see if the loaded pointer is a getelementptr of a global. + // TODO: Use SCEV instead of manually grubbing with GEPs. GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(LI->getOperand(0)); if (!GEP) return getCouldNotCompute(); @@ -4190,14 +4241,15 @@ const SCEV *ScalarEvolution::computeSCEVAtScope(const SCEV *V, const Loop *L) { } } - Constant *C; + Constant *C = 0; if (const CmpInst *CI = dyn_cast<CmpInst>(I)) C = ConstantFoldCompareInstOperands(CI->getPredicate(), Operands[0], Operands[1], TD); else C = ConstantFoldInstOperands(I->getOpcode(), I->getType(), &Operands[0], Operands.size(), TD); - return getSCEV(C); + if (C) + return getSCEV(C); } } @@ -4405,7 +4457,8 @@ SolveQuadraticEquation(const SCEVAddRecExpr *AddRec, ScalarEvolution &SE) { /// HowFarToZero - Return the number of times a backedge comparing the specified /// value to zero will execute. If not computable, return CouldNotCompute. -const SCEV *ScalarEvolution::HowFarToZero(const SCEV *V, const Loop *L) { +ScalarEvolution::BackedgeTakenInfo +ScalarEvolution::HowFarToZero(const SCEV *V, const Loop *L) { // If the value is a constant if (const SCEVConstant *C = dyn_cast<SCEVConstant>(V)) { // If the value is already zero, the branch will execute zero times. @@ -4485,7 +4538,8 @@ const SCEV *ScalarEvolution::HowFarToZero(const SCEV *V, const Loop *L) { /// HowFarToNonZero - Return the number of times a backedge checking the /// specified value for nonzero will execute. If not computable, return /// CouldNotCompute -const SCEV *ScalarEvolution::HowFarToNonZero(const SCEV *V, const Loop *L) { +ScalarEvolution::BackedgeTakenInfo +ScalarEvolution::HowFarToNonZero(const SCEV *V, const Loop *L) { // Loops that look like: while (X == 0) are very strange indeed. We don't // handle them yet except for the trivial case. This could be expanded in the // future as needed. @@ -4726,7 +4780,7 @@ bool ScalarEvolution::isImpliedCond(Value *CondValue, ICmpInst::Predicate Pred, const SCEV *LHS, const SCEV *RHS, bool Inverse) { - // Recursivly handle And and Or conditions. + // Recursively handle And and Or conditions. if (BinaryOperator *BO = dyn_cast<BinaryOperator>(CondValue)) { if (BO->getOpcode() == Instruction::And) { if (!Inverse) @@ -4929,7 +4983,7 @@ bool ScalarEvolution::isImpliedCond(Value *CondValue, } /// isImpliedCondOperands - Test whether the condition described by Pred, -/// LHS, and RHS is true whenever the condition desribed by Pred, FoundLHS, +/// LHS, and RHS is true whenever the condition described by Pred, FoundLHS, /// and FoundRHS is true. bool ScalarEvolution::isImpliedCondOperands(ICmpInst::Predicate Pred, const SCEV *LHS, const SCEV *RHS, @@ -4944,7 +4998,7 @@ bool ScalarEvolution::isImpliedCondOperands(ICmpInst::Predicate Pred, } /// isImpliedCondOperandsHelper - Test whether the condition described by -/// Pred, LHS, and RHS is true whenever the condition desribed by Pred, +/// Pred, LHS, and RHS is true whenever the condition described by Pred, /// FoundLHS, and FoundRHS is true. bool ScalarEvolution::isImpliedCondOperandsHelper(ICmpInst::Predicate Pred, @@ -5102,7 +5156,7 @@ ScalarEvolution::HowManyLessThans(const SCEV *LHS, const SCEV *RHS, // If MaxEnd is within a step of the maximum integer value in its type, // adjust it down to the minimum value which would produce the same effect. - // This allows the subsequent ceiling divison of (N+(step-1))/step to + // This allows the subsequent ceiling division of (N+(step-1))/step to // compute the correct value. const SCEV *StepMinusOne = getMinusSCEV(Step, getIntegerSCEV(1, Step->getType())); @@ -5319,8 +5373,8 @@ ScalarEvolution::ScalarEvolution() bool ScalarEvolution::runOnFunction(Function &F) { this->F = &F; LI = &getAnalysis<LoopInfo>(); - DT = &getAnalysis<DominatorTree>(); TD = getAnalysisIfAvailable<TargetData>(); + DT = &getAnalysis<DominatorTree>(); return false; } @@ -5379,7 +5433,7 @@ static void PrintLoopInfo(raw_ostream &OS, ScalarEvolution *SE, } void ScalarEvolution::print(raw_ostream &OS, const Module *) const { - // ScalarEvolution's implementaiton of the print method is to print + // ScalarEvolution's implementation of the print method is to print // out SCEV values of all instructions that are interesting. Doing // this potentially causes it to create new SCEV objects though, // which technically conflicts with the const qualifier. This isn't diff --git a/lib/Analysis/ScalarEvolutionAliasAnalysis.cpp b/lib/Analysis/ScalarEvolutionAliasAnalysis.cpp index 498c4a8..17b254f 100644 --- a/lib/Analysis/ScalarEvolutionAliasAnalysis.cpp +++ b/lib/Analysis/ScalarEvolutionAliasAnalysis.cpp @@ -10,6 +10,10 @@ // This file defines the ScalarEvolutionAliasAnalysis pass, which implements a // simple alias analysis implemented in terms of ScalarEvolution queries. // +// This differs from traditional loop dependence analysis in that it tests +// for dependencies within a single iteration of a loop, rather than +// dependences between different iterations. +// // ScalarEvolution has a more complete understanding of pointer arithmetic // than BasicAliasAnalysis' collection of ad-hoc analyses. // @@ -41,7 +45,7 @@ namespace { return (AliasAnalysis*)this; return this; } - + private: virtual void getAnalysisUsage(AnalysisUsage &AU) const; virtual bool runOnFunction(Function &F); @@ -89,7 +93,7 @@ ScalarEvolutionAliasAnalysis::GetBaseValue(const SCEV *S) { } else if (const SCEVAddExpr *A = dyn_cast<SCEVAddExpr>(S)) { // If there's a pointer operand, it'll be sorted at the end of the list. const SCEV *Last = A->getOperand(A->getNumOperands()-1); - if (isa<PointerType>(Last->getType())) + if (Last->getType()->isPointerTy()) return GetBaseValue(Last); } else if (const SCEVUnknown *U = dyn_cast<SCEVUnknown>(S)) { // This is a leaf node. diff --git a/lib/Analysis/ScalarEvolutionExpander.cpp b/lib/Analysis/ScalarEvolutionExpander.cpp index c2e1f89..f5f10c8 100644 --- a/lib/Analysis/ScalarEvolutionExpander.cpp +++ b/lib/Analysis/ScalarEvolutionExpander.cpp @@ -144,15 +144,34 @@ Value *SCEVExpander::InsertBinop(Instruction::BinaryOps Opcode, } } + // Save the original insertion point so we can restore it when we're done. + BasicBlock *SaveInsertBB = Builder.GetInsertBlock(); + BasicBlock::iterator SaveInsertPt = Builder.GetInsertPoint(); + + // Move the insertion point out of as many loops as we can. + while (const Loop *L = SE.LI->getLoopFor(Builder.GetInsertBlock())) { + if (!L->isLoopInvariant(LHS) || !L->isLoopInvariant(RHS)) break; + BasicBlock *Preheader = L->getLoopPreheader(); + if (!Preheader) break; + + // Ok, move up a level. + Builder.SetInsertPoint(Preheader, Preheader->getTerminator()); + } + // If we haven't found this binop, insert it. Value *BO = Builder.CreateBinOp(Opcode, LHS, RHS, "tmp"); rememberInstruction(BO); + + // Restore the original insert point. + if (SaveInsertBB) + restoreInsertPoint(SaveInsertBB, SaveInsertPt); + return BO; } /// FactorOutConstant - Test if S is divisible by Factor, using signed /// division. If so, update S with Factor divided out and return true. -/// S need not be evenly divisble if a reasonable remainder can be +/// S need not be evenly divisible if a reasonable remainder can be /// computed. /// TODO: When ScalarEvolution gets a SCEVSDivExpr, this can be made /// unnecessary; in its place, just signed-divide Ops[i] by the scale and @@ -462,7 +481,7 @@ Value *SCEVExpander::expandAddToGEP(const SCEV *const *op_begin, break; } - // If none of the operands were convertable to proper GEP indices, cast + // If none of the operands were convertible to proper GEP indices, cast // the base to i8* and do an ugly getelementptr with that. It's still // better than ptrtoint+arithmetic+inttoptr at least. if (!AnyNonZeroIndices) { @@ -493,12 +512,56 @@ Value *SCEVExpander::expandAddToGEP(const SCEV *const *op_begin, } } + // Save the original insertion point so we can restore it when we're done. + BasicBlock *SaveInsertBB = Builder.GetInsertBlock(); + BasicBlock::iterator SaveInsertPt = Builder.GetInsertPoint(); + + // Move the insertion point out of as many loops as we can. + while (const Loop *L = SE.LI->getLoopFor(Builder.GetInsertBlock())) { + if (!L->isLoopInvariant(V) || !L->isLoopInvariant(Idx)) break; + BasicBlock *Preheader = L->getLoopPreheader(); + if (!Preheader) break; + + // Ok, move up a level. + Builder.SetInsertPoint(Preheader, Preheader->getTerminator()); + } + // Emit a GEP. Value *GEP = Builder.CreateGEP(V, Idx, "uglygep"); rememberInstruction(GEP); + + // Restore the original insert point. + if (SaveInsertBB) + restoreInsertPoint(SaveInsertBB, SaveInsertPt); + return GEP; } + // Save the original insertion point so we can restore it when we're done. + BasicBlock *SaveInsertBB = Builder.GetInsertBlock(); + BasicBlock::iterator SaveInsertPt = Builder.GetInsertPoint(); + + // Move the insertion point out of as many loops as we can. + while (const Loop *L = SE.LI->getLoopFor(Builder.GetInsertBlock())) { + if (!L->isLoopInvariant(V)) break; + + bool AnyIndexNotLoopInvariant = false; + for (SmallVectorImpl<Value *>::const_iterator I = GepIndices.begin(), + E = GepIndices.end(); I != E; ++I) + if (!L->isLoopInvariant(*I)) { + AnyIndexNotLoopInvariant = true; + break; + } + if (AnyIndexNotLoopInvariant) + break; + + BasicBlock *Preheader = L->getLoopPreheader(); + if (!Preheader) break; + + // Ok, move up a level. + Builder.SetInsertPoint(Preheader, Preheader->getTerminator()); + } + // Insert a pretty getelementptr. Note that this GEP is not marked inbounds, // because ScalarEvolution may have changed the address arithmetic to // compute a value which is beyond the end of the allocated object. @@ -511,6 +574,11 @@ Value *SCEVExpander::expandAddToGEP(const SCEV *const *op_begin, "scevgep"); Ops.push_back(SE.getUnknown(GEP)); rememberInstruction(GEP); + + // Restore the original insert point. + if (SaveInsertBB) + restoreInsertPoint(SaveInsertBB, SaveInsertPt); + return expand(SE.getAddExpr(Ops)); } @@ -528,70 +596,179 @@ static bool isNonConstantNegative(const SCEV *F) { return SC->getValue()->getValue().isNegative(); } -Value *SCEVExpander::visitAddExpr(const SCEVAddExpr *S) { - int NumOperands = S->getNumOperands(); - const Type *Ty = SE.getEffectiveSCEVType(S->getType()); +/// PickMostRelevantLoop - Given two loops pick the one that's most relevant for +/// SCEV expansion. If they are nested, this is the most nested. If they are +/// neighboring, pick the later. +static const Loop *PickMostRelevantLoop(const Loop *A, const Loop *B, + DominatorTree &DT) { + if (!A) return B; + if (!B) return A; + if (A->contains(B)) return B; + if (B->contains(A)) return A; + if (DT.dominates(A->getHeader(), B->getHeader())) return B; + if (DT.dominates(B->getHeader(), A->getHeader())) return A; + return A; // Arbitrarily break the tie. +} - // Find the index of an operand to start with. Choose the operand with - // pointer type, if there is one, or the last operand otherwise. - int PIdx = 0; - for (; PIdx != NumOperands - 1; ++PIdx) - if (isa<PointerType>(S->getOperand(PIdx)->getType())) break; +/// GetRelevantLoop - Get the most relevant loop associated with the given +/// expression, according to PickMostRelevantLoop. +static const Loop *GetRelevantLoop(const SCEV *S, LoopInfo &LI, + DominatorTree &DT) { + if (isa<SCEVConstant>(S)) + return 0; + if (const SCEVUnknown *U = dyn_cast<SCEVUnknown>(S)) { + if (const Instruction *I = dyn_cast<Instruction>(U->getValue())) + return LI.getLoopFor(I->getParent()); + return 0; + } + if (const SCEVNAryExpr *N = dyn_cast<SCEVNAryExpr>(S)) { + const Loop *L = 0; + if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(S)) + L = AR->getLoop(); + for (SCEVNAryExpr::op_iterator I = N->op_begin(), E = N->op_end(); + I != E; ++I) + L = PickMostRelevantLoop(L, GetRelevantLoop(*I, LI, DT), DT); + return L; + } + if (const SCEVCastExpr *C = dyn_cast<SCEVCastExpr>(S)) + return GetRelevantLoop(C->getOperand(), LI, DT); + if (const SCEVUDivExpr *D = dyn_cast<SCEVUDivExpr>(S)) + return PickMostRelevantLoop(GetRelevantLoop(D->getLHS(), LI, DT), + GetRelevantLoop(D->getRHS(), LI, DT), + DT); + llvm_unreachable("Unexpected SCEV type!"); +} - // Expand code for the operand that we chose. - Value *V = expand(S->getOperand(PIdx)); +/// LoopCompare - Compare loops by PickMostRelevantLoop. +class LoopCompare { + DominatorTree &DT; +public: + explicit LoopCompare(DominatorTree &dt) : DT(dt) {} + + bool operator()(std::pair<const Loop *, const SCEV *> LHS, + std::pair<const Loop *, const SCEV *> RHS) const { + // Compare loops with PickMostRelevantLoop. + if (LHS.first != RHS.first) + return PickMostRelevantLoop(LHS.first, RHS.first, DT) != LHS.first; + + // If one operand is a non-constant negative and the other is not, + // put the non-constant negative on the right so that a sub can + // be used instead of a negate and add. + if (isNonConstantNegative(LHS.second)) { + if (!isNonConstantNegative(RHS.second)) + return false; + } else if (isNonConstantNegative(RHS.second)) + return true; - // Turn things like ptrtoint+arithmetic+inttoptr into GEP. See the - // comments on expandAddToGEP for details. - if (const PointerType *PTy = dyn_cast<PointerType>(V->getType())) { - // Take the operand at PIdx out of the list. - const SmallVectorImpl<const SCEV *> &Ops = S->getOperands(); - SmallVector<const SCEV *, 8> NewOps; - NewOps.insert(NewOps.end(), Ops.begin(), Ops.begin() + PIdx); - NewOps.insert(NewOps.end(), Ops.begin() + PIdx + 1, Ops.end()); - // Make a GEP. - return expandAddToGEP(NewOps.begin(), NewOps.end(), PTy, Ty, V); + // Otherwise they are equivalent according to this comparison. + return false; } +}; - // Otherwise, we'll expand the rest of the SCEVAddExpr as plain integer - // arithmetic. - V = InsertNoopCastOfTo(V, Ty); +Value *SCEVExpander::visitAddExpr(const SCEVAddExpr *S) { + const Type *Ty = SE.getEffectiveSCEVType(S->getType()); - // Emit a bunch of add instructions - for (int i = NumOperands-1; i >= 0; --i) { - if (i == PIdx) continue; - const SCEV *Op = S->getOperand(i); - if (isNonConstantNegative(Op)) { + // Collect all the add operands in a loop, along with their associated loops. + // Iterate in reverse so that constants are emitted last, all else equal, and + // so that pointer operands are inserted first, which the code below relies on + // to form more involved GEPs. + SmallVector<std::pair<const Loop *, const SCEV *>, 8> OpsAndLoops; + for (std::reverse_iterator<SCEVAddExpr::op_iterator> I(S->op_end()), + E(S->op_begin()); I != E; ++I) + OpsAndLoops.push_back(std::make_pair(GetRelevantLoop(*I, *SE.LI, *SE.DT), + *I)); + + // Sort by loop. Use a stable sort so that constants follow non-constants and + // pointer operands precede non-pointer operands. + std::stable_sort(OpsAndLoops.begin(), OpsAndLoops.end(), LoopCompare(*SE.DT)); + + // Emit instructions to add all the operands. Hoist as much as possible + // out of loops, and form meaningful getelementptrs where possible. + Value *Sum = 0; + for (SmallVectorImpl<std::pair<const Loop *, const SCEV *> >::iterator + I = OpsAndLoops.begin(), E = OpsAndLoops.end(); I != E; ) { + const Loop *CurLoop = I->first; + const SCEV *Op = I->second; + if (!Sum) { + // This is the first operand. Just expand it. + Sum = expand(Op); + ++I; + } else if (const PointerType *PTy = dyn_cast<PointerType>(Sum->getType())) { + // The running sum expression is a pointer. Try to form a getelementptr + // at this level with that as the base. + SmallVector<const SCEV *, 4> NewOps; + for (; I != E && I->first == CurLoop; ++I) + NewOps.push_back(I->second); + Sum = expandAddToGEP(NewOps.begin(), NewOps.end(), PTy, Ty, Sum); + } else if (const PointerType *PTy = dyn_cast<PointerType>(Op->getType())) { + // The running sum is an integer, and there's a pointer at this level. + // Try to form a getelementptr. + SmallVector<const SCEV *, 4> NewOps; + NewOps.push_back(SE.getUnknown(Sum)); + for (++I; I != E && I->first == CurLoop; ++I) + NewOps.push_back(I->second); + Sum = expandAddToGEP(NewOps.begin(), NewOps.end(), PTy, Ty, expand(Op)); + } else if (isNonConstantNegative(Op)) { + // Instead of doing a negate and add, just do a subtract. Value *W = expandCodeFor(SE.getNegativeSCEV(Op), Ty); - V = InsertBinop(Instruction::Sub, V, W); + Sum = InsertNoopCastOfTo(Sum, Ty); + Sum = InsertBinop(Instruction::Sub, Sum, W); + ++I; } else { + // A simple add. Value *W = expandCodeFor(Op, Ty); - V = InsertBinop(Instruction::Add, V, W); + Sum = InsertNoopCastOfTo(Sum, Ty); + // Canonicalize a constant to the RHS. + if (isa<Constant>(Sum)) std::swap(Sum, W); + Sum = InsertBinop(Instruction::Add, Sum, W); + ++I; } } - return V; + + return Sum; } Value *SCEVExpander::visitMulExpr(const SCEVMulExpr *S) { const Type *Ty = SE.getEffectiveSCEVType(S->getType()); - int FirstOp = 0; // Set if we should emit a subtract. - if (const SCEVConstant *SC = dyn_cast<SCEVConstant>(S->getOperand(0))) - if (SC->getValue()->isAllOnesValue()) - FirstOp = 1; - - int i = S->getNumOperands()-2; - Value *V = expandCodeFor(S->getOperand(i+1), Ty); - - // Emit a bunch of multiply instructions - for (; i >= FirstOp; --i) { - Value *W = expandCodeFor(S->getOperand(i), Ty); - V = InsertBinop(Instruction::Mul, V, W); + + // Collect all the mul operands in a loop, along with their associated loops. + // Iterate in reverse so that constants are emitted last, all else equal. + SmallVector<std::pair<const Loop *, const SCEV *>, 8> OpsAndLoops; + for (std::reverse_iterator<SCEVMulExpr::op_iterator> I(S->op_end()), + E(S->op_begin()); I != E; ++I) + OpsAndLoops.push_back(std::make_pair(GetRelevantLoop(*I, *SE.LI, *SE.DT), + *I)); + + // Sort by loop. Use a stable sort so that constants follow non-constants. + std::stable_sort(OpsAndLoops.begin(), OpsAndLoops.end(), LoopCompare(*SE.DT)); + + // Emit instructions to mul all the operands. Hoist as much as possible + // out of loops. + Value *Prod = 0; + for (SmallVectorImpl<std::pair<const Loop *, const SCEV *> >::iterator + I = OpsAndLoops.begin(), E = OpsAndLoops.end(); I != E; ) { + const SCEV *Op = I->second; + if (!Prod) { + // This is the first operand. Just expand it. + Prod = expand(Op); + ++I; + } else if (Op->isAllOnesValue()) { + // Instead of doing a multiply by negative one, just do a negate. + Prod = InsertNoopCastOfTo(Prod, Ty); + Prod = InsertBinop(Instruction::Sub, Constant::getNullValue(Ty), Prod); + ++I; + } else { + // A simple mul. + Value *W = expandCodeFor(Op, Ty); + Prod = InsertNoopCastOfTo(Prod, Ty); + // Canonicalize a constant to the RHS. + if (isa<Constant>(Prod)) std::swap(Prod, W); + Prod = InsertBinop(Instruction::Mul, Prod, W); + ++I; + } } - // -1 * ... ---> 0 - ... - if (FirstOp == 1) - V = InsertBinop(Instruction::Sub, Constant::getNullValue(Ty), V); - return V; + return Prod; } Value *SCEVExpander::visitUDivExpr(const SCEVUDivExpr *S) { @@ -658,6 +835,19 @@ SCEVExpander::getAddRecExprPHILiterally(const SCEVAddRecExpr *Normalized, IncV = 0; break; } + // If any of the operands don't dominate the insert position, bail. + // Addrec operands are always loop-invariant, so this can only happen + // if there are instructions which haven't been hoisted. + for (User::op_iterator OI = IncV->op_begin()+1, + OE = IncV->op_end(); OI != OE; ++OI) + if (Instruction *OInst = dyn_cast<Instruction>(OI)) + if (!SE.DT->dominates(OInst, IVIncInsertPos)) { + IncV = 0; + break; + } + if (!IncV) + break; + // Advance to the next instruction. IncV = dyn_cast<Instruction>(IncV->getOperand(0)); if (!IncV) break; @@ -702,7 +892,7 @@ SCEVExpander::getAddRecExprPHILiterally(const SCEVAddRecExpr *Normalized, // negative, insert a sub instead of an add for the increment (unless it's a // constant, because subtracts of constants are canonicalized to adds). const SCEV *Step = Normalized->getStepRecurrence(SE); - bool isPointer = isa<PointerType>(ExpandTy); + bool isPointer = ExpandTy->isPointerTy(); bool isNegative = !isPointer && isNonConstantNegative(Step); if (isNegative) Step = SE.getNegativeSCEV(Step); @@ -807,7 +997,7 @@ Value *SCEVExpander::expandAddRecExprLiterally(const SCEVAddRecExpr *S) { const Type *ExpandTy = PostLoopScale ? IntTy : STy; PHINode *PN = getAddRecExprPHILiterally(Normalized, L, ExpandTy, IntTy); - // Accomodate post-inc mode, if necessary. + // Accommodate post-inc mode, if necessary. Value *Result; if (L != PostIncLoop) Result = PN; @@ -852,7 +1042,7 @@ Value *SCEVExpander::visitAddRecExpr(const SCEVAddRecExpr *S) { PHINode *CanonicalIV = 0; if (PHINode *PN = L->getCanonicalInductionVariable()) if (SE.isSCEVable(PN->getType()) && - isa<IntegerType>(SE.getEffectiveSCEVType(PN->getType())) && + SE.getEffectiveSCEVType(PN->getType())->isIntegerTy() && SE.getTypeSizeInBits(PN->getType()) >= SE.getTypeSizeInBits(Ty)) CanonicalIV = PN; @@ -1074,7 +1264,7 @@ Value *SCEVExpander::expand(const SCEV *S) { // If the SCEV is computable at this level, insert it into the header // after the PHIs (and after any other instructions that we've inserted // there) so that it is guaranteed to dominate any user inside the loop. - if (L && S->hasComputableLoopEvolution(L)) + if (L && S->hasComputableLoopEvolution(L) && L != PostIncLoop) InsertPt = L->getHeader()->getFirstNonPHI(); while (isInsertedInstruction(InsertPt)) InsertPt = llvm::next(BasicBlock::iterator(InsertPt)); @@ -1118,7 +1308,7 @@ void SCEVExpander::rememberInstruction(Value *I) { } void SCEVExpander::restoreInsertPoint(BasicBlock *BB, BasicBlock::iterator I) { - // If we aquired more instructions since the old insert point was saved, + // If we acquired more instructions since the old insert point was saved, // advance past them. while (isInsertedInstruction(I)) ++I; diff --git a/lib/Analysis/ValueTracking.cpp b/lib/Analysis/ValueTracking.cpp index 7cc9c0d..09344a3 100644 --- a/lib/Analysis/ValueTracking.cpp +++ b/lib/Analysis/ValueTracking.cpp @@ -49,7 +49,7 @@ void llvm::ComputeMaskedBits(Value *V, const APInt &Mask, assert(V && "No Value?"); assert(Depth <= MaxDepth && "Limit Search Depth"); unsigned BitWidth = Mask.getBitWidth(); - assert((V->getType()->isIntOrIntVectorTy() || isa<PointerType>(V->getType())) + assert((V->getType()->isIntOrIntVectorTy() || V->getType()->isPointerTy()) && "Not integer or pointer type!"); assert((!TD || TD->getTypeSizeInBits(V->getType()->getScalarType()) == BitWidth) && @@ -249,7 +249,7 @@ void llvm::ComputeMaskedBits(Value *V, const APInt &Mask, unsigned SrcBitWidth; // Note that we handle pointer operands here because of inttoptr/ptrtoint // which fall through here. - if (isa<PointerType>(SrcTy)) + if (SrcTy->isPointerTy()) SrcBitWidth = TD->getTypeSizeInBits(SrcTy); else SrcBitWidth = SrcTy->getScalarSizeInBits(); @@ -269,10 +269,10 @@ void llvm::ComputeMaskedBits(Value *V, const APInt &Mask, } case Instruction::BitCast: { const Type *SrcTy = I->getOperand(0)->getType(); - if ((SrcTy->isIntegerTy() || isa<PointerType>(SrcTy)) && + if ((SrcTy->isIntegerTy() || SrcTy->isPointerTy()) && // TODO: For now, not handling conversions like: // (bitcast i64 %x to <2 x i32>) - !isa<VectorType>(I->getType())) { + !I->getType()->isVectorTy()) { ComputeMaskedBits(I->getOperand(0), Mask, KnownZero, KnownOne, TD, Depth+1); return; @@ -980,7 +980,7 @@ bool llvm::CannotBeNegativeZero(const Value *V, unsigned Depth) { /// may not be represented in the result. static Value *GetLinearExpression(Value *V, APInt &Scale, APInt &Offset, const TargetData *TD, unsigned Depth) { - assert(isa<IntegerType>(V->getType()) && "Not an integer value"); + assert(V->getType()->isIntegerTy() && "Not an integer value"); // Limit our recursion depth. if (Depth == 6) { @@ -1253,7 +1253,7 @@ Value *llvm::FindInsertedValue(Value *V, const unsigned *idx_begin, if (idx_begin == idx_end) return V; // We have indices, so V should have an indexable type - assert((isa<StructType>(V->getType()) || isa<ArrayType>(V->getType())) + assert((V->getType()->isStructTy() || V->getType()->isArrayTy()) && "Not looking at a struct or array?"); assert(ExtractValueInst::getIndexedType(V->getType(), idx_begin, idx_end) && "Invalid indices for type?"); diff --git a/lib/AsmParser/LLParser.cpp b/lib/AsmParser/LLParser.cpp index 9cae0d2..8083a07 100644 --- a/lib/AsmParser/LLParser.cpp +++ b/lib/AsmParser/LLParser.cpp @@ -614,7 +614,7 @@ bool LLParser::ParseAlias(const std::string &Name, LocTy NameLoc, Aliasee = ID.ConstantVal; } - if (!isa<PointerType>(Aliasee->getType())) + if (!Aliasee->getType()->isPointerTy()) return Error(AliaseeLoc, "alias must have pointer type"); // Okay, create the alias but do not insert it into the module yet. @@ -685,7 +685,7 @@ bool LLParser::ParseGlobal(const std::string &Name, LocTy NameLoc, return true; } - if (isa<FunctionType>(Ty) || Ty->isLabelTy()) + if (Ty->isFunctionTy() || Ty->isLabelTy()) return Error(TyLoc, "invalid type for global variable"); GlobalVariable *GV = 0; @@ -791,7 +791,7 @@ GlobalValue *LLParser::GetGlobalVal(const std::string &Name, const Type *Ty, GlobalValue *FwdVal; if (const FunctionType *FT = dyn_cast<FunctionType>(PTy->getElementType())) { // Function types can return opaque but functions can't. - if (isa<OpaqueType>(FT->getReturnType())) { + if (FT->getReturnType()->isOpaqueTy()) { Error(Loc, "function may not return opaque type"); return 0; } @@ -836,7 +836,7 @@ GlobalValue *LLParser::GetGlobalVal(unsigned ID, const Type *Ty, LocTy Loc) { GlobalValue *FwdVal; if (const FunctionType *FT = dyn_cast<FunctionType>(PTy->getElementType())) { // Function types can return opaque but functions can't. - if (isa<OpaqueType>(FT->getReturnType())) { + if (FT->getReturnType()->isOpaqueTy()) { Error(Loc, "function may not return opaque type"); return 0; } @@ -1515,7 +1515,7 @@ bool LLParser::ParseArgumentList(std::vector<ArgInfo> &ArgList, Name = ""; } - if (!ArgTy->isFirstClassType() && !isa<OpaqueType>(ArgTy)) + if (!ArgTy->isFirstClassType() && !ArgTy->isOpaqueTy()) return Error(TypeLoc, "invalid type for function argument"); ArgList.push_back(ArgInfo(TypeLoc, ArgTy, Attrs, Name)); @@ -1785,7 +1785,7 @@ Value *LLParser::PerFunctionState::GetVal(const std::string &Name, } // Don't make placeholders with invalid type. - if (!Ty->isFirstClassType() && !isa<OpaqueType>(Ty) && !Ty->isLabelTy()) { + if (!Ty->isFirstClassType() && !Ty->isOpaqueTy() && !Ty->isLabelTy()) { P.Error(Loc, "invalid use of a non-first-class type"); return 0; } @@ -1826,7 +1826,7 @@ Value *LLParser::PerFunctionState::GetVal(unsigned ID, const Type *Ty, return 0; } - if (!Ty->isFirstClassType() && !isa<OpaqueType>(Ty) && !Ty->isLabelTy()) { + if (!Ty->isFirstClassType() && !Ty->isOpaqueTy() && !Ty->isLabelTy()) { P.Error(Loc, "invalid use of a non-first-class type"); return 0; } @@ -2256,7 +2256,7 @@ bool LLParser::ParseValID(ValID &ID, PerFunctionState *PFS) { } else { assert(Opc == Instruction::ICmp && "Unexpected opcode for CmpInst!"); if (!Val0->getType()->isIntOrIntVectorTy() && - !isa<PointerType>(Val0->getType())) + !Val0->getType()->isPointerTy()) return Error(ID.Loc, "icmp requires pointer or integer operands"); ID.ConstantVal = ConstantExpr::getICmp(Pred, Val0, Val1); } @@ -2370,7 +2370,7 @@ bool LLParser::ParseValID(ValID &ID, PerFunctionState *PFS) { return true; if (Opc == Instruction::GetElementPtr) { - if (Elts.size() == 0 || !isa<PointerType>(Elts[0]->getType())) + if (Elts.size() == 0 || !Elts[0]->getType()->isPointerTy()) return Error(ID.Loc, "getelementptr requires pointer operand"); if (!GetElementPtrInst::getIndexedType(Elts[0]->getType(), @@ -2470,7 +2470,7 @@ bool LLParser::ParseGlobalValueVector(SmallVectorImpl<Constant*> &Elts) { bool LLParser::ConvertValIDToValue(const Type *Ty, ValID &ID, Value *&V, PerFunctionState *PFS) { - if (isa<FunctionType>(Ty)) + if (Ty->isFunctionTy()) return Error(ID.Loc, "functions are not values, refer to them as pointers"); switch (ID.Kind) { @@ -2509,7 +2509,7 @@ bool LLParser::ConvertValIDToValue(const Type *Ty, ValID &ID, Value *&V, V = GetGlobalVal(ID.UIntVal, Ty, ID.Loc); return V == 0; case ValID::t_APSInt: - if (!isa<IntegerType>(Ty)) + if (!Ty->isIntegerTy()) return Error(ID.Loc, "integer constant must have integer type"); ID.APSIntVal.extOrTrunc(Ty->getPrimitiveSizeInBits()); V = ConstantInt::get(Context, ID.APSIntVal); @@ -2535,19 +2535,19 @@ bool LLParser::ConvertValIDToValue(const Type *Ty, ValID &ID, Value *&V, return false; case ValID::t_Null: - if (!isa<PointerType>(Ty)) + if (!Ty->isPointerTy()) return Error(ID.Loc, "null must be a pointer type"); V = ConstantPointerNull::get(cast<PointerType>(Ty)); return false; case ValID::t_Undef: // FIXME: LabelTy should not be a first-class type. if ((!Ty->isFirstClassType() || Ty->isLabelTy()) && - !isa<OpaqueType>(Ty)) + !Ty->isOpaqueTy()) return Error(ID.Loc, "invalid type for undef constant"); V = UndefValue::get(Ty); return false; case ValID::t_EmptyArray: - if (!isa<ArrayType>(Ty) || cast<ArrayType>(Ty)->getNumElements() != 0) + if (!Ty->isArrayTy() || cast<ArrayType>(Ty)->getNumElements() != 0) return Error(ID.Loc, "invalid empty array initializer"); V = UndefValue::get(Ty); return false; @@ -2662,7 +2662,7 @@ bool LLParser::ParseFunctionHeader(Function *&Fn, bool isDefine) { } if (!FunctionType::isValidReturnType(RetType) || - isa<OpaqueType>(RetType)) + RetType->isOpaqueTy()) return Error(RetTypeLoc, "invalid function return type"); LocTy NameLoc = Lex.getLoc(); @@ -3186,7 +3186,7 @@ bool LLParser::ParseSwitch(Instruction *&Inst, PerFunctionState &PFS) { ParseToken(lltok::lsquare, "expected '[' with switch table")) return true; - if (!isa<IntegerType>(Cond->getType())) + if (!Cond->getType()->isIntegerTy()) return Error(CondLoc, "switch condition must have integer type"); // Parse the jump table pairs. @@ -3229,7 +3229,7 @@ bool LLParser::ParseIndirectBr(Instruction *&Inst, PerFunctionState &PFS) { ParseToken(lltok::lsquare, "expected '[' with indirectbr")) return true; - if (!isa<PointerType>(Address->getType())) + if (!Address->getType()->isPointerTy()) return Error(AddrLoc, "indirectbr address must have pointer type"); // Parse the destination list. @@ -3436,7 +3436,7 @@ bool LLParser::ParseCompare(Instruction *&Inst, PerFunctionState &PFS, } else { assert(Opc == Instruction::ICmp && "Unknown opcode for CmpInst!"); if (!LHS->getType()->isIntOrIntVectorTy() && - !isa<PointerType>(LHS->getType())) + !LHS->getType()->isPointerTy()) return Error(Loc, "icmp requires integer operands"); Inst = new ICmpInst(CmpInst::Predicate(Pred), LHS, RHS); } @@ -3761,7 +3761,7 @@ bool LLParser::ParseFree(Instruction *&Inst, PerFunctionState &PFS, BasicBlock* BB) { Value *Val; LocTy Loc; if (ParseTypeAndValue(Val, Loc, PFS)) return true; - if (!isa<PointerType>(Val->getType())) + if (!Val->getType()->isPointerTy()) return Error(Loc, "operand to free must be a pointer"); Inst = CallInst::CreateFree(Val, BB); return false; @@ -3778,7 +3778,7 @@ int LLParser::ParseLoad(Instruction *&Inst, PerFunctionState &PFS, ParseOptionalCommaAlign(Alignment, AteExtraComma)) return true; - if (!isa<PointerType>(Val->getType()) || + if (!Val->getType()->isPointerTy() || !cast<PointerType>(Val->getType())->getElementType()->isFirstClassType()) return Error(Loc, "load operand must be a pointer to a first class type"); @@ -3799,7 +3799,7 @@ int LLParser::ParseStore(Instruction *&Inst, PerFunctionState &PFS, ParseOptionalCommaAlign(Alignment, AteExtraComma)) return true; - if (!isa<PointerType>(Ptr->getType())) + if (!Ptr->getType()->isPointerTy()) return Error(PtrLoc, "store operand must be a pointer"); if (!Val->getType()->isFirstClassType()) return Error(Loc, "store operand must be a first class value"); @@ -3821,7 +3821,7 @@ bool LLParser::ParseGetResult(Instruction *&Inst, PerFunctionState &PFS) { ParseUInt32(Element, EltLoc)) return true; - if (!isa<StructType>(Val->getType()) && !isa<ArrayType>(Val->getType())) + if (!Val->getType()->isStructTy() && !Val->getType()->isArrayTy()) return Error(ValLoc, "getresult inst requires an aggregate operand"); if (!ExtractValueInst::getIndexedType(Val->getType(), Element)) return Error(EltLoc, "invalid getresult index for value"); @@ -3838,7 +3838,7 @@ int LLParser::ParseGetElementPtr(Instruction *&Inst, PerFunctionState &PFS) { if (ParseTypeAndValue(Ptr, Loc, PFS)) return true; - if (!isa<PointerType>(Ptr->getType())) + if (!Ptr->getType()->isPointerTy()) return Error(Loc, "base of getelementptr must be a pointer"); SmallVector<Value*, 16> Indices; @@ -3849,7 +3849,7 @@ int LLParser::ParseGetElementPtr(Instruction *&Inst, PerFunctionState &PFS) { break; } if (ParseTypeAndValue(Val, EltLoc, PFS)) return true; - if (!isa<IntegerType>(Val->getType())) + if (!Val->getType()->isIntegerTy()) return Error(EltLoc, "getelementptr index must be an integer"); Indices.push_back(Val); } diff --git a/lib/Bitcode/Reader/BitReader.cpp b/lib/Bitcode/Reader/BitReader.cpp index 7537435..15844c0 100644 --- a/lib/Bitcode/Reader/BitReader.cpp +++ b/lib/Bitcode/Reader/BitReader.cpp @@ -45,26 +45,44 @@ LLVMBool LLVMParseBitcodeInContext(LLVMContextRef ContextRef, /* Reads a module from the specified path, returning via the OutModule parameter a module provider which performs lazy deserialization. Returns 0 on success. Optionally returns a human-readable error message via OutMessage. */ -LLVMBool LLVMGetBitcodeModuleProvider(LLVMMemoryBufferRef MemBuf, - LLVMModuleProviderRef *OutMP, - char **OutMessage) { - return LLVMGetBitcodeModuleProviderInContext(wrap(&getGlobalContext()), - MemBuf, OutMP, OutMessage); -} - -LLVMBool LLVMGetBitcodeModuleProviderInContext(LLVMContextRef ContextRef, - LLVMMemoryBufferRef MemBuf, - LLVMModuleProviderRef *OutMP, - char **OutMessage) { +LLVMBool LLVMGetBitcodeModuleInContext(LLVMContextRef ContextRef, + LLVMMemoryBufferRef MemBuf, + LLVMModuleRef *OutM, + char **OutMessage) { std::string Message; - *OutMP = reinterpret_cast<LLVMModuleProviderRef>( - getLazyBitcodeModule(unwrap(MemBuf), *unwrap(ContextRef), &Message)); - if (!*OutMP) { + *OutM = wrap(getLazyBitcodeModule(unwrap(MemBuf), *unwrap(ContextRef), + &Message)); + if (!*OutM) { if (OutMessage) *OutMessage = strdup(Message.c_str()); return 1; } return 0; + +} + +LLVMBool LLVMGetBitcodeModule(LLVMMemoryBufferRef MemBuf, LLVMModuleRef *OutM, + char **OutMessage) { + return LLVMGetBitcodeModuleInContext(LLVMGetGlobalContext(), MemBuf, OutM, + OutMessage); +} + +/* Deprecated: Use LLVMGetBitcodeModuleInContext instead. */ +LLVMBool LLVMGetBitcodeModuleProviderInContext(LLVMContextRef ContextRef, + LLVMMemoryBufferRef MemBuf, + LLVMModuleProviderRef *OutMP, + char **OutMessage) { + return LLVMGetBitcodeModuleInContext(ContextRef, MemBuf, + reinterpret_cast<LLVMModuleRef*>(OutMP), + OutMessage); +} + +/* Deprecated: Use LLVMGetBitcodeModule instead. */ +LLVMBool LLVMGetBitcodeModuleProvider(LLVMMemoryBufferRef MemBuf, + LLVMModuleProviderRef *OutMP, + char **OutMessage) { + return LLVMGetBitcodeModuleProviderInContext(LLVMGetGlobalContext(), MemBuf, + OutMP, OutMessage); } diff --git a/lib/Bitcode/Reader/BitcodeReader.cpp b/lib/Bitcode/Reader/BitcodeReader.cpp index cebfbf6..a328837 100644 --- a/lib/Bitcode/Reader/BitcodeReader.cpp +++ b/lib/Bitcode/Reader/BitcodeReader.cpp @@ -963,12 +963,12 @@ bool BitcodeReader::ParseConstants() { V = Constant::getNullValue(CurTy); break; case bitc::CST_CODE_INTEGER: // INTEGER: [intval] - if (!isa<IntegerType>(CurTy) || Record.empty()) + if (!CurTy->isIntegerTy() || Record.empty()) return Error("Invalid CST_INTEGER record"); V = ConstantInt::get(CurTy, DecodeSignRotatedValue(Record[0])); break; case bitc::CST_CODE_WIDE_INTEGER: {// WIDE_INTEGER: [n x intval] - if (!isa<IntegerType>(CurTy) || Record.empty()) + if (!CurTy->isIntegerTy() || Record.empty()) return Error("Invalid WIDE_INTEGER record"); unsigned NumWords = Record.size(); @@ -1407,7 +1407,7 @@ bool BitcodeReader::ParseModule() { if (Record.size() < 6) return Error("Invalid MODULE_CODE_GLOBALVAR record"); const Type *Ty = getTypeByID(Record[0]); - if (!isa<PointerType>(Ty)) + if (!Ty->isPointerTy()) return Error("Global not a pointer type!"); unsigned AddressSpace = cast<PointerType>(Ty)->getAddressSpace(); Ty = cast<PointerType>(Ty)->getElementType(); @@ -1450,7 +1450,7 @@ bool BitcodeReader::ParseModule() { if (Record.size() < 8) return Error("Invalid MODULE_CODE_FUNCTION record"); const Type *Ty = getTypeByID(Record[0]); - if (!isa<PointerType>(Ty)) + if (!Ty->isPointerTy()) return Error("Function not a pointer type!"); const FunctionType *FTy = dyn_cast<FunctionType>(cast<PointerType>(Ty)->getElementType()); @@ -1491,7 +1491,7 @@ bool BitcodeReader::ParseModule() { if (Record.size() < 3) return Error("Invalid MODULE_ALIAS record"); const Type *Ty = getTypeByID(Record[0]); - if (!isa<PointerType>(Ty)) + if (!Ty->isPointerTy()) return Error("Function not a pointer type!"); GlobalAlias *NewGA = new GlobalAlias(Ty, GetDecodedLinkage(Record[2]), @@ -1622,6 +1622,7 @@ bool BitcodeReader::ParseFunctionBody(Function *F) { if (Stream.EnterSubBlock(bitc::FUNCTION_BLOCK_ID)) return Error("Malformed block record"); + InstructionList.clear(); unsigned ModuleValueListSize = ValueList.size(); // Add all the function arguments to the value table. @@ -1932,7 +1933,7 @@ bool BitcodeReader::ParseFunctionBody(Function *F) { const Type *ReturnType = F->getReturnType(); if (Vs.size() > 1 || - (isa<StructType>(ReturnType) && + (ReturnType->isStructTy() && (Vs.empty() || Vs[0]->getType() != ReturnType))) { Value *RV = UndefValue::get(ReturnType); for (unsigned i = 0, e = Vs.size(); i != e; ++i) { diff --git a/lib/Bitcode/Writer/ValueEnumerator.cpp b/lib/Bitcode/Writer/ValueEnumerator.cpp index 595497f..aa4c3af 100644 --- a/lib/Bitcode/Writer/ValueEnumerator.cpp +++ b/lib/Bitcode/Writer/ValueEnumerator.cpp @@ -27,7 +27,7 @@ static bool isSingleValueType(const std::pair<const llvm::Type*, } static bool isIntegerValue(const std::pair<const Value*, unsigned> &V) { - return isa<IntegerType>(V.first->getType()); + return V.first->getType()->isIntegerTy(); } static bool CompareByFrequency(const std::pair<const llvm::Type*, @@ -39,8 +39,6 @@ static bool CompareByFrequency(const std::pair<const llvm::Type*, /// ValueEnumerator - Enumerate module-level information. ValueEnumerator::ValueEnumerator(const Module *M) { - InstructionCount = 0; - // Enumerate the global variables. for (Module::const_global_iterator I = M->global_begin(), E = M->global_end(); I != E; ++I) @@ -377,6 +375,7 @@ void ValueEnumerator::EnumerateAttributes(const AttrListPtr &PAL) { void ValueEnumerator::incorporateFunction(const Function &F) { + InstructionCount = 0; NumModuleValues = Values.size(); // Adding function arguments to the value table. diff --git a/lib/CodeGen/AsmPrinter/AsmPrinter.cpp b/lib/CodeGen/AsmPrinter/AsmPrinter.cpp index fc08384..bd2b1b6 100644 --- a/lib/CodeGen/AsmPrinter/AsmPrinter.cpp +++ b/lib/CodeGen/AsmPrinter/AsmPrinter.cpp @@ -917,11 +917,10 @@ void AsmPrinter::EmitAlignment(unsigned NumBits, const GlobalValue *GV, if (NumBits == 0) return; // No need to emit alignment. - unsigned FillValue = 0; if (getCurrentSection()->getKind().isText()) - FillValue = MAI->getTextAlignFillValue(); - - OutStreamer.EmitValueToAlignment(1 << NumBits, FillValue, 1, 0); + OutStreamer.EmitCodeAlignment(1 << NumBits); + else + OutStreamer.EmitValueToAlignment(1 << NumBits, 0, 1, 0); } /// LowerConstant - Lower the specified LLVM Constant to an MCExpr. @@ -1717,7 +1716,7 @@ void AsmPrinter::EmitBasicBlockStart(const MachineBasicBlock *MBB) const { } // Print the main label for the block. - if (MBB->pred_empty() || MBB->isOnlyReachableByFallthrough()) { + if (MBB->pred_empty() || isBlockOnlyReachableByFallthrough(MBB)) { if (VerboseAsm) { // NOTE: Want this comment at start of line. O << MAI->getCommentString() << " BB#" << MBB->getNumber() << ':'; @@ -1764,6 +1763,39 @@ void AsmPrinter::printOffset(int64_t Offset) const { O << Offset; } +/// isBlockOnlyReachableByFallthough - Return true if the basic block has +/// exactly one predecessor and the control transfer mechanism between +/// the predecessor and this block is a fall-through. +bool AsmPrinter::isBlockOnlyReachableByFallthrough(const MachineBasicBlock *MBB) + const { + // If this is a landing pad, it isn't a fall through. If it has no preds, + // then nothing falls through to it. + if (MBB->isLandingPad() || MBB->pred_empty()) + return false; + + // If there isn't exactly one predecessor, it can't be a fall through. + MachineBasicBlock::const_pred_iterator PI = MBB->pred_begin(), PI2 = PI; + ++PI2; + if (PI2 != MBB->pred_end()) + return false; + + // The predecessor has to be immediately before this block. + const MachineBasicBlock *Pred = *PI; + + if (!Pred->isLayoutSuccessor(MBB)) + return false; + + // If the block is completely empty, then it definitely does fall through. + if (Pred->empty()) + return true; + + // Otherwise, check the last instruction. + const MachineInstr &LastInst = Pred->back(); + return !LastInst.getDesc().isBarrier(); +} + + + GCMetadataPrinter *AsmPrinter::GetOrCreateGCPrinter(GCStrategy *S) { if (!S->usesMetadata()) return 0; diff --git a/lib/CodeGen/AsmPrinter/DwarfDebug.cpp b/lib/CodeGen/AsmPrinter/DwarfDebug.cpp index 5093dd9..5ad1e5e 100644 --- a/lib/CodeGen/AsmPrinter/DwarfDebug.cpp +++ b/lib/CodeGen/AsmPrinter/DwarfDebug.cpp @@ -238,7 +238,18 @@ public: LIndex = DSI; } } - setLastInsn(LastInsn); + + unsigned CurrentLastInsnIndex = 0; + if (const MachineInstr *CL = getLastInsn()) + CurrentLastInsnIndex = MIIndexMap[CL]; + unsigned FIndex = MIIndexMap[getFirstInsn()]; + + // Set LastInsn as the last instruction for this scope only if + // it follows + // 1) this scope's first instruction and + // 2) current last instruction for this scope, if any. + if (LIndex >= CurrentLastInsnIndex && LIndex >= FIndex) + setLastInsn(LastInsn); } #ifndef NDEBUG @@ -1166,7 +1177,9 @@ DIE *DwarfDebug::createSubprogramDIE(const DISubprogram &SP, bool MakeDecl) { return SPDie; SPDie = new DIE(dwarf::DW_TAG_subprogram); - addString(SPDie, dwarf::DW_AT_name, dwarf::DW_FORM_string, SP.getName()); + // Constructors and operators for anonymous aggregates do not have names. + if (!SP.getName().empty()) + addString(SPDie, dwarf::DW_AT_name, dwarf::DW_FORM_string, SP.getName()); StringRef LinkageName = SP.getLinkageName(); if (!LinkageName.empty()) diff --git a/lib/CodeGen/AsmPrinter/DwarfException.cpp b/lib/CodeGen/AsmPrinter/DwarfException.cpp index c6c59f5..2b08ba4 100644 --- a/lib/CodeGen/AsmPrinter/DwarfException.cpp +++ b/lib/CodeGen/AsmPrinter/DwarfException.cpp @@ -638,18 +638,18 @@ void DwarfException::EmitExceptionTable() { const unsigned LandingPadSize = SizeOfEncodedValue(dwarf::DW_EH_PE_udata4); bool IsSJLJ = MAI->getExceptionHandlingType() == ExceptionHandling::SjLj; bool HaveTTData = IsSJLJ ? (!TypeInfos.empty() || !FilterIds.empty()) : true; - unsigned SizeSites; + unsigned CallSiteTableLength; if (IsSJLJ) - SizeSites = 0; + CallSiteTableLength = 0; else - SizeSites = CallSites.size() * + CallSiteTableLength = CallSites.size() * (SiteStartSize + SiteLengthSize + LandingPadSize); for (unsigned i = 0, e = CallSites.size(); i < e; ++i) { - SizeSites += MCAsmInfo::getULEB128Size(CallSites[i].Action); + CallSiteTableLength += MCAsmInfo::getULEB128Size(CallSites[i].Action); if (IsSJLJ) - SizeSites += MCAsmInfo::getULEB128Size(i); + CallSiteTableLength += MCAsmInfo::getULEB128Size(i); } // Type infos. @@ -698,7 +698,20 @@ void DwarfException::EmitExceptionTable() { Asm->OutStreamer.SwitchSection(LSDASection); Asm->EmitAlignment(2, 0, 0, false); + // Emit the LSDA. O << "GCC_except_table" << SubprogramCount << ":\n"; + EmitLabel("exception", SubprogramCount); + + if (IsSJLJ) { + SmallString<16> LSDAName; + raw_svector_ostream(LSDAName) << MAI->getPrivateGlobalPrefix() << + "_LSDA_" << Asm->getFunctionNumber(); + O << LSDAName.str() << ":\n"; + } + + // Emit the LSDA header. + EmitEncodingByte(dwarf::DW_EH_PE_omit, "@LPStart"); + EmitEncodingByte(TTypeEncoding, "@TType"); // The type infos need to be aligned. GCC does this by inserting padding just // before the type infos. However, this changes the size of the exception @@ -707,7 +720,7 @@ void DwarfException::EmitExceptionTable() { // So by increasing the size by inserting padding, you may increase the number // of bytes used for writing the size. If it increases, say by one byte, then // you now need to output one less byte of padding to get the type infos - // aligned. However this decreases the size of the exception table. This + // aligned. However this decreases the size of the exception table. This // changes the value you have to output for the exception table size. Due to // the variable length encoding, the number of bytes used for writing the // length may decrease. If so, you then have to increase the amount of @@ -716,41 +729,35 @@ void DwarfException::EmitExceptionTable() { // We chose another solution: don't output padding inside the table like GCC // does, instead output it before the table. unsigned SizeTypes = TypeInfos.size() * TypeFormatSize; - unsigned TyOffset = sizeof(int8_t) + // Call site format - MCAsmInfo::getULEB128Size(SizeSites) + // Call site table length - SizeSites + SizeActions + SizeTypes; - unsigned TotalSize = sizeof(int8_t) + // LPStart format - sizeof(int8_t) + // TType format - (HaveTTData ? - MCAsmInfo::getULEB128Size(TyOffset) : 0) + // TType base offset - TyOffset; + unsigned CallSiteTableLengthSize = + MCAsmInfo::getULEB128Size(CallSiteTableLength); + unsigned TTypeBaseOffset = + sizeof(int8_t) + // Call site format + CallSiteTableLengthSize + // Call site table length size + CallSiteTableLength + // Call site table length + SizeActions + // Actions size + SizeTypes; + unsigned TTypeBaseOffsetSize = MCAsmInfo::getULEB128Size(TTypeBaseOffset); + unsigned TotalSize = + sizeof(int8_t) + // LPStart format + sizeof(int8_t) + // TType format + (HaveTTData ? TTypeBaseOffsetSize : 0) + // TType base offset size + TTypeBaseOffset; // TType base offset unsigned SizeAlign = (4 - TotalSize) & 3; - for (unsigned i = 0; i != SizeAlign; ++i) { - Asm->EmitInt8(0); - EOL("Padding"); - } - - EmitLabel("exception", SubprogramCount); - - if (IsSJLJ) { - SmallString<16> LSDAName; - raw_svector_ostream(LSDAName) << MAI->getPrivateGlobalPrefix() << - "_LSDA_" << Asm->getFunctionNumber(); - O << LSDAName.str() << ":\n"; + if (HaveTTData) { + // Account for any extra padding that will be added to the call site table + // length. + EmitULEB128(TTypeBaseOffset, "@TType base offset", SizeAlign); + SizeAlign = 0; } - // Emit the header. - EmitEncodingByte(dwarf::DW_EH_PE_omit, "@LPStart"); - EmitEncodingByte(TTypeEncoding, "@TType"); - - if (HaveTTData) - EmitULEB128(TyOffset, "@TType base offset"); - // SjLj Exception handling if (IsSJLJ) { EmitEncodingByte(dwarf::DW_EH_PE_udata4, "Call site"); - EmitULEB128(SizeSites, "Call site table length"); + + // Add extra padding if it wasn't added to the TType base offset. + EmitULEB128(CallSiteTableLength, "Call site table length", SizeAlign); // Emit the landing pad site information. unsigned idx = 0; @@ -791,7 +798,9 @@ void DwarfException::EmitExceptionTable() { // Emit the landing pad call site table. EmitEncodingByte(dwarf::DW_EH_PE_udata4, "Call site"); - EmitULEB128(SizeSites, "Call site table length"); + + // Add extra padding if it wasn't added to the TType base offset. + EmitULEB128(CallSiteTableLength, "Call site table length", SizeAlign); for (SmallVectorImpl<CallSiteEntry>::const_iterator I = CallSites.begin(), E = CallSites.end(); I != E; ++I) { diff --git a/lib/CodeGen/AsmPrinter/DwarfPrinter.cpp b/lib/CodeGen/AsmPrinter/DwarfPrinter.cpp index 1299d04..28ff0eb 100644 --- a/lib/CodeGen/AsmPrinter/DwarfPrinter.cpp +++ b/lib/CodeGen/AsmPrinter/DwarfPrinter.cpp @@ -159,29 +159,35 @@ void DwarfPrinter::EmitSLEB128(int Value, const char *Desc) const { Value >>= 7; IsMore = Value != Sign || ((Byte ^ Sign) & 0x40) != 0; if (IsMore) Byte |= 0x80; - Asm->OutStreamer.EmitIntValue(Byte, 1, /*addrspace*/0); } while (IsMore); } /// EmitULEB128 - emit the specified signed leb128 value. -void DwarfPrinter::EmitULEB128(unsigned Value, const char *Desc) const { +void DwarfPrinter::EmitULEB128(unsigned Value, const char *Desc, + unsigned PadTo) const { if (Asm->VerboseAsm && Desc) Asm->OutStreamer.AddComment(Desc); - if (MAI->hasLEB128()) { + if (MAI->hasLEB128() && PadTo == 0) { O << "\t.uleb128\t" << Value; Asm->OutStreamer.AddBlankLine(); return; } - // If we don't have .uleb128, emit as .bytes. + // If we don't have .uleb128 or we want to emit padding, emit as .bytes. do { unsigned char Byte = static_cast<unsigned char>(Value & 0x7f); Value >>= 7; - if (Value) Byte |= 0x80; + if (Value || PadTo != 0) Byte |= 0x80; Asm->OutStreamer.EmitIntValue(Byte, 1, /*addrspace*/0); } while (Value); + + if (PadTo) { + if (PadTo > 1) + Asm->OutStreamer.EmitFill(PadTo - 1, 0x80/*fillval*/, 0/*addrspace*/); + Asm->OutStreamer.EmitFill(1, 0/*fillval*/, 0/*addrspace*/); + } } diff --git a/lib/CodeGen/AsmPrinter/DwarfPrinter.h b/lib/CodeGen/AsmPrinter/DwarfPrinter.h index 73de398..bd715f2 100644 --- a/lib/CodeGen/AsmPrinter/DwarfPrinter.h +++ b/lib/CodeGen/AsmPrinter/DwarfPrinter.h @@ -111,7 +111,8 @@ public: void EmitSLEB128(int Value, const char *Desc) const; /// EmitULEB128 - emit the specified unsigned leb128 value. - void EmitULEB128(unsigned Value, const char *Desc = 0) const; + void EmitULEB128(unsigned Value, const char *Desc = 0, + unsigned PadTo = 0) const; /// PrintLabelName - Print label name in form used by Dwarf writer. diff --git a/lib/CodeGen/CMakeLists.txt b/lib/CodeGen/CMakeLists.txt index 62cf339..d385b86 100644 --- a/lib/CodeGen/CMakeLists.txt +++ b/lib/CodeGen/CMakeLists.txt @@ -22,6 +22,7 @@ add_llvm_library(LLVMCodeGen LiveVariables.cpp LowerSubregs.cpp MachineBasicBlock.cpp + MachineCSE.cpp MachineDominators.cpp MachineFunction.cpp MachineFunctionAnalysis.cpp diff --git a/lib/CodeGen/CalcSpillWeights.cpp b/lib/CodeGen/CalcSpillWeights.cpp index 2bedd04..a328d0e 100644 --- a/lib/CodeGen/CalcSpillWeights.cpp +++ b/lib/CodeGen/CalcSpillWeights.cpp @@ -131,10 +131,7 @@ bool CalculateSpillWeights::runOnMachineFunction(MachineFunction &fn) { if (Hint.first || Hint.second) li.weight *= 1.01F; - // Divide the weight of the interval by its size. This encourages - // spilling of intervals that are large and have few uses, and - // discourages spilling of small intervals with many uses. - li.weight /= lis->getApproximateInstructionCount(li) * SlotIndex::NUM; + lis->normalizeSpillWeight(li); } } diff --git a/lib/CodeGen/CodePlacementOpt.cpp b/lib/CodeGen/CodePlacementOpt.cpp index a13a310..3ff2a04 100644 --- a/lib/CodeGen/CodePlacementOpt.cpp +++ b/lib/CodeGen/CodePlacementOpt.cpp @@ -102,13 +102,6 @@ bool CodePlacementOpt::HasAnalyzableTerminator(MachineBasicBlock *MBB) { // Conservatively ignore EH landing pads. if (MBB->isLandingPad()) return false; - // Ignore blocks which look like they might have EH-related control flow. - // At the time of this writing, there are blocks which AnalyzeBranch - // thinks end in single uncoditional branches, yet which have two CFG - // successors. Code in this file is not prepared to reason about such things. - if (!MBB->empty() && MBB->back().isEHLabel()) - return false; - // Aggressively handle return blocks and similar constructs. if (MBB->succ_empty()) return true; @@ -118,6 +111,14 @@ bool CodePlacementOpt::HasAnalyzableTerminator(MachineBasicBlock *MBB) { // Make sure the terminator is understood. if (TII->AnalyzeBranch(*MBB, TBB, FBB, Cond)) return false; + // Ignore blocks which look like they might have EH-related control flow. + // AnalyzeBranch thinks it knows how to analyze such things, but it doesn't + // recognize the possibility of a control transfer through an unwind. + // Such blocks contain EH_LABEL instructions, however they may be in the + // middle of the block. Instead of searching for them, just check to see + // if the CFG disagrees with AnalyzeBranch. + if (1u + !Cond.empty() != MBB->succ_size()) + return false; // Make sure we have the option of reversing the condition. if (!Cond.empty() && TII->ReverseBranchCondition(Cond)) return false; diff --git a/lib/CodeGen/DeadMachineInstructionElim.cpp b/lib/CodeGen/DeadMachineInstructionElim.cpp index a215a19..d69c995 100644 --- a/lib/CodeGen/DeadMachineInstructionElim.cpp +++ b/lib/CodeGen/DeadMachineInstructionElim.cpp @@ -55,7 +55,7 @@ FunctionPass *llvm::createDeadMachineInstructionElimPass() { bool DeadMachineInstructionElim::isDead(const MachineInstr *MI) const { // Don't delete instructions with side effects. bool SawStore = false; - if (!MI->isSafeToMove(TII, SawStore, 0) && !MI->isPHI()) + if (!MI->isSafeToMove(TII, 0, SawStore) && !MI->isPHI()) return false; // Examine each operand. diff --git a/lib/CodeGen/LLVMTargetMachine.cpp b/lib/CodeGen/LLVMTargetMachine.cpp index 278de02..fd442db 100644 --- a/lib/CodeGen/LLVMTargetMachine.cpp +++ b/lib/CodeGen/LLVMTargetMachine.cpp @@ -67,6 +67,9 @@ static cl::opt<bool> VerifyMachineCode("verify-machineinstrs", cl::Hidden, cl::desc("Verify generated machine code"), cl::init(getenv("LLVM_VERIFY_MACHINEINSTRS")!=NULL)); +static cl::opt<bool> EnableMachineCSE("machine-cse", cl::Hidden, + cl::desc("Enable Machine CSE")); + static cl::opt<cl::boolOrDefault> AsmVerbose("asm-verbose", cl::desc("Add comments to directives."), cl::init(cl::BOU_UNSET)); @@ -115,9 +118,10 @@ LLVMTargetMachine::setCodeModelForStatic() { bool LLVMTargetMachine::addPassesToEmitFile(PassManagerBase &PM, formatted_raw_ostream &Out, CodeGenFileType FileType, - CodeGenOpt::Level OptLevel) { + CodeGenOpt::Level OptLevel, + bool DisableVerify) { // Add common CodeGen passes. - if (addCommonCodeGenPasses(PM, OptLevel)) + if (addCommonCodeGenPasses(PM, OptLevel, DisableVerify)) return true; OwningPtr<MCContext> Context(new MCContext()); @@ -193,12 +197,13 @@ bool LLVMTargetMachine::addPassesToEmitFile(PassManagerBase &PM, /// bool LLVMTargetMachine::addPassesToEmitMachineCode(PassManagerBase &PM, JITCodeEmitter &JCE, - CodeGenOpt::Level OptLevel) { + CodeGenOpt::Level OptLevel, + bool DisableVerify) { // Make sure the code model is set. setCodeModelForJIT(); // Add common CodeGen passes. - if (addCommonCodeGenPasses(PM, OptLevel)) + if (addCommonCodeGenPasses(PM, OptLevel, DisableVerify)) return true; addCodeEmitter(PM, OptLevel, JCE); @@ -221,13 +226,19 @@ static void printAndVerify(PassManagerBase &PM, /// emitting to assembly files or machine code output. /// bool LLVMTargetMachine::addCommonCodeGenPasses(PassManagerBase &PM, - CodeGenOpt::Level OptLevel) { + CodeGenOpt::Level OptLevel, + bool DisableVerify) { // Standard LLVM-Level Passes. + // Before running any passes, run the verifier to determine if the input + // coming from the front-end and/or optimizer is valid. + if (!DisableVerify) + PM.add(createVerifierPass()); + // Optionally, tun split-GEPs and no-load GVN. if (EnableSplitGEPGVN) { PM.add(createGEPSplitterPass()); - PM.add(createGVNPass(/*NoPRE=*/false, /*NoLoads=*/true)); + PM.add(createGVNPass(/*NoLoads=*/true)); } // Run loop strength reduction before anything else. @@ -235,9 +246,6 @@ bool LLVMTargetMachine::addCommonCodeGenPasses(PassManagerBase &PM, PM.add(createLoopStrengthReducePass(getTargetLowering())); if (PrintLSR) PM.add(createPrintFunctionPass("\n\n*** Code after LSR ***\n", &dbgs())); -#ifndef NDEBUG - PM.add(createVerifierPass()); -#endif } // Turn exception handling constructs into something the code generators can @@ -277,6 +285,11 @@ bool LLVMTargetMachine::addCommonCodeGenPasses(PassManagerBase &PM, "*** Final LLVM Code input to ISel ***\n", &dbgs())); + // All passes which modify the LLVM IR are now complete; run the verifier + // to ensure that the IR is valid. + if (!DisableVerify) + PM.add(createVerifierPass()); + // Standard Lower-Level Passes. // Set up a MachineFunction for the rest of CodeGen to work on. @@ -307,6 +320,8 @@ bool LLVMTargetMachine::addCommonCodeGenPasses(PassManagerBase &PM, if (OptLevel != CodeGenOpt::None) { PM.add(createOptimizeExtsPass()); + if (EnableMachineCSE) + PM.add(createMachineCSEPass()); if (!DisableMachineLICM) PM.add(createMachineLICMPass()); if (!DisableMachineSink) diff --git a/lib/CodeGen/LiveIntervalAnalysis.cpp b/lib/CodeGen/LiveIntervalAnalysis.cpp index f6bf433..ccda66f 100644 --- a/lib/CodeGen/LiveIntervalAnalysis.cpp +++ b/lib/CodeGen/LiveIntervalAnalysis.cpp @@ -329,24 +329,43 @@ void LiveIntervals::handleVirtualRegisterDef(MachineBasicBlock *mbb, DEBUG(dbgs() << " +" << NewLR); interval.addRange(NewLR); - // Iterate over all of the blocks that the variable is completely - // live in, adding [insrtIndex(begin), instrIndex(end)+4) to the - // live interval. - for (SparseBitVector<>::iterator I = vi.AliveBlocks.begin(), - E = vi.AliveBlocks.end(); I != E; ++I) { - MachineBasicBlock *aliveBlock = mf_->getBlockNumbered(*I); - LiveRange LR(getMBBStartIdx(aliveBlock), getMBBEndIdx(aliveBlock), ValNo); - interval.addRange(LR); - DEBUG(dbgs() << " +" << LR); + bool PHIJoin = lv_->isPHIJoin(interval.reg); + + if (PHIJoin) { + // A phi join register is killed at the end of the MBB and revived as a new + // valno in the killing blocks. + assert(vi.AliveBlocks.empty() && "Phi join can't pass through blocks"); + DEBUG(dbgs() << " phi-join"); + ValNo->addKill(indexes_->getTerminatorGap(mbb)); + ValNo->setHasPHIKill(true); + } else { + // Iterate over all of the blocks that the variable is completely + // live in, adding [insrtIndex(begin), instrIndex(end)+4) to the + // live interval. + for (SparseBitVector<>::iterator I = vi.AliveBlocks.begin(), + E = vi.AliveBlocks.end(); I != E; ++I) { + MachineBasicBlock *aliveBlock = mf_->getBlockNumbered(*I); + LiveRange LR(getMBBStartIdx(aliveBlock), getMBBEndIdx(aliveBlock), ValNo); + interval.addRange(LR); + DEBUG(dbgs() << " +" << LR); + } } // Finally, this virtual register is live from the start of any killing // block to the 'use' slot of the killing instruction. for (unsigned i = 0, e = vi.Kills.size(); i != e; ++i) { MachineInstr *Kill = vi.Kills[i]; - SlotIndex killIdx = - getInstructionIndex(Kill).getDefIndex(); - LiveRange LR(getMBBStartIdx(Kill->getParent()), killIdx, ValNo); + SlotIndex Start = getMBBStartIdx(Kill->getParent()); + SlotIndex killIdx = getInstructionIndex(Kill).getDefIndex(); + + // Create interval with one of a NEW value number. Note that this value + // number isn't actually defined by an instruction, weird huh? :) + if (PHIJoin) { + ValNo = interval.getNextValue(SlotIndex(Start, true), 0, false, + VNInfoAllocator); + ValNo->setIsPHIDef(true); + } + LiveRange LR(Start, killIdx, ValNo); interval.addRange(LR); ValNo->addKill(killIdx); DEBUG(dbgs() << " +" << LR); @@ -409,48 +428,11 @@ void LiveIntervals::handleVirtualRegisterDef(MachineBasicBlock *mbb, interval.print(dbgs(), tri_); }); } else { - // Otherwise, this must be because of phi elimination. If this is the - // first redefinition of the vreg that we have seen, go back and change - // the live range in the PHI block to be a different value number. - if (interval.containsOneValue()) { - - VNInfo *VNI = interval.getValNumInfo(0); - // Phi elimination may have reused the register for multiple identical - // phi nodes. There will be a kill per phi. Remove the old ranges that - // we now know have an incorrect number. - for (unsigned ki=0, ke=vi.Kills.size(); ki != ke; ++ki) { - MachineInstr *Killer = vi.Kills[ki]; - SlotIndex Start = getMBBStartIdx(Killer->getParent()); - SlotIndex End = getInstructionIndex(Killer).getDefIndex(); - DEBUG({ - dbgs() << "\n\t\trenaming [" << Start << "," << End << "] in: "; - interval.print(dbgs(), tri_); - }); - interval.removeRange(Start, End); - - // Replace the interval with one of a NEW value number. Note that - // this value number isn't actually defined by an instruction, weird - // huh? :) - LiveRange LR(Start, End, - interval.getNextValue(SlotIndex(Start, true), - 0, false, VNInfoAllocator)); - LR.valno->setIsPHIDef(true); - interval.addRange(LR); - LR.valno->addKill(End); - } - - MachineBasicBlock *killMBB = getMBBFromIndex(VNI->def); - VNI->addKill(indexes_->getTerminatorGap(killMBB)); - VNI->setHasPHIKill(true); - DEBUG({ - dbgs() << " RESULT: "; - interval.print(dbgs(), tri_); - }); - } - + assert(lv_->isPHIJoin(interval.reg) && "Multiply defined register"); // In the case of PHI elimination, each variable definition is only // live until the end of the block. We've already taken care of the // rest of the live range. + SlotIndex defIndex = MIIdx.getDefIndex(); if (MO.isEarlyClobber()) defIndex = MIIdx.getUseIndex(); @@ -468,7 +450,7 @@ void LiveIntervals::handleVirtualRegisterDef(MachineBasicBlock *mbb, interval.addRange(LR); ValNo->addKill(indexes_->getTerminatorGap(mbb)); ValNo->setHasPHIKill(true); - DEBUG(dbgs() << " +" << LR); + DEBUG(dbgs() << " phi-join +" << LR); } } @@ -1358,11 +1340,9 @@ rewriteInstructionsForSpills(const LiveInterval &li, bool TrySplit, MachineBasicBlock *MBB = MI->getParent(); if (ImpUse && MI != ReMatDefMI) { - // Re-matting an instruction with virtual register use. Update the - // register interval's spill weight to HUGE_VALF to prevent it from - // being spilled. - LiveInterval &ImpLi = getInterval(ImpUse); - ImpLi.weight = HUGE_VALF; + // Re-matting an instruction with virtual register use. Prevent interval + // from being spilled. + getInterval(ImpUse).markNotSpillable(); } unsigned MBBId = MBB->getNumber(); @@ -1414,7 +1394,7 @@ rewriteInstructionsForSpills(const LiveInterval &li, bool TrySplit, LiveInterval &nI = getOrCreateInterval(NewVReg); if (!TrySplit) { // The spill weight is now infinity as it cannot be spilled again. - nI.weight = HUGE_VALF; + nI.markNotSpillable(); continue; } @@ -1562,6 +1542,28 @@ LiveIntervals::handleSpilledImpDefs(const LiveInterval &li, VirtRegMap &vrm, } } +float +LiveIntervals::getSpillWeight(bool isDef, bool isUse, unsigned loopDepth) { + // Limit the loop depth ridiculousness. + if (loopDepth > 200) + loopDepth = 200; + + // The loop depth is used to roughly estimate the number of times the + // instruction is executed. Something like 10^d is simple, but will quickly + // overflow a float. This expression behaves like 10^d for small d, but is + // more tempered for large d. At d=200 we get 6.7e33 which leaves a bit of + // headroom before overflow. + float lc = powf(1 + (100.0f / (loopDepth+10)), (float)loopDepth); + + return (isDef + isUse) * lc; +} + +void +LiveIntervals::normalizeSpillWeights(std::vector<LiveInterval*> &NewLIs) { + for (unsigned i = 0, e = NewLIs.size(); i != e; ++i) + normalizeSpillWeight(*NewLIs[i]); +} + std::vector<LiveInterval*> LiveIntervals:: addIntervalsForSpillsFast(const LiveInterval &li, const MachineLoopInfo *loopInfo, @@ -1570,8 +1572,7 @@ addIntervalsForSpillsFast(const LiveInterval &li, std::vector<LiveInterval*> added; - assert(li.weight != HUGE_VALF && - "attempt to spill already spilled interval!"); + assert(li.isSpillable() && "attempt to spill already spilled interval!"); DEBUG({ dbgs() << "\t\t\t\tadding intervals for spills for interval: "; @@ -1607,10 +1608,7 @@ addIntervalsForSpillsFast(const LiveInterval &li, // create a new register for this spill LiveInterval &nI = getOrCreateInterval(NewVReg); - - // the spill weight is now infinity as it - // cannot be spilled again - nI.weight = HUGE_VALF; + nI.markNotSpillable(); // Rewrite register operands to use the new vreg. for (SmallVectorImpl<unsigned>::iterator I = Indices.begin(), @@ -1664,8 +1662,7 @@ addIntervalsForSpills(const LiveInterval &li, if (EnableFastSpilling) return addIntervalsForSpillsFast(li, loopInfo, vrm); - assert(li.weight != HUGE_VALF && - "attempt to spill already spilled interval!"); + assert(li.isSpillable() && "attempt to spill already spilled interval!"); DEBUG({ dbgs() << "\t\t\t\tadding intervals for spills for interval: "; @@ -1739,6 +1736,7 @@ addIntervalsForSpills(const LiveInterval &li, } handleSpilledImpDefs(li, vrm, rc, NewLIs); + normalizeSpillWeights(NewLIs); return NewLIs; } @@ -1814,6 +1812,7 @@ addIntervalsForSpills(const LiveInterval &li, // Insert spills / restores if we are splitting. if (!TrySplit) { handleSpilledImpDefs(li, vrm, rc, NewLIs); + normalizeSpillWeights(NewLIs); return NewLIs; } @@ -1930,11 +1929,10 @@ addIntervalsForSpills(const LiveInterval &li, unsigned ImpUse = getReMatImplicitUse(li, ReMatDefMI); if (ImpUse) { // Re-matting an instruction with virtual register use. Add the - // register as an implicit use on the use MI and update the register - // interval's spill weight to HUGE_VALF to prevent it from being - // spilled. + // register as an implicit use on the use MI and mark the register + // interval as unspillable. LiveInterval &ImpLi = getInterval(ImpUse); - ImpLi.weight = HUGE_VALF; + ImpLi.markNotSpillable(); MI->addOperand(MachineOperand::CreateReg(ImpUse, false, true)); } } @@ -1973,6 +1971,7 @@ addIntervalsForSpills(const LiveInterval &li, } handleSpilledImpDefs(li, vrm, rc, RetNewLIs); + normalizeSpillWeights(RetNewLIs); return RetNewLIs; } diff --git a/lib/CodeGen/LiveVariables.cpp b/lib/CodeGen/LiveVariables.cpp index 8a124dc..68c8539 100644 --- a/lib/CodeGen/LiveVariables.cpp +++ b/lib/CodeGen/LiveVariables.cpp @@ -510,6 +510,7 @@ bool LiveVariables::runOnMachineFunction(MachineFunction &mf) { PHIVarInfo = new SmallVector<unsigned, 4>[MF->getNumBlockIDs()]; std::fill(PhysRegDef, PhysRegDef + NumRegs, (MachineInstr*)0); std::fill(PhysRegUse, PhysRegUse + NumRegs, (MachineInstr*)0); + PHIJoins.clear(); /// Get some space for a respectable number of registers. VirtRegInfo.resize(64); diff --git a/lib/CodeGen/MachineBasicBlock.cpp b/lib/CodeGen/MachineBasicBlock.cpp index 655a0bf..64134ce 100644 --- a/lib/CodeGen/MachineBasicBlock.cpp +++ b/lib/CodeGen/MachineBasicBlock.cpp @@ -143,36 +143,6 @@ MachineBasicBlock::iterator MachineBasicBlock::getFirstTerminator() { return I; } -/// isOnlyReachableViaFallthough - Return true if this basic block has -/// exactly one predecessor and the control transfer mechanism between -/// the predecessor and this block is a fall-through. -bool MachineBasicBlock::isOnlyReachableByFallthrough() const { - // If this is a landing pad, it isn't a fall through. If it has no preds, - // then nothing falls through to it. - if (isLandingPad() || pred_empty()) - return false; - - // If there isn't exactly one predecessor, it can't be a fall through. - const_pred_iterator PI = pred_begin(), PI2 = PI; - ++PI2; - if (PI2 != pred_end()) - return false; - - // The predecessor has to be immediately before this block. - const MachineBasicBlock *Pred = *PI; - - if (!Pred->isLayoutSuccessor(this)) - return false; - - // If the block is completely empty, then it definitely does fall through. - if (Pred->empty()) - return true; - - // Otherwise, check the last instruction. - const MachineInstr &LastInst = Pred->back(); - return !LastInst.getDesc().isBarrier(); -} - void MachineBasicBlock::dump() const { print(dbgs()); } diff --git a/lib/CodeGen/MachineCSE.cpp b/lib/CodeGen/MachineCSE.cpp new file mode 100644 index 0000000..023ace2 --- /dev/null +++ b/lib/CodeGen/MachineCSE.cpp @@ -0,0 +1,208 @@ +//===-- MachineCSE.cpp - Machine Common Subexpression Elimination Pass ----===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This pass performs global common subexpression elimination on machine +// instructions using a scoped hash table based value numbering scheme. It +// must be run while the machine function is still in SSA form. +// +//===----------------------------------------------------------------------===// + +#define DEBUG_TYPE "machine-cse" +#include "llvm/CodeGen/Passes.h" +#include "llvm/CodeGen/MachineDominators.h" +#include "llvm/CodeGen/MachineInstr.h" +#include "llvm/CodeGen/MachineRegisterInfo.h" +#include "llvm/Target/TargetInstrInfo.h" +#include "llvm/ADT/ScopedHashTable.h" +#include "llvm/ADT/Statistic.h" +#include "llvm/Support/Debug.h" + +using namespace llvm; + +namespace llvm { + template<> struct DenseMapInfo<MachineInstr*> { + static inline MachineInstr *getEmptyKey() { + return 0; + } + + static inline MachineInstr *getTombstoneKey() { + return reinterpret_cast<MachineInstr*>(-1); + } + + static unsigned getHashValue(const MachineInstr* const &MI) { + unsigned Hash = MI->getOpcode() * 37; + for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { + const MachineOperand &MO = MI->getOperand(i); + uint64_t Key = (uint64_t)MO.getType() << 32; + switch (MO.getType()) { + default: break; + case MachineOperand::MO_Register: + if (MO.isDef() && TargetRegisterInfo::isVirtualRegister(MO.getReg())) + continue; // Skip virtual register defs. + Key |= MO.getReg(); + break; + case MachineOperand::MO_Immediate: + Key |= MO.getImm(); + break; + case MachineOperand::MO_FrameIndex: + case MachineOperand::MO_ConstantPoolIndex: + case MachineOperand::MO_JumpTableIndex: + Key |= MO.getIndex(); + break; + case MachineOperand::MO_MachineBasicBlock: + Key |= DenseMapInfo<void*>::getHashValue(MO.getMBB()); + break; + case MachineOperand::MO_GlobalAddress: + Key |= DenseMapInfo<void*>::getHashValue(MO.getGlobal()); + break; + case MachineOperand::MO_BlockAddress: + Key |= DenseMapInfo<void*>::getHashValue(MO.getBlockAddress()); + break; + } + Key += ~(Key << 32); + Key ^= (Key >> 22); + Key += ~(Key << 13); + Key ^= (Key >> 8); + Key += (Key << 3); + Key ^= (Key >> 15); + Key += ~(Key << 27); + Key ^= (Key >> 31); + Hash = (unsigned)Key + Hash * 37; + } + return Hash; + } + + static bool isEqual(const MachineInstr* const &LHS, + const MachineInstr* const &RHS) { + if (RHS == getEmptyKey() || RHS == getTombstoneKey() || + LHS == getEmptyKey() || LHS == getTombstoneKey()) + return LHS == RHS; + return LHS->isIdenticalTo(RHS, MachineInstr::IgnoreVRegDefs); + } + }; +} // end llvm namespace + +namespace { + class MachineCSE : public MachineFunctionPass { + const TargetInstrInfo *TII; + MachineRegisterInfo *MRI; + MachineDominatorTree *DT; + ScopedHashTable<MachineInstr*, unsigned> VNT; + unsigned CurrVN; + public: + static char ID; // Pass identification + MachineCSE() : MachineFunctionPass(&ID), CurrVN(0) {} + + virtual bool runOnMachineFunction(MachineFunction &MF); + + virtual void getAnalysisUsage(AnalysisUsage &AU) const { + AU.setPreservesCFG(); + MachineFunctionPass::getAnalysisUsage(AU); + AU.addRequired<MachineDominatorTree>(); + AU.addPreserved<MachineDominatorTree>(); + } + + private: + bool PerformTrivialCoalescing(MachineInstr *MI, MachineBasicBlock *MBB); + bool ProcessBlock(MachineDomTreeNode *Node); + }; +} // end anonymous namespace + +char MachineCSE::ID = 0; +static RegisterPass<MachineCSE> +X("machine-cse", "Machine Common Subexpression Elimination"); + +FunctionPass *llvm::createMachineCSEPass() { return new MachineCSE(); } + +bool MachineCSE::PerformTrivialCoalescing(MachineInstr *MI, + MachineBasicBlock *MBB) { + bool Changed = false; + for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { + MachineOperand &MO = MI->getOperand(i); + if (MO.isReg() && MO.isUse()) { + unsigned Reg = MO.getReg(); + if (!Reg || TargetRegisterInfo::isPhysicalRegister(Reg)) + continue; + MachineInstr *DefMI = MRI->getVRegDef(Reg); + if (DefMI->getParent() == MBB) { + unsigned SrcReg, DstReg, SrcSubIdx, DstSubIdx; + if (TII->isMoveInstr(*DefMI, SrcReg, DstReg, SrcSubIdx, DstSubIdx) && + TargetRegisterInfo::isVirtualRegister(SrcReg) && + !SrcSubIdx && !DstSubIdx) { + MO.setReg(SrcReg); + Changed = true; + } + } + } + } + + return Changed; +} + +static bool hasLivePhysRegDefUse(MachineInstr *MI) { + for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { + MachineOperand &MO = MI->getOperand(i); + if (!MO.isReg()) + continue; + unsigned Reg = MO.getReg(); + if (!Reg) + continue; + if (TargetRegisterInfo::isPhysicalRegister(Reg) && + !(MO.isDef() && MO.isDead())) + return true; + } + return false; +} + +bool MachineCSE::ProcessBlock(MachineDomTreeNode *Node) { + bool Changed = false; + + ScopedHashTableScope<MachineInstr*, unsigned> VNTS(VNT); + MachineBasicBlock *MBB = Node->getBlock(); + for (MachineBasicBlock::iterator I = MBB->begin(), E = MBB->end(); I != E; + ++I) { + MachineInstr *MI = &*I; + bool SawStore = false; + if (!MI->isSafeToMove(TII, 0, SawStore)) + continue; + // Ignore copies or instructions that read / write physical registers + // (except for dead defs of physical registers). + unsigned SrcReg, DstReg, SrcSubIdx, DstSubIdx; + if (TII->isMoveInstr(*MI, SrcReg, DstReg, SrcSubIdx, DstSubIdx)) + continue; + if (hasLivePhysRegDefUse(MI)) + continue; + + bool FoundCSE = VNT.count(MI); + if (!FoundCSE) { + // Look for trivial copy coalescing opportunities. + if (PerformTrivialCoalescing(MI, MBB)) + FoundCSE = VNT.count(MI); + } + + if (FoundCSE) + DEBUG(dbgs() << "Found a common subexpression: " << *MI); + else + VNT.insert(MI, ++CurrVN); + } + + // Recursively call ProcessBlock with childred. + const std::vector<MachineDomTreeNode*> &Children = Node->getChildren(); + for (unsigned i = 0, e = Children.size(); i != e; ++i) + Changed |= ProcessBlock(Children[i]); + + return Changed; +} + +bool MachineCSE::runOnMachineFunction(MachineFunction &MF) { + TII = MF.getTarget().getInstrInfo(); + MRI = &MF.getRegInfo(); + DT = &getAnalysis<MachineDominatorTree>(); + return ProcessBlock(DT->getRootNode()); +} diff --git a/lib/CodeGen/MachineFunction.cpp b/lib/CodeGen/MachineFunction.cpp index f141c56..4377d5b 100644 --- a/lib/CodeGen/MachineFunction.cpp +++ b/lib/CodeGen/MachineFunction.cpp @@ -95,6 +95,9 @@ MachineFunction::MachineFunction(Function *F, const TargetMachine &TM, MFInfo = 0; FrameInfo = new (Allocator.Allocate<MachineFrameInfo>()) MachineFrameInfo(*TM.getFrameInfo()); + if (Fn->hasFnAttr(Attribute::StackAlignment)) + FrameInfo->setMaxAlignment(Attribute::getStackAlignmentFromAttrs( + Fn->getAttributes().getFnAttributes())); ConstantPool = new (Allocator.Allocate<MachineConstantPool>()) MachineConstantPool(TM.getTargetData()); Alignment = TM.getTargetLowering()->getFunctionAlignment(F); diff --git a/lib/CodeGen/MachineInstr.cpp b/lib/CodeGen/MachineInstr.cpp index b6d98e8..cba93f1 100644 --- a/lib/CodeGen/MachineInstr.cpp +++ b/lib/CodeGen/MachineInstr.cpp @@ -18,6 +18,7 @@ #include "llvm/Type.h" #include "llvm/Value.h" #include "llvm/Assembly/Writer.h" +#include "llvm/CodeGen/MachineConstantPool.h" #include "llvm/CodeGen/MachineFunction.h" #include "llvm/CodeGen/MachineMemOperand.h" #include "llvm/CodeGen/MachineRegisterInfo.h" @@ -701,6 +702,28 @@ void MachineInstr::addMemOperand(MachineFunction &MF, MemRefsEnd = NewMemRefsEnd; } +bool MachineInstr::isIdenticalTo(const MachineInstr *Other, + MICheckType Check) const { + if (Other->getOpcode() != getOpcode() || + Other->getNumOperands() != getNumOperands()) + return false; + for (unsigned i = 0, e = getNumOperands(); i != e; ++i) { + const MachineOperand &MO = getOperand(i); + const MachineOperand &OMO = Other->getOperand(i); + if (Check != CheckDefs && MO.isReg() && MO.isDef()) { + if (Check == IgnoreDefs) + continue; + // Check == IgnoreVRegDefs + if (TargetRegisterInfo::isPhysicalRegister(MO.getReg()) || + TargetRegisterInfo::isPhysicalRegister(OMO.getReg())) + if (MO.getReg() != OMO.getReg()) + return false; + } else if (!MO.isIdenticalTo(OMO)) + return false; + } + return true; +} + /// removeFromParent - This method unlinks 'this' from the containing basic /// block, and returns it, but does not delete it. MachineInstr *MachineInstr::removeFromParent() { @@ -959,8 +982,8 @@ void MachineInstr::copyPredicates(const MachineInstr *MI) { /// SawStore is set to true, it means that there is a store (or call) between /// the instruction's location and its intended destination. bool MachineInstr::isSafeToMove(const TargetInstrInfo *TII, - bool &SawStore, - AliasAnalysis *AA) const { + AliasAnalysis *AA, + bool &SawStore) const { // Ignore stuff that we obviously can't move. if (TID->mayStore() || TID->isCall()) { SawStore = true; @@ -985,11 +1008,11 @@ bool MachineInstr::isSafeToMove(const TargetInstrInfo *TII, /// isSafeToReMat - Return true if it's safe to rematerialize the specified /// instruction which defined the specified register instead of copying it. bool MachineInstr::isSafeToReMat(const TargetInstrInfo *TII, - unsigned DstReg, - AliasAnalysis *AA) const { + AliasAnalysis *AA, + unsigned DstReg) const { bool SawStore = false; if (!TII->isTriviallyReMaterializable(this, AA) || - !isSafeToMove(TII, SawStore, AA)) + !isSafeToMove(TII, AA, SawStore)) return false; for (unsigned i = 0, e = getNumOperands(); i != e; ++i) { const MachineOperand &MO = getOperand(i); diff --git a/lib/CodeGen/MachineLICM.cpp b/lib/CodeGen/MachineLICM.cpp index 92c84f3..0361694 100644 --- a/lib/CodeGen/MachineLICM.cpp +++ b/lib/CodeGen/MachineLICM.cpp @@ -252,32 +252,6 @@ bool MachineLICM::IsLoopInvariantInst(MachineInstr &I) { return false; } - DEBUG({ - dbgs() << "--- Checking if we can hoist " << I; - if (I.getDesc().getImplicitUses()) { - dbgs() << " * Instruction has implicit uses:\n"; - - const TargetRegisterInfo *TRI = TM->getRegisterInfo(); - for (const unsigned *ImpUses = I.getDesc().getImplicitUses(); - *ImpUses; ++ImpUses) - dbgs() << " -> " << TRI->getName(*ImpUses) << "\n"; - } - - if (I.getDesc().getImplicitDefs()) { - dbgs() << " * Instruction has implicit defines:\n"; - - const TargetRegisterInfo *TRI = TM->getRegisterInfo(); - for (const unsigned *ImpDefs = I.getDesc().getImplicitDefs(); - *ImpDefs; ++ImpDefs) - dbgs() << " -> " << TRI->getName(*ImpDefs) << "\n"; - } - }); - - if (I.getDesc().getImplicitDefs() || I.getDesc().getImplicitUses()) { - DEBUG(dbgs() << "Cannot hoist with implicit defines or uses\n"); - return false; - } - // The instruction is loop invariant if all of its operands are. for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i) { const MachineOperand &MO = I.getOperand(i); @@ -311,6 +285,10 @@ bool MachineLICM::IsLoopInvariantInst(MachineInstr &I) { } else if (!MO.isDead()) { // A def that isn't dead. We can't move it. return false; + } else if (CurLoop->getHeader()->isLiveIn(Reg)) { + // If the reg is live into the loop, we can't hoist an instruction + // which would clobber it. + return false; } } @@ -467,7 +445,7 @@ MachineLICM::LookForDuplicate(const MachineInstr *MI, std::vector<const MachineInstr*> &PrevMIs) { for (unsigned i = 0, e = PrevMIs.size(); i != e; ++i) { const MachineInstr *PrevMI = PrevMIs[i]; - if (TII->isIdentical(MI, PrevMI, RegInfo)) + if (TII->produceSameValue(MI, PrevMI)) return PrevMI; } return 0; @@ -480,9 +458,20 @@ bool MachineLICM::EliminateCSE(MachineInstr *MI, if (const MachineInstr *Dup = LookForDuplicate(MI, CI->second)) { DEBUG(dbgs() << "CSEing " << *MI << " with " << *Dup); + + // Replace virtual registers defined by MI by their counterparts defined + // by Dup. for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { const MachineOperand &MO = MI->getOperand(i); - if (MO.isReg() && MO.isDef()) + + // Physical registers may not differ here. + assert((!MO.isReg() || MO.getReg() == 0 || + !TargetRegisterInfo::isPhysicalRegister(MO.getReg()) || + MO.getReg() == Dup->getOperand(i).getReg()) && + "Instructions with different phys regs are not identical!"); + + if (MO.isReg() && MO.isDef() && + !TargetRegisterInfo::isPhysicalRegister(MO.getReg())) RegInfo->replaceRegWith(MO.getReg(), Dup->getOperand(i).getReg()); } MI->eraseFromParent(); diff --git a/lib/CodeGen/MachineSink.cpp b/lib/CodeGen/MachineSink.cpp index c391576..9ba7d14 100644 --- a/lib/CodeGen/MachineSink.cpp +++ b/lib/CodeGen/MachineSink.cpp @@ -149,7 +149,7 @@ bool MachineSinking::ProcessBlock(MachineBasicBlock &MBB) { /// instruction out of its current block into a successor. bool MachineSinking::SinkInstruction(MachineInstr *MI, bool &SawStore) { // Check if it's safe to move the instruction. - if (!MI->isSafeToMove(TII, SawStore, AA)) + if (!MI->isSafeToMove(TII, AA, SawStore)) return false; // FIXME: This should include support for sinking instructions within the diff --git a/lib/CodeGen/PBQP/HeuristicSolver.h b/lib/CodeGen/PBQP/HeuristicSolver.h index c156264..bd18b52 100644 --- a/lib/CodeGen/PBQP/HeuristicSolver.h +++ b/lib/CodeGen/PBQP/HeuristicSolver.h @@ -229,7 +229,7 @@ namespace PBQP { } /// \brief Apply rule R1. - /// @param nItr Node iterator for node to apply R1 to. + /// @param xnItr Node iterator for node to apply R1 to. /// /// Node will be automatically pushed to the solver stack. void applyR1(Graph::NodeItr xnItr) { @@ -277,7 +277,7 @@ namespace PBQP { } /// \brief Apply rule R2. - /// @param nItr Node iterator for node to apply R2 to. + /// @param xnItr Node iterator for node to apply R2 to. /// /// Node will be automatically pushed to the solver stack. void applyR2(Graph::NodeItr xnItr) { diff --git a/lib/CodeGen/PHIElimination.cpp b/lib/CodeGen/PHIElimination.cpp index b740c68..bdfd448 100644 --- a/lib/CodeGen/PHIElimination.cpp +++ b/lib/CodeGen/PHIElimination.cpp @@ -55,8 +55,6 @@ void llvm::PHIElimination::getAnalysisUsage(AnalysisUsage &AU) const { bool llvm::PHIElimination::runOnMachineFunction(MachineFunction &Fn) { MRI = &Fn.getRegInfo(); - PHIDefs.clear(); - PHIKills.clear(); bool Changed = false; // Split critical edges to help the coalescer @@ -215,10 +213,6 @@ void llvm::PHIElimination::LowerAtomicPHINode( TII->copyRegToReg(MBB, AfterPHIsIt, DestReg, IncomingReg, RC, RC); } - // Record PHI def. - assert(!hasPHIDef(DestReg) && "Vreg has multiple phi-defs?"); - PHIDefs[DestReg] = &MBB; - // Update live variable information if there is any. LiveVariables *LV = getAnalysisIfAvailable<LiveVariables>(); if (LV) { @@ -229,6 +223,7 @@ void llvm::PHIElimination::LowerAtomicPHINode( // Increment use count of the newly created virtual register. VI.NumUses++; + LV->setPHIJoin(IncomingReg); // When we are reusing the incoming register, it may already have been // killed in this block. The old kill will also have been inserted at @@ -276,9 +271,6 @@ void llvm::PHIElimination::LowerAtomicPHINode( // path the PHI. MachineBasicBlock &opBlock = *MPhi->getOperand(i*2+2).getMBB(); - // Record the kill. - PHIKills[SrcReg].insert(&opBlock); - // If source is defined by an implicit def, there is no need to insert a // copy. MachineInstr *DefMI = MRI->getVRegDef(SrcReg); diff --git a/lib/CodeGen/PHIElimination.h b/lib/CodeGen/PHIElimination.h index f3ab9e2..ff4aa20 100644 --- a/lib/CodeGen/PHIElimination.h +++ b/lib/CodeGen/PHIElimination.h @@ -22,17 +22,8 @@ namespace llvm { /// Lower PHI instructions to copies. class PHIElimination : public MachineFunctionPass { MachineRegisterInfo *MRI; // Machine register information - private: - - typedef SmallSet<MachineBasicBlock*, 4> PHIKillList; - typedef DenseMap<unsigned, PHIKillList> PHIKillMap; - typedef DenseMap<unsigned, MachineBasicBlock*> PHIDefMap; public: - - typedef PHIKillList::iterator phi_kill_iterator; - typedef PHIKillList::const_iterator const_phi_kill_iterator; - static char ID; // Pass identification, replacement for typeid PHIElimination() : MachineFunctionPass(&ID) {} @@ -40,38 +31,6 @@ namespace llvm { virtual void getAnalysisUsage(AnalysisUsage &AU) const; - /// Return true if the given vreg was defined by a PHI intsr prior to - /// lowering. - bool hasPHIDef(unsigned vreg) const { - return PHIDefs.count(vreg); - } - - /// Returns the block in which the PHI instruction which defined the - /// given vreg used to reside. - MachineBasicBlock* getPHIDefBlock(unsigned vreg) { - PHIDefMap::iterator phiDefItr = PHIDefs.find(vreg); - assert(phiDefItr != PHIDefs.end() && "vreg has no phi-def."); - return phiDefItr->second; - } - - /// Returns true if the given vreg was killed by a PHI instr. - bool hasPHIKills(unsigned vreg) const { - return PHIKills.count(vreg); - } - - /// Returns an iterator over the BasicBlocks which contained PHI - /// kills of this register prior to lowering. - phi_kill_iterator phiKillsBegin(unsigned vreg) { - PHIKillMap::iterator phiKillItr = PHIKills.find(vreg); - assert(phiKillItr != PHIKills.end() && "vreg has no phi-kills."); - return phiKillItr->second.begin(); - } - phi_kill_iterator phiKillsEnd(unsigned vreg) { - PHIKillMap::iterator phiKillItr = PHIKills.find(vreg); - assert(phiKillItr != PHIKills.end() && "vreg has no phi-kills."); - return phiKillItr->second.end(); - } - private: /// EliminatePHINodes - Eliminate phi nodes by inserting copy instructions /// in predecessor basic blocks. @@ -139,8 +98,6 @@ namespace llvm { typedef DenseMap<BBVRegPair, unsigned> VRegPHIUse; VRegPHIUse VRegPHIUseCount; - PHIDefMap PHIDefs; - PHIKillMap PHIKills; // Defs of PHI sources which are implicit_def. SmallPtrSet<MachineInstr*, 4> ImpDefs; diff --git a/lib/CodeGen/Passes.cpp b/lib/CodeGen/Passes.cpp index f67eb79..5ea2941 100644 --- a/lib/CodeGen/Passes.cpp +++ b/lib/CodeGen/Passes.cpp @@ -34,7 +34,7 @@ static cl::opt<RegisterRegAlloc::FunctionPassCtor, false, RegisterPassParser<RegisterRegAlloc> > RegAlloc("regalloc", cl::init(&createLinearScanRegisterAllocator), - cl::desc("Register allocator to use: (default = linearscan)")); + cl::desc("Register allocator to use (default=linearscan)")); //===---------------------------------------------------------------------===// diff --git a/lib/CodeGen/PrologEpilogInserter.cpp b/lib/CodeGen/PrologEpilogInserter.cpp index 040259e..138e711 100644 --- a/lib/CodeGen/PrologEpilogInserter.cpp +++ b/lib/CodeGen/PrologEpilogInserter.cpp @@ -175,9 +175,10 @@ void PEI::calculateCallsInformation(MachineFunction &Fn) { MachineBasicBlock::iterator I = *i; // If call frames are not being included as part of the stack frame, and - // there is no dynamic allocation (therefore referencing frame slots off - // sp), leave the pseudo ops alone. We'll eliminate them later. - if (RegInfo->hasReservedCallFrame(Fn) || RegInfo->hasFP(Fn)) + // the target doesn't indicate otherwise, remove the call frame pseudos + // here. The sub/add sp instruction pairs are still inserted, but we don't + // need to track the SP adjustment for frame index elimination. + if (RegInfo->canSimplifyCallFramePseudos(Fn)) RegInfo->eliminateCallFramePseudoInstr(Fn, *I->getParent(), I); } } diff --git a/lib/CodeGen/RegAllocLinearScan.cpp b/lib/CodeGen/RegAllocLinearScan.cpp index 8e44a57..5c5a394 100644 --- a/lib/CodeGen/RegAllocLinearScan.cpp +++ b/lib/CodeGen/RegAllocLinearScan.cpp @@ -334,10 +334,6 @@ namespace { SmallVector<unsigned, 256> &inactiveCounts, bool SkipDGRegs); - /// assignVirt2StackSlot - assigns this virtual register to a - /// stack slot. returns the stack slot - int assignVirt2StackSlot(unsigned virtReg); - void ComputeRelatedRegClasses(); template <typename ItTy> diff --git a/lib/CodeGen/RegAllocPBQP.cpp b/lib/CodeGen/RegAllocPBQP.cpp index 2701faf..81cfd8f 100644 --- a/lib/CodeGen/RegAllocPBQP.cpp +++ b/lib/CodeGen/RegAllocPBQP.cpp @@ -57,7 +57,7 @@ using namespace llvm; static RegisterRegAlloc -registerPBQPRepAlloc("pbqp", "PBQP register allocator.", +registerPBQPRepAlloc("pbqp", "PBQP register allocator", llvm::createPBQPRegisterAllocator); static cl::opt<bool> @@ -867,10 +867,6 @@ bool PBQPRegAlloc::runOnMachineFunction(MachineFunction &MF) { // Find the vreg intervals in need of allocation. findVRegIntervalsToAlloc(); - // If there aren't any then we're done here. - if (vregIntervalsToAlloc.empty() && emptyVRegIntervals.empty()) - return true; - // If there are non-empty intervals allocate them using pbqp. if (!vregIntervalsToAlloc.empty()) { diff --git a/lib/CodeGen/ScheduleDAGInstrs.cpp b/lib/CodeGen/ScheduleDAGInstrs.cpp index 56dd533..badf34e 100644 --- a/lib/CodeGen/ScheduleDAGInstrs.cpp +++ b/lib/CodeGen/ScheduleDAGInstrs.cpp @@ -72,7 +72,7 @@ static const Value *getUnderlyingObjectFromInt(const Value *V) { } else { return V; } - assert(isa<IntegerType>(V->getType()) && "Unexpected operand type!"); + assert(V->getType()->isIntegerTy() && "Unexpected operand type!"); } while (1); } @@ -87,7 +87,7 @@ static const Value *getUnderlyingObject(const Value *V) { break; const Value *O = getUnderlyingObjectFromInt(cast<User>(V)->getOperand(0)); // If that succeeded in finding a pointer, continue the search. - if (!isa<PointerType>(O->getType())) + if (!O->getType()->isPointerTy()) break; V = O; } while (1); diff --git a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp index 7da7848..e4ff44d 100644 --- a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp +++ b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp @@ -1064,7 +1064,7 @@ SDValue DAGCombiner::visitADD(SDNode *N) { if (VT.isInteger() && !VT.isVector()) { APInt LHSZero, LHSOne; APInt RHSZero, RHSOne; - APInt Mask = APInt::getAllOnesValue(VT.getSizeInBits()); + APInt Mask = APInt::getAllOnesValue(VT.getScalarType().getSizeInBits()); DAG.ComputeMaskedBits(N0, Mask, LHSZero, LHSOne); if (LHSZero.getBoolValue()) { @@ -1136,7 +1136,7 @@ SDValue DAGCombiner::visitADDC(SDNode *N) { // fold (addc a, b) -> (or a, b), CARRY_FALSE iff a and b share no bits. APInt LHSZero, LHSOne; APInt RHSZero, RHSOne; - APInt Mask = APInt::getAllOnesValue(VT.getSizeInBits()); + APInt Mask = APInt::getAllOnesValue(VT.getScalarType().getSizeInBits()); DAG.ComputeMaskedBits(N0, Mask, LHSZero, LHSOne); if (LHSZero.getBoolValue()) { @@ -1786,7 +1786,7 @@ SDValue DAGCombiner::visitAND(SDNode *N) { SDValue RAND = ReassociateOps(ISD::AND, N->getDebugLoc(), N0, N1); if (RAND.getNode() != 0) return RAND; - // fold (and (or x, 0xFFFF), 0xFF) -> 0xFF + // fold (and (or x, C), D) -> D if (C & D) == D if (N1C && N0.getOpcode() == ISD::OR) if (ConstantSDNode *ORI = dyn_cast<ConstantSDNode>(N0.getOperand(1))) if ((ORI->getAPIntValue() & N1C->getAPIntValue()) == N1C->getAPIntValue()) @@ -2025,13 +2025,15 @@ SDValue DAGCombiner::visitOR(SDNode *N) { if (ROR.getNode() != 0) return ROR; // Canonicalize (or (and X, c1), c2) -> (and (or X, c2), c1|c2) + // iff (c1 & c2) == 0. if (N1C && N0.getOpcode() == ISD::AND && N0.getNode()->hasOneUse() && isa<ConstantSDNode>(N0.getOperand(1))) { ConstantSDNode *C1 = cast<ConstantSDNode>(N0.getOperand(1)); - return DAG.getNode(ISD::AND, N->getDebugLoc(), VT, - DAG.getNode(ISD::OR, N0.getDebugLoc(), VT, - N0.getOperand(0), N1), - DAG.FoldConstantArithmetic(ISD::OR, VT, N1C, C1)); + if ((C1->getAPIntValue() & N1C->getAPIntValue()) != 0) + return DAG.getNode(ISD::AND, N->getDebugLoc(), VT, + DAG.getNode(ISD::OR, N0.getDebugLoc(), VT, + N0.getOperand(0), N1), + DAG.FoldConstantArithmetic(ISD::OR, VT, N1C, C1)); } // fold (or (setcc x), (setcc y)) -> (setcc (or x, y)) if (isSetCCEquivalent(N0, LL, LR, CC0) && isSetCCEquivalent(N1, RL, RR, CC1)){ @@ -2754,7 +2756,7 @@ SDValue DAGCombiner::visitSRL(SDNode *N) { if (N1C && N0.getOpcode() == ISD::CTLZ && N1C->getAPIntValue() == Log2_32(VT.getSizeInBits())) { APInt KnownZero, KnownOne; - APInt Mask = APInt::getAllOnesValue(VT.getSizeInBits()); + APInt Mask = APInt::getAllOnesValue(VT.getScalarType().getSizeInBits()); DAG.ComputeMaskedBits(N0.getOperand(0), Mask, KnownZero, KnownOne); // If any of the input bits are KnownOne, then the input couldn't be all @@ -4655,7 +4657,8 @@ SDValue DAGCombiner::visitBRCOND(SDNode *N) { DAG.DeleteNode(Trunc); } // Replace the uses of SRL with SETCC - DAG.ReplaceAllUsesOfValueWith(N1, SetCC); + WorkListRemover DeadNodes(*this); + DAG.ReplaceAllUsesOfValueWith(N1, SetCC, &DeadNodes); removeFromWorkList(N1.getNode()); DAG.DeleteNode(N1.getNode()); return SDValue(N, 0); // Return N so it doesn't get rechecked! @@ -4663,6 +4666,56 @@ SDValue DAGCombiner::visitBRCOND(SDNode *N) { } } } + + // Transform br(xor(x, y)) -> br(x != y) + // Transform br(xor(xor(x,y), 1)) -> br (x == y) + if (N1.hasOneUse() && N1.getOpcode() == ISD::XOR) { + SDNode *TheXor = N1.getNode(); + SDValue Op0 = TheXor->getOperand(0); + SDValue Op1 = TheXor->getOperand(1); + if (Op0.getOpcode() == Op1.getOpcode()) { + // Avoid missing important xor optimizations. + SDValue Tmp = visitXOR(TheXor); + if (Tmp.getNode()) { + DEBUG(dbgs() << "\nReplacing.8 "; + TheXor->dump(&DAG); + dbgs() << "\nWith: "; + Tmp.getNode()->dump(&DAG); + dbgs() << '\n'); + WorkListRemover DeadNodes(*this); + DAG.ReplaceAllUsesOfValueWith(N1, Tmp, &DeadNodes); + removeFromWorkList(TheXor); + DAG.DeleteNode(TheXor); + return DAG.getNode(ISD::BRCOND, N->getDebugLoc(), + MVT::Other, Chain, Tmp, N2); + } + } + + if (Op0.getOpcode() != ISD::SETCC && Op1.getOpcode() != ISD::SETCC) { + bool Equal = false; + if (ConstantSDNode *RHSCI = dyn_cast<ConstantSDNode>(Op0)) + if (RHSCI->getAPIntValue() == 1 && Op0.hasOneUse() && + Op0.getOpcode() == ISD::XOR) { + TheXor = Op0.getNode(); + Equal = true; + } + + EVT SetCCVT = N1.getValueType(); + if (LegalTypes) + SetCCVT = TLI.getSetCCResultType(SetCCVT); + SDValue SetCC = DAG.getSetCC(TheXor->getDebugLoc(), + SetCCVT, + Op0, Op1, + Equal ? ISD::SETEQ : ISD::SETNE); + // Replace the uses of XOR with SETCC + WorkListRemover DeadNodes(*this); + DAG.ReplaceAllUsesOfValueWith(N1, SetCC, &DeadNodes); + removeFromWorkList(N1.getNode()); + DAG.DeleteNode(N1.getNode()); + return DAG.getNode(ISD::BRCOND, N->getDebugLoc(), + MVT::Other, Chain, SetCC, N2); + } + } return SDValue(); } @@ -5012,7 +5065,7 @@ SDValue DAGCombiner::visitLOAD(SDNode *N) { assert(N->getValueType(2) == MVT::Other && "Malformed indexed loads?"); if (N->hasNUsesOfValue(0, 0) && N->hasNUsesOfValue(0, 1)) { SDValue Undef = DAG.getUNDEF(N->getValueType(0)); - DEBUG(dbgs() << "\nReplacing.6 "; + DEBUG(dbgs() << "\nReplacing.7 "; N->dump(&DAG); dbgs() << "\nWith: "; Undef.getNode()->dump(&DAG); diff --git a/lib/CodeGen/SelectionDAG/FastISel.cpp b/lib/CodeGen/SelectionDAG/FastISel.cpp index 35ef5b7..1d76c7c 100644 --- a/lib/CodeGen/SelectionDAG/FastISel.cpp +++ b/lib/CodeGen/SelectionDAG/FastISel.cpp @@ -350,6 +350,34 @@ bool FastISel::SelectCall(User *I) { (void)TargetSelectInstruction(cast<Instruction>(I)); return true; } + case Intrinsic::dbg_value: { + // This requires target support, but right now X86 is the only Fast target. + DbgValueInst *DI = cast<DbgValueInst>(I); + const TargetInstrDesc &II = TII.get(TargetOpcode::DBG_VALUE); + Value *V = DI->getValue(); + if (!V) { + // Currently the optimizer can produce this; insert an undef to + // help debugging. Probably the optimizer should not do this. + BuildMI(MBB, DL, II).addReg(0U).addImm(DI->getOffset()). + addMetadata(DI->getVariable()); + } else if (ConstantInt *CI = dyn_cast<ConstantInt>(V)) { + BuildMI(MBB, DL, II).addImm(CI->getZExtValue()).addImm(DI->getOffset()). + addMetadata(DI->getVariable()); + } else if (ConstantFP *CF = dyn_cast<ConstantFP>(V)) { + BuildMI(MBB, DL, II).addFPImm(CF).addImm(DI->getOffset()). + addMetadata(DI->getVariable()); + } else if (unsigned Reg = lookUpRegForValue(V)) { + BuildMI(MBB, DL, II).addReg(Reg, RegState::Debug).addImm(DI->getOffset()). + addMetadata(DI->getVariable()); + } else { + // We can't yet handle anything else here because it would require + // generating code, thus altering codegen because of debug info. + // Insert an undef so we can see what we dropped. + BuildMI(MBB, DL, II).addReg(0U).addImm(DI->getOffset()). + addMetadata(DI->getVariable()); + } + return true; + } case Intrinsic::eh_exception: { EVT VT = TLI.getValueType(I->getType()); switch (TLI.getOperationAction(ISD::EXCEPTIONADDR, VT)) { diff --git a/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp b/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp index e9321da..c7ab34f 100644 --- a/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp +++ b/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp @@ -1539,7 +1539,6 @@ SDValue SelectionDAGLegalize::ExpandVectorBuildThroughStack(SDNode* Node) { // the result as a vector. // Create the stack frame object. EVT VT = Node->getValueType(0); - EVT OpVT = Node->getOperand(0).getValueType(); EVT EltVT = VT.getVectorElementType(); DebugLoc dl = Node->getDebugLoc(); SDValue FIPtr = DAG.CreateStackTemporary(VT); @@ -1559,8 +1558,9 @@ SDValue SelectionDAGLegalize::ExpandVectorBuildThroughStack(SDNode* Node) { SDValue Idx = DAG.getConstant(Offset, FIPtr.getValueType()); Idx = DAG.getNode(ISD::ADD, dl, FIPtr.getValueType(), FIPtr, Idx); - // If EltVT smaller than OpVT, only store the bits necessary. - if (!OpVT.isVector() && EltVT.bitsLT(OpVT)) { + // If the destination vector element type is narrower than the source + // element type, only store the bits necessary. + if (EltVT.bitsLT(Node->getOperand(i).getValueType().getScalarType())) { Stores.push_back(DAG.getTruncStore(DAG.getEntryNode(), dl, Node->getOperand(i), Idx, SV, Offset, EltVT, false, false, 0)); @@ -1899,8 +1899,7 @@ SDValue SelectionDAGLegalize::ExpandLibCall(RTLIB::Libcall LC, SDNode *Node, TLI.LowerCallTo(InChain, RetTy, isSigned, !isSigned, false, false, 0, TLI.getLibcallCallingConv(LC), false, /*isReturnValueUsed=*/true, - Callee, Args, DAG, - Node->getDebugLoc(), DAG.GetOrdering(Node)); + Callee, Args, DAG, Node->getDebugLoc()); // Legalize the call sequence, starting with the chain. This will advance // the LastCALLSEQ_END to the legalized version of the CALLSEQ_END node that @@ -2308,7 +2307,7 @@ void SelectionDAGLegalize::ExpandNode(SDNode *Node, false, false, false, false, 0, CallingConv::C, false, /*isReturnValueUsed=*/true, DAG.getExternalSymbol("abort", TLI.getPointerTy()), - Args, DAG, dl, DAG.GetOrdering(Node)); + Args, DAG, dl); Results.push_back(CallResult.second); break; } diff --git a/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp b/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp index e4d123f..81f28ad 100644 --- a/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp +++ b/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp @@ -1080,8 +1080,8 @@ ExpandShiftWithKnownAmountBit(SDNode *N, SDValue &Lo, SDValue &Hi) { SDValue Amt = N->getOperand(1); EVT NVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0)); EVT ShTy = Amt.getValueType(); - unsigned ShBits = ShTy.getSizeInBits(); - unsigned NVTBits = NVT.getSizeInBits(); + unsigned ShBits = ShTy.getScalarType().getSizeInBits(); + unsigned NVTBits = NVT.getScalarType().getSizeInBits(); assert(isPowerOf2_32(NVTBits) && "Expanded integer type size not a power of two!"); DebugLoc dl = N->getDebugLoc(); diff --git a/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp b/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp index 0d929f1..f3e7ca4f 100644 --- a/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp +++ b/lib/CodeGen/SelectionDAG/LegalizeTypes.cpp @@ -1034,8 +1034,7 @@ SDValue DAGTypeLegalizer::MakeLibCall(RTLIB::Libcall LC, EVT RetVT, TLI.LowerCallTo(DAG.getEntryNode(), RetTy, isSigned, !isSigned, false, false, 0, TLI.getLibcallCallingConv(LC), false, /*isReturnValueUsed=*/true, - Callee, Args, DAG, dl, - DAG.GetOrdering(DAG.getEntryNode().getNode())); + Callee, Args, DAG, dl); return CallInfo.first; } diff --git a/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp b/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp index b51c61b..06e7b8c 100644 --- a/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp +++ b/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp @@ -218,8 +218,20 @@ void ScheduleDAGSDNodes::BuildSchedUnits() { // Check to see if the scheduler cares about latencies. bool UnitLatencies = ForceUnitLatencies(); - for (SelectionDAG::allnodes_iterator NI = DAG->allnodes_begin(), - E = DAG->allnodes_end(); NI != E; ++NI) { + // Add all nodes in depth first order. + SmallVector<SDNode*, 64> Worklist; + SmallPtrSet<SDNode*, 64> Visited; + Worklist.push_back(DAG->getRoot().getNode()); + Visited.insert(DAG->getRoot().getNode()); + + while (!Worklist.empty()) { + SDNode *NI = Worklist.pop_back_val(); + + // Add all operands to the worklist unless they've already been added. + for (unsigned i = 0, e = NI->getNumOperands(); i != e; ++i) + if (Visited.insert(NI->getOperand(i).getNode())) + Worklist.push_back(NI->getOperand(i).getNode()); + if (isPassiveNode(NI)) // Leaf node, e.g. a TargetImmediate. continue; diff --git a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp index 43cf37e..002bc68 100644 --- a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp +++ b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp @@ -468,18 +468,20 @@ static void AddNodeIDNode(FoldingSetNodeID &ID, const SDNode *N) { } /// encodeMemSDNodeFlags - Generic routine for computing a value for use in -/// the CSE map that carries volatility, indexing mode, and +/// the CSE map that carries volatility, temporalness, indexing mode, and /// extension/truncation information. /// static inline unsigned -encodeMemSDNodeFlags(int ConvType, ISD::MemIndexedMode AM, bool isVolatile) { +encodeMemSDNodeFlags(int ConvType, ISD::MemIndexedMode AM, bool isVolatile, + bool isNonTemporal) { assert((ConvType & 3) == ConvType && "ConvType may not require more than 2 bits!"); assert((AM & 7) == AM && "AM may not require more than 3 bits!"); return ConvType | (AM << 2) | - (isVolatile << 5); + (isVolatile << 5) | + (isNonTemporal << 6); } //===----------------------------------------------------------------------===// @@ -860,14 +862,14 @@ SDValue SelectionDAG::getZeroExtendInReg(SDValue Op, DebugLoc DL, EVT VT) { /// getNOT - Create a bitwise NOT operation as (XOR Val, -1). /// SDValue SelectionDAG::getNOT(DebugLoc DL, SDValue Val, EVT VT) { - EVT EltVT = VT.isVector() ? VT.getVectorElementType() : VT; + EVT EltVT = VT.getScalarType(); SDValue NegOne = getConstant(APInt::getAllOnesValue(EltVT.getSizeInBits()), VT); return getNode(ISD::XOR, DL, VT, Val, NegOne); } SDValue SelectionDAG::getConstant(uint64_t Val, EVT VT, bool isT) { - EVT EltVT = VT.isVector() ? VT.getVectorElementType() : VT; + EVT EltVT = VT.getScalarType(); assert((EltVT.getSizeInBits() >= 64 || (uint64_t)((int64_t)Val >> EltVT.getSizeInBits()) + 1 < 2) && "getConstant with a uint64_t value that doesn't fit in the type!"); @@ -881,7 +883,7 @@ SDValue SelectionDAG::getConstant(const APInt &Val, EVT VT, bool isT) { SDValue SelectionDAG::getConstant(const ConstantInt &Val, EVT VT, bool isT) { assert(VT.isInteger() && "Cannot create FP integer constant!"); - EVT EltVT = VT.isVector() ? VT.getVectorElementType() : VT; + EVT EltVT = VT.getScalarType(); assert(Val.getBitWidth() == EltVT.getSizeInBits() && "APInt size does not match type size!"); @@ -924,8 +926,7 @@ SDValue SelectionDAG::getConstantFP(const APFloat& V, EVT VT, bool isTarget) { SDValue SelectionDAG::getConstantFP(const ConstantFP& V, EVT VT, bool isTarget){ assert(VT.isFloatingPoint() && "Cannot create integer FP constant!"); - EVT EltVT = - VT.isVector() ? VT.getVectorElementType() : VT; + EVT EltVT = VT.getScalarType(); // Do the map lookup using the actual bit pattern for the floating point // value, so that we don't have problems with 0.0 comparing equal to -0.0, and @@ -959,8 +960,7 @@ SDValue SelectionDAG::getConstantFP(const ConstantFP& V, EVT VT, bool isTarget){ } SDValue SelectionDAG::getConstantFP(double Val, EVT VT, bool isTarget) { - EVT EltVT = - VT.isVector() ? VT.getVectorElementType() : VT; + EVT EltVT = VT.getScalarType(); if (EltVT==MVT::f32) return getConstantFP(APFloat((float)Val), VT, isTarget); else @@ -1345,7 +1345,7 @@ SDValue SelectionDAG::getBlockAddress(BlockAddress *BA, EVT VT, } SDValue SelectionDAG::getSrcValue(const Value *V) { - assert((!V || isa<PointerType>(V->getType())) && + assert((!V || V->getType()->isPointerTy()) && "SrcValue is not a pointer?"); FoldingSetNodeID ID; @@ -2233,6 +2233,29 @@ bool SelectionDAG::isKnownNeverNaN(SDValue Op) const { return false; } +bool SelectionDAG::isKnownNeverZero(SDValue Op) const { + // If the value is a constant, we can obviously see if it is a zero or not. + if (const ConstantFPSDNode *C = dyn_cast<ConstantFPSDNode>(Op)) + return !C->isZero(); + + // TODO: Recognize more cases here. + + return false; +} + +bool SelectionDAG::isEqualTo(SDValue A, SDValue B) const { + // Check the obvious case. + if (A == B) return true; + + // For for negative and positive zero. + if (const ConstantFPSDNode *CA = dyn_cast<ConstantFPSDNode>(A)) + if (const ConstantFPSDNode *CB = dyn_cast<ConstantFPSDNode>(B)) + if (CA->isZero() && CB->isZero()) return true; + + // Otherwise they may not be equal. + return false; +} + bool SelectionDAG::isVerifiedDebugInfoDesc(SDValue Op) const { GlobalAddressSDNode *GA = dyn_cast<GlobalAddressSDNode>(Op); if (!GA) return false; @@ -3081,8 +3104,7 @@ SDValue SelectionDAG::getStackArgumentTokenFactor(SDValue Chain) { /// operand. static SDValue getMemsetValue(SDValue Value, EVT VT, SelectionDAG &DAG, DebugLoc dl) { - unsigned NumBits = VT.isVector() ? - VT.getVectorElementType().getSizeInBits() : VT.getSizeInBits(); + unsigned NumBits = VT.getScalarType().getSizeInBits(); if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Value)) { APInt Val = APInt(NumBits, C->getZExtValue() & 255); unsigned Shift = 8; @@ -3474,7 +3496,7 @@ SDValue SelectionDAG::getMemcpy(SDValue Chain, DebugLoc dl, SDValue Dst, /*isReturnValueUsed=*/false, getExternalSymbol(TLI.getLibcallName(RTLIB::MEMCPY), TLI.getPointerTy()), - Args, *this, dl, GetOrdering(Chain.getNode())); + Args, *this, dl); return CallResult.second; } @@ -3523,7 +3545,7 @@ SDValue SelectionDAG::getMemmove(SDValue Chain, DebugLoc dl, SDValue Dst, /*isReturnValueUsed=*/false, getExternalSymbol(TLI.getLibcallName(RTLIB::MEMMOVE), TLI.getPointerTy()), - Args, *this, dl, GetOrdering(Chain.getNode())); + Args, *this, dl); return CallResult.second; } @@ -3582,7 +3604,7 @@ SDValue SelectionDAG::getMemset(SDValue Chain, DebugLoc dl, SDValue Dst, /*isReturnValueUsed=*/false, getExternalSymbol(TLI.getLibcallName(RTLIB::MEMSET), TLI.getPointerTy()), - Args, *this, dl, GetOrdering(Chain.getNode())); + Args, *this, dl); return CallResult.second; } @@ -3845,7 +3867,8 @@ SelectionDAG::getLoad(ISD::MemIndexedMode AM, DebugLoc dl, FoldingSetNodeID ID; AddNodeIDNode(ID, ISD::LOAD, VTs, Ops, 3); ID.AddInteger(MemVT.getRawBits()); - ID.AddInteger(encodeMemSDNodeFlags(ExtType, AM, MMO->isVolatile())); + ID.AddInteger(encodeMemSDNodeFlags(ExtType, AM, MMO->isVolatile(), + MMO->isNonTemporal())); void *IP = 0; if (SDNode *E = CSEMap.FindNodeOrInsertPos(ID, IP)) { cast<LoadSDNode>(E)->refineAlignment(MMO); @@ -3926,7 +3949,8 @@ SDValue SelectionDAG::getStore(SDValue Chain, DebugLoc dl, SDValue Val, FoldingSetNodeID ID; AddNodeIDNode(ID, ISD::STORE, VTs, Ops, 4); ID.AddInteger(VT.getRawBits()); - ID.AddInteger(encodeMemSDNodeFlags(false, ISD::UNINDEXED, MMO->isVolatile())); + ID.AddInteger(encodeMemSDNodeFlags(false, ISD::UNINDEXED, MMO->isVolatile(), + MMO->isNonTemporal())); void *IP = 0; if (SDNode *E = CSEMap.FindNodeOrInsertPos(ID, IP)) { cast<StoreSDNode>(E)->refineAlignment(MMO); @@ -3989,7 +4013,8 @@ SDValue SelectionDAG::getTruncStore(SDValue Chain, DebugLoc dl, SDValue Val, FoldingSetNodeID ID; AddNodeIDNode(ID, ISD::STORE, VTs, Ops, 4); ID.AddInteger(SVT.getRawBits()); - ID.AddInteger(encodeMemSDNodeFlags(true, ISD::UNINDEXED, MMO->isVolatile())); + ID.AddInteger(encodeMemSDNodeFlags(true, ISD::UNINDEXED, MMO->isVolatile(), + MMO->isNonTemporal())); void *IP = 0; if (SDNode *E = CSEMap.FindNodeOrInsertPos(ID, IP)) { cast<StoreSDNode>(E)->refineAlignment(MMO); @@ -4548,91 +4573,13 @@ SDNode *SelectionDAG::SelectNodeTo(SDNode *N, unsigned MachineOpc, SDNode *SelectionDAG::SelectNodeTo(SDNode *N, unsigned MachineOpc, SDVTList VTs, const SDValue *Ops, unsigned NumOps) { - return MorphNodeTo(N, ~MachineOpc, VTs, Ops, NumOps); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT) { - SDVTList VTs = getVTList(VT); - return MorphNodeTo(N, Opc, VTs, 0, 0); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT, SDValue Op1) { - SDVTList VTs = getVTList(VT); - SDValue Ops[] = { Op1 }; - return MorphNodeTo(N, Opc, VTs, Ops, 1); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT, SDValue Op1, - SDValue Op2) { - SDVTList VTs = getVTList(VT); - SDValue Ops[] = { Op1, Op2 }; - return MorphNodeTo(N, Opc, VTs, Ops, 2); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT, SDValue Op1, - SDValue Op2, SDValue Op3) { - SDVTList VTs = getVTList(VT); - SDValue Ops[] = { Op1, Op2, Op3 }; - return MorphNodeTo(N, Opc, VTs, Ops, 3); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT, const SDValue *Ops, - unsigned NumOps) { - SDVTList VTs = getVTList(VT); - return MorphNodeTo(N, Opc, VTs, Ops, NumOps); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT1, EVT VT2, const SDValue *Ops, - unsigned NumOps) { - SDVTList VTs = getVTList(VT1, VT2); - return MorphNodeTo(N, Opc, VTs, Ops, NumOps); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT1, EVT VT2) { - SDVTList VTs = getVTList(VT1, VT2); - return MorphNodeTo(N, Opc, VTs, (SDValue *)0, 0); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT1, EVT VT2, EVT VT3, - const SDValue *Ops, unsigned NumOps) { - SDVTList VTs = getVTList(VT1, VT2, VT3); - return MorphNodeTo(N, Opc, VTs, Ops, NumOps); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT1, EVT VT2, - SDValue Op1) { - SDVTList VTs = getVTList(VT1, VT2); - SDValue Ops[] = { Op1 }; - return MorphNodeTo(N, Opc, VTs, Ops, 1); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT1, EVT VT2, - SDValue Op1, SDValue Op2) { - SDVTList VTs = getVTList(VT1, VT2); - SDValue Ops[] = { Op1, Op2 }; - return MorphNodeTo(N, Opc, VTs, Ops, 2); -} - -SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, - EVT VT1, EVT VT2, - SDValue Op1, SDValue Op2, - SDValue Op3) { - SDVTList VTs = getVTList(VT1, VT2); - SDValue Ops[] = { Op1, Op2, Op3 }; - return MorphNodeTo(N, Opc, VTs, Ops, 3); + N = MorphNodeTo(N, ~MachineOpc, VTs, Ops, NumOps); + // Reset the NodeID to -1. + N->setNodeId(-1); + return N; } -/// MorphNodeTo - These *mutate* the specified node to have the specified +/// MorphNodeTo - This *mutates* the specified node to have the specified /// return type, opcode, and operands. /// /// Note that MorphNodeTo returns the resultant node. If there is already a @@ -4708,12 +4655,14 @@ SDNode *SelectionDAG::MorphNodeTo(SDNode *N, unsigned Opc, // Delete any nodes that are still dead after adding the uses for the // new operands. - SmallVector<SDNode *, 16> DeadNodes; - for (SmallPtrSet<SDNode *, 16>::iterator I = DeadNodeSet.begin(), - E = DeadNodeSet.end(); I != E; ++I) - if ((*I)->use_empty()) - DeadNodes.push_back(*I); - RemoveDeadNodes(DeadNodes); + if (!DeadNodeSet.empty()) { + SmallVector<SDNode *, 16> DeadNodes; + for (SmallPtrSet<SDNode *, 16>::iterator I = DeadNodeSet.begin(), + E = DeadNodeSet.end(); I != E; ++I) + if ((*I)->use_empty()) + DeadNodes.push_back(*I); + RemoveDeadNodes(DeadNodes); + } if (IP) CSEMap.InsertNode(N, IP); // Memoize the new node. @@ -5293,8 +5242,11 @@ GlobalAddressSDNode::GlobalAddressSDNode(unsigned Opc, const GlobalValue *GA, MemSDNode::MemSDNode(unsigned Opc, DebugLoc dl, SDVTList VTs, EVT memvt, MachineMemOperand *mmo) : SDNode(Opc, dl, VTs), MemoryVT(memvt), MMO(mmo) { - SubclassData = encodeMemSDNodeFlags(0, ISD::UNINDEXED, MMO->isVolatile()); + SubclassData = encodeMemSDNodeFlags(0, ISD::UNINDEXED, MMO->isVolatile(), + MMO->isNonTemporal()); assert(isVolatile() == MMO->isVolatile() && "Volatile encoding error!"); + assert(isNonTemporal() == MMO->isNonTemporal() && + "Non-temporal encoding error!"); assert(memvt.getStoreSize() == MMO->getSize() && "Size mismatch!"); } @@ -5303,7 +5255,8 @@ MemSDNode::MemSDNode(unsigned Opc, DebugLoc dl, SDVTList VTs, MachineMemOperand *mmo) : SDNode(Opc, dl, VTs, Ops, NumOps), MemoryVT(memvt), MMO(mmo) { - SubclassData = encodeMemSDNodeFlags(0, ISD::UNINDEXED, MMO->isVolatile()); + SubclassData = encodeMemSDNodeFlags(0, ISD::UNINDEXED, MMO->isVolatile(), + MMO->isNonTemporal()); assert(isVolatile() == MMO->isVolatile() && "Volatile encoding error!"); assert(memvt.getStoreSize() == MMO->getSize() && "Size mismatch!"); } @@ -5472,15 +5425,15 @@ std::string SDNode::getOperationName(const SelectionDAG *G) const { if (const TargetInstrInfo *TII = G->getTarget().getInstrInfo()) if (getMachineOpcode() < TII->getNumOpcodes()) return TII->get(getMachineOpcode()).getName(); - return "<<Unknown Machine Node>>"; + return "<<Unknown Machine Node #" + utostr(getOpcode()) + ">>"; } if (G) { const TargetLowering &TLI = G->getTargetLoweringInfo(); const char *Name = TLI.getTargetNodeName(getOpcode()); if (Name) return Name; - return "<<Unknown Target Node>>"; + return "<<Unknown Target Node #" + utostr(getOpcode()) + ">>"; } - return "<<Unknown Node>>"; + return "<<Unknown Node #" + utostr(getOpcode()) + ">>"; #ifndef NDEBUG case ISD::DELETED_NODE: @@ -5917,6 +5870,9 @@ void SDNode::print_details(raw_ostream &OS, const SelectionDAG *G) const { if (G) if (unsigned Order = G->GetOrdering(this)) OS << " [ORD=" << Order << ']'; + + if (getNodeId() != -1) + OS << " [ID=" << getNodeId() << ']'; } void SDNode::print(raw_ostream &OS, const SelectionDAG *G) const { @@ -6305,31 +6261,37 @@ bool ShuffleVectorSDNode::isSplatMask(const int *Mask, EVT VT) { return true; } +#ifdef XDEBUG static void checkForCyclesHelper(const SDNode *N, - std::set<const SDNode *> &visited) { - if (visited.find(N) != visited.end()) { + SmallPtrSet<const SDNode*, 32> &Visited, + SmallPtrSet<const SDNode*, 32> &Checked) { + // If this node has already been checked, don't check it again. + if (Checked.count(N)) + return; + + // If a node has already been visited on this depth-first walk, reject it as + // a cycle. + if (!Visited.insert(N)) { dbgs() << "Offending node:\n"; N->dumprFull(); - assert(0 && "Detected cycle in SelectionDAG"); + errs() << "Detected cycle in SelectionDAG\n"; + abort(); } - - std::set<const SDNode*>::iterator i; - bool inserted; - - tie(i, inserted) = visited.insert(N); - assert(inserted && "Missed cycle"); - - for(unsigned i = 0; i < N->getNumOperands(); ++i) { - checkForCyclesHelper(N->getOperand(i).getNode(), visited); - } - visited.erase(i); + + for(unsigned i = 0, e = N->getNumOperands(); i != e; ++i) + checkForCyclesHelper(N->getOperand(i).getNode(), Visited, Checked); + + Checked.insert(N); + Visited.erase(N); } +#endif void llvm::checkForCycles(const llvm::SDNode *N) { #ifdef XDEBUG assert(N && "Checking nonexistant SDNode"); - std::set<const SDNode *> visited; - checkForCyclesHelper(N, visited); + SmallPtrSet<const SDNode*, 32> visited; + SmallPtrSet<const SDNode*, 32> checked; + checkForCyclesHelper(N, visited, checked); #endif } diff --git a/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp b/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp index 85ecb95..308742b 100644 --- a/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp +++ b/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp @@ -155,7 +155,7 @@ namespace { /// this value and returns the result as a ValueVTs value. This uses /// Chain/Flag as the input and updates them for the output Chain/Flag. /// If the Flag pointer is NULL, no flag is used. - SDValue getCopyFromRegs(SelectionDAG &DAG, DebugLoc dl, unsigned Order, + SDValue getCopyFromRegs(SelectionDAG &DAG, DebugLoc dl, SDValue &Chain, SDValue *Flag) const; /// getCopyToRegs - Emit a series of CopyToReg nodes that copies the @@ -163,14 +163,14 @@ namespace { /// Chain/Flag as the input and updates them for the output Chain/Flag. /// If the Flag pointer is NULL, no flag is used. void getCopyToRegs(SDValue Val, SelectionDAG &DAG, DebugLoc dl, - unsigned Order, SDValue &Chain, SDValue *Flag) const; + SDValue &Chain, SDValue *Flag) const; /// AddInlineAsmOperands - Add this value to the specified inlineasm node /// operand list. This adds the code marker, matching input operand index /// (if applicable), and includes the number of values added into it. void AddInlineAsmOperands(unsigned Code, bool HasMatching, unsigned MatchingIdx, - SelectionDAG &DAG, unsigned Order, + SelectionDAG &DAG, std::vector<SDValue> &Ops) const; }; } @@ -180,7 +180,7 @@ namespace { /// larger then ValueVT then AssertOp can be used to specify whether the extra /// bits are known to be zero (ISD::AssertZext) or sign extended from ValueVT /// (ISD::AssertSext). -static SDValue getCopyFromParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, +static SDValue getCopyFromParts(SelectionDAG &DAG, DebugLoc dl, const SDValue *Parts, unsigned NumParts, EVT PartVT, EVT ValueVT, ISD::NodeType AssertOp = ISD::DELETED_NODE) { @@ -205,9 +205,9 @@ static SDValue getCopyFromParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, EVT HalfVT = EVT::getIntegerVT(*DAG.getContext(), RoundBits/2); if (RoundParts > 2) { - Lo = getCopyFromParts(DAG, dl, Order, Parts, RoundParts / 2, + Lo = getCopyFromParts(DAG, dl, Parts, RoundParts / 2, PartVT, HalfVT); - Hi = getCopyFromParts(DAG, dl, Order, Parts + RoundParts / 2, + Hi = getCopyFromParts(DAG, dl, Parts + RoundParts / 2, RoundParts / 2, PartVT, HalfVT); } else { Lo = DAG.getNode(ISD::BIT_CONVERT, dl, HalfVT, Parts[0]); @@ -223,7 +223,7 @@ static SDValue getCopyFromParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, // Assemble the trailing non-power-of-2 part. unsigned OddParts = NumParts - RoundParts; EVT OddVT = EVT::getIntegerVT(*DAG.getContext(), OddParts * PartBits); - Hi = getCopyFromParts(DAG, dl, Order, + Hi = getCopyFromParts(DAG, dl, Parts + RoundParts, OddParts, PartVT, OddVT); // Combine the round and odd parts. @@ -259,7 +259,7 @@ static SDValue getCopyFromParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, // If the register was not expanded, truncate or copy the value, // as appropriate. for (unsigned i = 0; i != NumParts; ++i) - Ops[i] = getCopyFromParts(DAG, dl, Order, &Parts[i], 1, + Ops[i] = getCopyFromParts(DAG, dl, &Parts[i], 1, PartVT, IntermediateVT); } else if (NumParts > 0) { // If the intermediate type was expanded, build the intermediate @@ -268,7 +268,7 @@ static SDValue getCopyFromParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, "Must expand into a divisible number of parts!"); unsigned Factor = NumParts / NumIntermediates; for (unsigned i = 0; i != NumIntermediates; ++i) - Ops[i] = getCopyFromParts(DAG, dl, Order, &Parts[i * Factor], Factor, + Ops[i] = getCopyFromParts(DAG, dl, &Parts[i * Factor], Factor, PartVT, IntermediateVT); } @@ -292,7 +292,7 @@ static SDValue getCopyFromParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, assert(ValueVT.isFloatingPoint() && PartVT.isInteger() && !PartVT.isVector() && "Unexpected split"); EVT IntVT = EVT::getIntegerVT(*DAG.getContext(), ValueVT.getSizeInBits()); - Val = getCopyFromParts(DAG, dl, Order, Parts, NumParts, PartVT, IntVT); + Val = getCopyFromParts(DAG, dl, Parts, NumParts, PartVT, IntVT); } } @@ -349,7 +349,7 @@ static SDValue getCopyFromParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, /// getCopyToParts - Create a series of nodes that contain the specified value /// split into legal parts. If the parts contain more bits than Val, then, for /// integers, ExtendKind can be used to specify how to generate the extra bits. -static void getCopyToParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, +static void getCopyToParts(SelectionDAG &DAG, DebugLoc dl, SDValue Val, SDValue *Parts, unsigned NumParts, EVT PartVT, ISD::NodeType ExtendKind = ISD::ANY_EXTEND) { @@ -417,7 +417,7 @@ static void getCopyToParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, SDValue OddVal = DAG.getNode(ISD::SRL, dl, ValueVT, Val, DAG.getConstant(RoundBits, TLI.getPointerTy())); - getCopyToParts(DAG, dl, Order, OddVal, Parts + RoundParts, + getCopyToParts(DAG, dl, OddVal, Parts + RoundParts, OddParts, PartVT); if (TLI.isBigEndian()) @@ -514,7 +514,7 @@ static void getCopyToParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, // If the register was not expanded, promote or copy the value, // as appropriate. for (unsigned i = 0; i != NumParts; ++i) - getCopyToParts(DAG, dl, Order, Ops[i], &Parts[i], 1, PartVT); + getCopyToParts(DAG, dl, Ops[i], &Parts[i], 1, PartVT); } else if (NumParts > 0) { // If the intermediate type was expanded, split each the value into // legal parts. @@ -522,7 +522,7 @@ static void getCopyToParts(SelectionDAG &DAG, DebugLoc dl, unsigned Order, "Must expand into a divisible number of parts!"); unsigned Factor = NumParts / NumIntermediates; for (unsigned i = 0; i != NumIntermediates; ++i) - getCopyToParts(DAG, dl, Order, Ops[i], &Parts[i*Factor], Factor, PartVT); + getCopyToParts(DAG, dl, Ops[i], &Parts[i*Factor], Factor, PartVT); } } @@ -680,7 +680,7 @@ SDValue SelectionDAGBuilder::getValue(const Value *V) { getCurDebugLoc()); } - if (isa<StructType>(C->getType()) || isa<ArrayType>(C->getType())) { + if (C->getType()->isStructTy() || C->getType()->isArrayTy()) { assert((isa<ConstantAggregateZero>(C) || isa<UndefValue>(C)) && "Unknown struct or array constant!"); @@ -747,8 +747,7 @@ SDValue SelectionDAGBuilder::getValue(const Value *V) { RegsForValue RFV(*DAG.getContext(), TLI, InReg, V->getType()); SDValue Chain = DAG.getEntryNode(); - return RFV.getCopyFromRegs(DAG, getCurDebugLoc(), - SDNodeOrder, Chain, NULL); + return RFV.getCopyFromRegs(DAG, getCurDebugLoc(), Chain, NULL); } /// Get the EVTs and ArgFlags collections that represent the legalized return @@ -849,14 +848,12 @@ void SelectionDAGBuilder::visitRet(ReturnInst &I) { Chain = DAG.getNode(ISD::TokenFactor, getCurDebugLoc(), MVT::Other, &Chains[0], NumValues); - } else { - for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i) { - SmallVector<EVT, 4> ValueVTs; - ComputeValueVTs(TLI, I.getOperand(i)->getType(), ValueVTs); - unsigned NumValues = ValueVTs.size(); - if (NumValues == 0) continue; - - SDValue RetOp = getValue(I.getOperand(i)); + } else if (I.getNumOperands() != 0) { + SmallVector<EVT, 4> ValueVTs; + ComputeValueVTs(TLI, I.getOperand(0)->getType(), ValueVTs); + unsigned NumValues = ValueVTs.size(); + if (NumValues) { + SDValue RetOp = getValue(I.getOperand(0)); for (unsigned j = 0, f = NumValues; j != f; ++j) { EVT VT = ValueVTs[j]; @@ -881,7 +878,7 @@ void SelectionDAGBuilder::visitRet(ReturnInst &I) { unsigned NumParts = TLI.getNumRegisters(*DAG.getContext(), VT); EVT PartVT = TLI.getRegisterType(*DAG.getContext(), VT); SmallVector<SDValue, 4> Parts(NumParts); - getCopyToParts(DAG, getCurDebugLoc(), SDNodeOrder, + getCopyToParts(DAG, getCurDebugLoc(), SDValue(RetOp.getNode(), RetOp.getResNo() + j), &Parts[0], NumParts, PartVT, ExtendKind); @@ -2080,7 +2077,7 @@ void SelectionDAGBuilder::visitIndirectBr(IndirectBrInst &I) { void SelectionDAGBuilder::visitFSub(User &I) { // -0.0 - X --> fneg const Type *Ty = I.getType(); - if (isa<VectorType>(Ty)) { + if (Ty->isVectorTy()) { if (ConstantVector *CV = dyn_cast<ConstantVector>(I.getOperand(0))) { const VectorType *DestTy = cast<VectorType>(I.getType()); const Type *ElTy = DestTy->getElementType(); @@ -2117,7 +2114,7 @@ void SelectionDAGBuilder::visitBinary(User &I, unsigned OpCode) { void SelectionDAGBuilder::visitShift(User &I, unsigned Opcode) { SDValue Op1 = getValue(I.getOperand(0)); SDValue Op2 = getValue(I.getOperand(1)); - if (!isa<VectorType>(I.getType()) && + if (!I.getType()->isVectorTy() && Op2.getValueType() != TLI.getShiftAmountTy()) { // If the operand is smaller than the shift count type, promote it. EVT PTy = TLI.getPointerTy(); @@ -2890,7 +2887,7 @@ void SelectionDAGBuilder::visitTargetIntrinsic(CallInst &I, /// /// where Op is the hexidecimal representation of floating point value. static SDValue -GetSignificand(SelectionDAG &DAG, SDValue Op, DebugLoc dl, unsigned Order) { +GetSignificand(SelectionDAG &DAG, SDValue Op, DebugLoc dl) { SDValue t1 = DAG.getNode(ISD::AND, dl, MVT::i32, Op, DAG.getConstant(0x007fffff, MVT::i32)); SDValue t2 = DAG.getNode(ISD::OR, dl, MVT::i32, t1, @@ -2905,7 +2902,7 @@ GetSignificand(SelectionDAG &DAG, SDValue Op, DebugLoc dl, unsigned Order) { /// where Op is the hexidecimal representation of floating point value. static SDValue GetExponent(SelectionDAG &DAG, SDValue Op, const TargetLowering &TLI, - DebugLoc dl, unsigned Order) { + DebugLoc dl) { SDValue t0 = DAG.getNode(ISD::AND, dl, MVT::i32, Op, DAG.getConstant(0x7f800000, MVT::i32)); SDValue t1 = DAG.getNode(ISD::SRL, dl, MVT::i32, t0, @@ -3089,13 +3086,13 @@ SelectionDAGBuilder::visitLog(CallInst &I) { SDValue Op1 = DAG.getNode(ISD::BIT_CONVERT, dl, MVT::i32, Op); // Scale the exponent by log(2) [0.69314718f]. - SDValue Exp = GetExponent(DAG, Op1, TLI, dl, SDNodeOrder); + SDValue Exp = GetExponent(DAG, Op1, TLI, dl); SDValue LogOfExponent = DAG.getNode(ISD::FMUL, dl, MVT::f32, Exp, getF32Constant(DAG, 0x3f317218)); // Get the significand and build it into a floating-point number with // exponent of 1. - SDValue X = GetSignificand(DAG, Op1, dl, SDNodeOrder); + SDValue X = GetSignificand(DAG, Op1, dl); if (LimitFloatPrecision <= 6) { // For floating-point precision of 6: @@ -3199,11 +3196,11 @@ SelectionDAGBuilder::visitLog2(CallInst &I) { SDValue Op1 = DAG.getNode(ISD::BIT_CONVERT, dl, MVT::i32, Op); // Get the exponent. - SDValue LogOfExponent = GetExponent(DAG, Op1, TLI, dl, SDNodeOrder); + SDValue LogOfExponent = GetExponent(DAG, Op1, TLI, dl); // Get the significand and build it into a floating-point number with // exponent of 1. - SDValue X = GetSignificand(DAG, Op1, dl, SDNodeOrder); + SDValue X = GetSignificand(DAG, Op1, dl); // Different possible minimax approximations of significand in // floating-point for various degrees of accuracy over [1,2]. @@ -3308,13 +3305,13 @@ SelectionDAGBuilder::visitLog10(CallInst &I) { SDValue Op1 = DAG.getNode(ISD::BIT_CONVERT, dl, MVT::i32, Op); // Scale the exponent by log10(2) [0.30102999f]. - SDValue Exp = GetExponent(DAG, Op1, TLI, dl, SDNodeOrder); + SDValue Exp = GetExponent(DAG, Op1, TLI, dl); SDValue LogOfExponent = DAG.getNode(ISD::FMUL, dl, MVT::f32, Exp, getF32Constant(DAG, 0x3e9a209a)); // Get the significand and build it into a floating-point number with // exponent of 1. - SDValue X = GetSignificand(DAG, Op1, dl, SDNodeOrder); + SDValue X = GetSignificand(DAG, Op1, dl); if (LimitFloatPrecision <= 6) { // For floating-point precision of 6: @@ -4287,8 +4284,8 @@ isInTailCallPosition(CallSite CS, Attributes CalleeRetAttr, // Check for a truly no-op bitcast. if (isa<BitCastInst>(U) && (U->getOperand(0)->getType() == U->getType() || - (isa<PointerType>(U->getOperand(0)->getType()) && - isa<PointerType>(U->getType())))) + (U->getOperand(0)->getType()->isPointerTy() && + U->getType()->isPointerTy()))) continue; // Otherwise it's not a true no-op. return false; @@ -4396,7 +4393,7 @@ void SelectionDAGBuilder::LowerCallTo(CallSite CS, SDValue Callee, CS.getCallingConv(), isTailCall, !CS.getInstruction()->use_empty(), - Callee, Args, DAG, getCurDebugLoc(), SDNodeOrder); + Callee, Args, DAG, getCurDebugLoc()); assert((isTailCall || Result.second.getNode()) && "Non-null chain expected with non-tail call!"); assert((Result.second.getNode() || !Result.first.getNode()) && @@ -4444,7 +4441,7 @@ void SelectionDAGBuilder::LowerCallTo(CallSite CS, SDValue Callee, unsigned NumRegs = TLI.getNumRegisters(RetTy->getContext(), VT); SDValue ReturnValue = - getCopyFromParts(DAG, getCurDebugLoc(), SDNodeOrder, &Values[CurReg], NumRegs, + getCopyFromParts(DAG, getCurDebugLoc(), &Values[CurReg], NumRegs, RegisterVT, VT, AssertOp); ReturnValues.push_back(ReturnValue); CurReg += NumRegs; @@ -4541,9 +4538,9 @@ bool SelectionDAGBuilder::visitMemCmpCall(CallInst &I) { return false; Value *LHS = I.getOperand(1), *RHS = I.getOperand(2); - if (!isa<PointerType>(LHS->getType()) || !isa<PointerType>(RHS->getType()) || - !isa<IntegerType>(I.getOperand(3)->getType()) || - !isa<IntegerType>(I.getType())) + if (!LHS->getType()->isPointerTy() || !RHS->getType()->isPointerTy() || + !I.getOperand(3)->getType()->isIntegerTy() || + !I.getType()->isIntegerTy()) return false; ConstantInt *Size = dyn_cast<ConstantInt>(I.getOperand(3)); @@ -4711,8 +4708,7 @@ void SelectionDAGBuilder::visitCall(CallInst &I) { /// Chain/Flag as the input and updates them for the output Chain/Flag. /// If the Flag pointer is NULL, no flag is used. SDValue RegsForValue::getCopyFromRegs(SelectionDAG &DAG, DebugLoc dl, - unsigned Order, SDValue &Chain, - SDValue *Flag) const { + SDValue &Chain, SDValue *Flag) const { // Assemble the legal parts into the final values. SmallVector<SDValue, 4> Values(ValueVTs.size()); SmallVector<SDValue, 8> Parts; @@ -4777,7 +4773,7 @@ SDValue RegsForValue::getCopyFromRegs(SelectionDAG &DAG, DebugLoc dl, Parts[i] = P; } - Values[Value] = getCopyFromParts(DAG, dl, Order, Parts.begin(), + Values[Value] = getCopyFromParts(DAG, dl, Parts.begin(), NumRegs, RegisterVT, ValueVT); Part += NumRegs; Parts.clear(); @@ -4793,8 +4789,7 @@ SDValue RegsForValue::getCopyFromRegs(SelectionDAG &DAG, DebugLoc dl, /// Chain/Flag as the input and updates them for the output Chain/Flag. /// If the Flag pointer is NULL, no flag is used. void RegsForValue::getCopyToRegs(SDValue Val, SelectionDAG &DAG, DebugLoc dl, - unsigned Order, SDValue &Chain, - SDValue *Flag) const { + SDValue &Chain, SDValue *Flag) const { // Get the list of the values's legal parts. unsigned NumRegs = Regs.size(); SmallVector<SDValue, 8> Parts(NumRegs); @@ -4803,7 +4798,7 @@ void RegsForValue::getCopyToRegs(SDValue Val, SelectionDAG &DAG, DebugLoc dl, unsigned NumParts = TLI->getNumRegisters(*DAG.getContext(), ValueVT); EVT RegisterVT = RegVTs[Value]; - getCopyToParts(DAG, dl, Order, + getCopyToParts(DAG, dl, Val.getValue(Val.getResNo() + Value), &Parts[Part], NumParts, RegisterVT); Part += NumParts; @@ -4844,7 +4839,7 @@ void RegsForValue::getCopyToRegs(SDValue Val, SelectionDAG &DAG, DebugLoc dl, /// values added into it. void RegsForValue::AddInlineAsmOperands(unsigned Code, bool HasMatching,unsigned MatchingIdx, - SelectionDAG &DAG, unsigned Order, + SelectionDAG &DAG, std::vector<SDValue> &Ops) const { assert(Regs.size() < (1 << 13) && "Too many inline asm outputs!"); unsigned Flag = Code | (Regs.size() << 3); @@ -5434,7 +5429,7 @@ void SelectionDAGBuilder::visitInlineAsm(CallSite CS) { 2 /* REGDEF */ , false, 0, - DAG, SDNodeOrder, + DAG, AsmNodeOperands); break; } @@ -5482,10 +5477,10 @@ void SelectionDAGBuilder::visitInlineAsm(CallSite CS) { // Use the produced MatchedRegs object to MatchedRegs.getCopyToRegs(InOperandVal, DAG, getCurDebugLoc(), - SDNodeOrder, Chain, &Flag); + Chain, &Flag); MatchedRegs.AddInlineAsmOperands(1 /*REGUSE*/, true, OpInfo.getMatchedOperand(), - DAG, SDNodeOrder, AsmNodeOperands); + DAG, AsmNodeOperands); break; } else { assert(((OpFlag & 7) == 4) && "Unknown matching constraint!"); @@ -5546,11 +5541,10 @@ void SelectionDAGBuilder::visitInlineAsm(CallSite CS) { } OpInfo.AssignedRegs.getCopyToRegs(InOperandVal, DAG, getCurDebugLoc(), - SDNodeOrder, Chain, &Flag); + Chain, &Flag); OpInfo.AssignedRegs.AddInlineAsmOperands(1/*REGUSE*/, false, 0, - DAG, SDNodeOrder, - AsmNodeOperands); + DAG, AsmNodeOperands); break; } case InlineAsm::isClobber: { @@ -5558,7 +5552,7 @@ void SelectionDAGBuilder::visitInlineAsm(CallSite CS) { // allocator is aware that the physreg got clobbered. if (!OpInfo.AssignedRegs.Regs.empty()) OpInfo.AssignedRegs.AddInlineAsmOperands(6 /* EARLYCLOBBER REGDEF */, - false, 0, DAG, SDNodeOrder, + false, 0, DAG, AsmNodeOperands); break; } @@ -5578,7 +5572,7 @@ void SelectionDAGBuilder::visitInlineAsm(CallSite CS) { // and set it as the value of the call. if (!RetValRegs.Regs.empty()) { SDValue Val = RetValRegs.getCopyFromRegs(DAG, getCurDebugLoc(), - SDNodeOrder, Chain, &Flag); + Chain, &Flag); // FIXME: Why don't we do this for inline asms with MRVs? if (CS.getType()->isSingleValueType() && CS.getType()->isSized()) { @@ -5618,7 +5612,7 @@ void SelectionDAGBuilder::visitInlineAsm(CallSite CS) { RegsForValue &OutRegs = IndirectStoresToEmit[i].first; Value *Ptr = IndirectStoresToEmit[i].second; SDValue OutVal = OutRegs.getCopyFromRegs(DAG, getCurDebugLoc(), - SDNodeOrder, Chain, &Flag); + Chain, &Flag); StoresToEmit.push_back(std::make_pair(OutVal, Ptr)); } @@ -5683,8 +5677,7 @@ TargetLowering::LowerCallTo(SDValue Chain, const Type *RetTy, CallingConv::ID CallConv, bool isTailCall, bool isReturnValueUsed, SDValue Callee, - ArgListTy &Args, SelectionDAG &DAG, DebugLoc dl, - unsigned Order) { + ArgListTy &Args, SelectionDAG &DAG, DebugLoc dl) { // Handle all of the outgoing arguments. SmallVector<ISD::OutputArg, 32> Outs; for (unsigned i = 0, e = Args.size(); i != e; ++i) { @@ -5735,7 +5728,7 @@ TargetLowering::LowerCallTo(SDValue Chain, const Type *RetTy, else if (Args[i].isZExt) ExtendKind = ISD::ZERO_EXTEND; - getCopyToParts(DAG, dl, Order, Op, &Parts[0], NumParts, + getCopyToParts(DAG, dl, Op, &Parts[0], NumParts, PartVT, ExtendKind); for (unsigned j = 0; j != NumParts; ++j) { @@ -5814,7 +5807,7 @@ TargetLowering::LowerCallTo(SDValue Chain, const Type *RetTy, EVT RegisterVT = getRegisterType(RetTy->getContext(), VT); unsigned NumRegs = getNumRegisters(RetTy->getContext(), VT); - ReturnValues.push_back(getCopyFromParts(DAG, dl, Order, &InVals[CurReg], + ReturnValues.push_back(getCopyFromParts(DAG, dl, &InVals[CurReg], NumRegs, RegisterVT, VT, AssertOp)); CurReg += NumRegs; @@ -5854,7 +5847,7 @@ void SelectionDAGBuilder::CopyValueToVirtualRegister(Value *V, unsigned Reg) { RegsForValue RFV(V->getContext(), TLI, Reg, V->getType()); SDValue Chain = DAG.getEntryNode(); - RFV.getCopyToRegs(Op, DAG, getCurDebugLoc(), SDNodeOrder, Chain, 0); + RFV.getCopyToRegs(Op, DAG, getCurDebugLoc(), Chain, 0); PendingExports.push_back(Chain); } @@ -5980,7 +5973,7 @@ void SelectionDAGISel::LowerArguments(BasicBlock *LLVMBB) { EVT VT = ValueVTs[0]; EVT RegVT = TLI.getRegisterType(*CurDAG->getContext(), VT); ISD::NodeType AssertOp = ISD::DELETED_NODE; - SDValue ArgValue = getCopyFromParts(DAG, dl, 0, &InVals[0], 1, + SDValue ArgValue = getCopyFromParts(DAG, dl, &InVals[0], 1, RegVT, VT, AssertOp); MachineFunction& MF = SDB->DAG.getMachineFunction(); @@ -6014,7 +6007,7 @@ void SelectionDAGISel::LowerArguments(BasicBlock *LLVMBB) { else if (F.paramHasAttr(Idx, Attribute::ZExt)) AssertOp = ISD::AssertZext; - ArgValues.push_back(getCopyFromParts(DAG, dl, 0, &InVals[i], + ArgValues.push_back(getCopyFromParts(DAG, dl, &InVals[i], NumParts, PartVT, VT, AssertOp)); } diff --git a/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp b/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp index eead526..2e2020d 100644 --- a/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp +++ b/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp @@ -723,9 +723,9 @@ void SelectionDAGISel::CodeGenAndEmitDAG() { // code to the MachineBasicBlock. if (TimePassesIsEnabled) { NamedRegionTimer T("Instruction Selection", GroupName); - InstructionSelect(); + DoInstructionSelection(); } else { - InstructionSelect(); + DoInstructionSelection(); } DEBUG(dbgs() << "Selected selection DAG:\n"); @@ -765,6 +765,66 @@ void SelectionDAGISel::CodeGenAndEmitDAG() { DEBUG(BB->dump()); } +void SelectionDAGISel::DoInstructionSelection() { + DEBUG(errs() << "===== Instruction selection begins:\n"); + + PreprocessISelDAG(); + + // Select target instructions for the DAG. + { + // Number all nodes with a topological order and set DAGSize. + DAGSize = CurDAG->AssignTopologicalOrder(); + + // Create a dummy node (which is not added to allnodes), that adds + // a reference to the root node, preventing it from being deleted, + // and tracking any changes of the root. + HandleSDNode Dummy(CurDAG->getRoot()); + ISelPosition = SelectionDAG::allnodes_iterator(CurDAG->getRoot().getNode()); + ++ISelPosition; + + // The AllNodes list is now topological-sorted. Visit the + // nodes by starting at the end of the list (the root of the + // graph) and preceding back toward the beginning (the entry + // node). + while (ISelPosition != CurDAG->allnodes_begin()) { + SDNode *Node = --ISelPosition; + // Skip dead nodes. DAGCombiner is expected to eliminate all dead nodes, + // but there are currently some corner cases that it misses. Also, this + // makes it theoretically possible to disable the DAGCombiner. + if (Node->use_empty()) + continue; + + SDNode *ResNode = Select(Node); + + // FIXME: This is pretty gross. 'Select' should be changed to not return + // anything at all and this code should be nuked with a tactical strike. + + // If node should not be replaced, continue with the next one. + if (ResNode == Node || Node->getOpcode() == ISD::DELETED_NODE) + continue; + // Replace node. + if (ResNode) + ReplaceUses(Node, ResNode); + + // If after the replacement this node is not used any more, + // remove this dead node. + if (Node->use_empty()) { // Don't delete EntryToken, etc. + ISelUpdater ISU(ISelPosition); + CurDAG->RemoveDeadNode(Node, &ISU); + } + } + + CurDAG->setRoot(Dummy.getValue()); + } + DEBUG(errs() << "===== Instruction selection ends:\n"); + + PostprocessISelDAG(); + + // FIXME: This shouldn't be needed, remove it. + CurDAG->RemoveDeadNodes(); +} + + void SelectionDAGISel::SelectAllBasicBlocks(Function &Fn, MachineFunction &MF, MachineModuleInfo *MMI, @@ -1316,13 +1376,29 @@ static SDNode *findFlagUse(SDNode *N) { /// This function recursively traverses up the operand chain, ignoring /// certain nodes. static bool findNonImmUse(SDNode *Use, SDNode* Def, SDNode *ImmedUse, - SDNode *Root, - SmallPtrSet<SDNode*, 16> &Visited) { - if (Use->getNodeId() < Def->getNodeId() || - !Visited.insert(Use)) + SDNode *Root, SmallPtrSet<SDNode*, 16> &Visited, + bool IgnoreChains) { + // The NodeID's are given uniques ID's where a node ID is guaranteed to be + // greater than all of its (recursive) operands. If we scan to a point where + // 'use' is smaller than the node we're scanning for, then we know we will + // never find it. + // + // The Use may be -1 (unassigned) if it is a newly allocated node. This can + // happen because we scan down to newly selected nodes in the case of flag + // uses. + if ((Use->getNodeId() < Def->getNodeId() && Use->getNodeId() != -1)) + return false; + + // Don't revisit nodes if we already scanned it and didn't fail, we know we + // won't fail if we scan it again. + if (!Visited.insert(Use)) return false; for (unsigned i = 0, e = Use->getNumOperands(); i != e; ++i) { + // Ignore chain uses, they are validated by HandleMergeInputChains. + if (Use->getOperand(i).getValueType() == MVT::Other && IgnoreChains) + continue; + SDNode *N = Use->getOperand(i).getNode(); if (N == Def) { if (Use == ImmedUse || Use == Root) @@ -1332,7 +1408,7 @@ static bool findNonImmUse(SDNode *Use, SDNode* Def, SDNode *ImmedUse, } // Traverse up the operand chain. - if (findNonImmUse(N, Def, ImmedUse, Root, Visited)) + if (findNonImmUse(N, Def, ImmedUse, Root, Visited, IgnoreChains)) return true; } return false; @@ -1347,9 +1423,10 @@ static bool findNonImmUse(SDNode *Use, SDNode* Def, SDNode *ImmedUse, /// have one non-chain use, we only need to watch out for load/op/store /// and load/op/cmp case where the root (store / cmp) may reach the load via /// its chain operand. -static inline bool isNonImmUse(SDNode *Root, SDNode *Def, SDNode *ImmedUse) { +static inline bool isNonImmUse(SDNode *Root, SDNode *Def, SDNode *ImmedUse, + bool IgnoreChains) { SmallPtrSet<SDNode*, 16> Visited; - return findNonImmUse(Root, Def, ImmedUse, Root, Visited); + return findNonImmUse(Root, Def, ImmedUse, Root, Visited, IgnoreChains); } /// IsProfitableToFold - Returns true if it's profitable to fold the specific @@ -1362,7 +1439,8 @@ bool SelectionDAGISel::IsProfitableToFold(SDValue N, SDNode *U, /// IsLegalToFold - Returns true if the specific operand node N of /// U can be folded during instruction selection that starts at Root. -bool SelectionDAGISel::IsLegalToFold(SDValue N, SDNode *U, SDNode *Root) const { +bool SelectionDAGISel::IsLegalToFold(SDValue N, SDNode *U, SDNode *Root, + bool IgnoreChains) const { if (OptLevel == CodeGenOpt::None) return false; // If Root use can somehow reach N through a path that that doesn't contain @@ -1416,7 +1494,7 @@ bool SelectionDAGISel::IsLegalToFold(SDValue N, SDNode *U, SDNode *Root) const { VT = Root->getValueType(Root->getNumValues()-1); } - return !isNonImmUse(Root, N.getNode(), U); + return !isNonImmUse(Root, N.getNode(), U, IgnoreChains); } SDNode *SelectionDAGISel::Select_INLINEASM(SDNode *N) { @@ -1428,6 +1506,7 @@ SDNode *SelectionDAGISel::Select_INLINEASM(SDNode *N) { VTs.push_back(MVT::Flag); SDValue New = CurDAG->getNode(ISD::INLINEASM, N->getDebugLoc(), VTs, &Ops[0], Ops.size()); + New->setNodeId(-1); return New.getNode(); } @@ -1443,7 +1522,1200 @@ SDNode *SelectionDAGISel::Select_EH_LABEL(SDNode *N) { MVT::Other, Tmp, Chain); } +/// GetVBR - decode a vbr encoding whose top bit is set. +ALWAYS_INLINE static uint64_t +GetVBR(uint64_t Val, const unsigned char *MatcherTable, unsigned &Idx) { + assert(Val >= 128 && "Not a VBR"); + Val &= 127; // Remove first vbr bit. + + unsigned Shift = 7; + uint64_t NextBits; + do { + NextBits = MatcherTable[Idx++]; + Val |= (NextBits&127) << Shift; + Shift += 7; + } while (NextBits & 128); + + return Val; +} + + +/// UpdateChainsAndFlags - When a match is complete, this method updates uses of +/// interior flag and chain results to use the new flag and chain results. +void SelectionDAGISel:: +UpdateChainsAndFlags(SDNode *NodeToMatch, SDValue InputChain, + const SmallVectorImpl<SDNode*> &ChainNodesMatched, + SDValue InputFlag, + const SmallVectorImpl<SDNode*> &FlagResultNodesMatched, + bool isMorphNodeTo) { + SmallVector<SDNode*, 4> NowDeadNodes; + + ISelUpdater ISU(ISelPosition); + + // Now that all the normal results are replaced, we replace the chain and + // flag results if present. + if (!ChainNodesMatched.empty()) { + assert(InputChain.getNode() != 0 && + "Matched input chains but didn't produce a chain"); + // Loop over all of the nodes we matched that produced a chain result. + // Replace all the chain results with the final chain we ended up with. + for (unsigned i = 0, e = ChainNodesMatched.size(); i != e; ++i) { + SDNode *ChainNode = ChainNodesMatched[i]; + + // If this node was already deleted, don't look at it. + if (ChainNode->getOpcode() == ISD::DELETED_NODE) + continue; + + // Don't replace the results of the root node if we're doing a + // MorphNodeTo. + if (ChainNode == NodeToMatch && isMorphNodeTo) + continue; + + SDValue ChainVal = SDValue(ChainNode, ChainNode->getNumValues()-1); + if (ChainVal.getValueType() == MVT::Flag) + ChainVal = ChainVal.getValue(ChainVal->getNumValues()-2); + assert(ChainVal.getValueType() == MVT::Other && "Not a chain?"); + CurDAG->ReplaceAllUsesOfValueWith(ChainVal, InputChain, &ISU); + + // If the node became dead, delete it. + if (ChainNode->use_empty()) + NowDeadNodes.push_back(ChainNode); + } + } + + // If the result produces a flag, update any flag results in the matched + // pattern with the flag result. + if (InputFlag.getNode() != 0) { + // Handle any interior nodes explicitly marked. + for (unsigned i = 0, e = FlagResultNodesMatched.size(); i != e; ++i) { + SDNode *FRN = FlagResultNodesMatched[i]; + + // If this node was already deleted, don't look at it. + if (FRN->getOpcode() == ISD::DELETED_NODE) + continue; + + assert(FRN->getValueType(FRN->getNumValues()-1) == MVT::Flag && + "Doesn't have a flag result"); + CurDAG->ReplaceAllUsesOfValueWith(SDValue(FRN, FRN->getNumValues()-1), + InputFlag, &ISU); + + // If the node became dead, delete it. + if (FRN->use_empty()) + NowDeadNodes.push_back(FRN); + } + } + + if (!NowDeadNodes.empty()) + CurDAG->RemoveDeadNodes(NowDeadNodes, &ISU); + + DEBUG(errs() << "ISEL: Match complete!\n"); +} + +enum ChainResult { + CR_Simple, + CR_InducesCycle, + CR_LeadsToInteriorNode +}; + +/// WalkChainUsers - Walk down the users of the specified chained node that is +/// part of the pattern we're matching, looking at all of the users we find. +/// This determines whether something is an interior node, whether we have a +/// non-pattern node in between two pattern nodes (which prevent folding because +/// it would induce a cycle) and whether we have a TokenFactor node sandwiched +/// between pattern nodes (in which case the TF becomes part of the pattern). +/// +/// The walk we do here is guaranteed to be small because we quickly get down to +/// already selected nodes "below" us. +static ChainResult +WalkChainUsers(SDNode *ChainedNode, + SmallVectorImpl<SDNode*> &ChainedNodesInPattern, + SmallVectorImpl<SDNode*> &InteriorChainedNodes) { + ChainResult Result = CR_Simple; + + for (SDNode::use_iterator UI = ChainedNode->use_begin(), + E = ChainedNode->use_end(); UI != E; ++UI) { + // Make sure the use is of the chain, not some other value we produce. + if (UI.getUse().getValueType() != MVT::Other) continue; + + SDNode *User = *UI; + + // If we see an already-selected machine node, then we've gone beyond the + // pattern that we're selecting down into the already selected chunk of the + // DAG. + if (User->isMachineOpcode() || + User->getOpcode() == ISD::HANDLENODE) // Root of the graph. + continue; + + if (User->getOpcode() == ISD::CopyToReg || + User->getOpcode() == ISD::CopyFromReg || + User->getOpcode() == ISD::INLINEASM) { + // If their node ID got reset to -1 then they've already been selected. + // Treat them like a MachineOpcode. + if (User->getNodeId() == -1) + continue; + } + + // If we have a TokenFactor, we handle it specially. + if (User->getOpcode() != ISD::TokenFactor) { + // If the node isn't a token factor and isn't part of our pattern, then it + // must be a random chained node in between two nodes we're selecting. + // This happens when we have something like: + // x = load ptr + // call + // y = x+4 + // store y -> ptr + // Because we structurally match the load/store as a read/modify/write, + // but the call is chained between them. We cannot fold in this case + // because it would induce a cycle in the graph. + if (!std::count(ChainedNodesInPattern.begin(), + ChainedNodesInPattern.end(), User)) + return CR_InducesCycle; + + // Otherwise we found a node that is part of our pattern. For example in: + // x = load ptr + // y = x+4 + // store y -> ptr + // This would happen when we're scanning down from the load and see the + // store as a user. Record that there is a use of ChainedNode that is + // part of the pattern and keep scanning uses. + Result = CR_LeadsToInteriorNode; + InteriorChainedNodes.push_back(User); + continue; + } + + // If we found a TokenFactor, there are two cases to consider: first if the + // TokenFactor is just hanging "below" the pattern we're matching (i.e. no + // uses of the TF are in our pattern) we just want to ignore it. Second, + // the TokenFactor can be sandwiched in between two chained nodes, like so: + // [Load chain] + // ^ + // | + // [Load] + // ^ ^ + // | \ DAG's like cheese + // / \ do you? + // / | + // [TokenFactor] [Op] + // ^ ^ + // | | + // \ / + // \ / + // [Store] + // + // In this case, the TokenFactor becomes part of our match and we rewrite it + // as a new TokenFactor. + // + // To distinguish these two cases, do a recursive walk down the uses. + switch (WalkChainUsers(User, ChainedNodesInPattern, InteriorChainedNodes)) { + case CR_Simple: + // If the uses of the TokenFactor are just already-selected nodes, ignore + // it, it is "below" our pattern. + continue; + case CR_InducesCycle: + // If the uses of the TokenFactor lead to nodes that are not part of our + // pattern that are not selected, folding would turn this into a cycle, + // bail out now. + return CR_InducesCycle; + case CR_LeadsToInteriorNode: + break; // Otherwise, keep processing. + } + + // Okay, we know we're in the interesting interior case. The TokenFactor + // is now going to be considered part of the pattern so that we rewrite its + // uses (it may have uses that are not part of the pattern) with the + // ultimate chain result of the generated code. We will also add its chain + // inputs as inputs to the ultimate TokenFactor we create. + Result = CR_LeadsToInteriorNode; + ChainedNodesInPattern.push_back(User); + InteriorChainedNodes.push_back(User); + continue; + } + + return Result; +} + +/// HandleMergeInputChains - This implements the OPC_EmitMergeInputChains +/// operation for when the pattern matched at least one node with a chains. The +/// input vector contains a list of all of the chained nodes that we match. We +/// must determine if this is a valid thing to cover (i.e. matching it won't +/// induce cycles in the DAG) and if so, creating a TokenFactor node. that will +/// be used as the input node chain for the generated nodes. +static SDValue +HandleMergeInputChains(SmallVectorImpl<SDNode*> &ChainNodesMatched, + SelectionDAG *CurDAG) { + // Walk all of the chained nodes we've matched, recursively scanning down the + // users of the chain result. This adds any TokenFactor nodes that are caught + // in between chained nodes to the chained and interior nodes list. + SmallVector<SDNode*, 3> InteriorChainedNodes; + for (unsigned i = 0, e = ChainNodesMatched.size(); i != e; ++i) { + if (WalkChainUsers(ChainNodesMatched[i], ChainNodesMatched, + InteriorChainedNodes) == CR_InducesCycle) + return SDValue(); // Would induce a cycle. + } + + // Okay, we have walked all the matched nodes and collected TokenFactor nodes + // that we are interested in. Form our input TokenFactor node. + SmallVector<SDValue, 3> InputChains; + for (unsigned i = 0, e = ChainNodesMatched.size(); i != e; ++i) { + // Add the input chain of this node to the InputChains list (which will be + // the operands of the generated TokenFactor) if it's not an interior node. + SDNode *N = ChainNodesMatched[i]; + if (N->getOpcode() != ISD::TokenFactor) { + if (std::count(InteriorChainedNodes.begin(),InteriorChainedNodes.end(),N)) + continue; + + // Otherwise, add the input chain. + SDValue InChain = ChainNodesMatched[i]->getOperand(0); + assert(InChain.getValueType() == MVT::Other && "Not a chain"); + InputChains.push_back(InChain); + continue; + } + + // If we have a token factor, we want to add all inputs of the token factor + // that are not part of the pattern we're matching. + for (unsigned op = 0, e = N->getNumOperands(); op != e; ++op) { + if (!std::count(ChainNodesMatched.begin(), ChainNodesMatched.end(), + N->getOperand(op).getNode())) + InputChains.push_back(N->getOperand(op)); + } + } + + SDValue Res; + if (InputChains.size() == 1) + return InputChains[0]; + return CurDAG->getNode(ISD::TokenFactor, ChainNodesMatched[0]->getDebugLoc(), + MVT::Other, &InputChains[0], InputChains.size()); +} + +/// MorphNode - Handle morphing a node in place for the selector. +SDNode *SelectionDAGISel:: +MorphNode(SDNode *Node, unsigned TargetOpc, SDVTList VTList, + const SDValue *Ops, unsigned NumOps, unsigned EmitNodeInfo) { + // It is possible we're using MorphNodeTo to replace a node with no + // normal results with one that has a normal result (or we could be + // adding a chain) and the input could have flags and chains as well. + // In this case we need to shifting the operands down. + // FIXME: This is a horrible hack and broken in obscure cases, no worse + // than the old isel though. We should sink this into MorphNodeTo. + int OldFlagResultNo = -1, OldChainResultNo = -1; + + unsigned NTMNumResults = Node->getNumValues(); + if (Node->getValueType(NTMNumResults-1) == MVT::Flag) { + OldFlagResultNo = NTMNumResults-1; + if (NTMNumResults != 1 && + Node->getValueType(NTMNumResults-2) == MVT::Other) + OldChainResultNo = NTMNumResults-2; + } else if (Node->getValueType(NTMNumResults-1) == MVT::Other) + OldChainResultNo = NTMNumResults-1; + + // Call the underlying SelectionDAG routine to do the transmogrification. Note + // that this deletes operands of the old node that become dead. + SDNode *Res = CurDAG->MorphNodeTo(Node, ~TargetOpc, VTList, Ops, NumOps); + + // MorphNodeTo can operate in two ways: if an existing node with the + // specified operands exists, it can just return it. Otherwise, it + // updates the node in place to have the requested operands. + if (Res == Node) { + // If we updated the node in place, reset the node ID. To the isel, + // this should be just like a newly allocated machine node. + Res->setNodeId(-1); + } + + unsigned ResNumResults = Res->getNumValues(); + // Move the flag if needed. + if ((EmitNodeInfo & OPFL_FlagOutput) && OldFlagResultNo != -1 && + (unsigned)OldFlagResultNo != ResNumResults-1) + CurDAG->ReplaceAllUsesOfValueWith(SDValue(Node, OldFlagResultNo), + SDValue(Res, ResNumResults-1)); + + if ((EmitNodeInfo & OPFL_FlagOutput) != 0) + --ResNumResults; + + // Move the chain reference if needed. + if ((EmitNodeInfo & OPFL_Chain) && OldChainResultNo != -1 && + (unsigned)OldChainResultNo != ResNumResults-1) + CurDAG->ReplaceAllUsesOfValueWith(SDValue(Node, OldChainResultNo), + SDValue(Res, ResNumResults-1)); + + // Otherwise, no replacement happened because the node already exists. Replace + // Uses of the old node with the new one. + if (Res != Node) + CurDAG->ReplaceAllUsesWith(Node, Res); + + return Res; +} + +/// CheckPatternPredicate - Implements OP_CheckPatternPredicate. +ALWAYS_INLINE static bool +CheckSame(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SDValue N, const SmallVectorImpl<SDValue> &RecordedNodes) { + // Accept if it is exactly the same as a previously recorded node. + unsigned RecNo = MatcherTable[MatcherIndex++]; + assert(RecNo < RecordedNodes.size() && "Invalid CheckSame"); + return N == RecordedNodes[RecNo]; +} + +/// CheckPatternPredicate - Implements OP_CheckPatternPredicate. +ALWAYS_INLINE static bool +CheckPatternPredicate(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SelectionDAGISel &SDISel) { + return SDISel.CheckPatternPredicate(MatcherTable[MatcherIndex++]); +} + +/// CheckNodePredicate - Implements OP_CheckNodePredicate. +ALWAYS_INLINE static bool +CheckNodePredicate(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SelectionDAGISel &SDISel, SDNode *N) { + return SDISel.CheckNodePredicate(N, MatcherTable[MatcherIndex++]); +} + +ALWAYS_INLINE static bool +CheckOpcode(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SDNode *N) { + return N->getOpcode() == MatcherTable[MatcherIndex++]; +} + +ALWAYS_INLINE static bool +CheckType(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SDValue N, const TargetLowering &TLI) { + MVT::SimpleValueType VT = (MVT::SimpleValueType)MatcherTable[MatcherIndex++]; + if (N.getValueType() == VT) return true; + + // Handle the case when VT is iPTR. + return VT == MVT::iPTR && N.getValueType() == TLI.getPointerTy(); +} + +ALWAYS_INLINE static bool +CheckChildType(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SDValue N, const TargetLowering &TLI, + unsigned ChildNo) { + if (ChildNo >= N.getNumOperands()) + return false; // Match fails if out of range child #. + return ::CheckType(MatcherTable, MatcherIndex, N.getOperand(ChildNo), TLI); +} + + +ALWAYS_INLINE static bool +CheckCondCode(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SDValue N) { + return cast<CondCodeSDNode>(N)->get() == + (ISD::CondCode)MatcherTable[MatcherIndex++]; +} + +ALWAYS_INLINE static bool +CheckValueType(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SDValue N, const TargetLowering &TLI) { + MVT::SimpleValueType VT = (MVT::SimpleValueType)MatcherTable[MatcherIndex++]; + if (cast<VTSDNode>(N)->getVT() == VT) + return true; + + // Handle the case when VT is iPTR. + return VT == MVT::iPTR && cast<VTSDNode>(N)->getVT() == TLI.getPointerTy(); +} + +ALWAYS_INLINE static bool +CheckInteger(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SDValue N) { + int64_t Val = MatcherTable[MatcherIndex++]; + if (Val & 128) + Val = GetVBR(Val, MatcherTable, MatcherIndex); + + ConstantSDNode *C = dyn_cast<ConstantSDNode>(N); + return C != 0 && C->getSExtValue() == Val; +} + +ALWAYS_INLINE static bool +CheckAndImm(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SDValue N, SelectionDAGISel &SDISel) { + int64_t Val = MatcherTable[MatcherIndex++]; + if (Val & 128) + Val = GetVBR(Val, MatcherTable, MatcherIndex); + + if (N->getOpcode() != ISD::AND) return false; + + ConstantSDNode *C = dyn_cast<ConstantSDNode>(N->getOperand(1)); + return C != 0 && SDISel.CheckAndMask(N.getOperand(0), C, Val); +} + +ALWAYS_INLINE static bool +CheckOrImm(const unsigned char *MatcherTable, unsigned &MatcherIndex, + SDValue N, SelectionDAGISel &SDISel) { + int64_t Val = MatcherTable[MatcherIndex++]; + if (Val & 128) + Val = GetVBR(Val, MatcherTable, MatcherIndex); + + if (N->getOpcode() != ISD::OR) return false; + + ConstantSDNode *C = dyn_cast<ConstantSDNode>(N->getOperand(1)); + return C != 0 && SDISel.CheckOrMask(N.getOperand(0), C, Val); +} + +/// IsPredicateKnownToFail - If we know how and can do so without pushing a +/// scope, evaluate the current node. If the current predicate is known to +/// fail, set Result=true and return anything. If the current predicate is +/// known to pass, set Result=false and return the MatcherIndex to continue +/// with. If the current predicate is unknown, set Result=false and return the +/// MatcherIndex to continue with. +static unsigned IsPredicateKnownToFail(const unsigned char *Table, + unsigned Index, SDValue N, + bool &Result, SelectionDAGISel &SDISel, + SmallVectorImpl<SDValue> &RecordedNodes){ + switch (Table[Index++]) { + default: + Result = false; + return Index-1; // Could not evaluate this predicate. + case SelectionDAGISel::OPC_CheckSame: + Result = !::CheckSame(Table, Index, N, RecordedNodes); + return Index; + case SelectionDAGISel::OPC_CheckPatternPredicate: + Result = !::CheckPatternPredicate(Table, Index, SDISel); + return Index; + case SelectionDAGISel::OPC_CheckPredicate: + Result = !::CheckNodePredicate(Table, Index, SDISel, N.getNode()); + return Index; + case SelectionDAGISel::OPC_CheckOpcode: + Result = !::CheckOpcode(Table, Index, N.getNode()); + return Index; + case SelectionDAGISel::OPC_CheckType: + Result = !::CheckType(Table, Index, N, SDISel.TLI); + return Index; + case SelectionDAGISel::OPC_CheckChild0Type: + case SelectionDAGISel::OPC_CheckChild1Type: + case SelectionDAGISel::OPC_CheckChild2Type: + case SelectionDAGISel::OPC_CheckChild3Type: + case SelectionDAGISel::OPC_CheckChild4Type: + case SelectionDAGISel::OPC_CheckChild5Type: + case SelectionDAGISel::OPC_CheckChild6Type: + case SelectionDAGISel::OPC_CheckChild7Type: + Result = !::CheckChildType(Table, Index, N, SDISel.TLI, + Table[Index-1] - SelectionDAGISel::OPC_CheckChild0Type); + return Index; + case SelectionDAGISel::OPC_CheckCondCode: + Result = !::CheckCondCode(Table, Index, N); + return Index; + case SelectionDAGISel::OPC_CheckValueType: + Result = !::CheckValueType(Table, Index, N, SDISel.TLI); + return Index; + case SelectionDAGISel::OPC_CheckInteger: + Result = !::CheckInteger(Table, Index, N); + return Index; + case SelectionDAGISel::OPC_CheckAndImm: + Result = !::CheckAndImm(Table, Index, N, SDISel); + return Index; + case SelectionDAGISel::OPC_CheckOrImm: + Result = !::CheckOrImm(Table, Index, N, SDISel); + return Index; + } +} + + +struct MatchScope { + /// FailIndex - If this match fails, this is the index to continue with. + unsigned FailIndex; + + /// NodeStack - The node stack when the scope was formed. + SmallVector<SDValue, 4> NodeStack; + + /// NumRecordedNodes - The number of recorded nodes when the scope was formed. + unsigned NumRecordedNodes; + + /// NumMatchedMemRefs - The number of matched memref entries. + unsigned NumMatchedMemRefs; + + /// InputChain/InputFlag - The current chain/flag + SDValue InputChain, InputFlag; + + /// HasChainNodesMatched - True if the ChainNodesMatched list is non-empty. + bool HasChainNodesMatched, HasFlagResultNodesMatched; +}; + +SDNode *SelectionDAGISel:: +SelectCodeCommon(SDNode *NodeToMatch, const unsigned char *MatcherTable, + unsigned TableSize) { + // FIXME: Should these even be selected? Handle these cases in the caller? + switch (NodeToMatch->getOpcode()) { + default: + break; + case ISD::EntryToken: // These nodes remain the same. + case ISD::BasicBlock: + case ISD::Register: + case ISD::HANDLENODE: + case ISD::TargetConstant: + case ISD::TargetConstantFP: + case ISD::TargetConstantPool: + case ISD::TargetFrameIndex: + case ISD::TargetExternalSymbol: + case ISD::TargetBlockAddress: + case ISD::TargetJumpTable: + case ISD::TargetGlobalTLSAddress: + case ISD::TargetGlobalAddress: + case ISD::TokenFactor: + case ISD::CopyFromReg: + case ISD::CopyToReg: + NodeToMatch->setNodeId(-1); // Mark selected. + return 0; + case ISD::AssertSext: + case ISD::AssertZext: + CurDAG->ReplaceAllUsesOfValueWith(SDValue(NodeToMatch, 0), + NodeToMatch->getOperand(0)); + return 0; + case ISD::INLINEASM: return Select_INLINEASM(NodeToMatch); + case ISD::EH_LABEL: return Select_EH_LABEL(NodeToMatch); + case ISD::UNDEF: return Select_UNDEF(NodeToMatch); + } + + assert(!NodeToMatch->isMachineOpcode() && "Node already selected!"); + + // Set up the node stack with NodeToMatch as the only node on the stack. + SmallVector<SDValue, 8> NodeStack; + SDValue N = SDValue(NodeToMatch, 0); + NodeStack.push_back(N); + + // MatchScopes - Scopes used when matching, if a match failure happens, this + // indicates where to continue checking. + SmallVector<MatchScope, 8> MatchScopes; + + // RecordedNodes - This is the set of nodes that have been recorded by the + // state machine. + SmallVector<SDValue, 8> RecordedNodes; + + // MatchedMemRefs - This is the set of MemRef's we've seen in the input + // pattern. + SmallVector<MachineMemOperand*, 2> MatchedMemRefs; + + // These are the current input chain and flag for use when generating nodes. + // Various Emit operations change these. For example, emitting a copytoreg + // uses and updates these. + SDValue InputChain, InputFlag; + + // ChainNodesMatched - If a pattern matches nodes that have input/output + // chains, the OPC_EmitMergeInputChains operation is emitted which indicates + // which ones they are. The result is captured into this list so that we can + // update the chain results when the pattern is complete. + SmallVector<SDNode*, 3> ChainNodesMatched; + SmallVector<SDNode*, 3> FlagResultNodesMatched; + + DEBUG(errs() << "ISEL: Starting pattern match on root node: "; + NodeToMatch->dump(CurDAG); + errs() << '\n'); + + // Determine where to start the interpreter. Normally we start at opcode #0, + // but if the state machine starts with an OPC_SwitchOpcode, then we + // accelerate the first lookup (which is guaranteed to be hot) with the + // OpcodeOffset table. + unsigned MatcherIndex = 0; + + if (!OpcodeOffset.empty()) { + // Already computed the OpcodeOffset table, just index into it. + if (N.getOpcode() < OpcodeOffset.size()) + MatcherIndex = OpcodeOffset[N.getOpcode()]; + DEBUG(errs() << " Initial Opcode index to " << MatcherIndex << "\n"); + + } else if (MatcherTable[0] == OPC_SwitchOpcode) { + // Otherwise, the table isn't computed, but the state machine does start + // with an OPC_SwitchOpcode instruction. Populate the table now, since this + // is the first time we're selecting an instruction. + unsigned Idx = 1; + while (1) { + // Get the size of this case. + unsigned CaseSize = MatcherTable[Idx++]; + if (CaseSize & 128) + CaseSize = GetVBR(CaseSize, MatcherTable, Idx); + if (CaseSize == 0) break; + + // Get the opcode, add the index to the table. + unsigned Opc = MatcherTable[Idx++]; + if (Opc >= OpcodeOffset.size()) + OpcodeOffset.resize((Opc+1)*2); + OpcodeOffset[Opc] = Idx; + Idx += CaseSize; + } + + // Okay, do the lookup for the first opcode. + if (N.getOpcode() < OpcodeOffset.size()) + MatcherIndex = OpcodeOffset[N.getOpcode()]; + } + + while (1) { + assert(MatcherIndex < TableSize && "Invalid index"); + BuiltinOpcodes Opcode = (BuiltinOpcodes)MatcherTable[MatcherIndex++]; + switch (Opcode) { + case OPC_Scope: { + // Okay, the semantics of this operation are that we should push a scope + // then evaluate the first child. However, pushing a scope only to have + // the first check fail (which then pops it) is inefficient. If we can + // determine immediately that the first check (or first several) will + // immediately fail, don't even bother pushing a scope for them. + unsigned FailIndex; + + while (1) { + unsigned NumToSkip = MatcherTable[MatcherIndex++]; + if (NumToSkip & 128) + NumToSkip = GetVBR(NumToSkip, MatcherTable, MatcherIndex); + // Found the end of the scope with no match. + if (NumToSkip == 0) { + FailIndex = 0; + break; + } + + FailIndex = MatcherIndex+NumToSkip; + + // If we can't evaluate this predicate without pushing a scope (e.g. if + // it is a 'MoveParent') or if the predicate succeeds on this node, we + // push the scope and evaluate the full predicate chain. + bool Result; + MatcherIndex = IsPredicateKnownToFail(MatcherTable, MatcherIndex, N, + Result, *this, RecordedNodes); + if (!Result) + break; + + DEBUG(errs() << " Skipped scope entry at index " << MatcherIndex + << " continuing at " << FailIndex << "\n"); + + + // Otherwise, we know that this case of the Scope is guaranteed to fail, + // move to the next case. + MatcherIndex = FailIndex; + } + + // If the whole scope failed to match, bail. + if (FailIndex == 0) break; + + // Push a MatchScope which indicates where to go if the first child fails + // to match. + MatchScope NewEntry; + NewEntry.FailIndex = FailIndex; + NewEntry.NodeStack.append(NodeStack.begin(), NodeStack.end()); + NewEntry.NumRecordedNodes = RecordedNodes.size(); + NewEntry.NumMatchedMemRefs = MatchedMemRefs.size(); + NewEntry.InputChain = InputChain; + NewEntry.InputFlag = InputFlag; + NewEntry.HasChainNodesMatched = !ChainNodesMatched.empty(); + NewEntry.HasFlagResultNodesMatched = !FlagResultNodesMatched.empty(); + MatchScopes.push_back(NewEntry); + continue; + } + case OPC_RecordNode: + // Remember this node, it may end up being an operand in the pattern. + RecordedNodes.push_back(N); + continue; + + case OPC_RecordChild0: case OPC_RecordChild1: + case OPC_RecordChild2: case OPC_RecordChild3: + case OPC_RecordChild4: case OPC_RecordChild5: + case OPC_RecordChild6: case OPC_RecordChild7: { + unsigned ChildNo = Opcode-OPC_RecordChild0; + if (ChildNo >= N.getNumOperands()) + break; // Match fails if out of range child #. + + RecordedNodes.push_back(N->getOperand(ChildNo)); + continue; + } + case OPC_RecordMemRef: + MatchedMemRefs.push_back(cast<MemSDNode>(N)->getMemOperand()); + continue; + + case OPC_CaptureFlagInput: + // If the current node has an input flag, capture it in InputFlag. + if (N->getNumOperands() != 0 && + N->getOperand(N->getNumOperands()-1).getValueType() == MVT::Flag) + InputFlag = N->getOperand(N->getNumOperands()-1); + continue; + + case OPC_MoveChild: { + unsigned ChildNo = MatcherTable[MatcherIndex++]; + if (ChildNo >= N.getNumOperands()) + break; // Match fails if out of range child #. + N = N.getOperand(ChildNo); + NodeStack.push_back(N); + continue; + } + + case OPC_MoveParent: + // Pop the current node off the NodeStack. + NodeStack.pop_back(); + assert(!NodeStack.empty() && "Node stack imbalance!"); + N = NodeStack.back(); + continue; + + case OPC_CheckSame: + if (!::CheckSame(MatcherTable, MatcherIndex, N, RecordedNodes)) break; + continue; + case OPC_CheckPatternPredicate: + if (!::CheckPatternPredicate(MatcherTable, MatcherIndex, *this)) break; + continue; + case OPC_CheckPredicate: + if (!::CheckNodePredicate(MatcherTable, MatcherIndex, *this, + N.getNode())) + break; + continue; + case OPC_CheckComplexPat: + if (!CheckComplexPattern(NodeToMatch, N, + MatcherTable[MatcherIndex++], RecordedNodes)) + break; + continue; + case OPC_CheckOpcode: + if (!::CheckOpcode(MatcherTable, MatcherIndex, N.getNode())) break; + continue; + + case OPC_CheckType: + if (!::CheckType(MatcherTable, MatcherIndex, N, TLI)) break; + continue; + + case OPC_SwitchOpcode: { + unsigned CurNodeOpcode = N.getOpcode(); + unsigned SwitchStart = MatcherIndex-1; (void)SwitchStart; + unsigned CaseSize; + while (1) { + // Get the size of this case. + CaseSize = MatcherTable[MatcherIndex++]; + if (CaseSize & 128) + CaseSize = GetVBR(CaseSize, MatcherTable, MatcherIndex); + if (CaseSize == 0) break; + + // If the opcode matches, then we will execute this case. + if (CurNodeOpcode == MatcherTable[MatcherIndex++]) + break; + + // Otherwise, skip over this case. + MatcherIndex += CaseSize; + } + + // If no cases matched, bail out. + if (CaseSize == 0) break; + + // Otherwise, execute the case we found. + DEBUG(errs() << " OpcodeSwitch from " << SwitchStart + << " to " << MatcherIndex << "\n"); + continue; + } + + case OPC_SwitchType: { + MVT::SimpleValueType CurNodeVT = N.getValueType().getSimpleVT().SimpleTy; + unsigned SwitchStart = MatcherIndex-1; (void)SwitchStart; + unsigned CaseSize; + while (1) { + // Get the size of this case. + CaseSize = MatcherTable[MatcherIndex++]; + if (CaseSize & 128) + CaseSize = GetVBR(CaseSize, MatcherTable, MatcherIndex); + if (CaseSize == 0) break; + + MVT::SimpleValueType CaseVT = + (MVT::SimpleValueType)MatcherTable[MatcherIndex++]; + if (CaseVT == MVT::iPTR) + CaseVT = TLI.getPointerTy().SimpleTy; + + // If the VT matches, then we will execute this case. + if (CurNodeVT == CaseVT) + break; + + // Otherwise, skip over this case. + MatcherIndex += CaseSize; + } + + // If no cases matched, bail out. + if (CaseSize == 0) break; + + // Otherwise, execute the case we found. + DEBUG(errs() << " TypeSwitch[" << EVT(CurNodeVT).getEVTString() + << "] from " << SwitchStart << " to " << MatcherIndex<<'\n'); + continue; + } + case OPC_CheckChild0Type: case OPC_CheckChild1Type: + case OPC_CheckChild2Type: case OPC_CheckChild3Type: + case OPC_CheckChild4Type: case OPC_CheckChild5Type: + case OPC_CheckChild6Type: case OPC_CheckChild7Type: + if (!::CheckChildType(MatcherTable, MatcherIndex, N, TLI, + Opcode-OPC_CheckChild0Type)) + break; + continue; + case OPC_CheckCondCode: + if (!::CheckCondCode(MatcherTable, MatcherIndex, N)) break; + continue; + case OPC_CheckValueType: + if (!::CheckValueType(MatcherTable, MatcherIndex, N, TLI)) break; + continue; + case OPC_CheckInteger: + if (!::CheckInteger(MatcherTable, MatcherIndex, N)) break; + continue; + case OPC_CheckAndImm: + if (!::CheckAndImm(MatcherTable, MatcherIndex, N, *this)) break; + continue; + case OPC_CheckOrImm: + if (!::CheckOrImm(MatcherTable, MatcherIndex, N, *this)) break; + continue; + + case OPC_CheckFoldableChainNode: { + assert(NodeStack.size() != 1 && "No parent node"); + // Verify that all intermediate nodes between the root and this one have + // a single use. + bool HasMultipleUses = false; + for (unsigned i = 1, e = NodeStack.size()-1; i != e; ++i) + if (!NodeStack[i].hasOneUse()) { + HasMultipleUses = true; + break; + } + if (HasMultipleUses) break; + + // Check to see that the target thinks this is profitable to fold and that + // we can fold it without inducing cycles in the graph. + if (!IsProfitableToFold(N, NodeStack[NodeStack.size()-2].getNode(), + NodeToMatch) || + !IsLegalToFold(N, NodeStack[NodeStack.size()-2].getNode(), + NodeToMatch, true/*We validate our own chains*/)) + break; + + continue; + } + case OPC_EmitInteger: { + MVT::SimpleValueType VT = + (MVT::SimpleValueType)MatcherTable[MatcherIndex++]; + int64_t Val = MatcherTable[MatcherIndex++]; + if (Val & 128) + Val = GetVBR(Val, MatcherTable, MatcherIndex); + RecordedNodes.push_back(CurDAG->getTargetConstant(Val, VT)); + continue; + } + case OPC_EmitRegister: { + MVT::SimpleValueType VT = + (MVT::SimpleValueType)MatcherTable[MatcherIndex++]; + unsigned RegNo = MatcherTable[MatcherIndex++]; + RecordedNodes.push_back(CurDAG->getRegister(RegNo, VT)); + continue; + } + + case OPC_EmitConvertToTarget: { + // Convert from IMM/FPIMM to target version. + unsigned RecNo = MatcherTable[MatcherIndex++]; + assert(RecNo < RecordedNodes.size() && "Invalid CheckSame"); + SDValue Imm = RecordedNodes[RecNo]; + + if (Imm->getOpcode() == ISD::Constant) { + int64_t Val = cast<ConstantSDNode>(Imm)->getZExtValue(); + Imm = CurDAG->getTargetConstant(Val, Imm.getValueType()); + } else if (Imm->getOpcode() == ISD::ConstantFP) { + const ConstantFP *Val=cast<ConstantFPSDNode>(Imm)->getConstantFPValue(); + Imm = CurDAG->getTargetConstantFP(*Val, Imm.getValueType()); + } + + RecordedNodes.push_back(Imm); + continue; + } + + case OPC_EmitMergeInputChains: { + assert(InputChain.getNode() == 0 && + "EmitMergeInputChains should be the first chain producing node"); + // This node gets a list of nodes we matched in the input that have + // chains. We want to token factor all of the input chains to these nodes + // together. However, if any of the input chains is actually one of the + // nodes matched in this pattern, then we have an intra-match reference. + // Ignore these because the newly token factored chain should not refer to + // the old nodes. + unsigned NumChains = MatcherTable[MatcherIndex++]; + assert(NumChains != 0 && "Can't TF zero chains"); + + assert(ChainNodesMatched.empty() && + "Should only have one EmitMergeInputChains per match"); + + // Read all of the chained nodes. + for (unsigned i = 0; i != NumChains; ++i) { + unsigned RecNo = MatcherTable[MatcherIndex++]; + assert(RecNo < RecordedNodes.size() && "Invalid CheckSame"); + ChainNodesMatched.push_back(RecordedNodes[RecNo].getNode()); + + // FIXME: What if other value results of the node have uses not matched + // by this pattern? + if (ChainNodesMatched.back() != NodeToMatch && + !RecordedNodes[RecNo].hasOneUse()) { + ChainNodesMatched.clear(); + break; + } + } + + // If the inner loop broke out, the match fails. + if (ChainNodesMatched.empty()) + break; + + // Merge the input chains if they are not intra-pattern references. + InputChain = HandleMergeInputChains(ChainNodesMatched, CurDAG); + + if (InputChain.getNode() == 0) + break; // Failed to merge. + + continue; + } + + case OPC_EmitCopyToReg: { + unsigned RecNo = MatcherTable[MatcherIndex++]; + assert(RecNo < RecordedNodes.size() && "Invalid CheckSame"); + unsigned DestPhysReg = MatcherTable[MatcherIndex++]; + + if (InputChain.getNode() == 0) + InputChain = CurDAG->getEntryNode(); + + InputChain = CurDAG->getCopyToReg(InputChain, NodeToMatch->getDebugLoc(), + DestPhysReg, RecordedNodes[RecNo], + InputFlag); + + InputFlag = InputChain.getValue(1); + continue; + } + + case OPC_EmitNodeXForm: { + unsigned XFormNo = MatcherTable[MatcherIndex++]; + unsigned RecNo = MatcherTable[MatcherIndex++]; + assert(RecNo < RecordedNodes.size() && "Invalid CheckSame"); + RecordedNodes.push_back(RunSDNodeXForm(RecordedNodes[RecNo], XFormNo)); + continue; + } + + case OPC_EmitNode: + case OPC_MorphNodeTo: { + uint16_t TargetOpc = MatcherTable[MatcherIndex++]; + TargetOpc |= (unsigned short)MatcherTable[MatcherIndex++] << 8; + unsigned EmitNodeInfo = MatcherTable[MatcherIndex++]; + // Get the result VT list. + unsigned NumVTs = MatcherTable[MatcherIndex++]; + SmallVector<EVT, 4> VTs; + for (unsigned i = 0; i != NumVTs; ++i) { + MVT::SimpleValueType VT = + (MVT::SimpleValueType)MatcherTable[MatcherIndex++]; + if (VT == MVT::iPTR) VT = TLI.getPointerTy().SimpleTy; + VTs.push_back(VT); + } + + if (EmitNodeInfo & OPFL_Chain) + VTs.push_back(MVT::Other); + if (EmitNodeInfo & OPFL_FlagOutput) + VTs.push_back(MVT::Flag); + + // This is hot code, so optimize the two most common cases of 1 and 2 + // results. + SDVTList VTList; + if (VTs.size() == 1) + VTList = CurDAG->getVTList(VTs[0]); + else if (VTs.size() == 2) + VTList = CurDAG->getVTList(VTs[0], VTs[1]); + else + VTList = CurDAG->getVTList(VTs.data(), VTs.size()); + + // Get the operand list. + unsigned NumOps = MatcherTable[MatcherIndex++]; + SmallVector<SDValue, 8> Ops; + for (unsigned i = 0; i != NumOps; ++i) { + unsigned RecNo = MatcherTable[MatcherIndex++]; + if (RecNo & 128) + RecNo = GetVBR(RecNo, MatcherTable, MatcherIndex); + + assert(RecNo < RecordedNodes.size() && "Invalid EmitNode"); + Ops.push_back(RecordedNodes[RecNo]); + } + + // If there are variadic operands to add, handle them now. + if (EmitNodeInfo & OPFL_VariadicInfo) { + // Determine the start index to copy from. + unsigned FirstOpToCopy = getNumFixedFromVariadicInfo(EmitNodeInfo); + FirstOpToCopy += (EmitNodeInfo & OPFL_Chain) ? 1 : 0; + assert(NodeToMatch->getNumOperands() >= FirstOpToCopy && + "Invalid variadic node"); + // Copy all of the variadic operands, not including a potential flag + // input. + for (unsigned i = FirstOpToCopy, e = NodeToMatch->getNumOperands(); + i != e; ++i) { + SDValue V = NodeToMatch->getOperand(i); + if (V.getValueType() == MVT::Flag) break; + Ops.push_back(V); + } + } + + // If this has chain/flag inputs, add them. + if (EmitNodeInfo & OPFL_Chain) + Ops.push_back(InputChain); + if ((EmitNodeInfo & OPFL_FlagInput) && InputFlag.getNode() != 0) + Ops.push_back(InputFlag); + + // Create the node. + SDNode *Res = 0; + if (Opcode != OPC_MorphNodeTo) { + // If this is a normal EmitNode command, just create the new node and + // add the results to the RecordedNodes list. + Res = CurDAG->getMachineNode(TargetOpc, NodeToMatch->getDebugLoc(), + VTList, Ops.data(), Ops.size()); + + // Add all the non-flag/non-chain results to the RecordedNodes list. + for (unsigned i = 0, e = VTs.size(); i != e; ++i) { + if (VTs[i] == MVT::Other || VTs[i] == MVT::Flag) break; + RecordedNodes.push_back(SDValue(Res, i)); + } + + } else { + Res = MorphNode(NodeToMatch, TargetOpc, VTList, Ops.data(), Ops.size(), + EmitNodeInfo); + } + + // If the node had chain/flag results, update our notion of the current + // chain and flag. + if (EmitNodeInfo & OPFL_FlagOutput) { + InputFlag = SDValue(Res, VTs.size()-1); + if (EmitNodeInfo & OPFL_Chain) + InputChain = SDValue(Res, VTs.size()-2); + } else if (EmitNodeInfo & OPFL_Chain) + InputChain = SDValue(Res, VTs.size()-1); + + // If the OPFL_MemRefs flag is set on this node, slap all of the + // accumulated memrefs onto it. + // + // FIXME: This is vastly incorrect for patterns with multiple outputs + // instructions that access memory and for ComplexPatterns that match + // loads. + if (EmitNodeInfo & OPFL_MemRefs) { + MachineSDNode::mmo_iterator MemRefs = + MF->allocateMemRefsArray(MatchedMemRefs.size()); + std::copy(MatchedMemRefs.begin(), MatchedMemRefs.end(), MemRefs); + cast<MachineSDNode>(Res) + ->setMemRefs(MemRefs, MemRefs + MatchedMemRefs.size()); + } + + DEBUG(errs() << " " + << (Opcode == OPC_MorphNodeTo ? "Morphed" : "Created") + << " node: "; Res->dump(CurDAG); errs() << "\n"); + + // If this was a MorphNodeTo then we're completely done! + if (Opcode == OPC_MorphNodeTo) { + // Update chain and flag uses. + UpdateChainsAndFlags(NodeToMatch, InputChain, ChainNodesMatched, + InputFlag, FlagResultNodesMatched, true); + return Res; + } + + continue; + } + + case OPC_MarkFlagResults: { + unsigned NumNodes = MatcherTable[MatcherIndex++]; + + // Read and remember all the flag-result nodes. + for (unsigned i = 0; i != NumNodes; ++i) { + unsigned RecNo = MatcherTable[MatcherIndex++]; + if (RecNo & 128) + RecNo = GetVBR(RecNo, MatcherTable, MatcherIndex); + + assert(RecNo < RecordedNodes.size() && "Invalid CheckSame"); + FlagResultNodesMatched.push_back(RecordedNodes[RecNo].getNode()); + } + continue; + } + + case OPC_CompleteMatch: { + // The match has been completed, and any new nodes (if any) have been + // created. Patch up references to the matched dag to use the newly + // created nodes. + unsigned NumResults = MatcherTable[MatcherIndex++]; + + for (unsigned i = 0; i != NumResults; ++i) { + unsigned ResSlot = MatcherTable[MatcherIndex++]; + if (ResSlot & 128) + ResSlot = GetVBR(ResSlot, MatcherTable, MatcherIndex); + + assert(ResSlot < RecordedNodes.size() && "Invalid CheckSame"); + SDValue Res = RecordedNodes[ResSlot]; + + // FIXME2: Eliminate this horrible hack by fixing the 'Gen' program + // after (parallel) on input patterns are removed. This would also + // allow us to stop encoding #results in OPC_CompleteMatch's table + // entry. + if (NodeToMatch->getNumValues() <= i || + NodeToMatch->getValueType(i) == MVT::Other || + NodeToMatch->getValueType(i) == MVT::Flag) + break; + assert((NodeToMatch->getValueType(i) == Res.getValueType() || + NodeToMatch->getValueType(i) == MVT::iPTR || + Res.getValueType() == MVT::iPTR || + NodeToMatch->getValueType(i).getSizeInBits() == + Res.getValueType().getSizeInBits()) && + "invalid replacement"); + CurDAG->ReplaceAllUsesOfValueWith(SDValue(NodeToMatch, i), Res); + } + + // If the root node defines a flag, add it to the flag nodes to update + // list. + if (NodeToMatch->getValueType(NodeToMatch->getNumValues()-1) == MVT::Flag) + FlagResultNodesMatched.push_back(NodeToMatch); + + // Update chain and flag uses. + UpdateChainsAndFlags(NodeToMatch, InputChain, ChainNodesMatched, + InputFlag, FlagResultNodesMatched, false); + + assert(NodeToMatch->use_empty() && + "Didn't replace all uses of the node?"); + + // FIXME: We just return here, which interacts correctly with SelectRoot + // above. We should fix this to not return an SDNode* anymore. + return 0; + } + } + + // If the code reached this point, then the match failed. See if there is + // another child to try in the current 'Scope', otherwise pop it until we + // find a case to check. + while (1) { + if (MatchScopes.empty()) { + CannotYetSelect(NodeToMatch); + return 0; + } + + // Restore the interpreter state back to the point where the scope was + // formed. + MatchScope &LastScope = MatchScopes.back(); + RecordedNodes.resize(LastScope.NumRecordedNodes); + NodeStack.clear(); + NodeStack.append(LastScope.NodeStack.begin(), LastScope.NodeStack.end()); + N = NodeStack.back(); + + DEBUG(errs() << " Match failed at index " << MatcherIndex + << " continuing at " << LastScope.FailIndex << "\n"); + + if (LastScope.NumMatchedMemRefs != MatchedMemRefs.size()) + MatchedMemRefs.resize(LastScope.NumMatchedMemRefs); + MatcherIndex = LastScope.FailIndex; + + InputChain = LastScope.InputChain; + InputFlag = LastScope.InputFlag; + if (!LastScope.HasChainNodesMatched) + ChainNodesMatched.clear(); + if (!LastScope.HasFlagResultNodesMatched) + FlagResultNodesMatched.clear(); + + // Check to see what the offset is at the new MatcherIndex. If it is zero + // we have reached the end of this scope, otherwise we have another child + // in the current scope to try. + unsigned NumToSkip = MatcherTable[MatcherIndex++]; + if (NumToSkip & 128) + NumToSkip = GetVBR(NumToSkip, MatcherTable, MatcherIndex); + + // If we have another child in this scope to match, update FailIndex and + // try it. + if (NumToSkip != 0) { + LastScope.FailIndex = MatcherIndex+NumToSkip; + break; + } + + // End of this scope, pop it and try the next child in the containing + // scope. + MatchScopes.pop_back(); + } + } +} + + + void SelectionDAGISel::CannotYetSelect(SDNode *N) { + if (N->getOpcode() == ISD::INTRINSIC_W_CHAIN || + N->getOpcode() == ISD::INTRINSIC_WO_CHAIN || + N->getOpcode() == ISD::INTRINSIC_VOID) + return CannotYetSelectIntrinsic(N); + std::string msg; raw_string_ostream Msg(msg); Msg << "Cannot yet select: "; diff --git a/lib/CodeGen/SelectionDAG/TargetLowering.cpp b/lib/CodeGen/SelectionDAG/TargetLowering.cpp index e88af4f..0e54ca4 100644 --- a/lib/CodeGen/SelectionDAG/TargetLowering.cpp +++ b/lib/CodeGen/SelectionDAG/TargetLowering.cpp @@ -1441,8 +1441,10 @@ bool TargetLowering::SimplifyDemandedBits(SDValue Op, case ISD::TRUNCATE: { // Simplify the input, using demanded bit information, and compute the known // zero/one bits live out. + unsigned OperandBitWidth = + Op.getOperand(0).getValueType().getScalarType().getSizeInBits(); APInt TruncMask = NewMask; - TruncMask.zext(Op.getOperand(0).getValueSizeInBits()); + TruncMask.zext(OperandBitWidth); if (SimplifyDemandedBits(Op.getOperand(0), TruncMask, KnownZero, KnownOne, TLO, Depth+1)) return true; @@ -1453,15 +1455,14 @@ bool TargetLowering::SimplifyDemandedBits(SDValue Op, // on the known demanded bits. if (Op.getOperand(0).getNode()->hasOneUse()) { SDValue In = Op.getOperand(0); - unsigned InBitWidth = In.getValueSizeInBits(); switch (In.getOpcode()) { default: break; case ISD::SRL: // Shrink SRL by a constant if none of the high bits shifted in are // demanded. if (ConstantSDNode *ShAmt = dyn_cast<ConstantSDNode>(In.getOperand(1))){ - APInt HighBits = APInt::getHighBitsSet(InBitWidth, - InBitWidth - BitWidth); + APInt HighBits = APInt::getHighBitsSet(OperandBitWidth, + OperandBitWidth - BitWidth); HighBits = HighBits.lshr(ShAmt->getZExtValue()); HighBits.trunc(BitWidth); @@ -1607,7 +1608,7 @@ static bool ValueHasExactlyOneBitSet(SDValue Val, const SelectionDAG &DAG) { // Fall back to ComputeMaskedBits to catch other known cases. EVT OpVT = Val.getValueType(); - unsigned BitWidth = OpVT.getSizeInBits(); + unsigned BitWidth = OpVT.getScalarType().getSizeInBits(); APInt Mask = APInt::getAllOnesValue(BitWidth); APInt KnownZero, KnownOne; DAG.ComputeMaskedBits(Val, Mask, KnownZero, KnownOne); @@ -1775,7 +1776,7 @@ TargetLowering::SimplifySetCC(EVT VT, SDValue N0, SDValue N1, break; // todo, be more careful with signed comparisons } } else if (N0.getOpcode() == ISD::SIGN_EXTEND_INREG && - (Cond == ISD::SETEQ || Cond == ISD::SETNE)) { + (Cond == ISD::SETEQ || Cond == ISD::SETNE)) { EVT ExtSrcTy = cast<VTSDNode>(N0.getOperand(1))->getVT(); unsigned ExtSrcTyBits = ExtSrcTy.getSizeInBits(); EVT ExtDstTy = N0.getValueType(); @@ -1809,22 +1810,21 @@ TargetLowering::SimplifySetCC(EVT VT, SDValue N0, SDValue N1, Cond); } else if ((N1C->isNullValue() || N1C->getAPIntValue() == 1) && (Cond == ISD::SETEQ || Cond == ISD::SETNE)) { - // SETCC (SETCC), [0|1], [EQ|NE] -> SETCC - if (N0.getOpcode() == ISD::SETCC) { + if (N0.getOpcode() == ISD::SETCC && + isTypeLegal(VT) && VT.bitsLE(N0.getValueType())) { bool TrueWhenTrue = (Cond == ISD::SETEQ) ^ (N1C->getAPIntValue() != 1); if (TrueWhenTrue) - return N0; - + return DAG.getNode(ISD::TRUNCATE, dl, VT, N0); // Invert the condition. ISD::CondCode CC = cast<CondCodeSDNode>(N0.getOperand(2))->get(); CC = ISD::getSetCCInverse(CC, N0.getOperand(0).getValueType().isInteger()); return DAG.getSetCC(dl, VT, N0.getOperand(0), N0.getOperand(1), CC); } - + if ((N0.getOpcode() == ISD::XOR || - (N0.getOpcode() == ISD::AND && + (N0.getOpcode() == ISD::AND && N0.getOperand(0).getOpcode() == ISD::XOR && N0.getOperand(1) == N0.getOperand(0).getOperand(1))) && isa<ConstantSDNode>(N0.getOperand(1)) && @@ -1847,9 +1847,36 @@ TargetLowering::SimplifySetCC(EVT VT, SDValue N0, SDValue N1, N0.getOperand(0).getOperand(0), N0.getOperand(1)); } + return DAG.getSetCC(dl, VT, Val, N1, Cond == ISD::SETEQ ? ISD::SETNE : ISD::SETEQ); } + } else if (N1C->getAPIntValue() == 1 && + (VT == MVT::i1 || + getBooleanContents() == ZeroOrOneBooleanContent)) { + SDValue Op0 = N0; + if (Op0.getOpcode() == ISD::TRUNCATE) + Op0 = Op0.getOperand(0); + + if ((Op0.getOpcode() == ISD::XOR) && + Op0.getOperand(0).getOpcode() == ISD::SETCC && + Op0.getOperand(1).getOpcode() == ISD::SETCC) { + // (xor (setcc), (setcc)) == / != 1 -> (setcc) != / == (setcc) + Cond = (Cond == ISD::SETEQ) ? ISD::SETNE : ISD::SETEQ; + return DAG.getSetCC(dl, VT, Op0.getOperand(0), Op0.getOperand(1), + Cond); + } else if (Op0.getOpcode() == ISD::AND && + isa<ConstantSDNode>(Op0.getOperand(1)) && + cast<ConstantSDNode>(Op0.getOperand(1))->getAPIntValue() == 1) { + // If this is (X&1) == / != 1, normalize it to (X&1) != / == 0. + if (Op0.getValueType() != VT) + Op0 = DAG.getNode(ISD::AND, dl, VT, + DAG.getNode(ISD::TRUNCATE, dl, VT, Op0.getOperand(0)), + DAG.getConstant(1, VT)); + return DAG.getSetCC(dl, VT, Op0, + DAG.getConstant(0, Op0.getValueType()), + Cond == ISD::SETEQ ? ISD::SETNE : ISD::SETEQ); + } } } diff --git a/lib/CodeGen/SimpleRegisterCoalescing.cpp b/lib/CodeGen/SimpleRegisterCoalescing.cpp index e7b0cff..ce72b2f 100644 --- a/lib/CodeGen/SimpleRegisterCoalescing.cpp +++ b/lib/CodeGen/SimpleRegisterCoalescing.cpp @@ -662,7 +662,7 @@ bool SimpleRegisterCoalescing::ReMaterializeTrivialDef(LiveInterval &SrcInt, if (!tii_->isTriviallyReMaterializable(DefMI, AA)) return false; bool SawStore = false; - if (!DefMI->isSafeToMove(tii_, SawStore, AA)) + if (!DefMI->isSafeToMove(tii_, AA, SawStore)) return false; if (TID.getNumDefs() != 1) return false; @@ -702,7 +702,8 @@ bool SimpleRegisterCoalescing::ReMaterializeTrivialDef(LiveInterval &SrcInt, for (const unsigned* SR = tri_->getSubRegisters(DstReg); *SR; ++SR) { if (!li_->hasInterval(*SR)) continue; - DLR = li_->getInterval(*SR).getLiveRangeContaining(DefIdx); + const LiveRange *DLR = + li_->getInterval(*SR).getLiveRangeContaining(DefIdx); if (DLR && DLR->valno->getCopy() == CopyMI) DLR->valno->setCopy(0); } @@ -741,9 +742,21 @@ bool SimpleRegisterCoalescing::ReMaterializeTrivialDef(LiveInterval &SrcInt, NewMI->addOperand(MO); if (MO.isDef() && li_->hasInterval(MO.getReg())) { unsigned Reg = MO.getReg(); - DLR = li_->getInterval(Reg).getLiveRangeContaining(DefIdx); + const LiveRange *DLR = + li_->getInterval(Reg).getLiveRangeContaining(DefIdx); if (DLR && DLR->valno->getCopy() == CopyMI) DLR->valno->setCopy(0); + // Handle subregs as well + if (TargetRegisterInfo::isPhysicalRegister(Reg)) { + for (const unsigned* SR = tri_->getSubRegisters(Reg); *SR; ++SR) { + if (!li_->hasInterval(*SR)) + continue; + const LiveRange *DLR = + li_->getInterval(*SR).getLiveRangeContaining(DefIdx); + if (DLR && DLR->valno->getCopy() == CopyMI) + DLR->valno->setCopy(0); + } + } } } @@ -752,6 +765,7 @@ bool SimpleRegisterCoalescing::ReMaterializeTrivialDef(LiveInterval &SrcInt, CopyMI->eraseFromParent(); ReMatCopies.insert(CopyMI); ReMatDefs.insert(DefMI); + DEBUG(dbgs() << "Remat: " << *NewMI); ++NumReMats; return true; } @@ -1705,6 +1719,7 @@ bool SimpleRegisterCoalescing::JoinCopy(CopyRec &TheCopy, bool &Again) { (AdjustCopiesBackFrom(SrcInt, DstInt, CopyMI) || RemoveCopyByCommutingDef(SrcInt, DstInt, CopyMI))) { JoinedCopies.insert(CopyMI); + DEBUG(dbgs() << "Trivial!\n"); return true; } @@ -1864,7 +1879,7 @@ static unsigned ComputeUltimateVN(VNInfo *VNI, // If the VN has already been computed, just return it. if (ThisValNoAssignments[VN] >= 0) return ThisValNoAssignments[VN]; -// assert(ThisValNoAssignments[VN] != -2 && "Cyclic case?"); + assert(ThisValNoAssignments[VN] != -2 && "Cyclic value numbers"); // If this val is not a copy from the other val, then it must be a new value // number in the destination. diff --git a/lib/CodeGen/TargetInstrInfoImpl.cpp b/lib/CodeGen/TargetInstrInfoImpl.cpp index a0fccab..e9e998f 100644 --- a/lib/CodeGen/TargetInstrInfoImpl.cpp +++ b/lib/CodeGen/TargetInstrInfoImpl.cpp @@ -150,6 +150,11 @@ void TargetInstrInfoImpl::reMaterialize(MachineBasicBlock &MBB, MBB.insert(I, MI); } +bool TargetInstrInfoImpl::produceSameValue(const MachineInstr *MI0, + const MachineInstr *MI1) const { + return MI0->isIdenticalTo(MI1, MachineInstr::IgnoreVRegDefs); +} + MachineInstr *TargetInstrInfoImpl::duplicate(MachineInstr *Orig, MachineFunction &MF) const { assert(!Orig->getDesc().isNotDuplicable() && @@ -157,37 +162,6 @@ MachineInstr *TargetInstrInfoImpl::duplicate(MachineInstr *Orig, return MF.CloneMachineInstr(Orig); } -bool -TargetInstrInfoImpl::isIdentical(const MachineInstr *MI, - const MachineInstr *Other, - const MachineRegisterInfo *MRI) const { - if (MI->getOpcode() != Other->getOpcode() || - MI->getNumOperands() != Other->getNumOperands()) - return false; - - for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { - const MachineOperand &MO = MI->getOperand(i); - const MachineOperand &OMO = Other->getOperand(i); - if (MO.isReg() && MO.isDef()) { - assert(OMO.isReg() && OMO.isDef()); - unsigned Reg = MO.getReg(); - if (TargetRegisterInfo::isPhysicalRegister(Reg)) { - if (Reg != OMO.getReg()) - return false; - } else if (MRI->getRegClass(MO.getReg()) != - MRI->getRegClass(OMO.getReg())) - return false; - - continue; - } - - if (!MO.isIdenticalTo(OMO)) - return false; - } - - return true; -} - unsigned TargetInstrInfoImpl::GetFunctionSizeInBytes(const MachineFunction &MF) const { unsigned FnSize = 0; diff --git a/lib/CodeGen/TargetLoweringObjectFileImpl.cpp b/lib/CodeGen/TargetLoweringObjectFileImpl.cpp index 190b533..ef6e129 100644 --- a/lib/CodeGen/TargetLoweringObjectFileImpl.cpp +++ b/lib/CodeGen/TargetLoweringObjectFileImpl.cpp @@ -33,6 +33,7 @@ #include "llvm/ADT/SmallString.h" #include "llvm/ADT/StringExtras.h" using namespace llvm; +using namespace dwarf; //===----------------------------------------------------------------------===// // ELF @@ -549,8 +550,8 @@ void TargetLoweringObjectFileMachO::Initialize(MCContext &Ctx, } // Exception Handling. - LSDASection = getMachOSection("__DATA", "__gcc_except_tab", 0, - SectionKind::getDataRel()); + LSDASection = getMachOSection("__TEXT", "__gcc_except_tab", 0, + SectionKind::getReadOnlyWithRel()); EHFrameSection = getMachOSection("__TEXT", "__eh_frame", MCSectionMachO::S_COALESCED | @@ -738,11 +739,23 @@ getSymbolForDwarfGlobalReference(const GlobalValue *GV, Mangler *Mang, MachineModuleInfo *MMI, unsigned Encoding) const { // The mach-o version of this method defaults to returning a stub reference. - if (Encoding & dwarf::DW_EH_PE_indirect) { + if (Encoding & DW_EH_PE_indirect) { + MachineModuleInfoMachO &MachOMMI = + MMI->getObjFileInfo<MachineModuleInfoMachO>(); + SmallString<128> Name; Mang->getNameWithPrefix(Name, GV, true); Name += "$non_lazy_ptr"; + + // Add information about the stub reference to MachOMMI so that the stub + // gets emitted by the asmprinter. MCSymbol *Sym = getContext().GetOrCreateSymbol(Name.str()); + MCSymbol *&StubSym = MachOMMI.getGVStubEntry(Sym); + if (StubSym == 0) { + Name.clear(); + Mang->getNameWithPrefix(Name, GV, false); + StubSym = getContext().GetOrCreateSymbol(Name.str()); + } return TargetLoweringObjectFile:: getSymbolForDwarfReference(Sym, MMI, @@ -753,6 +766,21 @@ getSymbolForDwarfGlobalReference(const GlobalValue *GV, Mangler *Mang, getSymbolForDwarfGlobalReference(GV, Mang, MMI, Encoding); } +unsigned TargetLoweringObjectFileMachO::getPersonalityEncoding() const { + return DW_EH_PE_indirect | DW_EH_PE_pcrel | DW_EH_PE_sdata4; +} + +unsigned TargetLoweringObjectFileMachO::getLSDAEncoding() const { + return DW_EH_PE_pcrel; +} + +unsigned TargetLoweringObjectFileMachO::getFDEEncoding() const { + return DW_EH_PE_pcrel; +} + +unsigned TargetLoweringObjectFileMachO::getTTypeEncoding() const { + return DW_EH_PE_indirect | DW_EH_PE_pcrel | DW_EH_PE_sdata4; +} //===----------------------------------------------------------------------===// // COFF diff --git a/lib/CodeGen/TwoAddressInstructionPass.cpp b/lib/CodeGen/TwoAddressInstructionPass.cpp index 6f4ca82..0ba3843 100644 --- a/lib/CodeGen/TwoAddressInstructionPass.cpp +++ b/lib/CodeGen/TwoAddressInstructionPass.cpp @@ -160,7 +160,7 @@ bool TwoAddressInstructionPass::Sink3AddrInstruction(MachineBasicBlock *MBB, MachineBasicBlock::iterator OldPos) { // Check if it's safe to move this instruction. bool SeenStore = true; // Be conservative. - if (!MI->isSafeToMove(TII, SeenStore, AA)) + if (!MI->isSafeToMove(TII, AA, SeenStore)) return false; unsigned DefReg = 0; @@ -1028,7 +1028,7 @@ bool TwoAddressInstructionPass::runOnMachineFunction(MachineFunction &MF) { // copying it. if (DefMI && DefMI->getDesc().isAsCheapAsAMove() && - DefMI->isSafeToReMat(TII, regB, AA) && + DefMI->isSafeToReMat(TII, AA, regB) && isProfitableToReMat(regB, rc, mi, DefMI, mbbi, Dist)){ DEBUG(dbgs() << "2addr: REMATTING : " << *DefMI << "\n"); unsigned regASubIdx = mi->getOperand(DstIdx).getSubReg(); diff --git a/lib/CodeGen/VirtRegMap.cpp b/lib/CodeGen/VirtRegMap.cpp index 5956b61..ed02696 100644 --- a/lib/CodeGen/VirtRegMap.cpp +++ b/lib/CodeGen/VirtRegMap.cpp @@ -261,19 +261,21 @@ bool VirtRegMap::FindUnusedRegisters(LiveIntervals* LIs) { void VirtRegMap::print(raw_ostream &OS, const Module* M) const { const TargetRegisterInfo* TRI = MF->getTarget().getRegisterInfo(); + const MachineRegisterInfo &MRI = MF->getRegInfo(); OS << "********** REGISTER MAP **********\n"; for (unsigned i = TargetRegisterInfo::FirstVirtualRegister, e = MF->getRegInfo().getLastVirtReg(); i <= e; ++i) { if (Virt2PhysMap[i] != (unsigned)VirtRegMap::NO_PHYS_REG) OS << "[reg" << i << " -> " << TRI->getName(Virt2PhysMap[i]) - << "]\n"; + << "] " << MRI.getRegClass(i)->getName() << "\n"; } for (unsigned i = TargetRegisterInfo::FirstVirtualRegister, e = MF->getRegInfo().getLastVirtReg(); i <= e; ++i) if (Virt2StackSlotMap[i] != VirtRegMap::NO_STACK_SLOT) - OS << "[reg" << i << " -> fi#" << Virt2StackSlotMap[i] << "]\n"; + OS << "[reg" << i << " -> fi#" << Virt2StackSlotMap[i] + << "] " << MRI.getRegClass(i)->getName() << "\n"; OS << '\n'; } diff --git a/lib/CodeGen/VirtRegRewriter.cpp b/lib/CodeGen/VirtRegRewriter.cpp index 84e0398..7aa0a91 100644 --- a/lib/CodeGen/VirtRegRewriter.cpp +++ b/lib/CodeGen/VirtRegRewriter.cpp @@ -46,7 +46,7 @@ namespace { static cl::opt<RewriterName> RewriterOpt("rewriter", - cl::desc("Rewriter to use: (default: local)"), + cl::desc("Rewriter to use (default=local)"), cl::Prefix, cl::values(clEnumVal(local, "local rewriter"), clEnumVal(trivial, "trivial rewriter"), diff --git a/lib/CompilerDriver/CompilationGraph.cpp b/lib/CompilerDriver/CompilationGraph.cpp index 524607b..7d1c7fe 100644 --- a/lib/CompilerDriver/CompilationGraph.cpp +++ b/lib/CompilerDriver/CompilationGraph.cpp @@ -34,7 +34,8 @@ namespace llvmc { const std::string& LanguageMap::GetLanguage(const sys::Path& File) const { StringRef suf = File.getSuffix(); - LanguageMap::const_iterator Lang = this->find(suf); + LanguageMap::const_iterator Lang = + this->find(suf.empty() ? "*empty*" : suf); if (Lang == this->end()) throw std::runtime_error("File '" + File.str() + "' has unknown suffix '" + suf.str() + '\''); @@ -313,7 +314,7 @@ int CompilationGraph::Build (const sys::Path& TempDir, JoinTool* JT = &dynamic_cast<JoinTool&>(*CurNode->ToolPtr.getPtr()); // Are there any files in the join list? - if (JT->JoinListEmpty()) + if (JT->JoinListEmpty() && !(JT->WorksOnEmpty() && InputFilenames.empty())) continue; Action CurAction = JT->GenerateAction(CurNode->HasChildren(), diff --git a/lib/CompilerDriver/Main.cpp b/lib/CompilerDriver/Main.cpp index 3a3487a..b5e507d 100644 --- a/lib/CompilerDriver/Main.cpp +++ b/lib/CompilerDriver/Main.cpp @@ -100,7 +100,8 @@ int Main(int argc, char** argv) { ProgramName = argv[0]; cl::ParseCommandLineOptions - (argc, argv, "LLVM Compiler Driver (Work In Progress)", true); + (argc, argv, "LLVM Compiler Driver (Work In Progress)", + /* ReadResponseFiles = */ false); PluginLoader Plugins; Plugins.RunInitialization(langMap, graph); @@ -126,10 +127,6 @@ int Main(int argc, char** argv) { return 0; } - if (InputFilenames.empty()) { - throw std::runtime_error("no input files"); - } - if (Time) { GlobalTimeLog = new std::stringstream; GlobalTimeLog->precision(2); diff --git a/lib/CompilerDriver/Tool.cpp b/lib/CompilerDriver/Tool.cpp index 9f4ab49..5e558ca 100644 --- a/lib/CompilerDriver/Tool.cpp +++ b/lib/CompilerDriver/Tool.cpp @@ -17,6 +17,8 @@ #include "llvm/ADT/StringExtras.h" #include "llvm/System/Path.h" +#include <algorithm> + using namespace llvm; using namespace llvmc; @@ -71,3 +73,22 @@ sys::Path Tool::OutFilename(const sys::Path& In, } return Out; } + +namespace { + template <class A, class B> + bool CompareFirst (std::pair<A,B> p1, std::pair<A,B> p2) { + return std::less<A>()(p1.first, p2.first); + } +} + +StrVector Tool::SortArgs(ArgsVector& Args) const { + StrVector Out; + + // HACK: this won't be needed when we'll migrate away from CommandLine. + std::stable_sort(Args.begin(), Args.end(), &CompareFirst<unsigned, std::string>); + for (ArgsVector::iterator B = Args.begin(), E = Args.end(); B != E; ++B) { + Out.push_back(B->second); + } + + return Out; +} diff --git a/lib/ExecutionEngine/ExecutionEngine.cpp b/lib/ExecutionEngine/ExecutionEngine.cpp index 6db3ef9..b2e2a04 100644 --- a/lib/ExecutionEngine/ExecutionEngine.cpp +++ b/lib/ExecutionEngine/ExecutionEngine.cpp @@ -344,7 +344,7 @@ int ExecutionEngine::runFunctionAsMain(Function *Fn, } // FALLS THROUGH case 0: - if (!isa<IntegerType>(FTy->getReturnType()) && + if (!FTy->getReturnType()->isIntegerTy() && !FTy->getReturnType()->isVoidTy()) { llvm_report_error("Invalid return type of main() supplied"); } @@ -614,7 +614,7 @@ GenericValue ExecutionEngine::getConstantValue(const Constant *C) { GV.IntVal.doubleToBits(GV.DoubleVal); break; case Type::PointerTyID: - assert(isa<PointerType>(DestTy) && "Invalid bitcast"); + assert(DestTy->isPointerTy() && "Invalid bitcast"); break; // getConstantValue(Op0) above already converted it } return GV; diff --git a/lib/ExecutionEngine/ExecutionEngineBindings.cpp b/lib/ExecutionEngine/ExecutionEngineBindings.cpp index 141cb27..c7495d4 100644 --- a/lib/ExecutionEngine/ExecutionEngineBindings.cpp +++ b/lib/ExecutionEngine/ExecutionEngineBindings.cpp @@ -87,11 +87,11 @@ void LLVMDisposeGenericValue(LLVMGenericValueRef GenVal) { /*===-- Operations on execution engines -----------------------------------===*/ -LLVMBool LLVMCreateExecutionEngine(LLVMExecutionEngineRef *OutEE, - LLVMModuleProviderRef MP, - char **OutError) { +LLVMBool LLVMCreateExecutionEngineForModule(LLVMExecutionEngineRef *OutEE, + LLVMModuleRef M, + char **OutError) { std::string Error; - EngineBuilder builder(unwrap(MP)); + EngineBuilder builder(unwrap(M)); builder.setEngineKind(EngineKind::Either) .setErrorStr(&Error); if (ExecutionEngine *EE = builder.create()){ @@ -102,11 +102,11 @@ LLVMBool LLVMCreateExecutionEngine(LLVMExecutionEngineRef *OutEE, return 1; } -LLVMBool LLVMCreateInterpreter(LLVMExecutionEngineRef *OutInterp, - LLVMModuleProviderRef MP, - char **OutError) { +LLVMBool LLVMCreateInterpreterForModule(LLVMExecutionEngineRef *OutInterp, + LLVMModuleRef M, + char **OutError) { std::string Error; - EngineBuilder builder(unwrap(MP)); + EngineBuilder builder(unwrap(M)); builder.setEngineKind(EngineKind::Interpreter) .setErrorStr(&Error); if (ExecutionEngine *Interp = builder.create()) { @@ -117,12 +117,12 @@ LLVMBool LLVMCreateInterpreter(LLVMExecutionEngineRef *OutInterp, return 1; } -LLVMBool LLVMCreateJITCompiler(LLVMExecutionEngineRef *OutJIT, - LLVMModuleProviderRef MP, - unsigned OptLevel, - char **OutError) { +LLVMBool LLVMCreateJITCompilerForModule(LLVMExecutionEngineRef *OutJIT, + LLVMModuleRef M, + unsigned OptLevel, + char **OutError) { std::string Error; - EngineBuilder builder(unwrap(MP)); + EngineBuilder builder(unwrap(M)); builder.setEngineKind(EngineKind::JIT) .setErrorStr(&Error) .setOptLevel((CodeGenOpt::Level)OptLevel); @@ -134,6 +134,35 @@ LLVMBool LLVMCreateJITCompiler(LLVMExecutionEngineRef *OutJIT, return 1; } +LLVMBool LLVMCreateExecutionEngine(LLVMExecutionEngineRef *OutEE, + LLVMModuleProviderRef MP, + char **OutError) { + /* The module provider is now actually a module. */ + return LLVMCreateExecutionEngineForModule(OutEE, + reinterpret_cast<LLVMModuleRef>(MP), + OutError); +} + +LLVMBool LLVMCreateInterpreter(LLVMExecutionEngineRef *OutInterp, + LLVMModuleProviderRef MP, + char **OutError) { + /* The module provider is now actually a module. */ + return LLVMCreateInterpreterForModule(OutInterp, + reinterpret_cast<LLVMModuleRef>(MP), + OutError); +} + +LLVMBool LLVMCreateJITCompiler(LLVMExecutionEngineRef *OutJIT, + LLVMModuleProviderRef MP, + unsigned OptLevel, + char **OutError) { + /* The module provider is now actually a module. */ + return LLVMCreateJITCompilerForModule(OutJIT, + reinterpret_cast<LLVMModuleRef>(MP), + OptLevel, OutError); +} + + void LLVMDisposeExecutionEngine(LLVMExecutionEngineRef EE) { delete unwrap(EE); } @@ -173,17 +202,29 @@ void LLVMFreeMachineCodeForFunction(LLVMExecutionEngineRef EE, LLVMValueRef F) { unwrap(EE)->freeMachineCodeForFunction(unwrap<Function>(F)); } +void LLVMAddModule(LLVMExecutionEngineRef EE, LLVMModuleRef M){ + unwrap(EE)->addModule(unwrap(M)); +} + void LLVMAddModuleProvider(LLVMExecutionEngineRef EE, LLVMModuleProviderRef MP){ - unwrap(EE)->addModule(unwrap(MP)); + /* The module provider is now actually a module. */ + LLVMAddModule(EE, reinterpret_cast<LLVMModuleRef>(MP)); +} + +LLVMBool LLVMRemoveModule(LLVMExecutionEngineRef EE, LLVMModuleRef M, + LLVMModuleRef *OutMod, char **OutError) { + Module *Mod = unwrap(M); + unwrap(EE)->removeModule(Mod); + *OutMod = wrap(Mod); + return 0; } LLVMBool LLVMRemoveModuleProvider(LLVMExecutionEngineRef EE, LLVMModuleProviderRef MP, LLVMModuleRef *OutMod, char **OutError) { - Module *M = unwrap(MP); - unwrap(EE)->removeModule(M); - *OutMod = wrap(M); - return 0; + /* The module provider is now actually a module. */ + return LLVMRemoveModule(EE, reinterpret_cast<LLVMModuleRef>(MP), OutMod, + OutError); } LLVMBool LLVMFindFunction(LLVMExecutionEngineRef EE, const char *Name, diff --git a/lib/ExecutionEngine/Interpreter/Execution.cpp b/lib/ExecutionEngine/Interpreter/Execution.cpp index e234cf1..a2aad5a 100644 --- a/lib/ExecutionEngine/Interpreter/Execution.cpp +++ b/lib/ExecutionEngine/Interpreter/Execution.cpp @@ -761,7 +761,7 @@ void Interpreter::visitAllocaInst(AllocaInst &I) { GenericValue Interpreter::executeGEPOperation(Value *Ptr, gep_type_iterator I, gep_type_iterator E, ExecutionContext &SF) { - assert(isa<PointerType>(Ptr->getType()) && + assert(Ptr->getType()->isPointerTy() && "Cannot getElementOffset of a nonpointer type!"); uint64_t Total = 0; @@ -1031,7 +1031,7 @@ GenericValue Interpreter::executePtrToIntInst(Value *SrcVal, const Type *DstTy, ExecutionContext &SF) { uint32_t DBitWidth = cast<IntegerType>(DstTy)->getBitWidth(); GenericValue Dest, Src = getOperandValue(SrcVal, SF); - assert(isa<PointerType>(SrcVal->getType()) && "Invalid PtrToInt instruction"); + assert(SrcVal->getType()->isPointerTy() && "Invalid PtrToInt instruction"); Dest.IntVal = APInt(DBitWidth, (intptr_t) Src.PointerVal); return Dest; @@ -1040,7 +1040,7 @@ GenericValue Interpreter::executePtrToIntInst(Value *SrcVal, const Type *DstTy, GenericValue Interpreter::executeIntToPtrInst(Value *SrcVal, const Type *DstTy, ExecutionContext &SF) { GenericValue Dest, Src = getOperandValue(SrcVal, SF); - assert(isa<PointerType>(DstTy) && "Invalid PtrToInt instruction"); + assert(DstTy->isPointerTy() && "Invalid PtrToInt instruction"); uint32_t PtrSize = TD.getPointerSizeInBits(); if (PtrSize != Src.IntVal.getBitWidth()) @@ -1055,8 +1055,8 @@ GenericValue Interpreter::executeBitCastInst(Value *SrcVal, const Type *DstTy, const Type *SrcTy = SrcVal->getType(); GenericValue Dest, Src = getOperandValue(SrcVal, SF); - if (isa<PointerType>(DstTy)) { - assert(isa<PointerType>(SrcTy) && "Invalid BitCast"); + if (DstTy->isPointerTy()) { + assert(SrcTy->isPointerTy() && "Invalid BitCast"); Dest.PointerVal = Src.PointerVal; } else if (DstTy->isIntegerTy()) { if (SrcTy->isFloatTy()) { diff --git a/lib/ExecutionEngine/JIT/JIT.cpp b/lib/ExecutionEngine/JIT/JIT.cpp index 18a996e..dd74d73 100644 --- a/lib/ExecutionEngine/JIT/JIT.cpp +++ b/lib/ExecutionEngine/JIT/JIT.cpp @@ -415,8 +415,8 @@ GenericValue JIT::runFunction(Function *F, switch (ArgValues.size()) { case 3: if (FTy->getParamType(0)->isIntegerTy(32) && - isa<PointerType>(FTy->getParamType(1)) && - isa<PointerType>(FTy->getParamType(2))) { + FTy->getParamType(1)->isPointerTy() && + FTy->getParamType(2)->isPointerTy()) { int (*PF)(int, char **, const char **) = (int(*)(int, char **, const char **))(intptr_t)FPtr; @@ -430,7 +430,7 @@ GenericValue JIT::runFunction(Function *F, break; case 2: if (FTy->getParamType(0)->isIntegerTy(32) && - isa<PointerType>(FTy->getParamType(1))) { + FTy->getParamType(1)->isPointerTy()) { int (*PF)(int, char **) = (int(*)(int, char **))(intptr_t)FPtr; // Call the function. diff --git a/lib/Linker/LinkModules.cpp b/lib/Linker/LinkModules.cpp index 7f441b0..8487c83 100644 --- a/lib/Linker/LinkModules.cpp +++ b/lib/Linker/LinkModules.cpp @@ -159,7 +159,7 @@ static bool RecursiveResolveTypesI(const Type *DstTy, const Type *SrcTy, if (DstTy == SrcTy) return false; // If already equal, noop // If we found our opaque type, resolve it now! - if (isa<OpaqueType>(DstTy) || isa<OpaqueType>(SrcTy)) + if (DstTy->isOpaqueTy() || SrcTy->isOpaqueTy()) return ResolveTypes(DstTy, SrcTy); // Two types cannot be resolved together if they are of different primitive diff --git a/lib/MC/CMakeLists.txt b/lib/MC/CMakeLists.txt index 9ead33b..4cf71dc 100644 --- a/lib/MC/CMakeLists.txt +++ b/lib/MC/CMakeLists.txt @@ -18,4 +18,5 @@ add_llvm_library(LLVMMC MCStreamer.cpp MCSymbol.cpp MCValue.cpp + TargetAsmBackend.cpp ) diff --git a/lib/MC/MCAsmStreamer.cpp b/lib/MC/MCAsmStreamer.cpp index 6add1b4..66a0a24 100644 --- a/lib/MC/MCAsmStreamer.cpp +++ b/lib/MC/MCAsmStreamer.cpp @@ -134,6 +134,9 @@ public: unsigned ValueSize = 1, unsigned MaxBytesToEmit = 0); + virtual void EmitCodeAlignment(unsigned ByteAlignment, + unsigned MaxBytesToEmit = 0); + virtual void EmitValueToOffset(const MCExpr *Offset, unsigned char Value = 0); @@ -513,6 +516,13 @@ void MCAsmStreamer::EmitValueToAlignment(unsigned ByteAlignment, int64_t Value, EmitEOL(); } +void MCAsmStreamer::EmitCodeAlignment(unsigned ByteAlignment, + unsigned MaxBytesToEmit) { + // Emit with a text fill value. + EmitValueToAlignment(ByteAlignment, MAI.getTextAlignFillValue(), + 1, MaxBytesToEmit); +} + void MCAsmStreamer::EmitValueToOffset(const MCExpr *Offset, unsigned char Value) { // FIXME: Verify that Offset is associated with the current section. diff --git a/lib/MC/MCAssembler.cpp b/lib/MC/MCAssembler.cpp index 653fbf2..96227db 100644 --- a/lib/MC/MCAssembler.cpp +++ b/lib/MC/MCAssembler.cpp @@ -42,6 +42,8 @@ STATISTIC(EmittedFragments, "Number of emitted assembler fragments"); static void WriteFileData(raw_ostream &OS, const MCSectionData &SD, MachObjectWriter &MOW); +static uint64_t WriteNopData(uint64_t Count, MachObjectWriter &MOW); + /// isVirtualSection - Check if this is a section which does not actually exist /// in the object file. static bool isVirtualSection(const MCSection &Section) { @@ -51,7 +53,7 @@ static bool isVirtualSection(const MCSection &Section) { return (Type == MCSectionMachO::S_ZEROFILL); } -static unsigned getFixupKindLog2Size(MCFixupKind Kind) { +static unsigned getFixupKindLog2Size(unsigned Kind) { switch (Kind) { default: llvm_unreachable("invalid fixup kind!"); case X86::reloc_pcrel_1byte: @@ -64,7 +66,7 @@ static unsigned getFixupKindLog2Size(MCFixupKind Kind) { } } -static bool isFixupKindPCRel(MCFixupKind Kind) { +static bool isFixupKindPCRel(unsigned Kind) { switch (Kind) { default: return false; @@ -439,6 +441,7 @@ public: std::vector<MachRelocationEntry> &Relocs) { uint32_t Address = Fragment.getOffset() + Fixup.Offset; unsigned IsPCRel = 0; + unsigned Log2Size = getFixupKindLog2Size(Fixup.Kind); unsigned Type = RIT_Vanilla; // See <reloc.h>. @@ -454,12 +457,10 @@ public: Value2 = SD->getFragment()->getAddress() + SD->getOffset(); } - unsigned Log2Size = getFixupKindLog2Size(Fixup.Kind); - // The value which goes in the fixup is current value of the expression. Fixup.FixedValue = Value - Value2 + Target.getConstant(); if (isFixupKindPCRel(Fixup.Kind)) { - Fixup.FixedValue -= Address + (1 << Log2Size); + Fixup.FixedValue -= Address; IsPCRel = 1; } @@ -507,6 +508,7 @@ public: uint32_t Value = 0; unsigned Index = 0; unsigned IsPCRel = 0; + unsigned Log2Size = getFixupKindLog2Size(Fixup.Kind); unsigned IsExtern = 0; unsigned Type = 0; @@ -544,10 +546,8 @@ public: // The value which goes in the fixup is current value of the expression. Fixup.FixedValue = Value + Target.getConstant(); - unsigned Log2Size = getFixupKindLog2Size(Fixup.Kind); - if (isFixupKindPCRel(Fixup.Kind)) { - Fixup.FixedValue -= Address + (1<<Log2Size); + Fixup.FixedValue -= Address; IsPCRel = 1; } @@ -1060,6 +1060,64 @@ void MCAssembler::LayoutSection(MCSectionData &SD) { SD.setFileSize(Address - SD.getAddress()); } +/// WriteNopData - Write optimal nops to the output file for the \arg Count +/// bytes. This returns the number of bytes written. It may return 0 if +/// the \arg Count is more than the maximum optimal nops. +/// +/// FIXME this is X86 32-bit specific and should move to a better place. +static uint64_t WriteNopData(uint64_t Count, MachObjectWriter &MOW) { + static const uint8_t Nops[16][16] = { + // nop + {0x90}, + // xchg %ax,%ax + {0x66, 0x90}, + // nopl (%[re]ax) + {0x0f, 0x1f, 0x00}, + // nopl 0(%[re]ax) + {0x0f, 0x1f, 0x40, 0x00}, + // nopl 0(%[re]ax,%[re]ax,1) + {0x0f, 0x1f, 0x44, 0x00, 0x00}, + // nopw 0(%[re]ax,%[re]ax,1) + {0x66, 0x0f, 0x1f, 0x44, 0x00, 0x00}, + // nopl 0L(%[re]ax) + {0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00}, + // nopl 0L(%[re]ax,%[re]ax,1) + {0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00}, + // nopw 0L(%[re]ax,%[re]ax,1) + {0x66, 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00}, + // nopw %cs:0L(%[re]ax,%[re]ax,1) + {0x66, 0x2e, 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00}, + // nopl 0(%[re]ax,%[re]ax,1) + // nopw 0(%[re]ax,%[re]ax,1) + {0x0f, 0x1f, 0x44, 0x00, 0x00, + 0x66, 0x0f, 0x1f, 0x44, 0x00, 0x00}, + // nopw 0(%[re]ax,%[re]ax,1) + // nopw 0(%[re]ax,%[re]ax,1) + {0x66, 0x0f, 0x1f, 0x44, 0x00, 0x00, + 0x66, 0x0f, 0x1f, 0x44, 0x00, 0x00}, + // nopw 0(%[re]ax,%[re]ax,1) + // nopl 0L(%[re]ax) */ + {0x66, 0x0f, 0x1f, 0x44, 0x00, 0x00, + 0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00}, + // nopl 0L(%[re]ax) + // nopl 0L(%[re]ax) + {0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00, + 0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00}, + // nopl 0L(%[re]ax) + // nopl 0L(%[re]ax,%[re]ax,1) + {0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00, + 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00} + }; + + if (Count > 15) + return 0; + + for (uint64_t i = 0; i < Count; i++) + MOW.Write8 (uint8_t(Nops[Count - 1][i])); + + return Count; +} + /// WriteFileData - Write the \arg F data to the output file. static void WriteFileData(raw_ostream &OS, const MCFragment &F, MachObjectWriter &MOW) { @@ -1083,6 +1141,14 @@ static void WriteFileData(raw_ostream &OS, const MCFragment &F, "' is not a divisor of padding size '" + Twine(AF.getFileSize()) + "'"); + // See if we are aligning with nops, and if so do that first to try to fill + // the Count bytes. Then if that did not fill any bytes or there are any + // bytes left to fill use the the Value and ValueSize to fill the rest. + if (AF.getEmitNops()) { + uint64_t NopByteCount = WriteNopData(Count, MOW); + Count -= NopByteCount; + } + for (uint64_t i = 0; i != Count; ++i) { switch (AF.getValueSize()) { default: diff --git a/lib/MC/MCMachOStreamer.cpp b/lib/MC/MCMachOStreamer.cpp index 0c9627d..a7a8a5d 100644 --- a/lib/MC/MCMachOStreamer.cpp +++ b/lib/MC/MCMachOStreamer.cpp @@ -137,6 +137,8 @@ public: virtual void EmitValueToAlignment(unsigned ByteAlignment, int64_t Value = 0, unsigned ValueSize = 1, unsigned MaxBytesToEmit = 0); + virtual void EmitCodeAlignment(unsigned ByteAlignment, + unsigned MaxBytesToEmit = 0); virtual void EmitValueToOffset(const MCExpr *Offset, unsigned char Value = 0); @@ -333,15 +335,13 @@ void MCMachOStreamer::EmitBytes(StringRef Data, unsigned AddrSpace) { void MCMachOStreamer::EmitValue(const MCExpr *Value, unsigned Size, unsigned AddrSpace) { - // Assume the front-end will have evaluate things absolute expressions, so - // just create data + fixup. MCDataFragment *DF = dyn_cast_or_null<MCDataFragment>(getCurrentFragment()); if (!DF) DF = new MCDataFragment(CurSectionData); // Avoid fixups when possible. int64_t AbsValue; - if (Value->EvaluateAsAbsolute(AbsValue)) { + if (AddValueSymbols(Value)->EvaluateAsAbsolute(AbsValue)) { // FIXME: Endianness assumption. for (unsigned i = 0; i != Size; ++i) DF->getContents().push_back(uint8_t(AbsValue >> (i * 8))); @@ -359,7 +359,20 @@ void MCMachOStreamer::EmitValueToAlignment(unsigned ByteAlignment, if (MaxBytesToEmit == 0) MaxBytesToEmit = ByteAlignment; new MCAlignFragment(ByteAlignment, Value, ValueSize, MaxBytesToEmit, - CurSectionData); + false /* EmitNops */, CurSectionData); + + // Update the maximum alignment on the current section if necessary. + if (ByteAlignment > CurSectionData->getAlignment()) + CurSectionData->setAlignment(ByteAlignment); +} + +void MCMachOStreamer::EmitCodeAlignment(unsigned ByteAlignment, + unsigned MaxBytesToEmit) { + if (MaxBytesToEmit == 0) + MaxBytesToEmit = ByteAlignment; + // FIXME the 0x90 is the default x86 1 byte nop opcode. + new MCAlignFragment(ByteAlignment, 0x90, 1, MaxBytesToEmit, + true /* EmitNops */, CurSectionData); // Update the maximum alignment on the current section if necessary. if (ByteAlignment > CurSectionData->getAlignment()) diff --git a/lib/MC/MCNullStreamer.cpp b/lib/MC/MCNullStreamer.cpp index 46e9ebf..ab61799 100644 --- a/lib/MC/MCNullStreamer.cpp +++ b/lib/MC/MCNullStreamer.cpp @@ -55,6 +55,9 @@ namespace { unsigned ValueSize = 1, unsigned MaxBytesToEmit = 0) {} + virtual void EmitCodeAlignment(unsigned ByteAlignment, + unsigned MaxBytesToEmit = 0) {} + virtual void EmitValueToOffset(const MCExpr *Offset, unsigned char Value = 0) {} diff --git a/lib/MC/MCParser/AsmParser.cpp b/lib/MC/MCParser/AsmParser.cpp index 51ad5d1..6185c30 100644 --- a/lib/MC/MCParser/AsmParser.cpp +++ b/lib/MC/MCParser/AsmParser.cpp @@ -146,7 +146,7 @@ bool AsmParser::Run() { // FIXME: Target hook & command line option for initial section. Out.SwitchSection(getMachOSection("__TEXT", "__text", MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS, - 0, SectionKind())); + 0, SectionKind::getText())); // Prime the lexer. @@ -914,8 +914,10 @@ bool AsmParser::ParseDirectiveDarwinSection() { return Error(Loc, ErrorStr.c_str()); // FIXME: Arch specific. + bool isText = Segment == "__TEXT"; // FIXME: Hack. Out.SwitchSection(getMachOSection(Segment, Section, TAA, StubSize, - SectionKind())); + isText ? SectionKind::getText() + : SectionKind::getDataRel())); return false; } @@ -929,8 +931,10 @@ bool AsmParser::ParseDirectiveSectionSwitch(const char *Segment, Lex(); // FIXME: Arch specific. + bool isText = StringRef(Segment) == "__TEXT"; // FIXME: Hack. Out.SwitchSection(getMachOSection(Segment, Section, TAA, StubSize, - SectionKind())); + isText ? SectionKind::getText() + : SectionKind::getDataRel())); // Set the implicit alignment, if any. // @@ -1237,8 +1241,14 @@ bool AsmParser::ParseDirectiveAlign(bool IsPow2, unsigned ValueSize) { } } - // FIXME: Target specific behavior about how the "extra" bytes are filled. - Out.EmitValueToAlignment(Alignment, FillExpr, ValueSize, MaxBytesToFill); + // FIXME: hard code the parser to use EmitCodeAlignment for text when using + // the TextAlignFillValue. + if(Out.getCurrentSection()->getKind().isText() && + Lexer.getMAI().getTextAlignFillValue() == FillExpr) + Out.EmitCodeAlignment(Alignment, MaxBytesToFill); + else + // FIXME: Target specific behavior about how the "extra" bytes are filled. + Out.EmitValueToAlignment(Alignment, FillExpr, ValueSize, MaxBytesToFill); return false; } @@ -1362,7 +1372,7 @@ bool AsmParser::ParseDirectiveComm(bool IsLocal) { if (IsLocal) { Out.EmitZerofill(getMachOSection("__DATA", "__bss", MCSectionMachO::S_ZEROFILL, 0, - SectionKind()), + SectionKind::getBSS()), Sym, Size, 1 << Pow2Alignment); return false; } @@ -1398,7 +1408,7 @@ bool AsmParser::ParseDirectiveDarwinZerofill() { // Create the zerofill section but no symbol Out.EmitZerofill(getMachOSection(Segment, Section, MCSectionMachO::S_ZEROFILL, 0, - SectionKind())); + SectionKind::getBSS())); return false; } @@ -1456,7 +1466,7 @@ bool AsmParser::ParseDirectiveDarwinZerofill() { // FIXME: Arch specific. Out.EmitZerofill(getMachOSection(Segment, Section, MCSectionMachO::S_ZEROFILL, 0, - SectionKind()), + SectionKind::getBSS()), Sym, Size, 1 << Pow2Alignment); return false; diff --git a/lib/MC/TargetAsmBackend.cpp b/lib/MC/TargetAsmBackend.cpp new file mode 100644 index 0000000..918d272 --- /dev/null +++ b/lib/MC/TargetAsmBackend.cpp @@ -0,0 +1,19 @@ +//===-- TargetAsmBackend.cpp - Target Assembly Backend ---------------------==// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "llvm/Target/TargetAsmBackend.h" +using namespace llvm; + +TargetAsmBackend::TargetAsmBackend(const Target &T) + : TheTarget(T) +{ +} + +TargetAsmBackend::~TargetAsmBackend() { +} diff --git a/lib/Support/APFloat.cpp b/lib/Support/APFloat.cpp index 1e6d22f..619f061 100644 --- a/lib/Support/APFloat.cpp +++ b/lib/Support/APFloat.cpp @@ -17,6 +17,7 @@ #include "llvm/ADT/FoldingSet.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Support/MathExtras.h" +#include <limits.h> #include <cstring> using namespace llvm; @@ -625,17 +626,58 @@ APFloat::copySignificand(const APFloat &rhs) /* Make this number a NaN, with an arbitrary but deterministic value for the significand. If double or longer, this is a signalling NaN, which may not be ideal. If float, this is QNaN(0). */ -void -APFloat::makeNaN(unsigned type) +void APFloat::makeNaN(bool SNaN, bool Negative, const APInt *fill) { category = fcNaN; - // FIXME: Add double and long double support for QNaN(0). - if (semantics->precision == 24 && semantics->maxExponent == 127) { - type |= 0x7fc00000U; - type &= ~0x80000000U; - } else - type = ~0U; - APInt::tcSet(significandParts(), type, partCount()); + sign = Negative; + + integerPart *significand = significandParts(); + unsigned numParts = partCount(); + + // Set the significand bits to the fill. + if (!fill || fill->getNumWords() < numParts) + APInt::tcSet(significand, 0, numParts); + if (fill) { + APInt::tcAssign(significand, fill->getRawData(), + std::min(fill->getNumWords(), numParts)); + + // Zero out the excess bits of the significand. + unsigned bitsToPreserve = semantics->precision - 1; + unsigned part = bitsToPreserve / 64; + bitsToPreserve %= 64; + significand[part] &= ((1ULL << bitsToPreserve) - 1); + for (part++; part != numParts; ++part) + significand[part] = 0; + } + + unsigned QNaNBit = semantics->precision - 2; + + if (SNaN) { + // We always have to clear the QNaN bit to make it an SNaN. + APInt::tcClearBit(significand, QNaNBit); + + // If there are no bits set in the payload, we have to set + // *something* to make it a NaN instead of an infinity; + // conventionally, this is the next bit down from the QNaN bit. + if (APInt::tcIsZero(significand, numParts)) + APInt::tcSetBit(significand, QNaNBit - 1); + } else { + // We always have to set the QNaN bit to make it a QNaN. + APInt::tcSetBit(significand, QNaNBit); + } + + // For x87 extended precision, we want to make a NaN, not a + // pseudo-NaN. Maybe we should expose the ability to make + // pseudo-NaNs? + if (semantics == &APFloat::x87DoubleExtended) + APInt::tcSetBit(significand, QNaNBit + 1); +} + +APFloat APFloat::makeNaN(const fltSemantics &Sem, bool SNaN, bool Negative, + const APInt *fill) { + APFloat value(Sem, uninitialized); + value.makeNaN(SNaN, Negative, fill); + return value; } APFloat & @@ -700,9 +742,14 @@ APFloat::APFloat(const fltSemantics &ourSemantics) { sign = false; } +APFloat::APFloat(const fltSemantics &ourSemantics, uninitializedTag tag) { + assertArithmeticOK(ourSemantics); + // Allocates storage if necessary but does not initialize it. + initialize(&ourSemantics); +} APFloat::APFloat(const fltSemantics &ourSemantics, - fltCategory ourCategory, bool negative, unsigned type) + fltCategory ourCategory, bool negative) { assertArithmeticOK(ourSemantics); initialize(&ourSemantics); @@ -711,7 +758,7 @@ APFloat::APFloat(const fltSemantics &ourSemantics, if (category == fcNormal) category = fcZero; else if (ourCategory == fcNaN) - makeNaN(type); + makeNaN(); } APFloat::APFloat(const fltSemantics &ourSemantics, const StringRef& text) @@ -2345,11 +2392,24 @@ APFloat::convertFromDecimalString(const StringRef &str, roundingMode rounding_mo if (decDigitValue(*D.firstSigDigit) >= 10U) { category = fcZero; fs = opOK; - } else if ((D.normalizedExponent + 1) * 28738 - <= 8651 * (semantics->minExponent - (int) semantics->precision)) { + + /* Check whether the normalized exponent is high enough to overflow + max during the log-rebasing in the max-exponent check below. */ + } else if (D.normalizedExponent - 1 > INT_MAX / 42039) { + fs = handleOverflow(rounding_mode); + + /* If it wasn't, then it also wasn't high enough to overflow max + during the log-rebasing in the min-exponent check. Check that it + won't overflow min in either check, then perform the min-exponent + check. */ + } else if (D.normalizedExponent - 1 < INT_MIN / 42039 || + (D.normalizedExponent + 1) * 28738 <= + 8651 * (semantics->minExponent - (int) semantics->precision)) { /* Underflow to zero and round. */ zeroSignificand(); fs = normalize(rounding_mode, lfLessThanHalf); + + /* We can finally safely perform the max-exponent check. */ } else if ((D.normalizedExponent - 1) * 42039 >= 12655 * semantics->maxExponent) { /* Overflow and round. */ diff --git a/lib/Support/APInt.cpp b/lib/Support/APInt.cpp index 3bce3f3..6a6384a 100644 --- a/lib/Support/APInt.cpp +++ b/lib/Support/APInt.cpp @@ -2344,13 +2344,21 @@ APInt::tcExtractBit(const integerPart *parts, unsigned int bit) & ((integerPart) 1 << bit % integerPartWidth)) != 0; } -/* Set the given bit of a bignum. */ +/* Set the given bit of a bignum. */ void APInt::tcSetBit(integerPart *parts, unsigned int bit) { parts[bit / integerPartWidth] |= (integerPart) 1 << (bit % integerPartWidth); } +/* Clears the given bit of a bignum. */ +void +APInt::tcClearBit(integerPart *parts, unsigned int bit) +{ + parts[bit / integerPartWidth] &= + ~((integerPart) 1 << (bit % integerPartWidth)); +} + /* Returns the bit number of the least significant set bit of a number. If the input number has no bits set -1U is returned. */ unsigned int diff --git a/lib/Support/CommandLine.cpp b/lib/Support/CommandLine.cpp index 961dc1f..2ab4103 100644 --- a/lib/Support/CommandLine.cpp +++ b/lib/Support/CommandLine.cpp @@ -650,7 +650,7 @@ void cl::ParseCommandLineOptions(int argc, char **argv, if (Handler == 0) { if (SinkOpts.empty()) { errs() << ProgramName << ": Unknown command line argument '" - << argv[i] << "'. Try: '" << argv[0] << " --help'\n"; + << argv[i] << "'. Try: '" << argv[0] << " -help'\n"; ErrorParsing = true; } else { for (SmallVectorImpl<Option*>::iterator I = SinkOpts.begin(), @@ -673,7 +673,7 @@ void cl::ParseCommandLineOptions(int argc, char **argv, errs() << ProgramName << ": Not enough positional command line arguments specified!\n" << "Must specify at least " << NumPositionalRequired - << " positional arguments: See: " << argv[0] << " --help\n"; + << " positional arguments: See: " << argv[0] << " -help\n"; ErrorParsing = true; } else if (!HasUnlimitedPositionals @@ -681,7 +681,7 @@ void cl::ParseCommandLineOptions(int argc, char **argv, errs() << ProgramName << ": Too many positional arguments specified!\n" << "Can specify at most " << PositionalOpts.size() - << " positional arguments: See: " << argv[0] << " --help\n"; + << " positional arguments: See: " << argv[0] << " -help\n"; ErrorParsing = true; } else if (ConsumeAfterOpt == 0) { @@ -1029,7 +1029,7 @@ void generic_parser_base::printOptionInfo(const Option &O, //===----------------------------------------------------------------------===// -// --help and --help-hidden option implementation +// -help and -help-hidden option implementation // static int OptNameCompare(const void *LHS, const void *RHS) { @@ -1134,7 +1134,7 @@ static HelpPrinter NormalPrinter(false); static HelpPrinter HiddenPrinter(true); static cl::opt<HelpPrinter, true, parser<bool> > -HOp("help", cl::desc("Display available options (--help-hidden for more)"), +HOp("help", cl::desc("Display available options (-help-hidden for more)"), cl::location(NormalPrinter), cl::ValueDisallowed); static cl::opt<HelpPrinter, true, parser<bool> > @@ -1222,8 +1222,8 @@ void cl::PrintHelpMessage() { // NormalPrinter variable is a HelpPrinter and the help gets printed when // its operator= is invoked. That's because the "normal" usages of the // help printer is to be assigned true/false depending on whether the - // --help option was given or not. Since we're circumventing that we have - // to make it look like --help was given, so we assign true. + // -help option was given or not. Since we're circumventing that we have + // to make it look like -help was given, so we assign true. NormalPrinter = true; } diff --git a/lib/Support/FormattedStream.cpp b/lib/Support/FormattedStream.cpp index 39b6cb3..c72b5a1 100644 --- a/lib/Support/FormattedStream.cpp +++ b/lib/Support/FormattedStream.cpp @@ -56,8 +56,6 @@ void formatted_raw_ostream::ComputeColumn(const char *Ptr, size_t Size) { /// PadToColumn - Align the output to some column number. /// /// \param NewCol - The column to move to. -/// \param MinPad - The minimum space to give after the most recent -/// I/O, even if the current column + minpad > newcol. /// formatted_raw_ostream &formatted_raw_ostream::PadToColumn(unsigned NewCol) { // Figure out what's in the buffer and add it to the column count. diff --git a/lib/Support/GraphWriter.cpp b/lib/Support/GraphWriter.cpp index c8bca6e..ec84f9b 100644 --- a/lib/Support/GraphWriter.cpp +++ b/lib/Support/GraphWriter.cpp @@ -137,7 +137,7 @@ void llvm::DisplayGraph(const sys::Path &Filename, bool wait, args.clear(); args.push_back(gv.c_str()); args.push_back(PSFilename.c_str()); - args.push_back("-spartan"); + args.push_back("--spartan"); args.push_back(0); ErrMsg.clear(); diff --git a/lib/Support/MemoryBuffer.cpp b/lib/Support/MemoryBuffer.cpp index 9253b01..eb046d0 100644 --- a/lib/Support/MemoryBuffer.cpp +++ b/lib/Support/MemoryBuffer.cpp @@ -174,7 +174,8 @@ MemoryBuffer *MemoryBuffer::getFile(StringRef Filename, std::string *ErrStr, #ifdef O_BINARY OpenFlags |= O_BINARY; // Open input file in binary mode on win32. #endif - int FD = ::open(Filename.str().c_str(), O_RDONLY|OpenFlags); + SmallString<256> PathBuf(Filename.begin(), Filename.end()); + int FD = ::open(PathBuf.c_str(), O_RDONLY|OpenFlags); if (FD == -1) { if (ErrStr) *ErrStr = strerror(errno); return 0; diff --git a/lib/Support/Regex.cpp b/lib/Support/Regex.cpp index 618ca05..a7631de 100644 --- a/lib/Support/Regex.cpp +++ b/lib/Support/Regex.cpp @@ -90,3 +90,79 @@ bool Regex::match(const StringRef &String, SmallVectorImpl<StringRef> *Matches){ return true; } + +std::string Regex::sub(StringRef Repl, StringRef String, + std::string *Error) { + SmallVector<StringRef, 8> Matches; + + // Reset error, if given. + if (Error && !Error->empty()) *Error = ""; + + // Return the input if there was no match. + if (!match(String, &Matches)) + return String; + + // Otherwise splice in the replacement string, starting with the prefix before + // the match. + std::string Res(String.begin(), Matches[0].begin()); + + // Then the replacement string, honoring possible substitutions. + while (!Repl.empty()) { + // Skip to the next escape. + std::pair<StringRef, StringRef> Split = Repl.split('\\'); + + // Add the skipped substring. + Res += Split.first; + + // Check for terminimation and trailing backslash. + if (Split.second.empty()) { + if (Repl.size() != Split.first.size() && + Error && Error->empty()) + *Error = "replacement string contained trailing backslash"; + break; + } + + // Otherwise update the replacement string and interpret escapes. + Repl = Split.second; + + // FIXME: We should have a StringExtras function for mapping C99 escapes. + switch (Repl[0]) { + // Treat all unrecognized characters as self-quoting. + default: + Res += Repl[0]; + Repl = Repl.substr(1); + break; + + // Single character escapes. + case 't': + Res += '\t'; + Repl = Repl.substr(1); + break; + case 'n': + Res += '\n'; + Repl = Repl.substr(1); + break; + + // Decimal escapes are backreferences. + case '0': case '1': case '2': case '3': case '4': + case '5': case '6': case '7': case '8': case '9': { + // Extract the backreference number. + StringRef Ref = Repl.slice(0, Repl.find_first_not_of("0123456789")); + Repl = Repl.substr(Ref.size()); + + unsigned RefValue; + if (!Ref.getAsInteger(10, RefValue) && + RefValue < Matches.size()) + Res += Matches[RefValue]; + else if (Error && Error->empty()) + *Error = "invalid backreference string '" + Ref.str() + "'"; + break; + } + } + } + + // And finally the suffix. + Res += StringRef(Matches[0].end(), String.end() - Matches[0].end()); + + return Res; +} diff --git a/lib/Support/StringRef.cpp b/lib/Support/StringRef.cpp index ae2640b..2b262dc 100644 --- a/lib/Support/StringRef.cpp +++ b/lib/Support/StringRef.cpp @@ -8,6 +8,7 @@ //===----------------------------------------------------------------------===// #include "llvm/ADT/StringRef.h" +#include "llvm/ADT/APInt.h" using namespace llvm; @@ -172,23 +173,28 @@ size_t StringRef::count(StringRef Str) const { return Count; } +static unsigned GetAutoSenseRadix(StringRef &Str) { + if (Str.startswith("0x")) { + Str = Str.substr(2); + return 16; + } else if (Str.startswith("0b")) { + Str = Str.substr(2); + return 2; + } else if (Str.startswith("0")) { + return 8; + } else { + return 10; + } +} + + /// GetAsUnsignedInteger - Workhorse method that converts a integer character /// sequence of radix up to 36 to an unsigned long long value. static bool GetAsUnsignedInteger(StringRef Str, unsigned Radix, unsigned long long &Result) { // Autosense radix if not specified. - if (Radix == 0) { - if (Str.startswith("0x")) { - Str = Str.substr(2); - Radix = 16; - } else if (Str.startswith("0b")) { - Str = Str.substr(2); - Radix = 2; - } else if (Str.startswith("0")) - Radix = 8; - else - Radix = 10; - } + if (Radix == 0) + Radix = GetAutoSenseRadix(Str); // Empty strings (after the radix autosense) are invalid. if (Str.empty()) return true; @@ -272,3 +278,78 @@ bool StringRef::getAsInteger(unsigned Radix, unsigned &Result) const { Result = Val; return false; } + +bool StringRef::getAsInteger(unsigned Radix, APInt &Result) const { + StringRef Str = *this; + + // Autosense radix if not specified. + if (Radix == 0) + Radix = GetAutoSenseRadix(Str); + + assert(Radix > 1 && Radix <= 36); + + // Empty strings (after the radix autosense) are invalid. + if (Str.empty()) return true; + + // Skip leading zeroes. This can be a significant improvement if + // it means we don't need > 64 bits. + while (!Str.empty() && Str.front() == '0') + Str = Str.substr(1); + + // If it was nothing but zeroes.... + if (Str.empty()) { + Result = APInt(64, 0); + return false; + } + + // (Over-)estimate the required number of bits. + unsigned Log2Radix = 0; + while ((1U << Log2Radix) < Radix) Log2Radix++; + bool IsPowerOf2Radix = ((1U << Log2Radix) == Radix); + + unsigned BitWidth = Log2Radix * Str.size(); + if (BitWidth < Result.getBitWidth()) + BitWidth = Result.getBitWidth(); // don't shrink the result + else + Result.zext(BitWidth); + + APInt RadixAP, CharAP; // unused unless !IsPowerOf2Radix + if (!IsPowerOf2Radix) { + // These must have the same bit-width as Result. + RadixAP = APInt(BitWidth, Radix); + CharAP = APInt(BitWidth, 0); + } + + // Parse all the bytes of the string given this radix. + Result = 0; + while (!Str.empty()) { + unsigned CharVal; + if (Str[0] >= '0' && Str[0] <= '9') + CharVal = Str[0]-'0'; + else if (Str[0] >= 'a' && Str[0] <= 'z') + CharVal = Str[0]-'a'+10; + else if (Str[0] >= 'A' && Str[0] <= 'Z') + CharVal = Str[0]-'A'+10; + else + return true; + + // If the parsed value is larger than the integer radix, the string is + // invalid. + if (CharVal >= Radix) + return true; + + // Add in this character. + if (IsPowerOf2Radix) { + Result <<= Log2Radix; + Result |= CharVal; + } else { + Result *= RadixAP; + CharAP = CharVal; + Result += CharAP; + } + + Str = Str.substr(1); + } + + return false; +} diff --git a/lib/Support/Triple.cpp b/lib/Support/Triple.cpp index 5a76184..61bf0a7 100644 --- a/lib/Support/Triple.cpp +++ b/lib/Support/Triple.cpp @@ -40,6 +40,7 @@ const char *Triple::getArchTypeName(ArchType Kind) { case x86: return "i386"; case x86_64: return "x86_64"; case xcore: return "xcore"; + case mblaze: return "mblaze"; } return "<invalid>"; @@ -62,6 +63,8 @@ const char *Triple::getArchTypePrefix(ArchType Kind) { case ppc64: case ppc: return "ppc"; + case mblaze: return "mblaze"; + case sparcv9: case sparc: return "sparc"; @@ -127,6 +130,8 @@ Triple::ArchType Triple::getArchTypeForLLVMName(StringRef Name) { return ppc64; if (Name == "ppc") return ppc; + if (Name == "mblaze") + return mblaze; if (Name == "sparc") return sparc; if (Name == "sparcv9") @@ -198,6 +203,8 @@ const char *Triple::getArchNameForAssembler() { return "ppc"; if (Str == "powerpc64") return "ppc64"; + if (Str == "mblaze" || Str == "microblaze") + return "mblaze"; if (Str == "arm") return "arm"; if (Str == "armv4t" || Str == "thumbv4t") @@ -234,6 +241,8 @@ void Triple::Parse() const { Arch = ppc; else if ((ArchName == "powerpc64") || (ArchName == "ppu")) Arch = ppc64; + else if (ArchName == "mblaze") + Arch = mblaze; else if (ArchName == "arm" || ArchName.startswith("armv") || ArchName == "xscale") diff --git a/lib/Target/ARM/ARMBaseInstrInfo.cpp b/lib/Target/ARM/ARMBaseInstrInfo.cpp index 6fe7c2c..8e537d8 100644 --- a/lib/Target/ARM/ARMBaseInstrInfo.cpp +++ b/lib/Target/ARM/ARMBaseInstrInfo.cpp @@ -643,6 +643,13 @@ ARMBaseInstrInfo::copyRegToReg(MachineBasicBlock &MBB, DebugLoc DL = DebugLoc::getUnknownLoc(); if (I != MBB.end()) DL = I->getDebugLoc(); + // tGPR is used sometimes in ARM instructions that need to avoid using + // certain registers. Just treat it as GPR here. + if (DestRC == ARM::tGPRRegisterClass) + DestRC = ARM::GPRRegisterClass; + if (SrcRC == ARM::tGPRRegisterClass) + SrcRC = ARM::GPRRegisterClass; + if (DestRC != SrcRC) { if (DestRC->getSize() != SrcRC->getSize()) return false; @@ -697,6 +704,11 @@ storeRegToStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, MFI.getObjectSize(FI), Align); + // tGPR is used sometimes in ARM instructions that need to avoid using + // certain registers. Just treat it as GPR here. + if (RC == ARM::tGPRRegisterClass) + RC = ARM::GPRRegisterClass; + if (RC == ARM::GPRRegisterClass) { AddDefaultPred(BuildMI(MBB, I, DL, get(ARM::STR)) .addReg(SrcReg, getKillRegState(isKill)) @@ -745,6 +757,11 @@ loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, MFI.getObjectSize(FI), Align); + // tGPR is used sometimes in ARM instructions that need to avoid using + // certain registers. Just treat it as GPR here. + if (RC == ARM::tGPRRegisterClass) + RC = ARM::GPRRegisterClass; + if (RC == ARM::GPRRegisterClass) { AddDefaultPred(BuildMI(MBB, I, DL, get(ARM::LDR), DestReg) .addFrameIndex(FI).addReg(0).addImm(0).addMemOperand(MMO)); @@ -1020,9 +1037,8 @@ ARMBaseInstrInfo::duplicate(MachineInstr *Orig, MachineFunction &MF) const { return MI; } -bool ARMBaseInstrInfo::isIdentical(const MachineInstr *MI0, - const MachineInstr *MI1, - const MachineRegisterInfo *MRI) const { +bool ARMBaseInstrInfo::produceSameValue(const MachineInstr *MI0, + const MachineInstr *MI1) const { int Opcode = MI0->getOpcode(); if (Opcode == ARM::t2LDRpci || Opcode == ARM::t2LDRpci_pic || @@ -1051,7 +1067,7 @@ bool ARMBaseInstrInfo::isIdentical(const MachineInstr *MI0, return ACPV0->hasSameValue(ACPV1); } - return TargetInstrInfoImpl::isIdentical(MI0, MI1, MRI); + return MI0->isIdenticalTo(MI1, MachineInstr::IgnoreVRegDefs); } /// getInstrPredicate - If instruction is predicated, returns its predicate diff --git a/lib/Target/ARM/ARMBaseInstrInfo.h b/lib/Target/ARM/ARMBaseInstrInfo.h index 0d9d4a7..e095acc 100644 --- a/lib/Target/ARM/ARMBaseInstrInfo.h +++ b/lib/Target/ARM/ARMBaseInstrInfo.h @@ -289,8 +289,8 @@ public: MachineInstr *duplicate(MachineInstr *Orig, MachineFunction &MF) const; - virtual bool isIdentical(const MachineInstr *MI, const MachineInstr *Other, - const MachineRegisterInfo *MRI) const; + virtual bool produceSameValue(const MachineInstr *MI0, + const MachineInstr *MI1) const; }; static inline diff --git a/lib/Target/ARM/ARMBaseRegisterInfo.cpp b/lib/Target/ARM/ARMBaseRegisterInfo.cpp index 91e3550..8044966 100644 --- a/lib/Target/ARM/ARMBaseRegisterInfo.cpp +++ b/lib/Target/ARM/ARMBaseRegisterInfo.cpp @@ -513,7 +513,7 @@ cannotEliminateFrame(const MachineFunction &MF) const { } /// estimateStackSize - Estimate and return the size of the frame. -static unsigned estimateStackSize(MachineFunction &MF, MachineFrameInfo *MFI) { +static unsigned estimateStackSize(MachineFunction &MF) { const MachineFrameInfo *FFI = MF.getFrameInfo(); int Offset = 0; for (int i = FFI->getObjectIndexBegin(); i != 0; ++i) { @@ -671,8 +671,16 @@ ARMBaseRegisterInfo::processFunctionBeforeCalleeSavedScan(MachineFunction &MF, } } + // If any of the stack slot references may be out of range of an immediate + // offset, make sure a register (or a spill slot) is available for the + // register scavenger. Note that if we're indexing off the frame pointer, the + // effective stack size is 4 bytes larger since the FP points to the stack + // slot of the previous FP. + bool BigStack = RS && + estimateStackSize(MF) + (hasFP(MF) ? 4 : 0) >= estimateRSStackSizeLimit(MF); + bool ExtraCSSpill = false; - if (!CanEliminateFrame || cannotEliminateFrame(MF)) { + if (BigStack || !CanEliminateFrame || cannotEliminateFrame(MF)) { AFI->setHasStackFrame(true); // If LR is not spilled, but at least one of R4, R5, R6, and R7 is spilled. @@ -727,51 +735,43 @@ ARMBaseRegisterInfo::processFunctionBeforeCalleeSavedScan(MachineFunction &MF, // callee-saved register or reserve a special spill slot to facilitate // register scavenging. Thumb1 needs a spill slot for stack pointer // adjustments also, even when the frame itself is small. - if (RS && !ExtraCSSpill) { - MachineFrameInfo *MFI = MF.getFrameInfo(); - // If any of the stack slot references may be out of range of an - // immediate offset, make sure a register (or a spill slot) is - // available for the register scavenger. Note that if we're indexing - // off the frame pointer, the effective stack size is 4 bytes larger - // since the FP points to the stack slot of the previous FP. - if (estimateStackSize(MF, MFI) + (hasFP(MF) ? 4 : 0) - >= estimateRSStackSizeLimit(MF)) { - // If any non-reserved CS register isn't spilled, just spill one or two - // extra. That should take care of it! - unsigned NumExtras = TargetAlign / 4; - SmallVector<unsigned, 2> Extras; - while (NumExtras && !UnspilledCS1GPRs.empty()) { - unsigned Reg = UnspilledCS1GPRs.back(); - UnspilledCS1GPRs.pop_back(); + if (BigStack && !ExtraCSSpill) { + // If any non-reserved CS register isn't spilled, just spill one or two + // extra. That should take care of it! + unsigned NumExtras = TargetAlign / 4; + SmallVector<unsigned, 2> Extras; + while (NumExtras && !UnspilledCS1GPRs.empty()) { + unsigned Reg = UnspilledCS1GPRs.back(); + UnspilledCS1GPRs.pop_back(); + if (!isReservedReg(MF, Reg)) { + Extras.push_back(Reg); + NumExtras--; + } + } + // For non-Thumb1 functions, also check for hi-reg CS registers + if (!AFI->isThumb1OnlyFunction()) { + while (NumExtras && !UnspilledCS2GPRs.empty()) { + unsigned Reg = UnspilledCS2GPRs.back(); + UnspilledCS2GPRs.pop_back(); if (!isReservedReg(MF, Reg)) { Extras.push_back(Reg); NumExtras--; } } - // For non-Thumb1 functions, also check for hi-reg CS registers - if (!AFI->isThumb1OnlyFunction()) { - while (NumExtras && !UnspilledCS2GPRs.empty()) { - unsigned Reg = UnspilledCS2GPRs.back(); - UnspilledCS2GPRs.pop_back(); - if (!isReservedReg(MF, Reg)) { - Extras.push_back(Reg); - NumExtras--; - } - } - } - if (Extras.size() && NumExtras == 0) { - for (unsigned i = 0, e = Extras.size(); i != e; ++i) { - MF.getRegInfo().setPhysRegUsed(Extras[i]); - AFI->setCSRegisterIsSpilled(Extras[i]); - } - } else if (!AFI->isThumb1OnlyFunction()) { - // note: Thumb1 functions spill to R12, not the stack. - // Reserve a slot closest to SP or frame pointer. - const TargetRegisterClass *RC = ARM::GPRRegisterClass; - RS->setScavengingFrameIndex(MFI->CreateStackObject(RC->getSize(), - RC->getAlignment(), - false)); + } + if (Extras.size() && NumExtras == 0) { + for (unsigned i = 0, e = Extras.size(); i != e; ++i) { + MF.getRegInfo().setPhysRegUsed(Extras[i]); + AFI->setCSRegisterIsSpilled(Extras[i]); } + } else if (!AFI->isThumb1OnlyFunction()) { + // note: Thumb1 functions spill to R12, not the stack. Reserve a slot + // closest to SP or frame pointer. + const TargetRegisterClass *RC = ARM::GPRRegisterClass; + MachineFrameInfo *MFI = MF.getFrameInfo(); + RS->setScavengingFrameIndex(MFI->CreateStackObject(RC->getSize(), + RC->getAlignment(), + false)); } } } @@ -1085,6 +1085,15 @@ hasReservedCallFrame(MachineFunction &MF) const { return !MF.getFrameInfo()->hasVarSizedObjects(); } +// canSimplifyCallFramePseudos - If there is a reserved call frame, the +// call frame pseudos can be simplified. Unlike most targets, having a FP +// is not sufficient here since we still may reference some objects via SP +// even when FP is available in Thumb2 mode. +bool ARMBaseRegisterInfo:: +canSimplifyCallFramePseudos(MachineFunction &MF) const { + return hasReservedCallFrame(MF) || MF.getFrameInfo()->hasVarSizedObjects(); +} + static void emitSPUpdate(bool isARM, MachineBasicBlock &MBB, MachineBasicBlock::iterator &MBBI, @@ -1119,13 +1128,14 @@ eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB, ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>(); assert(!AFI->isThumb1OnlyFunction() && - "This eliminateCallFramePseudoInstr does not suppor Thumb1!"); + "This eliminateCallFramePseudoInstr does not support Thumb1!"); bool isARM = !AFI->isThumbFunction(); // Replace the pseudo instruction with a new instruction... unsigned Opc = Old->getOpcode(); - ARMCC::CondCodes Pred = (ARMCC::CondCodes)Old->getOperand(1).getImm(); - // FIXME: Thumb2 version of ADJCALLSTACKUP and ADJCALLSTACKDOWN? + int PIdx = Old->findFirstPredOperandIdx(); + ARMCC::CondCodes Pred = (PIdx == -1) + ? ARMCC::AL : (ARMCC::CondCodes)Old->getOperand(PIdx).getImm(); if (Opc == ARM::ADJCALLSTACKDOWN || Opc == ARM::tADJCALLSTACKDOWN) { // Note: PredReg is operand 2 for ADJCALLSTACKDOWN. unsigned PredReg = Old->getOperand(2).getReg(); @@ -1149,7 +1159,6 @@ ARMBaseRegisterInfo::eliminateFrameIndex(MachineBasicBlock::iterator II, MachineInstr &MI = *II; MachineBasicBlock &MBB = *MI.getParent(); MachineFunction &MF = *MBB.getParent(); - const MachineFrameInfo *MFI = MF.getFrameInfo(); ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>(); assert(!AFI->isThumb1OnlyFunction() && "This eliminateFrameIndex does not support Thumb1!"); @@ -1160,12 +1169,12 @@ ARMBaseRegisterInfo::eliminateFrameIndex(MachineBasicBlock::iterator II, } int FrameIndex = MI.getOperand(i).getIndex(); - int Offset = MFI->getObjectOffset(FrameIndex) + MFI->getStackSize() + SPAdj; unsigned FrameReg; - Offset = getFrameIndexReference(MF, FrameIndex, FrameReg); + int Offset = getFrameIndexReference(MF, FrameIndex, FrameReg); if (FrameReg != ARM::SP) SPAdj = 0; + Offset += SPAdj; // Modify MI as necessary to handle as much of 'Offset' as possible bool Done = false; @@ -1256,7 +1265,7 @@ emitPrologue(MachineFunction &MF) const { MachineFrameInfo *MFI = MF.getFrameInfo(); ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>(); assert(!AFI->isThumb1OnlyFunction() && - "This emitPrologue does not suppor Thumb1!"); + "This emitPrologue does not support Thumb1!"); bool isARM = !AFI->isThumbFunction(); unsigned VARegSaveSize = AFI->getVarArgsRegSaveSize(); unsigned NumBytes = MFI->getStackSize(); @@ -1417,7 +1426,7 @@ emitEpilogue(MachineFunction &MF, MachineBasicBlock &MBB) const { MachineFrameInfo *MFI = MF.getFrameInfo(); ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>(); assert(!AFI->isThumb1OnlyFunction() && - "This emitEpilogue does not suppor Thumb1!"); + "This emitEpilogue does not support Thumb1!"); bool isARM = !AFI->isThumbFunction(); unsigned VARegSaveSize = AFI->getVarArgsRegSaveSize(); diff --git a/lib/Target/ARM/ARMBaseRegisterInfo.h b/lib/Target/ARM/ARMBaseRegisterInfo.h index 33ba21d..64f6ff1 100644 --- a/lib/Target/ARM/ARMBaseRegisterInfo.h +++ b/lib/Target/ARM/ARMBaseRegisterInfo.h @@ -138,6 +138,7 @@ public: virtual bool requiresFrameIndexScavenging(const MachineFunction &MF) const; virtual bool hasReservedCallFrame(MachineFunction &MF) const; + virtual bool canSimplifyCallFramePseudos(MachineFunction &MF) const; virtual void eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB, diff --git a/lib/Target/ARM/ARMISelDAGToDAG.cpp b/lib/Target/ARM/ARMISelDAGToDAG.cpp index df4ae70..013e00a 100644 --- a/lib/Target/ARM/ARMISelDAGToDAG.cpp +++ b/lib/Target/ARM/ARMISelDAGToDAG.cpp @@ -58,8 +58,6 @@ public: return "ARM Instruction Selection"; } - virtual void InstructionSelect(); - /// getI32Imm - Return a target constant of type i32 with the specified /// value. inline SDValue getI32Imm(unsigned Imm) { @@ -203,11 +201,6 @@ static bool isOpcWithIntImmediate(SDNode *N, unsigned Opc, unsigned& Imm) { } -void ARMDAGToDAGISel::InstructionSelect() { - SelectRoot(*CurDAG); - CurDAG->RemoveDeadNodes(); -} - bool ARMDAGToDAGISel::SelectShifterOperandReg(SDNode *Op, SDValue N, SDValue &BaseReg, diff --git a/lib/Target/ARM/ARMISelLowering.cpp b/lib/Target/ARM/ARMISelLowering.cpp index adf1644..6a2c6bb 100644 --- a/lib/Target/ARM/ARMISelLowering.cpp +++ b/lib/Target/ARM/ARMISelLowering.cpp @@ -294,6 +294,7 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM) setTargetDAGCombine(ISD::SIGN_EXTEND); setTargetDAGCombine(ISD::ZERO_EXTEND); setTargetDAGCombine(ISD::ANY_EXTEND); + setTargetDAGCombine(ISD::SELECT_CC); } computeRegisterProperties(); @@ -544,6 +545,8 @@ const char *ARMTargetLowering::getTargetNodeName(unsigned Opcode) const { case ARMISD::VZIP: return "ARMISD::VZIP"; case ARMISD::VUZP: return "ARMISD::VUZP"; case ARMISD::VTRN: return "ARMISD::VTRN"; + case ARMISD::FMAX: return "ARMISD::FMAX"; + case ARMISD::FMIN: return "ARMISD::FMIN"; } } @@ -921,7 +924,7 @@ ARMTargetLowering::LowerCall(SDValue Chain, SDValue Callee, // These operations are automatically eliminated by the prolog/epilog pass Chain = DAG.getCALLSEQ_START(Chain, DAG.getIntPtrConstant(NumBytes, true)); - SDValue StackPtr = DAG.getRegister(ARM::SP, MVT::i32); + SDValue StackPtr = DAG.getCopyFromReg(Chain, dl, ARM::SP, getPointerTy()); RegsToPassVector RegsToPass; SmallVector<SDValue, 8> MemOpChains; @@ -970,8 +973,6 @@ ARMTargetLowering::LowerCall(SDValue Chain, SDValue Callee, VA, ArgLocs[++i], StackPtr, MemOpChains, Flags); } else { assert(VA.isMemLoc()); - if (StackPtr.getNode() == 0) - StackPtr = DAG.getCopyFromReg(Chain, dl, ARM::SP, getPointerTy()); MemOpChains.push_back(LowerMemOpCallTo(Chain, StackPtr, Op1, dl, DAG, VA, Flags)); @@ -984,8 +985,6 @@ ARMTargetLowering::LowerCall(SDValue Chain, SDValue Callee, RegsToPass.push_back(std::make_pair(VA.getLocReg(), Arg)); } else { assert(VA.isMemLoc()); - if (StackPtr.getNode() == 0) - StackPtr = DAG.getCopyFromReg(Chain, dl, ARM::SP, getPointerTy()); MemOpChains.push_back(LowerMemOpCallTo(Chain, StackPtr, Arg, dl, DAG, VA, Flags)); @@ -1283,8 +1282,7 @@ ARMTargetLowering::LowerToTLSGeneralDynamicModel(GlobalAddressSDNode *GA, LowerCallTo(Chain, (const Type *) Type::getInt32Ty(*DAG.getContext()), false, false, false, false, 0, CallingConv::C, false, /*isReturnValueUsed=*/true, - DAG.getExternalSymbol("__tls_get_addr", PtrVT), Args, DAG, dl, - DAG.GetOrdering(Chain.getNode())); + DAG.getExternalSymbol("__tls_get_addr", PtrVT), Args, DAG, dl); return CallResult.first; } @@ -3856,23 +3854,106 @@ static SDValue PerformExtendCombine(SDNode *N, SelectionDAG &DAG, return SDValue(); } +/// PerformSELECT_CCCombine - Target-specific DAG combining for ISD::SELECT_CC +/// to match f32 max/min patterns to use NEON vmax/vmin instructions. +static SDValue PerformSELECT_CCCombine(SDNode *N, SelectionDAG &DAG, + const ARMSubtarget *ST) { + // If the target supports NEON, try to use vmax/vmin instructions for f32 + // selects like "x < y ? x : y". Unless the FiniteOnlyFPMath option is set, + // be careful about NaNs: NEON's vmax/vmin return NaN if either operand is + // a NaN; only do the transformation when it matches that behavior. + + // For now only do this when using NEON for FP operations; if using VFP, it + // is not obvious that the benefit outweighs the cost of switching to the + // NEON pipeline. + if (!ST->hasNEON() || !ST->useNEONForSinglePrecisionFP() || + N->getValueType(0) != MVT::f32) + return SDValue(); + + SDValue CondLHS = N->getOperand(0); + SDValue CondRHS = N->getOperand(1); + SDValue LHS = N->getOperand(2); + SDValue RHS = N->getOperand(3); + ISD::CondCode CC = cast<CondCodeSDNode>(N->getOperand(4))->get(); + + unsigned Opcode = 0; + bool IsReversed; + if (DAG.isEqualTo(LHS, CondLHS) && DAG.isEqualTo(RHS, CondRHS)) { + IsReversed = false; // x CC y ? x : y + } else if (DAG.isEqualTo(LHS, CondRHS) && DAG.isEqualTo(RHS, CondLHS)) { + IsReversed = true ; // x CC y ? y : x + } else { + return SDValue(); + } + + bool IsUnordered; + switch (CC) { + default: break; + case ISD::SETOLT: + case ISD::SETOLE: + case ISD::SETLT: + case ISD::SETLE: + case ISD::SETULT: + case ISD::SETULE: + // If LHS is NaN, an ordered comparison will be false and the result will + // be the RHS, but vmin(NaN, RHS) = NaN. Avoid this by checking that LHS + // != NaN. Likewise, for unordered comparisons, check for RHS != NaN. + IsUnordered = (CC == ISD::SETULT || CC == ISD::SETULE); + if (!DAG.isKnownNeverNaN(IsUnordered ? RHS : LHS)) + break; + // For less-than-or-equal comparisons, "+0 <= -0" will be true but vmin + // will return -0, so vmin can only be used for unsafe math or if one of + // the operands is known to be nonzero. + if ((CC == ISD::SETLE || CC == ISD::SETOLE || CC == ISD::SETULE) && + !UnsafeFPMath && + !(DAG.isKnownNeverZero(LHS) || DAG.isKnownNeverZero(RHS))) + break; + Opcode = IsReversed ? ARMISD::FMAX : ARMISD::FMIN; + break; + + case ISD::SETOGT: + case ISD::SETOGE: + case ISD::SETGT: + case ISD::SETGE: + case ISD::SETUGT: + case ISD::SETUGE: + // If LHS is NaN, an ordered comparison will be false and the result will + // be the RHS, but vmax(NaN, RHS) = NaN. Avoid this by checking that LHS + // != NaN. Likewise, for unordered comparisons, check for RHS != NaN. + IsUnordered = (CC == ISD::SETUGT || CC == ISD::SETUGE); + if (!DAG.isKnownNeverNaN(IsUnordered ? RHS : LHS)) + break; + // For greater-than-or-equal comparisons, "-0 >= +0" will be true but vmax + // will return +0, so vmax can only be used for unsafe math or if one of + // the operands is known to be nonzero. + if ((CC == ISD::SETGE || CC == ISD::SETOGE || CC == ISD::SETUGE) && + !UnsafeFPMath && + !(DAG.isKnownNeverZero(LHS) || DAG.isKnownNeverZero(RHS))) + break; + Opcode = IsReversed ? ARMISD::FMIN : ARMISD::FMAX; + break; + } + + if (!Opcode) + return SDValue(); + return DAG.getNode(Opcode, N->getDebugLoc(), N->getValueType(0), LHS, RHS); +} + SDValue ARMTargetLowering::PerformDAGCombine(SDNode *N, DAGCombinerInfo &DCI) const { switch (N->getOpcode()) { default: break; - case ISD::ADD: return PerformADDCombine(N, DCI); - case ISD::SUB: return PerformSUBCombine(N, DCI); + case ISD::ADD: return PerformADDCombine(N, DCI); + case ISD::SUB: return PerformSUBCombine(N, DCI); case ARMISD::VMOVRRD: return PerformVMOVRRDCombine(N, DCI); - case ISD::INTRINSIC_WO_CHAIN: - return PerformIntrinsicCombine(N, DCI.DAG); + case ISD::INTRINSIC_WO_CHAIN: return PerformIntrinsicCombine(N, DCI.DAG); case ISD::SHL: case ISD::SRA: - case ISD::SRL: - return PerformShiftCombine(N, DCI.DAG, Subtarget); + case ISD::SRL: return PerformShiftCombine(N, DCI.DAG, Subtarget); case ISD::SIGN_EXTEND: case ISD::ZERO_EXTEND: - case ISD::ANY_EXTEND: - return PerformExtendCombine(N, DCI.DAG, Subtarget); + case ISD::ANY_EXTEND: return PerformExtendCombine(N, DCI.DAG, Subtarget); + case ISD::SELECT_CC: return PerformSELECT_CCCombine(N, DCI.DAG, Subtarget); } return SDValue(); } diff --git a/lib/Target/ARM/ARMISelLowering.h b/lib/Target/ARM/ARMISelLowering.h index 3c5df45..f8f8adc 100644 --- a/lib/Target/ARM/ARMISelLowering.h +++ b/lib/Target/ARM/ARMISelLowering.h @@ -131,7 +131,11 @@ namespace llvm { VREV16, // reverse elements within 16-bit halfwords VZIP, // zip (interleave) VUZP, // unzip (deinterleave) - VTRN // transpose + VTRN, // transpose + + // Floating-point max and min: + FMAX, + FMIN }; } diff --git a/lib/Target/ARM/ARMInstrFormats.td b/lib/Target/ARM/ARMInstrFormats.td index db60458..76595fa 100644 --- a/lib/Target/ARM/ARMInstrFormats.td +++ b/lib/Target/ARM/ARMInstrFormats.td @@ -708,6 +708,20 @@ class AI3ldsbpr<dag oops, dag iops, Format f, InstrItinClass itin, let Inst{24} = 1; // P bit let Inst{27-25} = 0b000; } +class AI3lddpr<dag oops, dag iops, Format f, InstrItinClass itin, + string opc, string asm, string cstr, list<dag> pattern> + : I<oops, iops, AddrMode3, Size4Bytes, IndexModePre, f, itin, + opc, asm, cstr, pattern> { + let Inst{4} = 1; + let Inst{5} = 0; // H bit + let Inst{6} = 1; // S bit + let Inst{7} = 1; + let Inst{20} = 0; // L bit + let Inst{21} = 1; // W bit + let Inst{24} = 1; // P bit + let Inst{27-25} = 0b000; +} + // Pre-indexed stores class AI3sthpr<dag oops, dag iops, Format f, InstrItinClass itin, @@ -723,6 +737,19 @@ class AI3sthpr<dag oops, dag iops, Format f, InstrItinClass itin, let Inst{24} = 1; // P bit let Inst{27-25} = 0b000; } +class AI3stdpr<dag oops, dag iops, Format f, InstrItinClass itin, + string opc, string asm, string cstr, list<dag> pattern> + : I<oops, iops, AddrMode3, Size4Bytes, IndexModePre, f, itin, + opc, asm, cstr, pattern> { + let Inst{4} = 1; + let Inst{5} = 1; // H bit + let Inst{6} = 1; // S bit + let Inst{7} = 1; + let Inst{20} = 0; // L bit + let Inst{21} = 1; // W bit + let Inst{24} = 1; // P bit + let Inst{27-25} = 0b000; +} // Post-indexed loads class AI3ldhpo<dag oops, dag iops, Format f, InstrItinClass itin, @@ -734,7 +761,7 @@ class AI3ldhpo<dag oops, dag iops, Format f, InstrItinClass itin, let Inst{6} = 0; // S bit let Inst{7} = 1; let Inst{20} = 1; // L bit - let Inst{21} = 1; // W bit + let Inst{21} = 0; // W bit let Inst{24} = 0; // P bit let Inst{27-25} = 0b000; } @@ -747,7 +774,7 @@ class AI3ldshpo<dag oops, dag iops, Format f, InstrItinClass itin, let Inst{6} = 1; // S bit let Inst{7} = 1; let Inst{20} = 1; // L bit - let Inst{21} = 1; // W bit + let Inst{21} = 0; // W bit let Inst{24} = 0; // P bit let Inst{27-25} = 0b000; } @@ -760,7 +787,20 @@ class AI3ldsbpo<dag oops, dag iops, Format f, InstrItinClass itin, let Inst{6} = 1; // S bit let Inst{7} = 1; let Inst{20} = 1; // L bit - let Inst{21} = 1; // W bit + let Inst{21} = 0; // W bit + let Inst{24} = 0; // P bit + let Inst{27-25} = 0b000; +} +class AI3lddpo<dag oops, dag iops, Format f, InstrItinClass itin, + string opc, string asm, string cstr, list<dag> pattern> + : I<oops, iops, AddrMode3, Size4Bytes, IndexModePost, f, itin, + opc, asm, cstr, pattern> { + let Inst{4} = 1; + let Inst{5} = 0; // H bit + let Inst{6} = 1; // S bit + let Inst{7} = 1; + let Inst{20} = 0; // L bit + let Inst{21} = 0; // W bit let Inst{24} = 0; // P bit let Inst{27-25} = 0b000; } @@ -775,7 +815,20 @@ class AI3sthpo<dag oops, dag iops, Format f, InstrItinClass itin, let Inst{6} = 0; // S bit let Inst{7} = 1; let Inst{20} = 0; // L bit - let Inst{21} = 1; // W bit + let Inst{21} = 0; // W bit + let Inst{24} = 0; // P bit + let Inst{27-25} = 0b000; +} +class AI3stdpo<dag oops, dag iops, Format f, InstrItinClass itin, + string opc, string asm, string cstr, list<dag> pattern> + : I<oops, iops, AddrMode3, Size4Bytes, IndexModePost, f, itin, + opc, asm, cstr, pattern> { + let Inst{4} = 1; + let Inst{5} = 1; // H bit + let Inst{6} = 1; // S bit + let Inst{7} = 1; + let Inst{20} = 0; // L bit + let Inst{21} = 0; // W bit let Inst{24} = 0; // P bit let Inst{27-25} = 0b000; } @@ -1150,6 +1203,19 @@ class T2Iidxldst<bit signed, bits<2> opcod, bit load, bit pre, let Inst{8} = 1; // The W bit. } +// Helper class for disassembly only +// A6.3.16 & A6.3.17 +// T2Imac - Thumb2 multiply [accumulate, and absolute difference] instructions. +class T2I_mac<bit long, bits<3> op22_20, bits<4> op7_4, dag oops, dag iops, + InstrItinClass itin, string opc, string asm, list<dag> pattern> + : T2I<oops, iops, itin, opc, asm, pattern> { + let Inst{31-27} = 0b11111; + let Inst{26-24} = 0b011; + let Inst{23} = long; + let Inst{22-20} = op22_20; + let Inst{7-4} = op7_4; +} + // Tv5Pat - Same as Pat<>, but requires V5T Thumb mode. class Tv5Pat<dag pattern, dag result> : Pat<pattern, result> { list<Predicate> Predicates = [IsThumb1Only, HasV5T]; diff --git a/lib/Target/ARM/ARMInstrInfo.td b/lib/Target/ARM/ARMInstrInfo.td index 1c6f78a..87c6f6e 100644 --- a/lib/Target/ARM/ARMInstrInfo.td +++ b/lib/Target/ARM/ARMInstrInfo.td @@ -130,8 +130,6 @@ def IsThumb2 : Predicate<"Subtarget->isThumb2()">; def IsARM : Predicate<"!Subtarget->isThumb()">; def IsDarwin : Predicate<"Subtarget->isTargetDarwin()">; def IsNotDarwin : Predicate<"!Subtarget->isTargetDarwin()">; -def CarryDefIsUnused : Predicate<"!N->hasAnyUseOfValue(1)">; -def CarryDefIsUsed : Predicate<"N->hasAnyUseOfValue(1)">; // FIXME: Eventually this will be just "hasV6T2Ops". def UseMovt : Predicate<"Subtarget->useMovt()">; @@ -176,7 +174,7 @@ def imm16_31 : PatLeaf<(i32 imm), [{ return (int32_t)N->getZExtValue() >= 16 && (int32_t)N->getZExtValue() < 32; }]>; -def so_imm_neg : +def so_imm_neg : PatLeaf<(imm), [{ return ARM_AM::getSOImmVal(-(int)N->getZExtValue()) != -1; }], so_imm_neg_XFORM>; @@ -194,7 +192,7 @@ def sext_16_node : PatLeaf<(i32 GPR:$a), [{ /// bf_inv_mask_imm predicate - An AND mask to clear an arbitrary width bitfield /// e.g., 0xf000ffff def bf_inv_mask_imm : Operand<i32>, - PatLeaf<(imm), [{ + PatLeaf<(imm), [{ uint32_t v = (uint32_t)N->getZExtValue(); if (v == 0xffffffff) return 0; @@ -227,7 +225,7 @@ def lo16AllZero : PatLeaf<(i32 imm), [{ return (((uint32_t)N->getZExtValue()) & 0xFFFFUL) == 0; }], hi16>; -/// imm0_65535 predicate - True if the 32-bit immediate is in the range +/// imm0_65535 predicate - True if the 32-bit immediate is in the range /// [0.65535]. def imm0_65535 : PatLeaf<(i32 imm), [{ return (uint32_t)N->getZExtValue() < 65536; @@ -236,6 +234,21 @@ def imm0_65535 : PatLeaf<(i32 imm), [{ class BinOpFrag<dag res> : PatFrag<(ops node:$LHS, node:$RHS), res>; class UnOpFrag <dag res> : PatFrag<(ops node:$Src), res>; +/// adde and sube predicates - True based on whether the carry flag output +/// will be needed or not. +def adde_dead_carry : + PatFrag<(ops node:$LHS, node:$RHS), (adde node:$LHS, node:$RHS), + [{return !N->hasAnyUseOfValue(1);}]>; +def sube_dead_carry : + PatFrag<(ops node:$LHS, node:$RHS), (sube node:$LHS, node:$RHS), + [{return !N->hasAnyUseOfValue(1);}]>; +def adde_live_carry : + PatFrag<(ops node:$LHS, node:$RHS), (adde node:$LHS, node:$RHS), + [{return N->hasAnyUseOfValue(1);}]>; +def sube_live_carry : + PatFrag<(ops node:$LHS, node:$RHS), (sube node:$LHS, node:$RHS), + [{return N->hasAnyUseOfValue(1);}]>; + //===----------------------------------------------------------------------===// // Operand Definitions. // @@ -501,6 +514,22 @@ multiclass AI_unary_rrot<bits<8> opcod, string opc, PatFrag opnode> { } } +multiclass AI_unary_rrot_np<bits<8> opcod, string opc> { + def r : AExtI<opcod, (outs GPR:$dst), (ins GPR:$src), + IIC_iUNAr, opc, "\t$dst, $src", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV6]> { + let Inst{11-10} = 0b00; + let Inst{19-16} = 0b1111; + } + def r_rot : AExtI<opcod, (outs GPR:$dst), (ins GPR:$src, i32imm:$rot), + IIC_iUNAsi, opc, "\t$dst, $src, ror $rot", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV6]> { + let Inst{19-16} = 0b1111; + } +} + /// AI_bin_rrot - A binary operation with two forms: one whose operand is a /// register and one whose operand is a register rotated by 8/16/24. multiclass AI_bin_rrot<bits<8> opcod, string opc, PatFrag opnode> { @@ -510,13 +539,29 @@ multiclass AI_bin_rrot<bits<8> opcod, string opc, PatFrag opnode> { Requires<[IsARM, HasV6]> { let Inst{11-10} = 0b00; } - def rr_rot : AExtI<opcod, (outs GPR:$dst), (ins GPR:$LHS, GPR:$RHS, i32imm:$rot), + def rr_rot : AExtI<opcod, (outs GPR:$dst), (ins GPR:$LHS, GPR:$RHS, + i32imm:$rot), IIC_iALUsi, opc, "\t$dst, $LHS, $RHS, ror $rot", [(set GPR:$dst, (opnode GPR:$LHS, (rotr GPR:$RHS, rot_imm:$rot)))]>, Requires<[IsARM, HasV6]>; } +// For disassembly only. +multiclass AI_bin_rrot_np<bits<8> opcod, string opc> { + def rr : AExtI<opcod, (outs GPR:$dst), (ins GPR:$LHS, GPR:$RHS), + IIC_iALUr, opc, "\t$dst, $LHS, $RHS", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV6]> { + let Inst{11-10} = 0b00; + } + def rr_rot : AExtI<opcod, (outs GPR:$dst), (ins GPR:$LHS, GPR:$RHS, + i32imm:$rot), + IIC_iALUsi, opc, "\t$dst, $LHS, $RHS, ror $rot", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV6]>; +} + /// AI1_adde_sube_irs - Define instructions and patterns for adde and sube. let Uses = [CPSR] in { multiclass AI1_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, @@ -524,13 +569,13 @@ multiclass AI1_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, def ri : AsI1<opcod, (outs GPR:$dst), (ins GPR:$a, so_imm:$b), DPFrm, IIC_iALUi, opc, "\t$dst, $a, $b", [(set GPR:$dst, (opnode GPR:$a, so_imm:$b))]>, - Requires<[IsARM, CarryDefIsUnused]> { + Requires<[IsARM]> { let Inst{25} = 1; } def rr : AsI1<opcod, (outs GPR:$dst), (ins GPR:$a, GPR:$b), DPFrm, IIC_iALUr, opc, "\t$dst, $a, $b", [(set GPR:$dst, (opnode GPR:$a, GPR:$b))]>, - Requires<[IsARM, CarryDefIsUnused]> { + Requires<[IsARM]> { let isCommutable = Commutable; let Inst{11-4} = 0b00000000; let Inst{25} = 0; @@ -538,7 +583,7 @@ multiclass AI1_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, def rs : AsI1<opcod, (outs GPR:$dst), (ins GPR:$a, so_reg:$b), DPSoRegFrm, IIC_iALUsr, opc, "\t$dst, $a, $b", [(set GPR:$dst, (opnode GPR:$a, so_reg:$b))]>, - Requires<[IsARM, CarryDefIsUnused]> { + Requires<[IsARM]> { let Inst{25} = 0; } } @@ -549,16 +594,14 @@ multiclass AI1_adde_sube_s_irs<bits<4> opcod, string opc, PatFrag opnode, def Sri : AXI1<opcod, (outs GPR:$dst), (ins GPR:$a, so_imm:$b), DPFrm, IIC_iALUi, !strconcat(opc, "\t$dst, $a, $b"), [(set GPR:$dst, (opnode GPR:$a, so_imm:$b))]>, - Requires<[IsARM, CarryDefIsUsed]> { - let Defs = [CPSR]; + Requires<[IsARM]> { let Inst{20} = 1; let Inst{25} = 1; } def Srr : AXI1<opcod, (outs GPR:$dst), (ins GPR:$a, GPR:$b), DPFrm, IIC_iALUr, !strconcat(opc, "\t$dst, $a, $b"), [(set GPR:$dst, (opnode GPR:$a, GPR:$b))]>, - Requires<[IsARM, CarryDefIsUsed]> { - let Defs = [CPSR]; + Requires<[IsARM]> { let Inst{11-4} = 0b00000000; let Inst{20} = 1; let Inst{25} = 0; @@ -566,8 +609,7 @@ multiclass AI1_adde_sube_s_irs<bits<4> opcod, string opc, PatFrag opnode, def Srs : AXI1<opcod, (outs GPR:$dst), (ins GPR:$a, so_reg:$b), DPSoRegFrm, IIC_iALUsr, !strconcat(opc, "\t$dst, $a, $b"), [(set GPR:$dst, (opnode GPR:$a, so_reg:$b))]>, - Requires<[IsARM, CarryDefIsUsed]> { - let Defs = [CPSR]; + Requires<[IsARM]> { let Inst{20} = 1; let Inst{25} = 0; } @@ -593,13 +635,16 @@ PseudoInst<(outs), (ins cpinst_operand:$instid, cpinst_operand:$cpidx, i32imm:$size), NoItinerary, "${instid:label} ${cpidx:cpentry}", []>; -let Defs = [SP], Uses = [SP] in { +// FIXME: Marking these as hasSideEffects is necessary to prevent machine DCE +// from removing one half of the matched pairs. That breaks PEI, which assumes +// these will always be in pairs, and asserts if it finds otherwise. Better way? +let Defs = [SP], Uses = [SP], hasSideEffects = 1 in { def ADJCALLSTACKUP : PseudoInst<(outs), (ins i32imm:$amt1, i32imm:$amt2, pred:$p), NoItinerary, "@ ADJCALLSTACKUP $amt1", [(ARMcallseq_end timm:$amt1, timm:$amt2)]>; -def ADJCALLSTACKDOWN : +def ADJCALLSTACKDOWN : PseudoInst<(outs), (ins i32imm:$amt, pred:$p), NoItinerary, "@ ADJCALLSTACKDOWN $amt", [(ARMcallseq_start timm:$amt)]>; @@ -633,6 +678,14 @@ def WFI : AI<(outs), (ins), MiscFrm, NoItinerary, "wfi", "", let Inst{7-0} = 0b00000011; } +def SEL : AI<(outs GPR:$dst), (ins GPR:$a, GPR:$b), DPFrm, NoItinerary, "sel", + "\t$dst, $a, $b", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV6]> { + let Inst{27-20} = 0b01101000; + let Inst{7-4} = 0b1011; +} + def SEV : AI<(outs), (ins), MiscFrm, NoItinerary, "sev", "", [/* For disassembly only; pattern left blank */]>, Requires<[IsARM, HasV6T2]> { @@ -664,6 +717,34 @@ def CPS : AXI<(outs),(ins i32imm:$opt), MiscFrm, NoItinerary, "cps${opt:cps}", let Inst{5} = 0; } +// Preload signals the memory system of possible future data/instruction access. +// These are for disassembly only. +multiclass APreLoad<bit data, bit read, string opc> { + + def i : AXI<(outs), (ins GPR:$base, i32imm:$imm), MiscFrm, NoItinerary, + !strconcat(opc, "\t[$base, $imm]"), []> { + let Inst{31-26} = 0b111101; + let Inst{25} = 0; // 0 for immediate form + let Inst{24} = data; + let Inst{22} = read; + let Inst{21-20} = 0b01; + } + + def r : AXI<(outs), (ins addrmode2:$addr), MiscFrm, NoItinerary, + !strconcat(opc, "\t$addr"), []> { + let Inst{31-26} = 0b111101; + let Inst{25} = 1; // 1 for register form + let Inst{24} = data; + let Inst{22} = read; + let Inst{21-20} = 0b01; + let Inst{4} = 0; + } +} + +defm PLD : APreLoad<1, 1, "pld">; +defm PLDW : APreLoad<1, 0, "pldw">; +defm PLI : APreLoad<0, 1, "pli">; + def SETENDBE : AXI<(outs),(ins), MiscFrm, NoItinerary, "setend\tbe", [/* For disassembly only; pattern left blank */]>, Requires<[IsARM]> { @@ -761,7 +842,7 @@ def LEApcrelJT : AXI1<0x0, (outs GPR:$dst), "(${label}_${id}-(", "${:private}PCRELL${:uid}+8))\n"), !strconcat("${:private}PCRELL${:uid}:\n\t", - "add$p\t$dst, pc, #${:private}PCRELV${:uid}")), + "add$p\t$dst, pc, #${:private}PCRELV${:uid}")), []> { let Inst{25} = 1; } @@ -771,7 +852,7 @@ def LEApcrelJT : AXI1<0x0, (outs GPR:$dst), // let isReturn = 1, isTerminator = 1, isBarrier = 1 in - def BX_RET : AI<(outs), (ins), BrMiscFrm, IIC_Br, + def BX_RET : AI<(outs), (ins), BrMiscFrm, IIC_Br, "bx", "\tlr", [(ARMretflag)]> { let Inst{3-0} = 0b1110; let Inst{7-4} = 0b0001; @@ -828,9 +909,10 @@ let isCall = 1, } // ARMv4T - def BX : ABXIx2<(outs), (ins GPR:$func, variable_ops), + // Note: Restrict $func to the tGPR regclass to prevent it being in LR. + def BX : ABXIx2<(outs), (ins tGPR:$func, variable_ops), IIC_Br, "mov\tlr, pc\n\tbx\t$func", - [(ARMcall_nolink GPR:$func)]>, + [(ARMcall_nolink tGPR:$func)]>, Requires<[IsARM, IsNotDarwin]> { let Inst{7-4} = 0b0001; let Inst{19-8} = 0b111111111111; @@ -865,9 +947,10 @@ let isCall = 1, } // ARMv4T - def BXr9 : ABXIx2<(outs), (ins GPR:$func, variable_ops), + // Note: Restrict $func to the tGPR regclass to prevent it being in LR. + def BXr9 : ABXIx2<(outs), (ins tGPR:$func, variable_ops), IIC_Br, "mov\tlr, pc\n\tbx\t$func", - [(ARMcall_nolink GPR:$func)]>, Requires<[IsARM, IsDarwin]> { + [(ARMcall_nolink tGPR:$func)]>, Requires<[IsARM, IsDarwin]> { let Inst{7-4} = 0b0001; let Inst{19-8} = 0b111111111111; let Inst{27-20} = 0b00010010; @@ -917,7 +1000,7 @@ let isBranch = 1, isTerminator = 1 in { } // isBarrier = 1 // FIXME: should be able to write a pattern for ARMBrcond, but can't use - // a two-value operand where a dag node expects two operands. :( + // a two-value operand where a dag node expects two operands. :( def Bcc : ABI<0b1010, (outs), (ins brtarget:$target), IIC_Br, "b", "\t$target", [/*(ARMbrcond bb:$target, imm:$cc, CCR:$ccr)*/]>; @@ -931,25 +1014,61 @@ def BXJ : ABI<0b0001, (outs), (ins GPR:$func), NoItinerary, "bxj", "\t$func", let Inst{7-4} = 0b0010; } -// Supervisor call (software interrupt) -- for disassembly only +// Secure Monitor Call is a system instruction -- for disassembly only +def SMC : ABI<0b0001, (outs), (ins i32imm:$opt), NoItinerary, "smc", "\t$opt", + [/* For disassembly only; pattern left blank */]> { + let Inst{23-20} = 0b0110; + let Inst{7-4} = 0b0111; +} + +// Supervisor Call (Software Interrupt) -- for disassembly only let isCall = 1 in { def SVC : ABI<0b1111, (outs), (ins i32imm:$svc), IIC_Br, "svc", "\t$svc", [/* For disassembly only; pattern left blank */]>; } +// Store Return State is a system instruction -- for disassembly only +def SRSW : ABXI<{1,0,0,?}, (outs), (ins addrmode4:$addr, i32imm:$mode), + NoItinerary, "srs${addr:submode}\tsp!, $mode", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-28} = 0b1111; + let Inst{22-20} = 0b110; // W = 1 +} + +def SRS : ABXI<{1,0,0,?}, (outs), (ins addrmode4:$addr, i32imm:$mode), + NoItinerary, "srs${addr:submode}\tsp, $mode", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-28} = 0b1111; + let Inst{22-20} = 0b100; // W = 0 +} + +// Return From Exception is a system instruction -- for disassembly only +def RFEW : ABXI<{1,0,0,?}, (outs), (ins addrmode4:$addr, GPR:$base), + NoItinerary, "rfe${addr:submode}\t$base!", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-28} = 0b1111; + let Inst{22-20} = 0b011; // W = 1 +} + +def RFE : ABXI<{1,0,0,?}, (outs), (ins addrmode4:$addr, GPR:$base), + NoItinerary, "rfe${addr:submode}\t$base", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-28} = 0b1111; + let Inst{22-20} = 0b001; // W = 0 +} + //===----------------------------------------------------------------------===// // Load / store Instructions. // // Load -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def LDR : AI2ldw<(outs GPR:$dst), (ins addrmode2:$addr), LdFrm, IIC_iLoadr, "ldr", "\t$dst, $addr", [(set GPR:$dst, (load addrmode2:$addr))]>; // Special LDR for loads from non-pc-relative constpools. -let canFoldAsLoad = 1, mayLoad = 1, isReMaterializable = 1, - mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, mayLoad = 1, isReMaterializable = 1 in def LDRcp : AI2ldw<(outs GPR:$dst), (ins addrmode2:$addr), LdFrm, IIC_iLoadr, "ldr", "\t$dst, $addr", []>; @@ -958,7 +1077,7 @@ def LDRH : AI3ldh<(outs GPR:$dst), (ins addrmode3:$addr), LdMiscFrm, IIC_iLoadr, "ldrh", "\t$dst, $addr", [(set GPR:$dst, (zextloadi16 addrmode3:$addr))]>; -def LDRB : AI2ldb<(outs GPR:$dst), (ins addrmode2:$addr), LdFrm, +def LDRB : AI2ldb<(outs GPR:$dst), (ins addrmode2:$addr), LdFrm, IIC_iLoadr, "ldrb", "\t$dst, $addr", [(set GPR:$dst, (zextloadi8 addrmode2:$addr))]>; @@ -1017,9 +1136,22 @@ def LDRSB_PRE : AI3ldsbpr<(outs GPR:$dst, GPR:$base_wb), def LDRSB_POST: AI3ldsbpo<(outs GPR:$dst, GPR:$base_wb), (ins GPR:$base,am3offset:$offset), LdMiscFrm, IIC_iLoadru, "ldrsb", "\t$dst, [$base], $offset", "$base = $base_wb", []>; + +// For disassembly only +def LDRD_PRE : AI3lddpr<(outs GPR:$dst1, GPR:$dst2, GPR:$base_wb), + (ins addrmode3:$addr), LdMiscFrm, IIC_iLoadr, + "ldrd", "\t$dst1, $dst2, $addr!", "$addr.base = $base_wb", []>, + Requires<[IsARM, HasV5TE]>; + +// For disassembly only +def LDRD_POST : AI3lddpo<(outs GPR:$dst1, GPR:$dst2, GPR:$base_wb), + (ins GPR:$base,am3offset:$offset), LdMiscFrm, IIC_iLoadr, + "ldrd", "\t$dst1, $dst2, [$base], $offset", "$base = $base_wb", []>, + Requires<[IsARM, HasV5TE]>; + } -// LDRT and LDRBT are for disassembly only. +// LDRT, LDRBT, LDRSBT, LDRHT, LDRSHT are for disassembly only. def LDRT : AI2ldwpo<(outs GPR:$dst, GPR:$base_wb), (ins GPR:$base, am2offset:$offset), LdFrm, IIC_iLoadru, @@ -1028,8 +1160,26 @@ def LDRT : AI2ldwpo<(outs GPR:$dst, GPR:$base_wb), } def LDRBT : AI2ldbpo<(outs GPR:$dst, GPR:$base_wb), - (ins GPR:$base,am2offset:$offset), LdFrm, IIC_iLoadru, - "ldrb", "\t$dst, [$base], $offset", "$base = $base_wb", []> { + (ins GPR:$base,am2offset:$offset), LdFrm, IIC_iLoadru, + "ldrbt", "\t$dst, [$base], $offset", "$base = $base_wb", []> { + let Inst{21} = 1; // overwrite +} + +def LDRSBT : AI3ldsbpo<(outs GPR:$dst, GPR:$base_wb), + (ins GPR:$base,am2offset:$offset), LdMiscFrm, IIC_iLoadru, + "ldrsbt", "\t$dst, [$base], $offset", "$base = $base_wb", []> { + let Inst{21} = 1; // overwrite +} + +def LDRHT : AI3ldhpo<(outs GPR:$dst, GPR:$base_wb), + (ins GPR:$base, am3offset:$offset), LdMiscFrm, IIC_iLoadru, + "ldrht", "\t$dst, [$base], $offset", "$base = $base_wb", []> { + let Inst{21} = 1; // overwrite +} + +def LDRSHT : AI3ldshpo<(outs GPR:$dst, GPR:$base_wb), + (ins GPR:$base,am3offset:$offset), LdMiscFrm, IIC_iLoadru, + "ldrsht", "\t$dst, [$base], $offset", "$base = $base_wb", []> { let Inst{21} = 1; // overwrite } @@ -1039,8 +1189,8 @@ def STR : AI2stw<(outs), (ins GPR:$src, addrmode2:$addr), StFrm, IIC_iStorer, [(store GPR:$src, addrmode2:$addr)]>; // Stores with truncate -def STRH : AI3sth<(outs), (ins GPR:$src, addrmode3:$addr), StMiscFrm, IIC_iStorer, - "strh", "\t$src, $addr", +def STRH : AI3sth<(outs), (ins GPR:$src, addrmode3:$addr), StMiscFrm, + IIC_iStorer, "strh", "\t$src, $addr", [(truncstorei16 GPR:$src, addrmode3:$addr)]>; def STRB : AI2stb<(outs), (ins GPR:$src, addrmode2:$addr), StFrm, IIC_iStorer, @@ -1055,51 +1205,65 @@ def STRD : AI3std<(outs), (ins GPR:$src1, GPR:$src2, addrmode3:$addr), // Indexed stores def STR_PRE : AI2stwpr<(outs GPR:$base_wb), - (ins GPR:$src, GPR:$base, am2offset:$offset), + (ins GPR:$src, GPR:$base, am2offset:$offset), StFrm, IIC_iStoreru, "str", "\t$src, [$base, $offset]!", "$base = $base_wb", [(set GPR:$base_wb, (pre_store GPR:$src, GPR:$base, am2offset:$offset))]>; def STR_POST : AI2stwpo<(outs GPR:$base_wb), - (ins GPR:$src, GPR:$base,am2offset:$offset), + (ins GPR:$src, GPR:$base,am2offset:$offset), StFrm, IIC_iStoreru, "str", "\t$src, [$base], $offset", "$base = $base_wb", [(set GPR:$base_wb, (post_store GPR:$src, GPR:$base, am2offset:$offset))]>; def STRH_PRE : AI3sthpr<(outs GPR:$base_wb), - (ins GPR:$src, GPR:$base,am3offset:$offset), + (ins GPR:$src, GPR:$base,am3offset:$offset), StMiscFrm, IIC_iStoreru, "strh", "\t$src, [$base, $offset]!", "$base = $base_wb", [(set GPR:$base_wb, (pre_truncsti16 GPR:$src, GPR:$base,am3offset:$offset))]>; def STRH_POST: AI3sthpo<(outs GPR:$base_wb), - (ins GPR:$src, GPR:$base,am3offset:$offset), + (ins GPR:$src, GPR:$base,am3offset:$offset), StMiscFrm, IIC_iStoreru, "strh", "\t$src, [$base], $offset", "$base = $base_wb", [(set GPR:$base_wb, (post_truncsti16 GPR:$src, GPR:$base, am3offset:$offset))]>; def STRB_PRE : AI2stbpr<(outs GPR:$base_wb), - (ins GPR:$src, GPR:$base,am2offset:$offset), + (ins GPR:$src, GPR:$base,am2offset:$offset), StFrm, IIC_iStoreru, "strb", "\t$src, [$base, $offset]!", "$base = $base_wb", [(set GPR:$base_wb, (pre_truncsti8 GPR:$src, GPR:$base, am2offset:$offset))]>; def STRB_POST: AI2stbpo<(outs GPR:$base_wb), - (ins GPR:$src, GPR:$base,am2offset:$offset), + (ins GPR:$src, GPR:$base,am2offset:$offset), StFrm, IIC_iStoreru, "strb", "\t$src, [$base], $offset", "$base = $base_wb", [(set GPR:$base_wb, (post_truncsti8 GPR:$src, GPR:$base, am2offset:$offset))]>; -// STRT and STRBT are for disassembly only. +// For disassembly only +def STRD_PRE : AI3stdpr<(outs GPR:$base_wb), + (ins GPR:$src1, GPR:$src2, GPR:$base, am3offset:$offset), + StMiscFrm, IIC_iStoreru, + "strd", "\t$src1, $src2, [$base, $offset]!", + "$base = $base_wb", []>; + +// For disassembly only +def STRD_POST: AI3stdpo<(outs GPR:$base_wb), + (ins GPR:$src1, GPR:$src2, GPR:$base, am3offset:$offset), + StMiscFrm, IIC_iStoreru, + "strd", "\t$src1, $src2, [$base], $offset", + "$base = $base_wb", []>; + +// STRT, STRBT, and STRHT are for disassembly only. def STRT : AI2stwpo<(outs GPR:$base_wb), - (ins GPR:$src, GPR:$base,am2offset:$offset), + (ins GPR:$src, GPR:$base,am2offset:$offset), StFrm, IIC_iStoreru, "strt", "\t$src, [$base], $offset", "$base = $base_wb", [/* For disassembly only; pattern left blank */]> { @@ -1107,13 +1271,21 @@ def STRT : AI2stwpo<(outs GPR:$base_wb), } def STRBT : AI2stbpo<(outs GPR:$base_wb), - (ins GPR:$src, GPR:$base,am2offset:$offset), + (ins GPR:$src, GPR:$base,am2offset:$offset), StFrm, IIC_iStoreru, "strbt", "\t$src, [$base], $offset", "$base = $base_wb", [/* For disassembly only; pattern left blank */]> { let Inst{21} = 1; // overwrite } +def STRHT: AI3sthpo<(outs GPR:$base_wb), + (ins GPR:$src, GPR:$base,am3offset:$offset), + StMiscFrm, IIC_iStoreru, + "strht", "\t$src, [$base], $offset", "$base = $base_wb", + [/* For disassembly only; pattern left blank */]> { + let Inst{21} = 1; // overwrite +} + //===----------------------------------------------------------------------===// // Load / store multiple Instructions. // @@ -1141,7 +1313,7 @@ def MOVr : AsI1<0b1101, (outs GPR:$dst), (ins GPR:$src), DPFrm, IIC_iMOVr, let Inst{25} = 0; } -def MOVs : AsI1<0b1101, (outs GPR:$dst), (ins so_reg:$src), +def MOVs : AsI1<0b1101, (outs GPR:$dst), (ins so_reg:$src), DPSoRegFrm, IIC_iMOVsr, "mov", "\t$dst, $src", [(set GPR:$dst, so_reg:$src)]>, UnaryDP { let Inst{25} = 0; @@ -1154,7 +1326,7 @@ def MOVi : AsI1<0b1101, (outs GPR:$dst), (ins so_imm:$src), DPFrm, IIC_iMOVi, } let isReMaterializable = 1, isAsCheapAsAMove = 1 in -def MOVi16 : AI1<0b1000, (outs GPR:$dst), (ins i32imm:$src), +def MOVi16 : AI1<0b1000, (outs GPR:$dst), (ins i32imm:$src), DPFrm, IIC_iMOVi, "movw", "\t$dst, $src", [(set GPR:$dst, imm0_65535:$src)]>, @@ -1168,7 +1340,7 @@ def MOVTi16 : AI1<0b1010, (outs GPR:$dst), (ins GPR:$src, i32imm:$imm), DPFrm, IIC_iMOVi, "movt", "\t$dst, $imm", [(set GPR:$dst, - (or (and GPR:$src, 0xffff), + (or (and GPR:$src, 0xffff), lo16AllZero:$imm))]>, UnaryDP, Requires<[IsARM, HasV6T2]> { let Inst{20} = 0; @@ -1187,7 +1359,7 @@ def MOVrx : AsI1<0b1101, (outs GPR:$dst), (ins GPR:$src), Pseudo, IIC_iMOVsi, // due to flag operands. let Defs = [CPSR] in { -def MOVsrl_flag : AI1<0b1101, (outs GPR:$dst), (ins GPR:$src), Pseudo, +def MOVsrl_flag : AI1<0b1101, (outs GPR:$dst), (ins GPR:$src), Pseudo, IIC_iMOVsi, "movs", "\t$dst, $src, lsr #1", [(set GPR:$dst, (ARMsrl_flag GPR:$src))]>, UnaryDP; def MOVsra_flag : AI1<0b1101, (outs GPR:$dst), (ins GPR:$src), Pseudo, @@ -1211,7 +1383,11 @@ defm SXTAB : AI_bin_rrot<0b01101010, defm SXTAH : AI_bin_rrot<0b01101011, "sxtah", BinOpFrag<(add node:$LHS, (sext_inreg node:$RHS,i16))>>; -// TODO: SXT(A){B|H}16 +// For disassembly only +defm SXTB16 : AI_unary_rrot_np<0b01101000, "sxtb16">; + +// For disassembly only +defm SXTAB16 : AI_bin_rrot_np<0b01101000, "sxtab16">; // Zero extenders @@ -1235,9 +1411,9 @@ defm UXTAH : AI_bin_rrot<0b01101111, "uxtah", } // This isn't safe in general, the add is two 16-bit units, not a 32-bit add. -//defm UXTAB16 : xxx<"uxtab16", 0xff00ff>; +// For disassembly only +defm UXTAB16 : AI_bin_rrot_np<0b01101100, "uxtab16">; -// TODO: UXT(A){B|H}16 def SBFX : I<(outs GPR:$dst), (ins GPR:$src, imm0_31:$lsb, imm0_31:$width), @@ -1273,13 +1449,13 @@ defm SUBS : AI1_bin_s_irs<0b0010, "subs", BinOpFrag<(subc node:$LHS, node:$RHS)>>; defm ADC : AI1_adde_sube_irs<0b0101, "adc", - BinOpFrag<(adde node:$LHS, node:$RHS)>, 1>; + BinOpFrag<(adde_dead_carry node:$LHS, node:$RHS)>, 1>; defm SBC : AI1_adde_sube_irs<0b0110, "sbc", - BinOpFrag<(sube node:$LHS, node:$RHS)>>; + BinOpFrag<(sube_dead_carry node:$LHS, node:$RHS)>>; defm ADCS : AI1_adde_sube_s_irs<0b0101, "adcs", - BinOpFrag<(adde node:$LHS, node:$RHS)>, 1>; + BinOpFrag<(adde_live_carry node:$LHS, node:$RHS)>, 1>; defm SBCS : AI1_adde_sube_s_irs<0b0110, "sbcs", - BinOpFrag<(sube node:$LHS, node:$RHS)>>; + BinOpFrag<(sube_live_carry node:$LHS, node:$RHS) >>; // These don't define reg/reg forms, because they are handled above. def RSBri : AsI1<0b0011, (outs GPR:$dst), (ins GPR:$a, so_imm:$b), DPFrm, @@ -1313,14 +1489,14 @@ def RSBSrs : AI1<0b0011, (outs GPR:$dst), (ins GPR:$a, so_reg:$b), DPSoRegFrm, let Uses = [CPSR] in { def RSCri : AsI1<0b0111, (outs GPR:$dst), (ins GPR:$a, so_imm:$b), DPFrm, IIC_iALUi, "rsc", "\t$dst, $a, $b", - [(set GPR:$dst, (sube so_imm:$b, GPR:$a))]>, - Requires<[IsARM, CarryDefIsUnused]> { + [(set GPR:$dst, (sube_dead_carry so_imm:$b, GPR:$a))]>, + Requires<[IsARM]> { let Inst{25} = 1; } def RSCrs : AsI1<0b0111, (outs GPR:$dst), (ins GPR:$a, so_reg:$b), DPSoRegFrm, IIC_iALUsr, "rsc", "\t$dst, $a, $b", - [(set GPR:$dst, (sube so_reg:$b, GPR:$a))]>, - Requires<[IsARM, CarryDefIsUnused]> { + [(set GPR:$dst, (sube_dead_carry so_reg:$b, GPR:$a))]>, + Requires<[IsARM]> { let Inst{25} = 0; } } @@ -1329,15 +1505,15 @@ def RSCrs : AsI1<0b0111, (outs GPR:$dst), (ins GPR:$a, so_reg:$b), let Defs = [CPSR], Uses = [CPSR] in { def RSCSri : AXI1<0b0111, (outs GPR:$dst), (ins GPR:$a, so_imm:$b), DPFrm, IIC_iALUi, "rscs\t$dst, $a, $b", - [(set GPR:$dst, (sube so_imm:$b, GPR:$a))]>, - Requires<[IsARM, CarryDefIsUnused]> { + [(set GPR:$dst, (sube_dead_carry so_imm:$b, GPR:$a))]>, + Requires<[IsARM]> { let Inst{20} = 1; let Inst{25} = 1; } def RSCSrs : AXI1<0b0111, (outs GPR:$dst), (ins GPR:$a, so_reg:$b), DPSoRegFrm, IIC_iALUsr, "rscs\t$dst, $a, $b", - [(set GPR:$dst, (sube so_reg:$b, GPR:$a))]>, - Requires<[IsARM, CarryDefIsUnused]> { + [(set GPR:$dst, (sube_dead_carry so_reg:$b, GPR:$a))]>, + Requires<[IsARM]> { let Inst{20} = 1; let Inst{25} = 0; } @@ -1358,10 +1534,9 @@ def : ARMPat<(add GPR:$src, so_imm_neg:$imm), // (mul X, 2^n+1) -> (add (X << n), X) // (mul X, 2^n-1) -> (rsb X, (X << n)) -// Saturating adds/subtracts -- for disassembly only - +// ARM Arithmetic Instruction -- for disassembly only // GPR:$dst = GPR:$a op GPR:$b -class AQI<bits<8> op27_20, bits<4> op7_4, string opc> +class AAI<bits<8> op27_20, bits<4> op7_4, string opc> : AI<(outs GPR:$dst), (ins GPR:$a, GPR:$b), DPFrm, IIC_iALUr, opc, "\t$dst, $a, $b", [/* For disassembly only; pattern left blank */]> { @@ -1369,22 +1544,116 @@ class AQI<bits<8> op27_20, bits<4> op7_4, string opc> let Inst{7-4} = op7_4; } -def QADD : AQI<0b00010000, 0b0101, "qadd">; -def QADD16 : AQI<0b01100010, 0b0001, "qadd16">; -def QADD8 : AQI<0b01100010, 0b1001, "qadd8">; -def QASX : AQI<0b01100010, 0b0011, "qasx">; -def QDADD : AQI<0b00010100, 0b0101, "qdadd">; -def QDSUB : AQI<0b00010110, 0b0101, "qdsub">; -def QSAX : AQI<0b01100010, 0b0101, "qsax">; -def QSUB : AQI<0b00010010, 0b0101, "qsub">; -def QSUB16 : AQI<0b01100010, 0b0111, "qsub16">; -def QSUB8 : AQI<0b01100010, 0b1111, "qsub8">; -def UQADD16 : AQI<0b01100110, 0b0001, "uqadd16">; -def UQADD8 : AQI<0b01100110, 0b1001, "uqadd8">; -def UQASX : AQI<0b01100110, 0b0011, "uqasx">; -def UQSAX : AQI<0b01100110, 0b0101, "uqsax">; -def UQSUB16 : AQI<0b01100110, 0b0111, "uqsub16">; -def UQSUB8 : AQI<0b01100110, 0b1111, "uqsub8">; +// Saturating add/subtract -- for disassembly only + +def QADD : AAI<0b00010000, 0b0101, "qadd">; +def QADD16 : AAI<0b01100010, 0b0001, "qadd16">; +def QADD8 : AAI<0b01100010, 0b1001, "qadd8">; +def QASX : AAI<0b01100010, 0b0011, "qasx">; +def QDADD : AAI<0b00010100, 0b0101, "qdadd">; +def QDSUB : AAI<0b00010110, 0b0101, "qdsub">; +def QSAX : AAI<0b01100010, 0b0101, "qsax">; +def QSUB : AAI<0b00010010, 0b0101, "qsub">; +def QSUB16 : AAI<0b01100010, 0b0111, "qsub16">; +def QSUB8 : AAI<0b01100010, 0b1111, "qsub8">; +def UQADD16 : AAI<0b01100110, 0b0001, "uqadd16">; +def UQADD8 : AAI<0b01100110, 0b1001, "uqadd8">; +def UQASX : AAI<0b01100110, 0b0011, "uqasx">; +def UQSAX : AAI<0b01100110, 0b0101, "uqsax">; +def UQSUB16 : AAI<0b01100110, 0b0111, "uqsub16">; +def UQSUB8 : AAI<0b01100110, 0b1111, "uqsub8">; + +// Signed/Unsigned add/subtract -- for disassembly only + +def SASX : AAI<0b01100001, 0b0011, "sasx">; +def SADD16 : AAI<0b01100001, 0b0001, "sadd16">; +def SADD8 : AAI<0b01100001, 0b1001, "sadd8">; +def SSAX : AAI<0b01100001, 0b0101, "ssax">; +def SSUB16 : AAI<0b01100001, 0b0111, "ssub16">; +def SSUB8 : AAI<0b01100001, 0b1111, "ssub8">; +def UASX : AAI<0b01100101, 0b0011, "uasx">; +def UADD16 : AAI<0b01100101, 0b0001, "uadd16">; +def UADD8 : AAI<0b01100101, 0b1001, "uadd8">; +def USAX : AAI<0b01100101, 0b0101, "usax">; +def USUB16 : AAI<0b01100101, 0b0111, "usub16">; +def USUB8 : AAI<0b01100101, 0b1111, "usub8">; + +// Signed/Unsigned halving add/subtract -- for disassembly only + +def SHASX : AAI<0b01100011, 0b0011, "shasx">; +def SHADD16 : AAI<0b01100011, 0b0001, "shadd16">; +def SHADD8 : AAI<0b01100011, 0b1001, "shadd8">; +def SHSAX : AAI<0b01100011, 0b0101, "shsax">; +def SHSUB16 : AAI<0b01100011, 0b0111, "shsub16">; +def SHSUB8 : AAI<0b01100011, 0b1111, "shsub8">; +def UHASX : AAI<0b01100111, 0b0011, "uhasx">; +def UHADD16 : AAI<0b01100111, 0b0001, "uhadd16">; +def UHADD8 : AAI<0b01100111, 0b1001, "uhadd8">; +def UHSAX : AAI<0b01100111, 0b0101, "uhsax">; +def UHSUB16 : AAI<0b01100111, 0b0111, "uhsub16">; +def UHSUB8 : AAI<0b01100111, 0b1111, "uhsub8">; + +// Unsigned Sum of Absolute Differences [and Accumulate] -- for disassembly only + +def USAD8 : AI<(outs GPR:$dst), (ins GPR:$a, GPR:$b), + MulFrm /* for convenience */, NoItinerary, "usad8", + "\t$dst, $a, $b", []>, + Requires<[IsARM, HasV6]> { + let Inst{27-20} = 0b01111000; + let Inst{15-12} = 0b1111; + let Inst{7-4} = 0b0001; +} +def USADA8 : AI<(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$acc), + MulFrm /* for convenience */, NoItinerary, "usada8", + "\t$dst, $a, $b, $acc", []>, + Requires<[IsARM, HasV6]> { + let Inst{27-20} = 0b01111000; + let Inst{7-4} = 0b0001; +} + +// Signed/Unsigned saturate -- for disassembly only + +def SSATlsl : AI<(outs GPR:$dst), (ins i32imm:$bit_pos, GPR:$a, i32imm:$shamt), + DPFrm, NoItinerary, "ssat", "\t$dst, $bit_pos, $a, lsl $shamt", + [/* For disassembly only; pattern left blank */]> { + let Inst{27-21} = 0b0110101; + let Inst{6-4} = 0b001; +} + +def SSATasr : AI<(outs GPR:$dst), (ins i32imm:$bit_pos, GPR:$a, i32imm:$shamt), + DPFrm, NoItinerary, "ssat", "\t$dst, $bit_pos, $a, asr $shamt", + [/* For disassembly only; pattern left blank */]> { + let Inst{27-21} = 0b0110101; + let Inst{6-4} = 0b101; +} + +def SSAT16 : AI<(outs GPR:$dst), (ins i32imm:$bit_pos, GPR:$a), DPFrm, + NoItinerary, "ssat16", "\t$dst, $bit_pos, $a", + [/* For disassembly only; pattern left blank */]> { + let Inst{27-20} = 0b01101010; + let Inst{7-4} = 0b0011; +} + +def USATlsl : AI<(outs GPR:$dst), (ins i32imm:$bit_pos, GPR:$a, i32imm:$shamt), + DPFrm, NoItinerary, "usat", "\t$dst, $bit_pos, $a, lsl $shamt", + [/* For disassembly only; pattern left blank */]> { + let Inst{27-21} = 0b0110111; + let Inst{6-4} = 0b001; +} + +def USATasr : AI<(outs GPR:$dst), (ins i32imm:$bit_pos, GPR:$a, i32imm:$shamt), + DPFrm, NoItinerary, "usat", "\t$dst, $bit_pos, $a, asr $shamt", + [/* For disassembly only; pattern left blank */]> { + let Inst{27-21} = 0b0110111; + let Inst{6-4} = 0b101; +} + +def USAT16 : AI<(outs GPR:$dst), (ins i32imm:$bit_pos, GPR:$a), DPFrm, + NoItinerary, "usat16", "\t$dst, $bit_pos, $a", + [/* For disassembly only; pattern left blank */]> { + let Inst{27-20} = 0b01101110; + let Inst{7-4} = 0b0011; +} //===----------------------------------------------------------------------===// // Bitwise Instructions. @@ -1408,6 +1677,17 @@ def BFC : I<(outs GPR:$dst), (ins GPR:$src, bf_inv_mask_imm:$imm), let Inst{6-0} = 0b0011111; } +// A8.6.18 BFI - Bitfield insert (Encoding A1) +// Added for disassembler with the pattern field purposely left blank. +def BFI : I<(outs GPR:$dst), (ins GPR:$src, bf_inv_mask_imm:$imm), + AddrMode1, Size4Bytes, IndexModeNone, DPFrm, IIC_iUNAsi, + "bfi", "\t$dst, $src, $imm", "", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV6T2]> { + let Inst{27-21} = 0b0111110; + let Inst{6-4} = 0b001; // Rn: Inst{3-0} != 15 +} + def MVNr : AsI1<0b1111, (outs GPR:$dst), (ins GPR:$src), DPFrm, IIC_iMOVr, "mvn", "\t$dst, $src", [(set GPR:$dst, (not GPR:$src))]>, UnaryDP { @@ -1420,7 +1700,7 @@ def MVNs : AsI1<0b1111, (outs GPR:$dst), (ins so_reg:$src), DPSoRegFrm, let Inst{25} = 0; } let isReMaterializable = 1, isAsCheapAsAMove = 1 in -def MVNi : AsI1<0b1111, (outs GPR:$dst), (ins so_imm:$imm), DPFrm, +def MVNi : AsI1<0b1111, (outs GPR:$dst), (ins so_imm:$imm), DPFrm, IIC_iMOVi, "mvn", "\t$dst, $imm", [(set GPR:$dst, so_imm_not:$imm)]>,UnaryDP { let Inst{25} = 1; @@ -1483,6 +1763,14 @@ def SMMUL : AMul2I <0b0111010, (outs GPR:$dst), (ins GPR:$a, GPR:$b), let Inst{15-12} = 0b1111; } +def SMMULR : AMul2I <0b0111010, (outs GPR:$dst), (ins GPR:$a, GPR:$b), + IIC_iMUL32, "smmulr", "\t$dst, $a, $b", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV6]> { + let Inst{7-4} = 0b0011; // R = 1 + let Inst{15-12} = 0b1111; +} + def SMMLA : AMul2I <0b0111010, (outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), IIC_iMAC32, "smmla", "\t$dst, $a, $b, $c", [(set GPR:$dst, (add (mulhs GPR:$a, GPR:$b), GPR:$c))]>, @@ -1490,6 +1778,12 @@ def SMMLA : AMul2I <0b0111010, (outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), let Inst{7-4} = 0b0001; } +def SMMLAR : AMul2I <0b0111010, (outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), + IIC_iMAC32, "smmlar", "\t$dst, $a, $b, $c", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV6]> { + let Inst{7-4} = 0b0011; // R = 1 +} def SMMLS : AMul2I <0b0111010, (outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), IIC_iMAC32, "smmls", "\t$dst, $a, $b, $c", @@ -1498,6 +1792,13 @@ def SMMLS : AMul2I <0b0111010, (outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), let Inst{7-4} = 0b1101; } +def SMMLSR : AMul2I <0b0111010, (outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), + IIC_iMAC32, "smmlsr", "\t$dst, $a, $b, $c", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV6]> { + let Inst{7-4} = 0b1111; // R = 1 +} + multiclass AI_smul<string opc, PatFrag opnode> { def BB : AMulxyI<0b0001011, (outs GPR:$dst), (ins GPR:$a, GPR:$b), IIC_iMUL32, !strconcat(opc, "bb"), "\t$dst, $a, $b", @@ -1569,7 +1870,7 @@ multiclass AI_smla<string opc, PatFrag opnode> { def BT : AMulxyI<0b0001000, (outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$acc), IIC_iMAC16, !strconcat(opc, "bt"), "\t$dst, $a, $b, $acc", [(set GPR:$dst, (add GPR:$acc, (opnode (sext_inreg GPR:$a, i16), - (sra GPR:$b, (i32 16)))))]>, + (sra GPR:$b, (i32 16)))))]>, Requires<[IsARM, HasV5TE]> { let Inst{5} = 0; let Inst{6} = 1; @@ -1648,7 +1949,54 @@ def SMLALTT : AMulxyI<0b0001010,(outs GPR:$ldst,GPR:$hdst),(ins GPR:$a,GPR:$b), let Inst{6} = 1; } -// TODO: Dual halfword multiple: SMUAD, SMUSD, SMLAD, SMLSD, SMLALD, SMLSLD +// Helper class for AI_smld -- for disassembly only +class AMulDualI<bit long, bit sub, bit swap, dag oops, dag iops, + InstrItinClass itin, string opc, string asm> + : AI<oops, iops, MulFrm, itin, opc, asm, []>, Requires<[IsARM, HasV6]> { + let Inst{4} = 1; + let Inst{5} = swap; + let Inst{6} = sub; + let Inst{7} = 0; + let Inst{21-20} = 0b00; + let Inst{22} = long; + let Inst{27-23} = 0b01110; +} + +multiclass AI_smld<bit sub, string opc> { + + def D : AMulDualI<0, sub, 0, (outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$acc), + NoItinerary, !strconcat(opc, "d"), "\t$dst, $a, $b, $acc">; + + def DX : AMulDualI<0, sub, 1, (outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$acc), + NoItinerary, !strconcat(opc, "dx"), "\t$dst, $a, $b, $acc">; + + def LD : AMulDualI<1, sub, 0, (outs GPR:$ldst,GPR:$hdst), (ins GPR:$a,GPR:$b), + NoItinerary, !strconcat(opc, "ld"), "\t$ldst, $hdst, $a, $b">; + + def LDX : AMulDualI<1, sub, 1, (outs GPR:$ldst,GPR:$hdst),(ins GPR:$a,GPR:$b), + NoItinerary, !strconcat(opc, "ldx"),"\t$ldst, $hdst, $a, $b">; + +} + +defm SMLA : AI_smld<0, "smla">; +defm SMLS : AI_smld<1, "smls">; + +multiclass AI_sdml<bit sub, string opc> { + + def D : AMulDualI<0, sub, 0, (outs GPR:$dst), (ins GPR:$a, GPR:$b), + NoItinerary, !strconcat(opc, "d"), "\t$dst, $a, $b"> { + let Inst{15-12} = 0b1111; + } + + def DX : AMulDualI<0, sub, 1, (outs GPR:$dst), (ins GPR:$a, GPR:$b), + NoItinerary, !strconcat(opc, "dx"), "\t$dst, $a, $b"> { + let Inst{15-12} = 0b1111; + } + +} + +defm SMUA : AI_sdml<0, "smua">; +defm SMUS : AI_sdml<1, "smus">; //===----------------------------------------------------------------------===// // Misc. Arithmetic Instructions. @@ -1706,7 +2054,7 @@ def REVSH : AMiscA1I<0b01101111, (outs GPR:$dst), (ins GPR:$src), IIC_iUNAr, def PKHBT : AMiscA1I<0b01101000, (outs GPR:$dst), (ins GPR:$src1, GPR:$src2, i32imm:$shamt), - IIC_iALUsi, "pkhbt", "\t$dst, $src1, $src2, LSL $shamt", + IIC_iALUsi, "pkhbt", "\t$dst, $src1, $src2, lsl $shamt", [(set GPR:$dst, (or (and GPR:$src1, 0xFFFF), (and (shl GPR:$src2, (i32 imm:$shamt)), 0xFFFF0000)))]>, @@ -1723,7 +2071,7 @@ def : ARMV6Pat<(or (and GPR:$src1, 0xFFFF), (shl GPR:$src2, imm16_31:$shamt)), def PKHTB : AMiscA1I<0b01101000, (outs GPR:$dst), (ins GPR:$src1, GPR:$src2, i32imm:$shamt), - IIC_iALUsi, "pkhtb", "\t$dst, $src1, $src2, ASR $shamt", + IIC_iALUsi, "pkhtb", "\t$dst, $src1, $src2, asr $shamt", [(set GPR:$dst, (or (and GPR:$src1, 0xFFFF0000), (and (sra GPR:$src2, imm16_31:$shamt), 0xFFFF)))]>, Requires<[IsARM, HasV6]> { @@ -1769,7 +2117,7 @@ def : ARMPat<(ARMcmpZ GPR:$src, so_imm_neg:$imm), // Conditional moves // FIXME: should be able to write a pattern for ARMcmov, but can't use -// a two-value operand where a dag node expects two operands. :( +// a two-value operand where a dag node expects two operands. :( def MOVCCr : AI1<0b1101, (outs GPR:$dst), (ins GPR:$false, GPR:$true), DPFrm, IIC_iCMOVr, "mov", "\t$dst, $true", [/*(set GPR:$dst, (ARMcmov GPR:$false, GPR:$true, imm:$cc, CCR:$ccr))*/]>, @@ -1807,6 +2155,7 @@ def Int_MemBarrierV7 : AInoP<(outs), (ins), Requires<[IsARM, HasV7]> { let Inst{31-4} = 0xf57ff05; // FIXME: add support for options other than a full system DMB + // See DMB disassembly-only variants below. let Inst{3-0} = 0b1111; } @@ -1817,6 +2166,7 @@ def Int_SyncBarrierV7 : AInoP<(outs), (ins), Requires<[IsARM, HasV7]> { let Inst{31-4} = 0xf57ff04; // FIXME: add support for options other than a full system DSB + // See DSB disassembly-only variants below. let Inst{3-0} = 0b1111; } @@ -1839,6 +2189,64 @@ def Int_SyncBarrierV6 : AInoP<(outs), (ins GPR:$zero), } } +// Helper class for multiclass MemB -- for disassembly only +class AMBI<string opc, string asm> + : AInoP<(outs), (ins), MiscFrm, NoItinerary, opc, asm, + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV7]> { + let Inst{31-20} = 0xf57; +} + +multiclass MemB<bits<4> op7_4, string opc> { + + def st : AMBI<opc, "\tst"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b1110; + } + + def ish : AMBI<opc, "\tish"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b1011; + } + + def ishst : AMBI<opc, "\tishst"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b1010; + } + + def nsh : AMBI<opc, "\tnsh"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b0111; + } + + def nshst : AMBI<opc, "\tnshst"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b0110; + } + + def osh : AMBI<opc, "\tosh"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b0011; + } + + def oshst : AMBI<opc, "\toshst"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b0010; + } +} + +// These DMB variants are for disassembly only. +defm DMB : MemB<0b0101, "dmb">; + +// These DSB variants are for disassembly only. +defm DSB : MemB<0b0100, "dsb">; + +// ISB has only full system option -- for disassembly only +def ISBsy : AMBI<"isb", ""> { + let Inst{7-4} = 0b0110; + let Inst{3-0} = 0b1111; +} + let usesCustomInserter = 1 in { let Uses = [CPSR] in { def ATOMIC_LOAD_ADD_I8 : PseudoInst< @@ -1978,6 +2386,14 @@ def STREXD : AIstrex<0b01, (outs GPR:$success), []>; } +// Clear-Exclusive is for disassembly only. +def CLREX : AXI<(outs), (ins), MiscFrm, NoItinerary, "clrex", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV7]> { + let Inst{31-20} = 0xf57; + let Inst{7-4} = 0b0001; +} + // SWP/SWPB are deprecated in V6/V7 and for disassembly only. let mayLoad = 1 in { def SWP : AI<(outs GPR:$dst), (ins GPR:$src, GPR:$ptr), LdStExFrm, NoItinerary, @@ -2049,7 +2465,7 @@ let Defs = // Two piece so_imms. let isReMaterializable = 1 in -def MOVi2pieces : AI1x2<(outs GPR:$dst), (ins so_imm2part:$src), +def MOVi2pieces : AI1x2<(outs GPR:$dst), (ins so_imm2part:$src), Pseudo, IIC_iMOVi, "mov", "\t$dst, $src", [(set GPR:$dst, so_imm2part:$src)]>, @@ -2074,7 +2490,7 @@ def : ARMPat<(add GPR:$LHS, so_neg_imm2part:$RHS), // FIXME: Remove this when we can do generalized remat. let isReMaterializable = 1 in def MOVi32imm : AI1x2<(outs GPR:$dst), (ins i32imm:$src), Pseudo, IIC_iMOVi, - "movw", "\t$dst, ${src:lo16}\n\tmovt${p}\t$dst, ${src:hi16}", + "movw", "\t$dst, ${src:lo16}\n\tmovt${p}\t$dst, ${src:hi16}", [(set GPR:$dst, (i32 imm:$src))]>, Requires<[IsARM, HasV6T2]>; @@ -2201,6 +2617,102 @@ def CDP2 : ABXI<0b1110, (outs), (ins nohash_imm:$cop, i32imm:$opc1, let Inst{4} = 0; } +class ACI<dag oops, dag iops, string opc, string asm> + : I<oops, iops, AddrModeNone, Size4Bytes, IndexModeNone, BrFrm, NoItinerary, + opc, asm, "", [/* For disassembly only; pattern left blank */]> { + let Inst{27-25} = 0b110; +} + +multiclass LdStCop<bits<4> op31_28, bit load, string opc> { + + def _OFFSET : ACI<(outs), + (ins nohash_imm:$cop, nohash_imm:$CRd, addrmode2:$addr), + opc, "\tp$cop, cr$CRd, $addr"> { + let Inst{31-28} = op31_28; + let Inst{24} = 1; // P = 1 + let Inst{21} = 0; // W = 0 + let Inst{22} = 0; // D = 0 + let Inst{20} = load; + } + + def _PRE : ACI<(outs), + (ins nohash_imm:$cop, nohash_imm:$CRd, addrmode2:$addr), + opc, "\tp$cop, cr$CRd, $addr!"> { + let Inst{31-28} = op31_28; + let Inst{24} = 1; // P = 1 + let Inst{21} = 1; // W = 1 + let Inst{22} = 0; // D = 0 + let Inst{20} = load; + } + + def _POST : ACI<(outs), + (ins nohash_imm:$cop, nohash_imm:$CRd, GPR:$base, am2offset:$offset), + opc, "\tp$cop, cr$CRd, [$base], $offset"> { + let Inst{31-28} = op31_28; + let Inst{24} = 0; // P = 0 + let Inst{21} = 1; // W = 1 + let Inst{22} = 0; // D = 0 + let Inst{20} = load; + } + + def _OPTION : ACI<(outs), + (ins nohash_imm:$cop, nohash_imm:$CRd, GPR:$base, i32imm:$option), + opc, "\tp$cop, cr$CRd, [$base], $option"> { + let Inst{31-28} = op31_28; + let Inst{24} = 0; // P = 0 + let Inst{23} = 1; // U = 1 + let Inst{21} = 0; // W = 0 + let Inst{22} = 0; // D = 0 + let Inst{20} = load; + } + + def L_OFFSET : ACI<(outs), + (ins nohash_imm:$cop, nohash_imm:$CRd, addrmode2:$addr), + opc, "l\tp$cop, cr$CRd, $addr"> { + let Inst{31-28} = op31_28; + let Inst{24} = 1; // P = 1 + let Inst{21} = 0; // W = 0 + let Inst{22} = 1; // D = 1 + let Inst{20} = load; + } + + def L_PRE : ACI<(outs), + (ins nohash_imm:$cop, nohash_imm:$CRd, addrmode2:$addr), + opc, "l\tp$cop, cr$CRd, $addr!"> { + let Inst{31-28} = op31_28; + let Inst{24} = 1; // P = 1 + let Inst{21} = 1; // W = 1 + let Inst{22} = 1; // D = 1 + let Inst{20} = load; + } + + def L_POST : ACI<(outs), + (ins nohash_imm:$cop, nohash_imm:$CRd, GPR:$base, am2offset:$offset), + opc, "l\tp$cop, cr$CRd, [$base], $offset"> { + let Inst{31-28} = op31_28; + let Inst{24} = 0; // P = 0 + let Inst{21} = 1; // W = 1 + let Inst{22} = 1; // D = 1 + let Inst{20} = load; + } + + def L_OPTION : ACI<(outs), + (ins nohash_imm:$cop, nohash_imm:$CRd, GPR:$base, nohash_imm:$option), + opc, "l\tp$cop, cr$CRd, [$base], $option"> { + let Inst{31-28} = op31_28; + let Inst{24} = 0; // P = 0 + let Inst{23} = 1; // U = 1 + let Inst{21} = 0; // W = 0 + let Inst{22} = 1; // D = 1 + let Inst{20} = load; + } +} + +defm LDC : LdStCop<{?,?,?,?}, 1, "ldc">; +defm LDC2 : LdStCop<0b1111, 1, "ldc2">; +defm STC : LdStCop<{?,?,?,?}, 0, "stc">; +defm STC2 : LdStCop<0b1111, 0, "stc2">; + def MCR : ABI<0b1110, (outs), (ins nohash_imm:$cop, i32imm:$opc1, GPR:$Rt, nohash_imm:$CRn, nohash_imm:$CRm, i32imm:$opc2), NoItinerary, "mcr", "\tp$cop, $opc1, $Rt, cr$CRn, cr$CRm, $opc2", @@ -2282,14 +2794,28 @@ def MRSsys : ABI<0b0001,(outs GPR:$dst),(ins), NoItinerary,"mrs","\t$dst, spsr", } // FIXME: mask is ignored for the time being. -def MSR : ABI<0b0001,(outs),(ins GPR:$src), NoItinerary, "mrs", "\tcpsr, $src", +def MSR : ABI<0b0001,(outs),(ins GPR:$src), NoItinerary, "msr", "\tcpsr, $src", [/* For disassembly only; pattern left blank */]> { let Inst{23-20} = 0b0010; let Inst{7-4} = 0b0000; } // FIXME: mask is ignored for the time being. -def MSRsys : ABI<0b0001,(outs),(ins GPR:$src),NoItinerary,"mrs","\tspsr, $src", +def MSRi : ABI<0b0011,(outs),(ins so_imm:$a), NoItinerary, "msr", "\tcpsr, $a", + [/* For disassembly only; pattern left blank */]> { + let Inst{23-20} = 0b0010; + let Inst{7-4} = 0b0000; +} + +// FIXME: mask is ignored for the time being. +def MSRsys : ABI<0b0001,(outs),(ins GPR:$src),NoItinerary,"msr","\tspsr, $src", + [/* For disassembly only; pattern left blank */]> { + let Inst{23-20} = 0b0110; + let Inst{7-4} = 0b0000; +} + +// FIXME: mask is ignored for the time being. +def MSRsysi : ABI<0b0011,(outs),(ins so_imm:$a),NoItinerary,"msr","\tspsr, $a", [/* For disassembly only; pattern left blank */]> { let Inst{23-20} = 0b0110; let Inst{7-4} = 0b0000; diff --git a/lib/Target/ARM/ARMInstrNEON.td b/lib/Target/ARM/ARMInstrNEON.td index e2be7ba..3aa0810 100644 --- a/lib/Target/ARM/ARMInstrNEON.td +++ b/lib/Target/ARM/ARMInstrNEON.td @@ -83,11 +83,17 @@ def NEONvrev32 : SDNode<"ARMISD::VREV32", SDTARMVSHUF>; def NEONvrev16 : SDNode<"ARMISD::VREV16", SDTARMVSHUF>; def SDTARMVSHUF2 : SDTypeProfile<2, 2, [SDTCisVec<0>, SDTCisSameAs<0, 1>, - SDTCisSameAs<0, 2>, SDTCisSameAs<0, 3>]>; + SDTCisSameAs<0, 2>, + SDTCisSameAs<0, 3>]>; def NEONzip : SDNode<"ARMISD::VZIP", SDTARMVSHUF2>; def NEONuzp : SDNode<"ARMISD::VUZP", SDTARMVSHUF2>; def NEONtrn : SDNode<"ARMISD::VTRN", SDTARMVSHUF2>; +def SDTARMFMAX : SDTypeProfile<1, 2, [SDTCisVT<0, f32>, SDTCisSameAs<0, 1>, + SDTCisSameAs<0, 2>]>; +def NEONfmax : SDNode<"ARMISD::FMAX", SDTARMFMAX>; +def NEONfmin : SDNode<"ARMISD::FMIN", SDTARMFMAX>; + //===----------------------------------------------------------------------===// // NEON operand definitions //===----------------------------------------------------------------------===// @@ -123,9 +129,7 @@ def h64imm : Operand<i64> { let mayLoad = 1, hasExtraDefRegAllocReq = 1 in { def VLDMD : NI<(outs), (ins addrmode_neonldstm:$addr, reglist:$dst1, variable_ops), - IIC_fpLoadm, - "vldm", "${addr:submode} ${addr:base}, $dst1", - []> { + IIC_fpLoadm, "vldm", "${addr:submode} ${addr:base}, $dst1", []> { let Inst{27-25} = 0b110; let Inst{20} = 1; let Inst{11-9} = 0b101; @@ -133,9 +137,7 @@ def VLDMD : NI<(outs), def VLDMS : NI<(outs), (ins addrmode_neonldstm:$addr, reglist:$dst1, variable_ops), - IIC_fpLoadm, - "vldm", "${addr:submode} ${addr:base}, $dst1", - []> { + IIC_fpLoadm, "vldm", "${addr:submode} ${addr:base}, $dst1", []> { let Inst{27-25} = 0b110; let Inst{20} = 1; let Inst{11-9} = 0b101; @@ -144,10 +146,9 @@ def VLDMS : NI<(outs), */ // Use vldmia to load a Q register as a D register pair. -def VLDRQ : NI4<(outs QPR:$dst), (ins addrmode4:$addr), - IIC_fpLoadm, - "vldmia", "$addr, ${dst:dregpair}", - [(set QPR:$dst, (v2f64 (load addrmode4:$addr)))]> { +def VLDRQ : NI4<(outs QPR:$dst), (ins addrmode4:$addr), IIC_fpLoadm, + "vldmia", "$addr, ${dst:dregpair}", + [(set QPR:$dst, (v2f64 (load addrmode4:$addr)))]> { let Inst{27-25} = 0b110; let Inst{24} = 0; // P bit let Inst{23} = 1; // U bit @@ -156,10 +157,9 @@ def VLDRQ : NI4<(outs QPR:$dst), (ins addrmode4:$addr), } // Use vstmia to store a Q register as a D register pair. -def VSTRQ : NI4<(outs), (ins QPR:$src, addrmode4:$addr), - IIC_fpStorem, - "vstmia", "$addr, ${src:dregpair}", - [(store (v2f64 QPR:$src), addrmode4:$addr)]> { +def VSTRQ : NI4<(outs), (ins QPR:$src, addrmode4:$addr), IIC_fpStorem, + "vstmia", "$addr, ${src:dregpair}", + [(store (v2f64 QPR:$src), addrmode4:$addr)]> { let Inst{27-25} = 0b110; let Inst{24} = 0; // P bit let Inst{23} = 1; // U bit @@ -191,6 +191,29 @@ def VLD1q32 : VLD1Q<0b1000, "vld1", "32", v4i32, int_arm_neon_vld1>; def VLD1qf : VLD1Q<0b1000, "vld1", "32", v4f32, int_arm_neon_vld1>; def VLD1q64 : VLD1Q<0b1100, "vld1", "64", v2i64, int_arm_neon_vld1>; +// These (dreg triple/quadruple) are for disassembly only. +class VLD1D3<bits<4> op7_4, string OpcodeStr, string Dt> + : NLdSt<0, 0b10, 0b0110, op7_4, (outs DPR:$dst1, DPR:$dst2, DPR:$dst3), + (ins addrmode6:$addr), IIC_VLD1, OpcodeStr, Dt, + "\\{$dst1, $dst2, $dst3\\}, $addr", "", + [/* For disassembly only; pattern left blank */]>; +class VLD1D4<bits<4> op7_4, string OpcodeStr, string Dt> + : NLdSt<0,0b10,0b0010,op7_4,(outs DPR:$dst1, DPR:$dst2, DPR:$dst3, DPR:$dst4), + (ins addrmode6:$addr), IIC_VLD1, OpcodeStr, Dt, + "\\{$dst1, $dst2, $dst3, $dst4\\}, $addr", "", + [/* For disassembly only; pattern left blank */]>; + +def VLD1d8T : VLD1D3<0b0000, "vld1", "8">; +def VLD1d16T : VLD1D3<0b0100, "vld1", "16">; +def VLD1d32T : VLD1D3<0b1000, "vld1", "32">; +//def VLD1d64T : VLD1D3<0b1100, "vld1", "64">; + +def VLD1d8Q : VLD1D4<0b0000, "vld1", "8">; +def VLD1d16Q : VLD1D4<0b0100, "vld1", "16">; +def VLD1d32Q : VLD1D4<0b1000, "vld1", "32">; +//def VLD1d64Q : VLD1D4<0b1100, "vld1", "64">; + + let mayLoad = 1, hasExtraDefRegAllocReq = 1 in { // VLD2 : Vector Load (multiple 2-element structures) @@ -216,6 +239,16 @@ def VLD2q8 : VLD2Q<0b0000, "vld2", "8">; def VLD2q16 : VLD2Q<0b0100, "vld2", "16">; def VLD2q32 : VLD2Q<0b1000, "vld2", "32">; +// These (double-spaced dreg pair) are for disassembly only. +class VLD2Ddbl<bits<4> op7_4, string OpcodeStr, string Dt> + : NLdSt<0,0b10,0b1001,op7_4, (outs DPR:$dst1, DPR:$dst2), + (ins addrmode6:$addr), IIC_VLD2, + OpcodeStr, Dt, "\\{$dst1, $dst2\\}, $addr", "", []>; + +def VLD2d8D : VLD2Ddbl<0b0000, "vld2", "8">; +def VLD2d16D : VLD2Ddbl<0b0100, "vld2", "16">; +def VLD2d32D : VLD2Ddbl<0b1000, "vld2", "32">; + // VLD3 : Vector Load (multiple 3-element structures) class VLD3D<bits<4> op7_4, string OpcodeStr, string Dt> : NLdSt<0,0b10,0b0100,op7_4, (outs DPR:$dst1, DPR:$dst2, DPR:$dst3), @@ -285,105 +318,64 @@ def VLD4q32b : VLD4WB<0b1000, "vld4", "32">; class VLD2LN<bits<4> op11_8, string OpcodeStr, string Dt> : NLdSt<1,0b10,op11_8,{?,?,?,?}, (outs DPR:$dst1, DPR:$dst2), (ins addrmode6:$addr, DPR:$src1, DPR:$src2, nohash_imm:$lane), - IIC_VLD2, - OpcodeStr, Dt, "\\{$dst1[$lane], $dst2[$lane]\\}, $addr", + IIC_VLD2, OpcodeStr, Dt, "\\{$dst1[$lane], $dst2[$lane]\\}, $addr", "$src1 = $dst1, $src2 = $dst2", []>; // vld2 to single-spaced registers. def VLD2LNd8 : VLD2LN<0b0001, "vld2", "8">; -def VLD2LNd16 : VLD2LN<0b0101, "vld2", "16"> { - let Inst{5} = 0; -} -def VLD2LNd32 : VLD2LN<0b1001, "vld2", "32"> { - let Inst{6} = 0; -} +def VLD2LNd16 : VLD2LN<0b0101, "vld2", "16"> { let Inst{5} = 0; } +def VLD2LNd32 : VLD2LN<0b1001, "vld2", "32"> { let Inst{6} = 0; } // vld2 to double-spaced even registers. -def VLD2LNq16a: VLD2LN<0b0101, "vld2", "16"> { - let Inst{5} = 1; -} -def VLD2LNq32a: VLD2LN<0b1001, "vld2", "32"> { - let Inst{6} = 1; -} +def VLD2LNq16a: VLD2LN<0b0101, "vld2", "16"> { let Inst{5} = 1; } +def VLD2LNq32a: VLD2LN<0b1001, "vld2", "32"> { let Inst{6} = 1; } // vld2 to double-spaced odd registers. -def VLD2LNq16b: VLD2LN<0b0101, "vld2", "16"> { - let Inst{5} = 1; -} -def VLD2LNq32b: VLD2LN<0b1001, "vld2", "32"> { - let Inst{6} = 1; -} +def VLD2LNq16b: VLD2LN<0b0101, "vld2", "16"> { let Inst{5} = 1; } +def VLD2LNq32b: VLD2LN<0b1001, "vld2", "32"> { let Inst{6} = 1; } // VLD3LN : Vector Load (single 3-element structure to one lane) class VLD3LN<bits<4> op11_8, string OpcodeStr, string Dt> : NLdSt<1,0b10,op11_8,{?,?,?,?}, (outs DPR:$dst1, DPR:$dst2, DPR:$dst3), (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, - nohash_imm:$lane), IIC_VLD3, - OpcodeStr, Dt, + nohash_imm:$lane), IIC_VLD3, OpcodeStr, Dt, "\\{$dst1[$lane], $dst2[$lane], $dst3[$lane]\\}, $addr", "$src1 = $dst1, $src2 = $dst2, $src3 = $dst3", []>; // vld3 to single-spaced registers. -def VLD3LNd8 : VLD3LN<0b0010, "vld3", "8"> { - let Inst{4} = 0; -} -def VLD3LNd16 : VLD3LN<0b0110, "vld3", "16"> { - let Inst{5-4} = 0b00; -} -def VLD3LNd32 : VLD3LN<0b1010, "vld3", "32"> { - let Inst{6-4} = 0b000; -} +def VLD3LNd8 : VLD3LN<0b0010, "vld3", "8"> { let Inst{4} = 0; } +def VLD3LNd16 : VLD3LN<0b0110, "vld3", "16"> { let Inst{5-4} = 0b00; } +def VLD3LNd32 : VLD3LN<0b1010, "vld3", "32"> { let Inst{6-4} = 0b000; } // vld3 to double-spaced even registers. -def VLD3LNq16a: VLD3LN<0b0110, "vld3", "16"> { - let Inst{5-4} = 0b10; -} -def VLD3LNq32a: VLD3LN<0b1010, "vld3", "32"> { - let Inst{6-4} = 0b100; -} +def VLD3LNq16a: VLD3LN<0b0110, "vld3", "16"> { let Inst{5-4} = 0b10; } +def VLD3LNq32a: VLD3LN<0b1010, "vld3", "32"> { let Inst{6-4} = 0b100; } // vld3 to double-spaced odd registers. -def VLD3LNq16b: VLD3LN<0b0110, "vld3", "16"> { - let Inst{5-4} = 0b10; -} -def VLD3LNq32b: VLD3LN<0b1010, "vld3", "32"> { - let Inst{6-4} = 0b100; -} +def VLD3LNq16b: VLD3LN<0b0110, "vld3", "16"> { let Inst{5-4} = 0b10; } +def VLD3LNq32b: VLD3LN<0b1010, "vld3", "32"> { let Inst{6-4} = 0b100; } // VLD4LN : Vector Load (single 4-element structure to one lane) class VLD4LN<bits<4> op11_8, string OpcodeStr, string Dt> : NLdSt<1,0b10,op11_8,{?,?,?,?}, (outs DPR:$dst1, DPR:$dst2, DPR:$dst3, DPR:$dst4), (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, DPR:$src4, - nohash_imm:$lane), IIC_VLD4, - OpcodeStr, Dt, + nohash_imm:$lane), IIC_VLD4, OpcodeStr, Dt, "\\{$dst1[$lane], $dst2[$lane], $dst3[$lane], $dst4[$lane]\\}, $addr", "$src1 = $dst1, $src2 = $dst2, $src3 = $dst3, $src4 = $dst4", []>; // vld4 to single-spaced registers. def VLD4LNd8 : VLD4LN<0b0011, "vld4", "8">; -def VLD4LNd16 : VLD4LN<0b0111, "vld4", "16"> { - let Inst{5} = 0; -} -def VLD4LNd32 : VLD4LN<0b1011, "vld4", "32"> { - let Inst{6} = 0; -} +def VLD4LNd16 : VLD4LN<0b0111, "vld4", "16"> { let Inst{5} = 0; } +def VLD4LNd32 : VLD4LN<0b1011, "vld4", "32"> { let Inst{6} = 0; } // vld4 to double-spaced even registers. -def VLD4LNq16a: VLD4LN<0b0111, "vld4", "16"> { - let Inst{5} = 1; -} -def VLD4LNq32a: VLD4LN<0b1011, "vld4", "32"> { - let Inst{6} = 1; -} +def VLD4LNq16a: VLD4LN<0b0111, "vld4", "16"> { let Inst{5} = 1; } +def VLD4LNq32a: VLD4LN<0b1011, "vld4", "32"> { let Inst{6} = 1; } // vld4 to double-spaced odd registers. -def VLD4LNq16b: VLD4LN<0b0111, "vld4", "16"> { - let Inst{5} = 1; -} -def VLD4LNq32b: VLD4LN<0b1011, "vld4", "32"> { - let Inst{6} = 1; -} +def VLD4LNq16b: VLD4LN<0b0111, "vld4", "16"> { let Inst{5} = 1; } +def VLD4LNq32b: VLD4LN<0b1011, "vld4", "32"> { let Inst{6} = 1; } // VLD1DUP : Vector Load (single element to all lanes) // VLD2DUP : Vector Load (single 2-element structure to all lanes) @@ -418,6 +410,31 @@ def VST1qf : VST1Q<0b1000, "vst1", "32", v4f32, int_arm_neon_vst1>; def VST1q64 : VST1Q<0b1100, "vst1", "64", v2i64, int_arm_neon_vst1>; } // hasExtraSrcRegAllocReq +// These (dreg triple/quadruple) are for disassembly only. +class VST1D3<bits<4> op7_4, string OpcodeStr, string Dt> + : NLdSt<0, 0b00, 0b0110, op7_4, (outs), + (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3), IIC_VST, + OpcodeStr, Dt, + "\\{$src1, $src2, $src3\\}, $addr", "", + [/* For disassembly only; pattern left blank */]>; +class VST1D4<bits<4> op7_4, string OpcodeStr, string Dt> + : NLdSt<0, 0b00, 0b0010, op7_4, (outs), + (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, DPR:$src4), + IIC_VST, OpcodeStr, Dt, + "\\{$src1, $src2, $src3, $src4\\}, $addr", "", + [/* For disassembly only; pattern left blank */]>; + +def VST1d8T : VST1D3<0b0000, "vst1", "8">; +def VST1d16T : VST1D3<0b0100, "vst1", "16">; +def VST1d32T : VST1D3<0b1000, "vst1", "32">; +//def VST1d64T : VST1D3<0b1100, "vst1", "64">; + +def VST1d8Q : VST1D4<0b0000, "vst1", "8">; +def VST1d16Q : VST1D4<0b0100, "vst1", "16">; +def VST1d32Q : VST1D4<0b1000, "vst1", "32">; +//def VST1d64Q : VST1D4<0b1100, "vst1", "64">; + + let mayStore = 1, hasExtraSrcRegAllocReq = 1 in { // VST2 : Vector Store (multiple 2-element structures) @@ -428,8 +445,7 @@ class VST2D<bits<4> op7_4, string OpcodeStr, string Dt> class VST2Q<bits<4> op7_4, string OpcodeStr, string Dt> : NLdSt<0,0b00,0b0011,op7_4, (outs), (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, DPR:$src4), - IIC_VST, - OpcodeStr, Dt, "\\{$src1, $src2, $src3, $src4\\}, $addr", + IIC_VST, OpcodeStr, Dt, "\\{$src1, $src2, $src3, $src4\\}, $addr", "", []>; def VST2d8 : VST2D<0b0000, "vst2", "8">; @@ -443,6 +459,16 @@ def VST2q8 : VST2Q<0b0000, "vst2", "8">; def VST2q16 : VST2Q<0b0100, "vst2", "16">; def VST2q32 : VST2Q<0b1000, "vst2", "32">; +// These (double-spaced dreg pair) are for disassembly only. +class VST2Ddbl<bits<4> op7_4, string OpcodeStr, string Dt> + : NLdSt<0, 0b00, 0b1001, op7_4, (outs), + (ins addrmode6:$addr, DPR:$src1, DPR:$src2), IIC_VST, + OpcodeStr, Dt, "\\{$src1, $src2\\}, $addr", "", []>; + +def VST2d8D : VST2Ddbl<0b0000, "vst2", "8">; +def VST2d16D : VST2Ddbl<0b0100, "vst2", "16">; +def VST2d32D : VST2Ddbl<0b1000, "vst2", "32">; + // VST3 : Vector Store (multiple 3-element structures) class VST3D<bits<4> op7_4, string OpcodeStr, string Dt> : NLdSt<0,0b00,0b0100,op7_4, (outs), @@ -476,14 +502,12 @@ def VST3q32b : VST3WB<0b1000, "vst3", "32">; class VST4D<bits<4> op7_4, string OpcodeStr, string Dt> : NLdSt<0,0b00,0b0000,op7_4, (outs), (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, DPR:$src4), - IIC_VST, - OpcodeStr, Dt, "\\{$src1, $src2, $src3, $src4\\}, $addr", + IIC_VST, OpcodeStr, Dt, "\\{$src1, $src2, $src3, $src4\\}, $addr", "", []>; class VST4WB<bits<4> op7_4, string OpcodeStr, string Dt> : NLdSt<0,0b00,0b0001,op7_4, (outs GPR:$wb), (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, DPR:$src4), - IIC_VST, - OpcodeStr, Dt, "\\{$src1, $src2, $src3, $src4\\}, $addr", + IIC_VST, OpcodeStr, Dt, "\\{$src1, $src2, $src3, $src4\\}, $addr", "$addr.addr = $wb", []>; def VST4d8 : VST4D<0b0000, "vst4", "8">; @@ -511,104 +535,63 @@ def VST4q32b : VST4WB<0b1000, "vst4", "32">; // VST2LN : Vector Store (single 2-element structure from one lane) class VST2LN<bits<4> op11_8, string OpcodeStr, string Dt> : NLdSt<1,0b00,op11_8,{?,?,?,?}, (outs), - (ins addrmode6:$addr, DPR:$src1, DPR:$src2, nohash_imm:$lane), - IIC_VST, - OpcodeStr, Dt, "\\{$src1[$lane], $src2[$lane]\\}, $addr", - "", []>; + (ins addrmode6:$addr, DPR:$src1, DPR:$src2, nohash_imm:$lane), + IIC_VST, OpcodeStr, Dt, "\\{$src1[$lane], $src2[$lane]\\}, $addr", + "", []>; // vst2 to single-spaced registers. def VST2LNd8 : VST2LN<0b0001, "vst2", "8">; -def VST2LNd16 : VST2LN<0b0101, "vst2", "16"> { - let Inst{5} = 0; -} -def VST2LNd32 : VST2LN<0b1001, "vst2", "32"> { - let Inst{6} = 0; -} +def VST2LNd16 : VST2LN<0b0101, "vst2", "16"> { let Inst{5} = 0; } +def VST2LNd32 : VST2LN<0b1001, "vst2", "32"> { let Inst{6} = 0; } // vst2 to double-spaced even registers. -def VST2LNq16a: VST2LN<0b0101, "vst2", "16"> { - let Inst{5} = 1; -} -def VST2LNq32a: VST2LN<0b1001, "vst2", "32"> { - let Inst{6} = 1; -} +def VST2LNq16a: VST2LN<0b0101, "vst2", "16"> { let Inst{5} = 1; } +def VST2LNq32a: VST2LN<0b1001, "vst2", "32"> { let Inst{6} = 1; } // vst2 to double-spaced odd registers. -def VST2LNq16b: VST2LN<0b0101, "vst2", "16"> { - let Inst{5} = 1; -} -def VST2LNq32b: VST2LN<0b1001, "vst2", "32"> { - let Inst{6} = 1; -} +def VST2LNq16b: VST2LN<0b0101, "vst2", "16"> { let Inst{5} = 1; } +def VST2LNq32b: VST2LN<0b1001, "vst2", "32"> { let Inst{6} = 1; } // VST3LN : Vector Store (single 3-element structure from one lane) class VST3LN<bits<4> op11_8, string OpcodeStr, string Dt> : NLdSt<1,0b00,op11_8,{?,?,?,?}, (outs), - (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, - nohash_imm:$lane), IIC_VST, - OpcodeStr, Dt, - "\\{$src1[$lane], $src2[$lane], $src3[$lane]\\}, $addr", "", []>; + (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, + nohash_imm:$lane), IIC_VST, OpcodeStr, Dt, + "\\{$src1[$lane], $src2[$lane], $src3[$lane]\\}, $addr", "", []>; // vst3 to single-spaced registers. -def VST3LNd8 : VST3LN<0b0010, "vst3", "8"> { - let Inst{4} = 0; -} -def VST3LNd16 : VST3LN<0b0110, "vst3", "16"> { - let Inst{5-4} = 0b00; -} -def VST3LNd32 : VST3LN<0b1010, "vst3", "32"> { - let Inst{6-4} = 0b000; -} +def VST3LNd8 : VST3LN<0b0010, "vst3", "8"> { let Inst{4} = 0; } +def VST3LNd16 : VST3LN<0b0110, "vst3", "16"> { let Inst{5-4} = 0b00; } +def VST3LNd32 : VST3LN<0b1010, "vst3", "32"> { let Inst{6-4} = 0b000; } // vst3 to double-spaced even registers. -def VST3LNq16a: VST3LN<0b0110, "vst3", "16"> { - let Inst{5-4} = 0b10; -} -def VST3LNq32a: VST3LN<0b1010, "vst3", "32"> { - let Inst{6-4} = 0b100; -} +def VST3LNq16a: VST3LN<0b0110, "vst3", "16"> { let Inst{5-4} = 0b10; } +def VST3LNq32a: VST3LN<0b1010, "vst3", "32"> { let Inst{6-4} = 0b100; } // vst3 to double-spaced odd registers. -def VST3LNq16b: VST3LN<0b0110, "vst3", "16"> { - let Inst{5-4} = 0b10; -} -def VST3LNq32b: VST3LN<0b1010, "vst3", "32"> { - let Inst{6-4} = 0b100; -} +def VST3LNq16b: VST3LN<0b0110, "vst3", "16"> { let Inst{5-4} = 0b10; } +def VST3LNq32b: VST3LN<0b1010, "vst3", "32"> { let Inst{6-4} = 0b100; } // VST4LN : Vector Store (single 4-element structure from one lane) class VST4LN<bits<4> op11_8, string OpcodeStr, string Dt> : NLdSt<1,0b00,op11_8,{?,?,?,?}, (outs), - (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, DPR:$src4, - nohash_imm:$lane), IIC_VST, - OpcodeStr, Dt, + (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, DPR:$src4, + nohash_imm:$lane), IIC_VST, OpcodeStr, Dt, "\\{$src1[$lane], $src2[$lane], $src3[$lane], $src4[$lane]\\}, $addr", - "", []>; + "", []>; // vst4 to single-spaced registers. def VST4LNd8 : VST4LN<0b0011, "vst4", "8">; -def VST4LNd16 : VST4LN<0b0111, "vst4", "16"> { - let Inst{5} = 0; -} -def VST4LNd32 : VST4LN<0b1011, "vst4", "32"> { - let Inst{6} = 0; -} +def VST4LNd16 : VST4LN<0b0111, "vst4", "16"> { let Inst{5} = 0; } +def VST4LNd32 : VST4LN<0b1011, "vst4", "32"> { let Inst{6} = 0; } // vst4 to double-spaced even registers. -def VST4LNq16a: VST4LN<0b0111, "vst4", "16"> { - let Inst{5} = 1; -} -def VST4LNq32a: VST4LN<0b1011, "vst4", "32"> { - let Inst{6} = 1; -} +def VST4LNq16a: VST4LN<0b0111, "vst4", "16"> { let Inst{5} = 1; } +def VST4LNq32a: VST4LN<0b1011, "vst4", "32"> { let Inst{6} = 1; } // vst4 to double-spaced odd registers. -def VST4LNq16b: VST4LN<0b0111, "vst4", "16"> { - let Inst{5} = 1; -} -def VST4LNq32b: VST4LN<0b1011, "vst4", "32"> { - let Inst{6} = 1; -} +def VST4LNq16b: VST4LN<0b0111, "vst4", "16"> { let Inst{5} = 1; } +def VST4LNq32b: VST4LN<0b1011, "vst4", "32"> { let Inst{6} = 1; } } // mayStore = 1, hasExtraSrcRegAllocReq = 1 @@ -656,34 +639,26 @@ def SubReg_i32_lane : SDNodeXForm<imm, [{ // Instruction Classes //===----------------------------------------------------------------------===// -// Basic 2-register operations, both double- and quad-register. +// Basic 2-register operations: single-, double- and quad-register. +class N2VS<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18, + bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr, + string Dt, ValueType ResTy, ValueType OpTy, SDNode OpNode> + : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, + (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src), + IIC_VUNAD, OpcodeStr, Dt, "$dst, $src", "", []>; class N2VD<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18, - bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr,string Dt, - ValueType ResTy, ValueType OpTy, SDNode OpNode> + bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr, + string Dt, ValueType ResTy, ValueType OpTy, SDNode OpNode> : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, (outs DPR:$dst), (ins DPR:$src), IIC_VUNAD, OpcodeStr, Dt, "$dst, $src", "", [(set DPR:$dst, (ResTy (OpNode (OpTy DPR:$src))))]>; class N2VQ<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18, - bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr,string Dt, - ValueType ResTy, ValueType OpTy, SDNode OpNode> + bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr, + string Dt, ValueType ResTy, ValueType OpTy, SDNode OpNode> : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 1, op4, (outs QPR:$dst), (ins QPR:$src), IIC_VUNAQ, OpcodeStr, Dt, "$dst, $src", "", [(set QPR:$dst, (ResTy (OpNode (OpTy QPR:$src))))]>; -// Basic 2-register operations, scalar single-precision. -class N2VDs<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18, - bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr,string Dt, - ValueType ResTy, ValueType OpTy, SDNode OpNode> - : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, - (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src), - IIC_VUNAD, OpcodeStr, Dt, "$dst, $src", "", []>; - -class N2VDsPat<SDNode OpNode, ValueType ResTy, ValueType OpTy, NeonI Inst> - : NEONFPPat<(ResTy (OpNode SPR:$a)), - (EXTRACT_SUBREG - (Inst (INSERT_SUBREG (OpTy (IMPLICIT_DEF)), SPR:$a, arm_ssubreg_0)), - arm_ssubreg_0)>; - // Basic 2-register intrinsics, both double- and quad-register. class N2VDInt<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18, bits<2> op17_16, bits<5> op11_7, bit op4, @@ -700,21 +675,6 @@ class N2VQInt<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18, (ins QPR:$src), itin, OpcodeStr, Dt, "$dst, $src", "", [(set QPR:$dst, (ResTy (IntOp (OpTy QPR:$src))))]>; -// Basic 2-register intrinsics, scalar single-precision -class N2VDInts<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18, - bits<2> op17_16, bits<5> op11_7, bit op4, - InstrItinClass itin, string OpcodeStr, string Dt, - ValueType ResTy, ValueType OpTy, Intrinsic IntOp> - : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, - (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src), itin, - OpcodeStr, Dt, "$dst, $src", "", []>; - -class N2VDIntsPat<SDNode OpNode, NeonI Inst> - : NEONFPPat<(f32 (OpNode SPR:$a)), - (EXTRACT_SUBREG - (Inst (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), SPR:$a, arm_ssubreg_0)), - arm_ssubreg_0)>; - // Narrow 2-register intrinsics. class N2VNInt<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18, bits<2> op17_16, bits<5> op11_7, bit op6, bit op4, @@ -742,15 +702,22 @@ class N2VDShuffle<bits<2> op19_18, bits<5> op11_7, string OpcodeStr, string Dt> class N2VQShuffle<bits<2> op19_18, bits<5> op11_7, InstrItinClass itin, string OpcodeStr, string Dt> : N2V<0b11, 0b11, op19_18, 0b10, op11_7, 1, 0, (outs QPR:$dst1, QPR:$dst2), - (ins QPR:$src1, QPR:$src2), itin, - OpcodeStr, Dt, "$dst1, $dst2", + (ins QPR:$src1, QPR:$src2), itin, OpcodeStr, Dt, "$dst1, $dst2", "$src1 = $dst1, $src2 = $dst2", []>; -// Basic 3-register operations, both double- and quad-register. +// Basic 3-register operations: single-, double- and quad-register. +class N3VS<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, + string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy, + SDNode OpNode, bit Commutable> + : N3V<op24, op23, op21_20, op11_8, 0, op4, + (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src1, DPR_VFP2:$src2), IIC_VBIND, + OpcodeStr, Dt, "$dst, $src1, $src2", "", []> { + let isCommutable = Commutable; +} + class N3VD<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, InstrItinClass itin, string OpcodeStr, string Dt, - ValueType ResTy, ValueType OpTy, - SDNode OpNode, bit Commutable> + ValueType ResTy, ValueType OpTy, SDNode OpNode, bit Commutable> : N3V<op24, op23, op21_20, op11_8, 0, op4, (outs DPR:$dst), (ins DPR:$src1, DPR:$src2), itin, OpcodeStr, Dt, "$dst, $src1, $src2", "", @@ -763,9 +730,9 @@ class N3VDX<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, ValueType ResTy, ValueType OpTy, SDNode OpNode, bit Commutable> : N3VX<op24, op23, op21_20, op11_8, 0, op4, - (outs DPR:$dst), (ins DPR:$src1, DPR:$src2), itin, - OpcodeStr, "$dst, $src1, $src2", "", - [(set DPR:$dst, (ResTy (OpNode (OpTy DPR:$src1), (OpTy DPR:$src2))))]> { + (outs DPR:$dst), (ins DPR:$src1, DPR:$src2), itin, + OpcodeStr, "$dst, $src1, $src2", "", + [(set DPR:$dst, (ResTy (OpNode (OpTy DPR:$src1), (OpTy DPR:$src2))))]>{ let isCommutable = Commutable; } class N3VDSL<bits<2> op21_20, bits<4> op11_8, @@ -776,27 +743,23 @@ class N3VDSL<bits<2> op21_20, bits<4> op11_8, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "", [(set (Ty DPR:$dst), (Ty (ShOp (Ty DPR:$src1), - (Ty (NEONvduplane (Ty DPR_VFP2:$src2), - imm:$lane)))))]> { + (Ty (NEONvduplane (Ty DPR_VFP2:$src2), imm:$lane)))))]>{ let isCommutable = 0; } class N3VDSL16<bits<2> op21_20, bits<4> op11_8, string OpcodeStr, string Dt, ValueType Ty, SDNode ShOp> : N3V<0, 1, op21_20, op11_8, 1, 0, (outs DPR:$dst), (ins DPR:$src1, DPR_8:$src2, nohash_imm:$lane), - IIC_VMULi16D, - OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "", + IIC_VMULi16D, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "", [(set (Ty DPR:$dst), (Ty (ShOp (Ty DPR:$src1), - (Ty (NEONvduplane (Ty DPR_8:$src2), - imm:$lane)))))]> { + (Ty (NEONvduplane (Ty DPR_8:$src2), imm:$lane)))))]> { let isCommutable = 0; } class N3VQ<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, InstrItinClass itin, string OpcodeStr, string Dt, - ValueType ResTy, ValueType OpTy, - SDNode OpNode, bit Commutable> + ValueType ResTy, ValueType OpTy, SDNode OpNode, bit Commutable> : N3V<op24, op23, op21_20, op11_8, 1, op4, (outs QPR:$dst), (ins QPR:$src1, QPR:$src2), itin, OpcodeStr, Dt, "$dst, $src1, $src2", "", @@ -805,12 +768,11 @@ class N3VQ<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, } class N3VQX<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, InstrItinClass itin, string OpcodeStr, - ValueType ResTy, ValueType OpTy, - SDNode OpNode, bit Commutable> + ValueType ResTy, ValueType OpTy, SDNode OpNode, bit Commutable> : N3VX<op24, op23, op21_20, op11_8, 1, op4, - (outs QPR:$dst), (ins QPR:$src1, QPR:$src2), itin, - OpcodeStr, "$dst, $src1, $src2", "", - [(set QPR:$dst, (ResTy (OpNode (OpTy QPR:$src1), (OpTy QPR:$src2))))]> { + (outs QPR:$dst), (ins QPR:$src1, QPR:$src2), itin, + OpcodeStr, "$dst, $src1, $src2", "", + [(set QPR:$dst, (ResTy (OpNode (OpTy QPR:$src1), (OpTy QPR:$src2))))]>{ let isCommutable = Commutable; } class N3VQSL<bits<2> op21_20, bits<4> op11_8, @@ -825,13 +787,11 @@ class N3VQSL<bits<2> op21_20, bits<4> op11_8, imm:$lane)))))]> { let isCommutable = 0; } -class N3VQSL16<bits<2> op21_20, bits<4> op11_8, - string OpcodeStr, string Dt, +class N3VQSL16<bits<2> op21_20, bits<4> op11_8, string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy, SDNode ShOp> : N3V<1, 1, op21_20, op11_8, 1, 0, (outs QPR:$dst), (ins QPR:$src1, DPR_8:$src2, nohash_imm:$lane), - IIC_VMULi16Q, - OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "", + IIC_VMULi16Q, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "", [(set (ResTy QPR:$dst), (ResTy (ShOp (ResTy QPR:$src1), (ResTy (NEONvduplane (OpTy DPR_8:$src2), @@ -839,27 +799,10 @@ class N3VQSL16<bits<2> op21_20, bits<4> op11_8, let isCommutable = 0; } -// Basic 3-register operations, scalar single-precision -class N3VDs<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, - string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy, - SDNode OpNode, bit Commutable> - : N3V<op24, op23, op21_20, op11_8, 0, op4, - (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src1, DPR_VFP2:$src2), IIC_VBIND, - OpcodeStr, Dt, "$dst, $src1, $src2", "", []> { - let isCommutable = Commutable; -} -class N3VDsPat<SDNode OpNode, NeonI Inst> - : NEONFPPat<(f32 (OpNode SPR:$a, SPR:$b)), - (EXTRACT_SUBREG - (Inst (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), SPR:$a, arm_ssubreg_0), - (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), SPR:$b, arm_ssubreg_0)), - arm_ssubreg_0)>; - // Basic 3-register intrinsics, both double- and quad-register. class N3VDInt<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, InstrItinClass itin, string OpcodeStr, string Dt, - ValueType ResTy, ValueType OpTy, - Intrinsic IntOp, bit Commutable> + ValueType ResTy, ValueType OpTy, Intrinsic IntOp, bit Commutable> : N3V<op24, op23, op21_20, op11_8, 0, op4, (outs DPR:$dst), (ins DPR:$src1, DPR:$src2), itin, OpcodeStr, Dt, "$dst, $src1, $src2", "", @@ -891,8 +834,7 @@ class N3VDIntSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin, class N3VQInt<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, InstrItinClass itin, string OpcodeStr, string Dt, - ValueType ResTy, ValueType OpTy, - Intrinsic IntOp, bit Commutable> + ValueType ResTy, ValueType OpTy, Intrinsic IntOp, bit Commutable> : N3V<op24, op23, op21_20, op11_8, 1, op4, (outs QPR:$dst), (ins QPR:$src1, QPR:$src2), itin, OpcodeStr, Dt, "$dst, $src1, $src2", "", @@ -924,7 +866,15 @@ class N3VQIntSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin, let isCommutable = 0; } -// Multiply-Add/Sub operations, both double- and quad-register. +// Multiply-Add/Sub operations: single-, double- and quad-register. +class N3VSMulOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, + InstrItinClass itin, string OpcodeStr, string Dt, + ValueType Ty, SDNode MulOp, SDNode OpNode> + : N3V<op24, op23, op21_20, op11_8, 0, op4, + (outs DPR_VFP2:$dst), + (ins DPR_VFP2:$src1, DPR_VFP2:$src2, DPR_VFP2:$src3), itin, + OpcodeStr, Dt, "$dst, $src2, $src3", "$src1 = $dst", []>; + class N3VDMulOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, InstrItinClass itin, string OpcodeStr, string Dt, ValueType Ty, SDNode MulOp, SDNode OpNode> @@ -976,8 +926,8 @@ class N3VQMulOpSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin, [(set (ResTy QPR:$dst), (ResTy (ShOp (ResTy QPR:$src1), (ResTy (MulOp QPR:$src2, - (ResTy (NEONvduplane (OpTy DPR_VFP2:$src3), - imm:$lane)))))))]>; + (ResTy (NEONvduplane (OpTy DPR_VFP2:$src3), + imm:$lane)))))))]>; class N3VQMulOpSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin, string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy, @@ -989,25 +939,8 @@ class N3VQMulOpSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin, [(set (ResTy QPR:$dst), (ResTy (ShOp (ResTy QPR:$src1), (ResTy (MulOp QPR:$src2, - (ResTy (NEONvduplane (OpTy DPR_8:$src3), - imm:$lane)))))))]>; - -// Multiply-Add/Sub operations, scalar single-precision -class N3VDMulOps<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, - InstrItinClass itin, string OpcodeStr, string Dt, - ValueType Ty, SDNode MulOp, SDNode OpNode> - : N3V<op24, op23, op21_20, op11_8, 0, op4, - (outs DPR_VFP2:$dst), - (ins DPR_VFP2:$src1, DPR_VFP2:$src2, DPR_VFP2:$src3), itin, - OpcodeStr, Dt, "$dst, $src2, $src3", "$src1 = $dst", []>; - -class N3VDMulOpsPat<SDNode MulNode, SDNode OpNode, NeonI Inst> - : NEONFPPat<(f32 (OpNode SPR:$acc, (f32 (MulNode SPR:$a, SPR:$b)))), - (EXTRACT_SUBREG - (Inst (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), SPR:$acc, arm_ssubreg_0), - (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), SPR:$a, arm_ssubreg_0), - (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), SPR:$b, arm_ssubreg_0)), - arm_ssubreg_0)>; + (ResTy (NEONvduplane (OpTy DPR_8:$src3), + imm:$lane)))))))]>; // Neon 3-argument intrinsics, both double- and quad-register. // The destination register is also used as the first source operand register. @@ -1050,9 +983,9 @@ class N3VLInt3SL<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass itin, (OpTy DPR:$src2), (OpTy (NEONvduplane (OpTy DPR_VFP2:$src3), imm:$lane)))))]>; -class N3VLInt3SL16<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass itin, - string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy, - Intrinsic IntOp> +class N3VLInt3SL16<bit op24, bits<2> op21_20, bits<4> op11_8, + InstrItinClass itin, string OpcodeStr, string Dt, + ValueType ResTy, ValueType OpTy, Intrinsic IntOp> : N3V<op24, 1, op21_20, op11_8, 1, 0, (outs QPR:$dst), (ins QPR:$src1, DPR:$src2, DPR_8:$src3, nohash_imm:$lane), itin, @@ -1063,7 +996,6 @@ class N3VLInt3SL16<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass iti (OpTy (NEONvduplane (OpTy DPR_8:$src3), imm:$lane)))))]>; - // Narrowing 3-register intrinsics. class N3VNInt<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4, string OpcodeStr, string Dt, ValueType TyD, ValueType TyQ, @@ -1095,9 +1027,9 @@ class N3VLIntSL<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass itin, (ResTy (IntOp (OpTy DPR:$src1), (OpTy (NEONvduplane (OpTy DPR_VFP2:$src2), imm:$lane)))))]>; -class N3VLIntSL16<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass itin, - string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy, - Intrinsic IntOp> +class N3VLIntSL16<bit op24, bits<2> op21_20, bits<4> op11_8, + InstrItinClass itin, string OpcodeStr, string Dt, + ValueType ResTy, ValueType OpTy, Intrinsic IntOp> : N3V<op24, 1, op21_20, op11_8, 1, 0, (outs QPR:$dst), (ins DPR:$src1, DPR_8:$src2, nohash_imm:$lane), itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "", @@ -1249,6 +1181,45 @@ class N2VCvtQ<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4, // S = single int (32 bit) elements // D = double int (64 bit) elements +// Neon 2-register vector operations -- for disassembly only. + +// First with only element sizes of 8, 16 and 32 bits: +multiclass N2V_QHS_cmp<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16, + bits<5> op11_7, bit op4, string opc, string Dt, + string asm> { + // 64-bit vector types. + def v8i8 : N2V<op24_23, op21_20, 0b00, op17_16, op11_7, 0, op4, + (outs DPR:$dst), (ins DPR:$src), NoItinerary, + opc, !strconcat(Dt, "8"), asm, "", []>; + def v4i16 : N2V<op24_23, op21_20, 0b01, op17_16, op11_7, 0, op4, + (outs DPR:$dst), (ins DPR:$src), NoItinerary, + opc, !strconcat(Dt, "16"), asm, "", []>; + def v2i32 : N2V<op24_23, op21_20, 0b10, op17_16, op11_7, 0, op4, + (outs DPR:$dst), (ins DPR:$src), NoItinerary, + opc, !strconcat(Dt, "32"), asm, "", []>; + def v2f32 : N2V<op24_23, op21_20, 0b10, op17_16, op11_7, 0, op4, + (outs DPR:$dst), (ins DPR:$src), NoItinerary, + opc, "f32", asm, "", []> { + let Inst{10} = 1; // overwrite F = 1 + } + + // 128-bit vector types. + def v16i8 : N2V<op24_23, op21_20, 0b00, op17_16, op11_7, 1, op4, + (outs QPR:$dst), (ins QPR:$src), NoItinerary, + opc, !strconcat(Dt, "8"), asm, "", []>; + def v8i16 : N2V<op24_23, op21_20, 0b01, op17_16, op11_7, 1, op4, + (outs QPR:$dst), (ins QPR:$src), NoItinerary, + opc, !strconcat(Dt, "16"), asm, "", []>; + def v4i32 : N2V<op24_23, op21_20, 0b10, op17_16, op11_7, 1, op4, + (outs QPR:$dst), (ins QPR:$src), NoItinerary, + opc, !strconcat(Dt, "32"), asm, "", []>; + def v4f32 : N2V<op24_23, op21_20, 0b10, op17_16, op11_7, 1, op4, + (outs QPR:$dst), (ins QPR:$src), NoItinerary, + opc, "f32", asm, "", []> { + let Inst{10} = 1; // overwrite F = 1 + } +} + // Neon 3-register vector operations. // First with only element sizes of 8, 16 and 32 bits: @@ -1262,22 +1233,22 @@ multiclass N3V_QHS<bit op24, bit op23, bits<4> op11_8, bit op4, OpcodeStr, !strconcat(Dt, "8"), v8i8, v8i8, OpNode, Commutable>; def v4i16 : N3VD<op24, op23, 0b01, op11_8, op4, itinD16, - OpcodeStr, !strconcat(Dt, "16"), - v4i16, v4i16, OpNode, Commutable>; + OpcodeStr, !strconcat(Dt, "16"), + v4i16, v4i16, OpNode, Commutable>; def v2i32 : N3VD<op24, op23, 0b10, op11_8, op4, itinD32, - OpcodeStr, !strconcat(Dt, "32"), - v2i32, v2i32, OpNode, Commutable>; + OpcodeStr, !strconcat(Dt, "32"), + v2i32, v2i32, OpNode, Commutable>; // 128-bit vector types. def v16i8 : N3VQ<op24, op23, 0b00, op11_8, op4, itinQ16, - OpcodeStr, !strconcat(Dt, "8"), - v16i8, v16i8, OpNode, Commutable>; + OpcodeStr, !strconcat(Dt, "8"), + v16i8, v16i8, OpNode, Commutable>; def v8i16 : N3VQ<op24, op23, 0b01, op11_8, op4, itinQ16, - OpcodeStr, !strconcat(Dt, "16"), - v8i16, v8i16, OpNode, Commutable>; + OpcodeStr, !strconcat(Dt, "16"), + v8i16, v8i16, OpNode, Commutable>; def v4i32 : N3VQ<op24, op23, 0b10, op11_8, op4, itinQ32, - OpcodeStr, !strconcat(Dt, "32"), - v4i32, v4i32, OpNode, Commutable>; + OpcodeStr, !strconcat(Dt, "32"), + v4i32, v4i32, OpNode, Commutable>; } multiclass N3VSL_HS<bits<4> op11_8, string OpcodeStr, string Dt, SDNode ShOp> { @@ -1372,7 +1343,7 @@ multiclass N3VIntSL_HS<bits<4> op11_8, def v2i32 : N3VDIntSL<0b10, op11_8, itinD32, OpcodeStr, !strconcat(Dt, "32"), v2i32, IntOp>; def v8i16 : N3VQIntSL16<0b01, op11_8, itinQ16, - OpcodeStr, !strconcat(Dt, "16"), v8i16, v4i16, IntOp>; + OpcodeStr, !strconcat(Dt, "16"), v8i16, v4i16, IntOp>; def v4i32 : N3VQIntSL<0b10, op11_8, itinQ32, OpcodeStr, !strconcat(Dt, "32"), v4i32, v2i32, IntOp>; } @@ -1386,8 +1357,8 @@ multiclass N3VInt_QHS<bit op24, bit op23, bits<4> op11_8, bit op4, : N3VInt_HS<op24, op23, op11_8, op4, itinD16, itinD32, itinQ16, itinQ32, OpcodeStr, Dt, IntOp, Commutable> { def v8i8 : N3VDInt<op24, op23, 0b00, op11_8, op4, itinD16, - OpcodeStr, !strconcat(Dt, "8"), - v8i8, v8i8, IntOp, Commutable>; + OpcodeStr, !strconcat(Dt, "8"), + v8i8, v8i8, IntOp, Commutable>; def v16i8 : N3VQInt<op24, op23, 0b00, op11_8, op4, itinQ16, OpcodeStr, !strconcat(Dt, "8"), v16i8, v16i8, IntOp, Commutable>; @@ -1402,11 +1373,11 @@ multiclass N3VInt_QHSD<bit op24, bit op23, bits<4> op11_8, bit op4, : N3VInt_QHS<op24, op23, op11_8, op4, itinD16, itinD32, itinQ16, itinQ32, OpcodeStr, Dt, IntOp, Commutable> { def v1i64 : N3VDInt<op24, op23, 0b11, op11_8, op4, itinD32, - OpcodeStr, !strconcat(Dt, "64"), - v1i64, v1i64, IntOp, Commutable>; + OpcodeStr, !strconcat(Dt, "64"), + v1i64, v1i64, IntOp, Commutable>; def v2i64 : N3VQInt<op24, op23, 0b11, op11_8, op4, itinQ32, - OpcodeStr, !strconcat(Dt, "64"), - v2i64, v2i64, IntOp, Commutable>; + OpcodeStr, !strconcat(Dt, "64"), + v2i64, v2i64, IntOp, Commutable>; } @@ -1511,9 +1482,11 @@ multiclass N3VMulOpSL_HS<bits<4> op11_8, def v2i32 : N3VDMulOpSL<0b10, op11_8, itinD32, OpcodeStr, !strconcat(Dt, "32"), v2i32, mul, ShOp>; def v8i16 : N3VQMulOpSL16<0b01, op11_8, itinQ16, - OpcodeStr, !strconcat(Dt, "16"), v8i16, v4i16, mul, ShOp>; + OpcodeStr, !strconcat(Dt, "16"), v8i16, v4i16, + mul, ShOp>; def v4i32 : N3VQMulOpSL<0b10, op11_8, itinQ32, - OpcodeStr, !strconcat(Dt, "32"), v4i32, v2i32, mul, ShOp>; + OpcodeStr, !strconcat(Dt, "32"), v4i32, v2i32, + mul, ShOp>; } // Neon 3-argument intrinsics, @@ -1522,19 +1495,19 @@ multiclass N3VInt3_QHS<bit op24, bit op23, bits<4> op11_8, bit op4, string OpcodeStr, string Dt, Intrinsic IntOp> { // 64-bit vector types. def v8i8 : N3VDInt3<op24, op23, 0b00, op11_8, op4, IIC_VMACi16D, - OpcodeStr, !strconcat(Dt, "8"), v8i8, v8i8, IntOp>; + OpcodeStr, !strconcat(Dt, "8"), v8i8, v8i8, IntOp>; def v4i16 : N3VDInt3<op24, op23, 0b01, op11_8, op4, IIC_VMACi16D, - OpcodeStr, !strconcat(Dt, "16"), v4i16, v4i16, IntOp>; + OpcodeStr, !strconcat(Dt, "16"), v4i16, v4i16, IntOp>; def v2i32 : N3VDInt3<op24, op23, 0b10, op11_8, op4, IIC_VMACi32D, - OpcodeStr, !strconcat(Dt, "32"), v2i32, v2i32, IntOp>; + OpcodeStr, !strconcat(Dt, "32"), v2i32, v2i32, IntOp>; // 128-bit vector types. def v16i8 : N3VQInt3<op24, op23, 0b00, op11_8, op4, IIC_VMACi16Q, - OpcodeStr, !strconcat(Dt, "8"), v16i8, v16i8, IntOp>; + OpcodeStr, !strconcat(Dt, "8"), v16i8, v16i8, IntOp>; def v8i16 : N3VQInt3<op24, op23, 0b01, op11_8, op4, IIC_VMACi16Q, - OpcodeStr, !strconcat(Dt, "16"), v8i16, v8i16, IntOp>; + OpcodeStr, !strconcat(Dt, "16"), v8i16, v8i16, IntOp>; def v4i32 : N3VQInt3<op24, op23, 0b10, op11_8, op4, IIC_VMACi32Q, - OpcodeStr, !strconcat(Dt, "32"), v4i32, v4i32, IntOp>; + OpcodeStr, !strconcat(Dt, "32"), v4i32, v4i32, IntOp>; } @@ -1576,17 +1549,17 @@ multiclass N2VInt_QHS<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16, def v8i8 : N2VDInt<op24_23, op21_20, 0b00, op17_16, op11_7, op4, itinD, OpcodeStr, !strconcat(Dt, "8"), v8i8, v8i8, IntOp>; def v4i16 : N2VDInt<op24_23, op21_20, 0b01, op17_16, op11_7, op4, - itinD, OpcodeStr, !strconcat(Dt, "16"), v4i16, v4i16, IntOp>; + itinD, OpcodeStr, !strconcat(Dt, "16"),v4i16,v4i16,IntOp>; def v2i32 : N2VDInt<op24_23, op21_20, 0b10, op17_16, op11_7, op4, - itinD, OpcodeStr, !strconcat(Dt, "32"), v2i32, v2i32, IntOp>; + itinD, OpcodeStr, !strconcat(Dt, "32"),v2i32,v2i32,IntOp>; // 128-bit vector types. def v16i8 : N2VQInt<op24_23, op21_20, 0b00, op17_16, op11_7, op4, - itinQ, OpcodeStr, !strconcat(Dt, "8"), v16i8, v16i8, IntOp>; + itinQ, OpcodeStr, !strconcat(Dt, "8"), v16i8,v16i8,IntOp>; def v8i16 : N2VQInt<op24_23, op21_20, 0b01, op17_16, op11_7, op4, - itinQ, OpcodeStr, !strconcat(Dt, "16"), v8i16, v8i16, IntOp>; + itinQ, OpcodeStr, !strconcat(Dt, "16"),v8i16,v8i16,IntOp>; def v4i32 : N2VQInt<op24_23, op21_20, 0b10, op17_16, op11_7, op4, - itinQ, OpcodeStr, !strconcat(Dt, "32"), v4i32, v4i32, IntOp>; + itinQ, OpcodeStr, !strconcat(Dt, "32"),v4i32,v4i32,IntOp>; } @@ -1846,29 +1819,31 @@ def VMULpd : N3VDInt<1, 0, 0b00, 0b1001, 1, IIC_VMULi16D, "vmul", "p8", def VMULpq : N3VQInt<1, 0, 0b00, 0b1001, 1, IIC_VMULi16Q, "vmul", "p8", v16i8, v16i8, int_arm_neon_vmulp, 1>; def VMULfd : N3VD<1, 0, 0b00, 0b1101, 1, IIC_VBIND, "vmul", "f32", - v2f32, v2f32, fmul, 1>; + v2f32, v2f32, fmul, 1>; def VMULfq : N3VQ<1, 0, 0b00, 0b1101, 1, IIC_VBINQ, "vmul", "f32", - v4f32, v4f32, fmul, 1>; -defm VMULsl : N3VSL_HS<0b1000, "vmul", "i", mul>; -def VMULslfd : N3VDSL<0b10, 0b1001, IIC_VBIND, "vmul", "f32", v2f32, fmul>; -def VMULslfq : N3VQSL<0b10, 0b1001, IIC_VBINQ, "vmul", "f32", v4f32, v2f32, fmul>; + v4f32, v4f32, fmul, 1>; +defm VMULsl : N3VSL_HS<0b1000, "vmul", "i", mul>; +def VMULslfd : N3VDSL<0b10, 0b1001, IIC_VBIND, "vmul", "f32", v2f32, fmul>; +def VMULslfq : N3VQSL<0b10, 0b1001, IIC_VBINQ, "vmul", "f32", v4f32, + v2f32, fmul>; + def : Pat<(v8i16 (mul (v8i16 QPR:$src1), (v8i16 (NEONvduplane (v8i16 QPR:$src2), imm:$lane)))), (v8i16 (VMULslv8i16 (v8i16 QPR:$src1), (v4i16 (EXTRACT_SUBREG QPR:$src2, - (DSubReg_i16_reg imm:$lane))), + (DSubReg_i16_reg imm:$lane))), (SubReg_i16_lane imm:$lane)))>; def : Pat<(v4i32 (mul (v4i32 QPR:$src1), (v4i32 (NEONvduplane (v4i32 QPR:$src2), imm:$lane)))), (v4i32 (VMULslv4i32 (v4i32 QPR:$src1), (v2i32 (EXTRACT_SUBREG QPR:$src2, - (DSubReg_i32_reg imm:$lane))), + (DSubReg_i32_reg imm:$lane))), (SubReg_i32_lane imm:$lane)))>; def : Pat<(v4f32 (fmul (v4f32 QPR:$src1), (v4f32 (NEONvduplane (v4f32 QPR:$src2), imm:$lane)))), (v4f32 (VMULslfq (v4f32 QPR:$src1), (v2f32 (EXTRACT_SUBREG QPR:$src2, - (DSubReg_i32_reg imm:$lane))), + (DSubReg_i32_reg imm:$lane))), (SubReg_i32_lane imm:$lane)))>; // VQDMULH : Vector Saturating Doubling Multiply Returning High Half @@ -1883,14 +1858,14 @@ def : Pat<(v8i16 (int_arm_neon_vqdmulh (v8i16 QPR:$src1), imm:$lane)))), (v8i16 (VQDMULHslv8i16 (v8i16 QPR:$src1), (v4i16 (EXTRACT_SUBREG QPR:$src2, - (DSubReg_i16_reg imm:$lane))), + (DSubReg_i16_reg imm:$lane))), (SubReg_i16_lane imm:$lane)))>; def : Pat<(v4i32 (int_arm_neon_vqdmulh (v4i32 QPR:$src1), (v4i32 (NEONvduplane (v4i32 QPR:$src2), imm:$lane)))), (v4i32 (VQDMULHslv4i32 (v4i32 QPR:$src1), (v2i32 (EXTRACT_SUBREG QPR:$src2, - (DSubReg_i32_reg imm:$lane))), + (DSubReg_i32_reg imm:$lane))), (SubReg_i32_lane imm:$lane)))>; // VQRDMULH : Vector Rounding Saturating Doubling Multiply Returning High Half @@ -1905,14 +1880,14 @@ def : Pat<(v8i16 (int_arm_neon_vqrdmulh (v8i16 QPR:$src1), imm:$lane)))), (v8i16 (VQRDMULHslv8i16 (v8i16 QPR:$src1), (v4i16 (EXTRACT_SUBREG QPR:$src2, - (DSubReg_i16_reg imm:$lane))), + (DSubReg_i16_reg imm:$lane))), (SubReg_i16_lane imm:$lane)))>; def : Pat<(v4i32 (int_arm_neon_vqrdmulh (v4i32 QPR:$src1), (v4i32 (NEONvduplane (v4i32 QPR:$src2), imm:$lane)))), (v4i32 (VQRDMULHslv4i32 (v4i32 QPR:$src1), (v2i32 (EXTRACT_SUBREG QPR:$src2, - (DSubReg_i32_reg imm:$lane))), + (DSubReg_i32_reg imm:$lane))), (SubReg_i32_lane imm:$lane)))>; // VMULL : Vector Multiply Long (integer and polynomial) (Q = D * D) @@ -1950,30 +1925,28 @@ def VMLAslfq : N3VQMulOpSL<0b10, 0b0001, IIC_VMACQ, "vmla", "f32", v4f32, v2f32, fmul, fadd>; def : Pat<(v8i16 (add (v8i16 QPR:$src1), - (mul (v8i16 QPR:$src2), - (v8i16 (NEONvduplane (v8i16 QPR:$src3), imm:$lane))))), - (v8i16 (VMLAslv8i16 (v8i16 QPR:$src1), - (v8i16 QPR:$src2), + (mul (v8i16 QPR:$src2), + (v8i16 (NEONvduplane (v8i16 QPR:$src3), imm:$lane))))), + (v8i16 (VMLAslv8i16 (v8i16 QPR:$src1), (v8i16 QPR:$src2), (v4i16 (EXTRACT_SUBREG QPR:$src3, - (DSubReg_i16_reg imm:$lane))), + (DSubReg_i16_reg imm:$lane))), (SubReg_i16_lane imm:$lane)))>; def : Pat<(v4i32 (add (v4i32 QPR:$src1), - (mul (v4i32 QPR:$src2), - (v4i32 (NEONvduplane (v4i32 QPR:$src3), imm:$lane))))), - (v4i32 (VMLAslv4i32 (v4i32 QPR:$src1), - (v4i32 QPR:$src2), + (mul (v4i32 QPR:$src2), + (v4i32 (NEONvduplane (v4i32 QPR:$src3), imm:$lane))))), + (v4i32 (VMLAslv4i32 (v4i32 QPR:$src1), (v4i32 QPR:$src2), (v2i32 (EXTRACT_SUBREG QPR:$src3, - (DSubReg_i32_reg imm:$lane))), + (DSubReg_i32_reg imm:$lane))), (SubReg_i32_lane imm:$lane)))>; def : Pat<(v4f32 (fadd (v4f32 QPR:$src1), - (fmul (v4f32 QPR:$src2), - (v4f32 (NEONvduplane (v4f32 QPR:$src3), imm:$lane))))), + (fmul (v4f32 QPR:$src2), + (v4f32 (NEONvduplane (v4f32 QPR:$src3), imm:$lane))))), (v4f32 (VMLAslfq (v4f32 QPR:$src1), (v4f32 QPR:$src2), (v2f32 (EXTRACT_SUBREG QPR:$src3, - (DSubReg_i32_reg imm:$lane))), + (DSubReg_i32_reg imm:$lane))), (SubReg_i32_lane imm:$lane)))>; // VMLAL : Vector Multiply Accumulate Long (Q += D * D) @@ -2003,30 +1976,27 @@ def VMLSslfq : N3VQMulOpSL<0b10, 0b0101, IIC_VMACQ, "vmls", "f32", v4f32, v2f32, fmul, fsub>; def : Pat<(v8i16 (sub (v8i16 QPR:$src1), - (mul (v8i16 QPR:$src2), - (v8i16 (NEONvduplane (v8i16 QPR:$src3), imm:$lane))))), - (v8i16 (VMLSslv8i16 (v8i16 QPR:$src1), - (v8i16 QPR:$src2), + (mul (v8i16 QPR:$src2), + (v8i16 (NEONvduplane (v8i16 QPR:$src3), imm:$lane))))), + (v8i16 (VMLSslv8i16 (v8i16 QPR:$src1), (v8i16 QPR:$src2), (v4i16 (EXTRACT_SUBREG QPR:$src3, - (DSubReg_i16_reg imm:$lane))), + (DSubReg_i16_reg imm:$lane))), (SubReg_i16_lane imm:$lane)))>; def : Pat<(v4i32 (sub (v4i32 QPR:$src1), - (mul (v4i32 QPR:$src2), - (v4i32 (NEONvduplane (v4i32 QPR:$src3), imm:$lane))))), - (v4i32 (VMLSslv4i32 (v4i32 QPR:$src1), - (v4i32 QPR:$src2), + (mul (v4i32 QPR:$src2), + (v4i32 (NEONvduplane (v4i32 QPR:$src3), imm:$lane))))), + (v4i32 (VMLSslv4i32 (v4i32 QPR:$src1), (v4i32 QPR:$src2), (v2i32 (EXTRACT_SUBREG QPR:$src3, - (DSubReg_i32_reg imm:$lane))), + (DSubReg_i32_reg imm:$lane))), (SubReg_i32_lane imm:$lane)))>; def : Pat<(v4f32 (fsub (v4f32 QPR:$src1), - (fmul (v4f32 QPR:$src2), - (v4f32 (NEONvduplane (v4f32 QPR:$src3), imm:$lane))))), - (v4f32 (VMLSslfq (v4f32 QPR:$src1), - (v4f32 QPR:$src2), + (fmul (v4f32 QPR:$src2), + (v4f32 (NEONvduplane (v4f32 QPR:$src3), imm:$lane))))), + (v4f32 (VMLSslfq (v4f32 QPR:$src1), (v4f32 QPR:$src2), (v2f32 (EXTRACT_SUBREG QPR:$src3, - (DSubReg_i32_reg imm:$lane))), + (DSubReg_i32_reg imm:$lane))), (SubReg_i32_lane imm:$lane)))>; // VMLSL : Vector Multiply Subtract Long (Q -= D * D) @@ -2088,6 +2058,10 @@ def VCEQfd : N3VD<0,0,0b00,0b1110,0, IIC_VBIND, "vceq", "f32", v2i32, v2f32, NEONvceq, 1>; def VCEQfq : N3VQ<0,0,0b00,0b1110,0, IIC_VBINQ, "vceq", "f32", v4i32, v4f32, NEONvceq, 1>; +// For disassembly only. +defm VCEQz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00010, 0, "vceq", "i", + "$dst, $src, #0">; + // VCGE : Vector Compare Greater Than or Equal defm VCGEs : N3V_QHS<0, 0, 0b0011, 1, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, "vcge", "s", NEONvcge, 0>; @@ -2097,6 +2071,13 @@ def VCGEfd : N3VD<1,0,0b00,0b1110,0, IIC_VBIND, "vcge", "f32", v2i32, v2f32, NEONvcge, 0>; def VCGEfq : N3VQ<1,0,0b00,0b1110,0, IIC_VBINQ, "vcge", "f32", v4i32, v4f32, NEONvcge, 0>; +// For disassembly only. +defm VCGEz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00001, 0, "vcge", "s", + "$dst, $src, #0">; +// For disassembly only. +defm VCLEz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00011, 0, "vcle", "s", + "$dst, $src, #0">; + // VCGT : Vector Compare Greater Than defm VCGTs : N3V_QHS<0, 0, 0b0011, 0, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, "vcgt", "s", NEONvcgt, 0>; @@ -2106,6 +2087,13 @@ def VCGTfd : N3VD<1,0,0b10,0b1110,0, IIC_VBIND, "vcgt", "f32", v2i32, v2f32, NEONvcgt, 0>; def VCGTfq : N3VQ<1,0,0b10,0b1110,0, IIC_VBINQ, "vcgt", "f32", v4i32, v4f32, NEONvcgt, 0>; +// For disassembly only. +defm VCGTz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00000, 0, "vcgt", "s", + "$dst, $src, #0">; +// For disassembly only. +defm VCLTz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00100, 0, "vclt", "s", + "$dst, $src, #0">; + // VACGE : Vector Absolute Compare Greater Than or Equal (aka VCAGE) def VACGEd : N3VDInt<1, 0, 0b00, 0b1110, 1, IIC_VBIND, "vacge", "f32", v2i32, v2f32, int_arm_neon_vacged, 0>; @@ -2247,9 +2235,9 @@ defm VABALu : N3VLInt3_QHS<1,1,0b0101,0, "vabal", "u", int_arm_neon_vabalu>; // Vector Maximum and Minimum. // VMAX : Vector Maximum -defm VMAXs : N3VInt_QHS<0, 0, 0b0110, 0, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, +defm VMAXs : N3VInt_QHS<0,0,0b0110,0, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, "vmax", "s", int_arm_neon_vmaxs, 1>; -defm VMAXu : N3VInt_QHS<1, 0, 0b0110, 0, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, +defm VMAXu : N3VInt_QHS<1,0,0b0110,0, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, "vmax", "u", int_arm_neon_vmaxu, 1>; def VMAXfd : N3VDInt<0, 0, 0b00, 0b1111, 0, IIC_VBIND, "vmax", "f32", v2f32, v2f32, int_arm_neon_vmaxs, 1>; @@ -2257,9 +2245,9 @@ def VMAXfq : N3VQInt<0, 0, 0b00, 0b1111, 0, IIC_VBINQ, "vmax", "f32", v4f32, v4f32, int_arm_neon_vmaxs, 1>; // VMIN : Vector Minimum -defm VMINs : N3VInt_QHS<0, 0, 0b0110, 1, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, +defm VMINs : N3VInt_QHS<0,0,0b0110,1, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, "vmin", "s", int_arm_neon_vmins, 1>; -defm VMINu : N3VInt_QHS<1, 0, 0b0110, 1, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, +defm VMINu : N3VInt_QHS<1,0,0b0110,1, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, "vmin", "u", int_arm_neon_vminu, 1>; def VMINfd : N3VDInt<0, 0, 0b10, 0b1111, 0, IIC_VBIND, "vmin", "f32", v2f32, v2f32, int_arm_neon_vmins, 1>; @@ -2401,16 +2389,17 @@ def VSHLLi32 : N2VLShMax<1, 1, 0b111010, 0b0011, 0, 0, 0, "vshll", "i32", v2i64, v2i32, NEONvshlli>; // VSHRN : Vector Shift Right and Narrow -defm VSHRN : N2VNSh_HSD<0,1,0b1000,0,0,1, IIC_VSHLiD, "vshrn", "i", NEONvshrn>; +defm VSHRN : N2VNSh_HSD<0,1,0b1000,0,0,1, IIC_VSHLiD, "vshrn", "i", + NEONvshrn>; // VRSHL : Vector Rounding Shift defm VRSHLs : N3VInt_QHSD<0,0,0b0101,0, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, - IIC_VSHLi4Q, "vrshl", "s", int_arm_neon_vrshifts, 0>; + IIC_VSHLi4Q, "vrshl", "s", int_arm_neon_vrshifts,0>; defm VRSHLu : N3VInt_QHSD<1,0,0b0101,0, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, - IIC_VSHLi4Q, "vrshl", "u", int_arm_neon_vrshiftu, 0>; + IIC_VSHLi4Q, "vrshl", "u", int_arm_neon_vrshiftu,0>; // VRSHR : Vector Rounding Shift Right -defm VRSHRs : N2VSh_QHSD<0, 1, 0b0010, 1, IIC_VSHLi4D, "vrshr", "s", NEONvrshrs>; -defm VRSHRu : N2VSh_QHSD<1, 1, 0b0010, 1, IIC_VSHLi4D, "vrshr", "u", NEONvrshru>; +defm VRSHRs : N2VSh_QHSD<0,1,0b0010,1, IIC_VSHLi4D, "vrshr", "s", NEONvrshrs>; +defm VRSHRu : N2VSh_QHSD<1,1,0b0010,1, IIC_VSHLi4D, "vrshr", "u", NEONvrshru>; // VRSHRN : Vector Rounding Shift Right and Narrow defm VRSHRN : N2VNSh_HSD<0, 1, 0b1000, 0, 1, 1, IIC_VSHLi4D, "vrshrn", "i", @@ -2418,14 +2407,14 @@ defm VRSHRN : N2VNSh_HSD<0, 1, 0b1000, 0, 1, 1, IIC_VSHLi4D, "vrshrn", "i", // VQSHL : Vector Saturating Shift defm VQSHLs : N3VInt_QHSD<0,0,0b0100,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, - IIC_VSHLi4Q, "vqshl", "s", int_arm_neon_vqshifts, 0>; + IIC_VSHLi4Q, "vqshl", "s", int_arm_neon_vqshifts,0>; defm VQSHLu : N3VInt_QHSD<1,0,0b0100,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, - IIC_VSHLi4Q, "vqshl", "u", int_arm_neon_vqshiftu, 0>; + IIC_VSHLi4Q, "vqshl", "u", int_arm_neon_vqshiftu,0>; // VQSHL : Vector Saturating Shift Left (Immediate) -defm VQSHLsi : N2VSh_QHSD<0, 1, 0b0111, 1, IIC_VSHLi4D, "vqshl", "s", NEONvqshls>; -defm VQSHLui : N2VSh_QHSD<1, 1, 0b0111, 1, IIC_VSHLi4D, "vqshl", "u", NEONvqshlu>; +defm VQSHLsi : N2VSh_QHSD<0,1,0b0111,1, IIC_VSHLi4D, "vqshl", "s", NEONvqshls>; +defm VQSHLui : N2VSh_QHSD<1,1,0b0111,1, IIC_VSHLi4D, "vqshl", "u", NEONvqshlu>; // VQSHLU : Vector Saturating Shift Left (Immediate, Unsigned) -defm VQSHLsu : N2VSh_QHSD<1, 1, 0b0110, 1, IIC_VSHLi4D, "vqshlu", "s", NEONvqshlsu>; +defm VQSHLsu : N2VSh_QHSD<1,1,0b0110,1, IIC_VSHLi4D, "vqshlu","s",NEONvqshlsu>; // VQSHRN : Vector Saturating Shift Right and Narrow defm VQSHRNs : N2VNSh_HSD<0, 1, 0b1001, 0, 0, 1, IIC_VSHLi4D, "vqshrn", "s", @@ -2438,10 +2427,10 @@ defm VQSHRUN : N2VNSh_HSD<1, 1, 0b1000, 0, 0, 1, IIC_VSHLi4D, "vqshrun", "s", NEONvqshrnsu>; // VQRSHL : Vector Saturating Rounding Shift -defm VQRSHLs : N3VInt_QHSD<0, 0, 0b0101, 1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, +defm VQRSHLs : N3VInt_QHSD<0,0,0b0101,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, IIC_VSHLi4Q, "vqrshl", "s", int_arm_neon_vqrshifts, 0>; -defm VQRSHLu : N3VInt_QHSD<1, 0, 0b0101, 1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, +defm VQRSHLu : N3VInt_QHSD<1,0,0b0101,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, IIC_VSHLi4Q, "vqrshl", "u", int_arm_neon_vqrshiftu, 0>; @@ -2508,7 +2497,7 @@ def VNEGs16q : VNEGQ<0b01, "vneg", "s16", v8i16>; def VNEGs32q : VNEGQ<0b10, "vneg", "s32", v4i32>; // VNEG : Vector Negate (floating-point) -def VNEGf32d : N2V<0b11, 0b11, 0b10, 0b01, 0b01111, 0, 0, +def VNEGfd : N2V<0b11, 0b11, 0b10, 0b01, 0b01111, 0, 0, (outs DPR:$dst), (ins DPR:$src), IIC_VUNAD, "vneg", "f32", "$dst, $src", "", [(set DPR:$dst, (v2f32 (fneg DPR:$src)))]>; @@ -2547,6 +2536,14 @@ def VCNTq : N2VQInt<0b11, 0b11, 0b00, 0b00, 0b01010, 0, IIC_VCNTiQ, "vcnt", "8", v16i8, v16i8, int_arm_neon_vcnt>; +// Vector Swap -- for disassembly only. +def VSWPd : N2VX<0b11, 0b11, 0b00, 0b10, 0b00000, 0, 0, + (outs DPR:$dst), (ins DPR:$src), NoItinerary, + "vswp", "$dst, $src", "", []>; +def VSWPq : N2VX<0b11, 0b11, 0b00, 0b10, 0b00000, 1, 0, + (outs QPR:$dst), (ins QPR:$src), NoItinerary, + "vswp", "$dst, $src", "", []>; + // Vector Move Operations. // VMOV : Vector Move (Register) @@ -2678,10 +2675,10 @@ def : Pat<(extractelt (v4i32 QPR:$src), imm:$lane), (DSubReg_i32_reg imm:$lane))), (SubReg_i32_lane imm:$lane))>; def : Pat<(extractelt (v2f32 DPR:$src1), imm:$src2), - (EXTRACT_SUBREG (v2f32 (COPY_TO_REGCLASS (v2f32 DPR:$src1), DPR_VFP2)), + (EXTRACT_SUBREG (v2f32 (COPY_TO_REGCLASS (v2f32 DPR:$src1),DPR_VFP2)), (SSubReg_f32_reg imm:$src2))>; def : Pat<(extractelt (v4f32 QPR:$src1), imm:$src2), - (EXTRACT_SUBREG (v4f32 (COPY_TO_REGCLASS (v4f32 QPR:$src1), QPR_VFP2)), + (EXTRACT_SUBREG (v4f32 (COPY_TO_REGCLASS (v4f32 QPR:$src1),QPR_VFP2)), (SSubReg_f32_reg imm:$src2))>; //def : Pat<(extractelt (v2i64 QPR:$src1), imm:$src2), // (EXTRACT_SUBREG QPR:$src1, (DSubReg_f64_reg imm:$src2))>; @@ -2849,11 +2846,13 @@ def VDUPfqf : N2V<0b11, 0b11, {?,1}, {0,0}, 0b11000, 1, 0, def : Pat<(v2i64 (NEONvduplane (v2i64 QPR:$src), imm:$lane)), (INSERT_SUBREG QPR:$src, - (i64 (EXTRACT_SUBREG QPR:$src, (DSubReg_f64_reg imm:$lane))), + (i64 (EXTRACT_SUBREG QPR:$src, + (DSubReg_f64_reg imm:$lane))), (DSubReg_f64_other_reg imm:$lane))>; def : Pat<(v2f64 (NEONvduplane (v2f64 QPR:$src), imm:$lane)), (INSERT_SUBREG QPR:$src, - (f64 (EXTRACT_SUBREG QPR:$src, (DSubReg_f64_reg imm:$lane))), + (f64 (EXTRACT_SUBREG QPR:$src, + (DSubReg_f64_reg imm:$lane))), (DSubReg_f64_other_reg imm:$lane))>; // VMOVN : Vector Narrowing Move @@ -3092,70 +3091,110 @@ def VTBX4 // NEON instructions for single-precision FP math //===----------------------------------------------------------------------===// +class N2VSPat<SDNode OpNode, ValueType ResTy, ValueType OpTy, NeonI Inst> + : NEONFPPat<(ResTy (OpNode SPR:$a)), + (EXTRACT_SUBREG (Inst (INSERT_SUBREG (OpTy (IMPLICIT_DEF)), + SPR:$a, arm_ssubreg_0)), + arm_ssubreg_0)>; + +class N3VSPat<SDNode OpNode, NeonI Inst> + : NEONFPPat<(f32 (OpNode SPR:$a, SPR:$b)), + (EXTRACT_SUBREG (Inst (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), + SPR:$a, arm_ssubreg_0), + (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), + SPR:$b, arm_ssubreg_0)), + arm_ssubreg_0)>; + +class N3VSMulOpPat<SDNode MulNode, SDNode OpNode, NeonI Inst> + : NEONFPPat<(f32 (OpNode SPR:$acc, (f32 (MulNode SPR:$a, SPR:$b)))), + (EXTRACT_SUBREG (Inst (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), + SPR:$acc, arm_ssubreg_0), + (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), + SPR:$a, arm_ssubreg_0), + (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), + SPR:$b, arm_ssubreg_0)), + arm_ssubreg_0)>; + // These need separate instructions because they must use DPR_VFP2 register // class which have SPR sub-registers. // Vector Add Operations used for single-precision FP let neverHasSideEffects = 1 in -def VADDfd_sfp : N3VDs<0, 0, 0b00, 0b1101, 0, "vadd", "f32", v2f32, v2f32, fadd,1>; -def : N3VDsPat<fadd, VADDfd_sfp>; +def VADDfd_sfp : N3VS<0,0,0b00,0b1101,0, "vadd", "f32", v2f32, v2f32, fadd, 1>; +def : N3VSPat<fadd, VADDfd_sfp>; // Vector Sub Operations used for single-precision FP let neverHasSideEffects = 1 in -def VSUBfd_sfp : N3VDs<0, 0, 0b10, 0b1101, 0, "vsub", "f32", v2f32, v2f32, fsub,0>; -def : N3VDsPat<fsub, VSUBfd_sfp>; +def VSUBfd_sfp : N3VS<0,0,0b10,0b1101,0, "vsub", "f32", v2f32, v2f32, fsub, 0>; +def : N3VSPat<fsub, VSUBfd_sfp>; // Vector Multiply Operations used for single-precision FP let neverHasSideEffects = 1 in -def VMULfd_sfp : N3VDs<1, 0, 0b00, 0b1101, 1, "vmul", "f32", v2f32, v2f32, fmul,1>; -def : N3VDsPat<fmul, VMULfd_sfp>; +def VMULfd_sfp : N3VS<1,0,0b00,0b1101,1, "vmul", "f32", v2f32, v2f32, fmul, 1>; +def : N3VSPat<fmul, VMULfd_sfp>; // Vector Multiply-Accumulate/Subtract used for single-precision FP // vml[as].f32 can cause 4-8 cycle stalls in following ASIMD instructions, so // we want to avoid them for now. e.g., alternating vmla/vadd instructions. //let neverHasSideEffects = 1 in -//def VMLAfd_sfp : N3VDMulOps<0, 0, 0b00, 0b1101, 1, IIC_VMACD, "vmla", "f32", v2f32,fmul,fadd>; -//def : N3VDMulOpsPat<fmul, fadd, VMLAfd_sfp>; +//def VMLAfd_sfp : N3VSMulOp<0,0,0b00,0b1101,1, IIC_VMACD, "vmla", "f32", +// v2f32, fmul, fadd>; +//def : N3VSMulOpPat<fmul, fadd, VMLAfd_sfp>; //let neverHasSideEffects = 1 in -//def VMLSfd_sfp : N3VDMulOps<0, 0, 0b10, 0b1101, 1, IIC_VMACD, "vmls", "f32", v2f32,fmul,fsub>; -//def : N3VDMulOpsPat<fmul, fsub, VMLSfd_sfp>; +//def VMLSfd_sfp : N3VSMulOp<0,0,0b10,0b1101,1, IIC_VMACD, "vmls", "f32", +// v2f32, fmul, fsub>; +//def : N3VSMulOpPat<fmul, fsub, VMLSfd_sfp>; // Vector Absolute used for single-precision FP let neverHasSideEffects = 1 in -def VABSfd_sfp : N2VDInts<0b11, 0b11, 0b10, 0b01, 0b01110, 0, - IIC_VUNAD, "vabs", "f32", - v2f32, v2f32, int_arm_neon_vabs>; -def : N2VDIntsPat<fabs, VABSfd_sfp>; +def VABSfd_sfp : N2V<0b11, 0b11, 0b10, 0b01, 0b01110, 0, 0, + (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src), IIC_VUNAD, + "vabs", "f32", "$dst, $src", "", []>; +def : N2VSPat<fabs, f32, v2f32, VABSfd_sfp>; // Vector Negate used for single-precision FP let neverHasSideEffects = 1 in -def VNEGf32d_sfp : N2V<0b11, 0b11, 0b10, 0b01, 0b01111, 0, 0, - (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src), IIC_VUNAD, - "vneg", "f32", "$dst, $src", "", []>; -def : N2VDIntsPat<fneg, VNEGf32d_sfp>; +def VNEGfd_sfp : N2V<0b11, 0b11, 0b10, 0b01, 0b01111, 0, 0, + (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src), IIC_VUNAD, + "vneg", "f32", "$dst, $src", "", []>; +def : N2VSPat<fneg, f32, v2f32, VNEGfd_sfp>; + +// Vector Maximum used for single-precision FP +let neverHasSideEffects = 1 in +def VMAXfd_sfp : N3V<0, 0, 0b00, 0b1111, 0, 0, (outs DPR_VFP2:$dst), + (ins DPR_VFP2:$src1, DPR_VFP2:$src2), IIC_VBIND, + "vmax", "f32", "$dst, $src1, $src2", "", []>; +def : N3VSPat<NEONfmax, VMAXfd_sfp>; + +// Vector Minimum used for single-precision FP +let neverHasSideEffects = 1 in +def VMINfd_sfp : N3V<0, 0, 0b00, 0b1111, 0, 0, (outs DPR_VFP2:$dst), + (ins DPR_VFP2:$src1, DPR_VFP2:$src2), IIC_VBIND, + "vmin", "f32", "$dst, $src1, $src2", "", []>; +def : N3VSPat<NEONfmin, VMINfd_sfp>; // Vector Convert between single-precision FP and integer let neverHasSideEffects = 1 in -def VCVTf2sd_sfp : N2VDs<0b11, 0b11, 0b10, 0b11, 0b01110, 0, "vcvt", "s32.f32", - v2i32, v2f32, fp_to_sint>; -def : N2VDsPat<arm_ftosi, f32, v2f32, VCVTf2sd_sfp>; +def VCVTf2sd_sfp : N2VS<0b11, 0b11, 0b10, 0b11, 0b01110, 0, "vcvt", "s32.f32", + v2i32, v2f32, fp_to_sint>; +def : N2VSPat<arm_ftosi, f32, v2f32, VCVTf2sd_sfp>; let neverHasSideEffects = 1 in -def VCVTf2ud_sfp : N2VDs<0b11, 0b11, 0b10, 0b11, 0b01111, 0, "vcvt", "u32.f32", - v2i32, v2f32, fp_to_uint>; -def : N2VDsPat<arm_ftoui, f32, v2f32, VCVTf2ud_sfp>; +def VCVTf2ud_sfp : N2VS<0b11, 0b11, 0b10, 0b11, 0b01111, 0, "vcvt", "u32.f32", + v2i32, v2f32, fp_to_uint>; +def : N2VSPat<arm_ftoui, f32, v2f32, VCVTf2ud_sfp>; let neverHasSideEffects = 1 in -def VCVTs2fd_sfp : N2VDs<0b11, 0b11, 0b10, 0b11, 0b01100, 0, "vcvt", "f32.s32", - v2f32, v2i32, sint_to_fp>; -def : N2VDsPat<arm_sitof, f32, v2i32, VCVTs2fd_sfp>; +def VCVTs2fd_sfp : N2VS<0b11, 0b11, 0b10, 0b11, 0b01100, 0, "vcvt", "f32.s32", + v2f32, v2i32, sint_to_fp>; +def : N2VSPat<arm_sitof, f32, v2i32, VCVTs2fd_sfp>; let neverHasSideEffects = 1 in -def VCVTu2fd_sfp : N2VDs<0b11, 0b11, 0b10, 0b11, 0b01101, 0, "vcvt", "f32.u32", - v2f32, v2i32, uint_to_fp>; -def : N2VDsPat<arm_uitof, f32, v2i32, VCVTu2fd_sfp>; +def VCVTu2fd_sfp : N2VS<0b11, 0b11, 0b10, 0b11, 0b01101, 0, "vcvt", "f32.u32", + v2f32, v2i32, uint_to_fp>; +def : N2VSPat<arm_uitof, f32, v2i32, VCVTu2fd_sfp>; //===----------------------------------------------------------------------===// // Non-Instruction Patterns diff --git a/lib/Target/ARM/ARMInstrThumb.td b/lib/Target/ARM/ARMInstrThumb.td index 64142ad..1c77f27 100644 --- a/lib/Target/ARM/ARMInstrThumb.td +++ b/lib/Target/ARM/ARMInstrThumb.td @@ -120,7 +120,10 @@ def t_addrmode_sp : Operand<i32>, // Miscellaneous Instructions. // -let Defs = [SP], Uses = [SP] in { +// FIXME: Marking these as hasSideEffects is necessary to prevent machine DCE +// from removing one half of the matched pairs. That breaks PEI, which assumes +// these will always be in pairs, and asserts if it finds otherwise. Better way? +let Defs = [SP], Uses = [SP], hasSideEffects = 1 in { def tADJCALLSTACKUP : PseudoInst<(outs), (ins i32imm:$amt1, i32imm:$amt2), NoItinerary, "@ tADJCALLSTACKUP $amt1", @@ -132,6 +135,55 @@ PseudoInst<(outs), (ins i32imm:$amt), NoItinerary, [(ARMcallseq_start imm:$amt)]>, Requires<[IsThumb1Only]>; } +def tNOP : T1pI<(outs), (ins), NoItinerary, "nop", "", + [/* For disassembly only; pattern left blank */]>, + T1Encoding<0b101111> { + let Inst{9-8} = 0b11; + let Inst{7-0} = 0b00000000; +} + +def tYIELD : T1pI<(outs), (ins), NoItinerary, "yield", "", + [/* For disassembly only; pattern left blank */]>, + T1Encoding<0b101111> { + let Inst{9-8} = 0b11; + let Inst{7-0} = 0b00010000; +} + +def tWFE : T1pI<(outs), (ins), NoItinerary, "wfe", "", + [/* For disassembly only; pattern left blank */]>, + T1Encoding<0b101111> { + let Inst{9-8} = 0b11; + let Inst{7-0} = 0b00100000; +} + +def tWFI : T1pI<(outs), (ins), NoItinerary, "wfi", "", + [/* For disassembly only; pattern left blank */]>, + T1Encoding<0b101111> { + let Inst{9-8} = 0b11; + let Inst{7-0} = 0b00110000; +} + +def tSEV : T1pI<(outs), (ins), NoItinerary, "sev", "", + [/* For disassembly only; pattern left blank */]>, + T1Encoding<0b101111> { + let Inst{9-8} = 0b11; + let Inst{7-0} = 0b01000000; +} + +def tSETENDBE : T1I<(outs), (ins), NoItinerary, "setend\tbe", + [/* For disassembly only; pattern left blank */]>, + T1Encoding<0b101101> { + let Inst{9-5} = 0b10010; + let Inst{3} = 1; +} + +def tSETENDLE : T1I<(outs), (ins), NoItinerary, "setend\tle", + [/* For disassembly only; pattern left blank */]>, + T1Encoding<0b101101> { + let Inst{9-5} = 0b10010; + let Inst{3} = 0; +} + // The i32imm operand $val can be used by a debugger to store more information // about the breakpoint. def tBKPT : T1I<(outs), (ins i32imm:$val), NoItinerary, "bkpt\t$val", @@ -140,6 +192,19 @@ def tBKPT : T1I<(outs), (ins i32imm:$val), NoItinerary, "bkpt\t$val", let Inst{9-8} = 0b10; } +// Change Processor State is a system instruction -- for disassembly only. +// The singleton $opt operand contains the following information: +// opt{4-0} = mode ==> don't care +// opt{5} = changemode ==> 0 (false for 16-bit Thumb instr) +// opt{8-6} = AIF from Inst{2-0} +// opt{10-9} = 1:imod from Inst{4} with 0b10 as enable and 0b11 as disable +// +// The opt{4-0} and opt{5} sub-fields are to accommodate 32-bit Thumb and ARM +// CPS which has more options. +def tCPS : T1I<(outs), (ins i32imm:$opt), NoItinerary, "cps${opt:cps}", + [/* For disassembly only; pattern left blank */]>, + T1Misc<0b0110011>; + // For both thumb1 and thumb2. let isNotDuplicable = 1 in def tPICADD : TIt<(outs GPR:$dst), (ins GPR:$lhs, pclabel:$cp), IIC_iALUr, @@ -208,7 +273,7 @@ let isReturn = 1, isTerminator = 1, isBarrier = 1 in { let Inst{6-3} = 0b1110; // Rm = lr } // Alternative return instruction used by vararg functions. - def tBX_RET_vararg : TI<(outs), (ins tGPR:$target), IIC_Br, "bx\t$target", []>, + def tBX_RET_vararg : TI<(outs), (ins tGPR:$target), IIC_Br, "bx\t$target",[]>, T1Special<{1,1,0,?}>; // A6.2.3 & A8.6.25 } @@ -236,20 +301,20 @@ let isCall = 1, D24, D25, D26, D27, D28, D29, D30, D31, CPSR, FPSCR] in { // Also used for Thumb2 def tBL : TIx2<0b11110, 0b11, 1, - (outs), (ins i32imm:$func, variable_ops), IIC_Br, + (outs), (ins i32imm:$func, variable_ops), IIC_Br, "bl\t${func:call}", [(ARMtcall tglobaladdr:$func)]>, Requires<[IsThumb, IsNotDarwin]>; // ARMv5T and above, also used for Thumb2 def tBLXi : TIx2<0b11110, 0b11, 0, - (outs), (ins i32imm:$func, variable_ops), IIC_Br, + (outs), (ins i32imm:$func, variable_ops), IIC_Br, "blx\t${func:call}", [(ARMcall tglobaladdr:$func)]>, Requires<[IsThumb, HasV5T, IsNotDarwin]>; // Also used for Thumb2 - def tBLXr : TI<(outs), (ins GPR:$func, variable_ops), IIC_Br, + def tBLXr : TI<(outs), (ins GPR:$func, variable_ops), IIC_Br, "blx\t$func", [(ARMtcall GPR:$func)]>, Requires<[IsThumb, HasV5T, IsNotDarwin]>, @@ -257,7 +322,7 @@ let isCall = 1, // ARMv4T def tBX : TIx2<{?,?,?,?,?}, {?,?}, ?, - (outs), (ins tGPR:$func, variable_ops), IIC_Br, + (outs), (ins tGPR:$func, variable_ops), IIC_Br, "mov\tlr, pc\n\tbx\t$func", [(ARMcall_nolink tGPR:$func)]>, Requires<[IsThumb1Only, IsNotDarwin]>; @@ -271,20 +336,20 @@ let isCall = 1, D24, D25, D26, D27, D28, D29, D30, D31, CPSR, FPSCR] in { // Also used for Thumb2 def tBLr9 : TIx2<0b11110, 0b11, 1, - (outs), (ins i32imm:$func, variable_ops), IIC_Br, + (outs), (ins i32imm:$func, variable_ops), IIC_Br, "bl\t${func:call}", [(ARMtcall tglobaladdr:$func)]>, Requires<[IsThumb, IsDarwin]>; // ARMv5T and above, also used for Thumb2 def tBLXi_r9 : TIx2<0b11110, 0b11, 0, - (outs), (ins i32imm:$func, variable_ops), IIC_Br, + (outs), (ins i32imm:$func, variable_ops), IIC_Br, "blx\t${func:call}", [(ARMcall tglobaladdr:$func)]>, Requires<[IsThumb, HasV5T, IsDarwin]>; // Also used for Thumb2 - def tBLXr_r9 : TI<(outs), (ins GPR:$func, variable_ops), IIC_Br, + def tBLXr_r9 : TI<(outs), (ins GPR:$func, variable_ops), IIC_Br, "blx\t$func", [(ARMtcall GPR:$func)]>, Requires<[IsThumb, HasV5T, IsDarwin]>, @@ -292,7 +357,7 @@ let isCall = 1, // ARMv4T def tBXr9 : TIx2<{?,?,?,?,?}, {?,?}, ?, - (outs), (ins tGPR:$func, variable_ops), IIC_Br, + (outs), (ins tGPR:$func, variable_ops), IIC_Br, "mov\tlr, pc\n\tbx\t$func", [(ARMcall_nolink tGPR:$func)]>, Requires<[IsThumb1Only, IsDarwin]>; @@ -307,7 +372,7 @@ let isBranch = 1, isTerminator = 1 in { // Far jump let Defs = [LR] in - def tBfar : TIx2<0b11110, 0b11, 1, (outs), (ins brtarget:$target), IIC_Br, + def tBfar : TIx2<0b11110, 0b11, 1, (outs), (ins brtarget:$target), IIC_Br, "bl\t$target\t@ far jump",[]>; def tBR_JTr : T1JTI<(outs), @@ -340,16 +405,34 @@ let isBranch = 1, isTerminator = 1 in { T1Misc<{1,0,?,1,?,?,?}>; } +// A8.6.218 Supervisor Call (Software Interrupt) -- for disassembly only +// A8.6.16 B: Encoding T1 +// If Inst{11-8} == 0b1111 then SEE SVC +let isCall = 1 in { +def tSVC : T1pI<(outs), (ins i32imm:$svc), IIC_Br, "svc", "\t$svc", []>, + Encoding16 { + let Inst{15-12} = 0b1101; + let Inst{11-8} = 0b1111; +} +} + +// A8.6.16 B: Encoding T1 -- for disassembly only +// If Inst{11-8} == 0b1110 then UNDEFINED +def tTRAP : T1I<(outs), (ins), IIC_Br, "trap", []>, Encoding16 { + let Inst{15-12} = 0b1101; + let Inst{11-8} = 0b1110; +} + //===----------------------------------------------------------------------===// // Load Store Instructions. // -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in -def tLDR : T1pI4<(outs tGPR:$dst), (ins t_addrmode_s4:$addr), IIC_iLoadr, +let canFoldAsLoad = 1, isReMaterializable = 1 in +def tLDR : T1pI4<(outs tGPR:$dst), (ins t_addrmode_s4:$addr), IIC_iLoadr, "ldr", "\t$dst, $addr", [(set tGPR:$dst, (load t_addrmode_s4:$addr))]>, T1LdSt<0b100>; -def tLDRi: T1pI4<(outs tGPR:$dst), (ins t_addrmode_s4:$addr), IIC_iLoadr, +def tLDRi: T1pI4<(outs tGPR:$dst), (ins t_addrmode_s4:$addr), IIC_iLoadr, "ldr", "\t$dst, $addr", []>, T1LdSt4Imm<{1,?,?}>; @@ -399,15 +482,14 @@ def tRestore : T1pIs<(outs tGPR:$dst), (ins t_addrmode_sp:$addr), IIC_iLoadi, // Load tconstpool // FIXME: Use ldr.n to work around a Darwin assembler bug. -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def tLDRpci : T1pIs<(outs tGPR:$dst), (ins i32imm:$addr), IIC_iLoadi, "ldr", ".n\t$dst, $addr", [(set tGPR:$dst, (load (ARMWrapper tconstpool:$addr)))]>, T1Encoding<{0,1,0,0,1,?}>; // A6.2 & A8.6.59 // Special LDR for loads from non-pc-relative constpools. -let canFoldAsLoad = 1, mayLoad = 1, isReMaterializable = 1, - mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, mayLoad = 1, isReMaterializable = 1 in def tLDRcp : T1pIs<(outs tGPR:$dst), (ins i32imm:$addr), IIC_iLoadi, "ldr", "\t$dst, $addr", []>, T1LdStSP<{1,?,?}>; @@ -769,7 +851,7 @@ def tUXTH : T1pI<(outs tGPR:$dst), (ins tGPR:$src), IIC_iUNAr, T1Misc<{0,0,1,0,1,0,?}>; -// Conditional move tMOVCCr - Used to implement the Thumb SELECT_CC DAG operation. +// Conditional move tMOVCCr - Used to implement the Thumb SELECT_CC operation. // Expanded after instruction selection into a branch sequence. let usesCustomInserter = 1 in // Expanded after instruction selection. def tMOVCCr_pseudo : diff --git a/lib/Target/ARM/ARMInstrThumb2.td b/lib/Target/ARM/ARMInstrThumb2.td index 55c7aa2..316567d 100644 --- a/lib/Target/ARM/ARMInstrThumb2.td +++ b/lib/Target/ARM/ARMInstrThumb2.td @@ -13,7 +13,7 @@ // IT block predicate field def it_pred : Operand<i32> { - let PrintMethod = "printPredicateOperand"; + let PrintMethod = "printMandatoryPredicateOperand"; } // IT block condition mask @@ -53,10 +53,10 @@ def t2_so_imm_neg_XFORM : SDNodeXForm<imm, [{ // bits [bits 0-7], the 4-bit shift/splat amount is the next 4 bits [bits 8-11]. def t2_so_imm : Operand<i32>, PatLeaf<(imm), [{ - return ARM_AM::getT2SOImmVal((uint32_t)N->getZExtValue()) != -1; + return ARM_AM::getT2SOImmVal((uint32_t)N->getZExtValue()) != -1; }]>; -// t2_so_imm_not - Match an immediate that is a complement +// t2_so_imm_not - Match an immediate that is a complement // of a t2_so_imm. def t2_so_imm_not : Operand<i32>, PatLeaf<(imm), [{ @@ -114,13 +114,13 @@ def imm0_4095 : Operand<i32>, return (uint32_t)N->getZExtValue() < 4096; }]>; -def imm0_4095_neg : PatLeaf<(i32 imm), [{ - return (uint32_t)(-N->getZExtValue()) < 4096; -}], imm_neg_XFORM>; +def imm0_4095_neg : PatLeaf<(i32 imm), [{ + return (uint32_t)(-N->getZExtValue()) < 4096; +}], imm_neg_XFORM>; def imm0_255_neg : PatLeaf<(i32 imm), [{ return (uint32_t)(-N->getZExtValue()) < 255; -}], imm_neg_XFORM>; +}], imm_neg_XFORM>; // Define Thumb2 specific addressing modes. @@ -208,7 +208,7 @@ multiclass T2I_un_irs<bits<4> opcod, string opc, PatFrag opnode, /// T2I_bin_irs - Defines a set of (op reg, {so_imm|r|so_reg}) patterns for a // binary operation that produces a value. These are predicable and can be /// changed to modify CPSR. -multiclass T2I_bin_irs<bits<4> opcod, string opc, PatFrag opnode, +multiclass T2I_bin_irs<bits<4> opcod, string opc, PatFrag opnode, bit Commutable = 0, string wide =""> { // shifted imm def ri : T2sI<(outs GPR:$dst), (ins GPR:$lhs, t2_so_imm:$rhs), IIC_iALUi, @@ -368,15 +368,16 @@ multiclass T2I_bin_ii12rs<bits<3> op23_21, string opc, PatFrag opnode, } /// T2I_adde_sube_irs - Defines a set of (op reg, {so_imm|r|so_reg}) patterns -/// for a binary operation that produces a value and use and define the carry +/// for a binary operation that produces a value and use the carry /// bit. It's not predicable. let Uses = [CPSR] in { -multiclass T2I_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, bit Commutable = 0> { +multiclass T2I_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, + bit Commutable = 0> { // shifted imm def ri : T2sI<(outs GPR:$dst), (ins GPR:$lhs, t2_so_imm:$rhs), IIC_iALUi, opc, "\t$dst, $lhs, $rhs", [(set GPR:$dst, (opnode GPR:$lhs, t2_so_imm:$rhs))]>, - Requires<[IsThumb2, CarryDefIsUnused]> { + Requires<[IsThumb2]> { let Inst{31-27} = 0b11110; let Inst{25} = 0; let Inst{24-21} = opcod; @@ -387,7 +388,7 @@ multiclass T2I_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, bit Comm def rr : T2sI<(outs GPR:$dst), (ins GPR:$lhs, GPR:$rhs), IIC_iALUr, opc, ".w\t$dst, $lhs, $rhs", [(set GPR:$dst, (opnode GPR:$lhs, GPR:$rhs))]>, - Requires<[IsThumb2, CarryDefIsUnused]> { + Requires<[IsThumb2]> { let isCommutable = Commutable; let Inst{31-27} = 0b11101; let Inst{26-25} = 0b01; @@ -401,19 +402,23 @@ multiclass T2I_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, bit Comm def rs : T2sI<(outs GPR:$dst), (ins GPR:$lhs, t2_so_reg:$rhs), IIC_iALUsi, opc, ".w\t$dst, $lhs, $rhs", [(set GPR:$dst, (opnode GPR:$lhs, t2_so_reg:$rhs))]>, - Requires<[IsThumb2, CarryDefIsUnused]> { + Requires<[IsThumb2]> { let Inst{31-27} = 0b11101; let Inst{26-25} = 0b01; let Inst{24-21} = opcod; let Inst{20} = 0; // The S bit. } - // Carry setting variants +} + +// Carry setting variants +let Defs = [CPSR] in { +multiclass T2I_adde_sube_s_irs<bits<4> opcod, string opc, PatFrag opnode, + bit Commutable = 0> { // shifted imm - def Sri : T2XI<(outs GPR:$dst), (ins GPR:$lhs, t2_so_imm:$rhs), IIC_iALUi, - !strconcat(opc, "s\t$dst, $lhs, $rhs"), - [(set GPR:$dst, (opnode GPR:$lhs, t2_so_imm:$rhs))]>, - Requires<[IsThumb2, CarryDefIsUsed]> { - let Defs = [CPSR]; + def ri : T2sI<(outs GPR:$dst), (ins GPR:$lhs, t2_so_imm:$rhs), IIC_iALUi, + opc, "\t$dst, $lhs, $rhs", + [(set GPR:$dst, (opnode GPR:$lhs, t2_so_imm:$rhs))]>, + Requires<[IsThumb2]> { let Inst{31-27} = 0b11110; let Inst{25} = 0; let Inst{24-21} = opcod; @@ -421,11 +426,10 @@ multiclass T2I_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, bit Comm let Inst{15} = 0; } // register - def Srr : T2XI<(outs GPR:$dst), (ins GPR:$lhs, GPR:$rhs), IIC_iALUr, - !strconcat(opc, "s.w\t$dst, $lhs, $rhs"), - [(set GPR:$dst, (opnode GPR:$lhs, GPR:$rhs))]>, - Requires<[IsThumb2, CarryDefIsUsed]> { - let Defs = [CPSR]; + def rr : T2sI<(outs GPR:$dst), (ins GPR:$lhs, GPR:$rhs), IIC_iALUr, + opc, ".w\t$dst, $lhs, $rhs", + [(set GPR:$dst, (opnode GPR:$lhs, GPR:$rhs))]>, + Requires<[IsThumb2]> { let isCommutable = Commutable; let Inst{31-27} = 0b11101; let Inst{26-25} = 0b01; @@ -436,11 +440,10 @@ multiclass T2I_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, bit Comm let Inst{5-4} = 0b00; // type } // shifted register - def Srs : T2XI<(outs GPR:$dst), (ins GPR:$lhs, t2_so_reg:$rhs), IIC_iALUsi, - !strconcat(opc, "s.w\t$dst, $lhs, $rhs"), - [(set GPR:$dst, (opnode GPR:$lhs, t2_so_reg:$rhs))]>, - Requires<[IsThumb2, CarryDefIsUsed]> { - let Defs = [CPSR]; + def rs : T2sI<(outs GPR:$dst), (ins GPR:$lhs, t2_so_reg:$rhs), IIC_iALUsi, + opc, ".w\t$dst, $lhs, $rhs", + [(set GPR:$dst, (opnode GPR:$lhs, t2_so_reg:$rhs))]>, + Requires<[IsThumb2]> { let Inst{31-27} = 0b11101; let Inst{26-25} = 0b01; let Inst{24-21} = opcod; @@ -448,6 +451,7 @@ multiclass T2I_adde_sube_irs<bits<4> opcod, string opc, PatFrag opnode, bit Comm } } } +} /// T2I_rbin_s_is - Same as T2I_rbin_is except sets 's' bit. let Defs = [CPSR] in { @@ -626,19 +630,6 @@ multiclass T2I_st<bits<2> opcod, string opc, PatFrag opnode> { } } -/// T2I_picld - Defines the PIC load pattern. -class T2I_picld<string opc, PatFrag opnode> : - T2I<(outs GPR:$dst), (ins addrmodepc:$addr), IIC_iLoadi, - !strconcat("\n${addr:label}:\n\t", opc), "\t$dst, $addr", - [(set GPR:$dst, (opnode addrmodepc:$addr))]>; - -/// T2I_picst - Defines the PIC store pattern. -class T2I_picst<string opc, PatFrag opnode> : - T2I<(outs), (ins GPR:$src, addrmodepc:$addr), IIC_iStorer, - !strconcat("\n${addr:label}:\n\t", opc), "\t$src, $addr", - [(opnode GPR:$src, addrmodepc:$addr)]>; - - /// T2I_unary_rrot - A unary operation with two forms: one whose operand is a /// register and one whose operand is a register rotated by 8/16/24. multiclass T2I_unary_rrot<bits<3> opcod, string opc, PatFrag opnode> { @@ -666,6 +657,31 @@ multiclass T2I_unary_rrot<bits<3> opcod, string opc, PatFrag opnode> { } } +// DO variant - disassembly only, no pattern + +multiclass T2I_unary_rrot_DO<bits<3> opcod, string opc> { + def r : T2I<(outs GPR:$dst), (ins GPR:$src), IIC_iUNAr, + opc, "\t$dst, $src", []> { + let Inst{31-27} = 0b11111; + let Inst{26-23} = 0b0100; + let Inst{22-20} = opcod; + let Inst{19-16} = 0b1111; // Rn + let Inst{15-12} = 0b1111; + let Inst{7} = 1; + let Inst{5-4} = 0b00; // rotate + } + def r_rot : T2I<(outs GPR:$dst), (ins GPR:$src, i32imm:$rot), IIC_iUNAsi, + opc, "\t$dst, $src, ror $rot", []> { + let Inst{31-27} = 0b11111; + let Inst{26-23} = 0b0100; + let Inst{22-20} = opcod; + let Inst{19-16} = 0b1111; // Rn + let Inst{15-12} = 0b1111; + let Inst{7} = 1; + let Inst{5-4} = {?,?}; // rotate + } +} + /// T2I_bin_rrot - A binary operation with two forms: one whose operand is a /// register and one whose operand is a register rotated by 8/16/24. multiclass T2I_bin_rrot<bits<3> opcod, string opc, PatFrag opnode> { @@ -692,6 +708,29 @@ multiclass T2I_bin_rrot<bits<3> opcod, string opc, PatFrag opnode> { } } +// DO variant - disassembly only, no pattern + +multiclass T2I_bin_rrot_DO<bits<3> opcod, string opc> { + def rr : T2I<(outs GPR:$dst), (ins GPR:$LHS, GPR:$RHS), IIC_iALUr, + opc, "\t$dst, $LHS, $RHS", []> { + let Inst{31-27} = 0b11111; + let Inst{26-23} = 0b0100; + let Inst{22-20} = opcod; + let Inst{15-12} = 0b1111; + let Inst{7} = 1; + let Inst{5-4} = 0b00; // rotate + } + def rr_rot : T2I<(outs GPR:$dst), (ins GPR:$LHS, GPR:$RHS, i32imm:$rot), + IIC_iALUsr, opc, "\t$dst, $LHS, $RHS, ror $rot", []> { + let Inst{31-27} = 0b11111; + let Inst{26-23} = 0b0100; + let Inst{22-20} = opcod; + let Inst{15-12} = 0b1111; + let Inst{7} = 1; + let Inst{5-4} = {?,?}; // rotate + } +} + //===----------------------------------------------------------------------===// // Instructions //===----------------------------------------------------------------------===// @@ -734,7 +773,7 @@ def t2ADDrSPi : T2sI<(outs GPR:$dst), (ins GPR:$sp, t2_so_imm:$imm), let Inst{19-16} = 0b1101; // Rn = sp let Inst{15} = 0; } -def t2ADDrSPi12 : T2I<(outs GPR:$dst), (ins GPR:$sp, imm0_4095:$imm), +def t2ADDrSPi12 : T2I<(outs GPR:$dst), (ins GPR:$sp, imm0_4095:$imm), IIC_iALUi, "addw", "\t$dst, $sp, $imm", []> { let Inst{31-27} = 0b11110; let Inst{25} = 1; @@ -787,6 +826,25 @@ def t2SUBrSPs : T2sI<(outs GPR:$dst), (ins GPR:$sp, t2_so_reg:$rhs), let Inst{15} = 0; } +// Signed and unsigned division, for disassembly only +def t2SDIV : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b), IIC_iALUi, + "sdiv", "\t$dst, $a, $b", []> { + let Inst{31-27} = 0b11111; + let Inst{26-21} = 0b011100; + let Inst{20} = 0b1; + let Inst{15-12} = 0b1111; + let Inst{7-4} = 0b1111; +} + +def t2UDIV : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b), IIC_iALUi, + "udiv", "\t$dst, $a, $b", []> { + let Inst{31-27} = 0b11111; + let Inst{26-21} = 0b011101; + let Inst{20} = 0b1; + let Inst{15-12} = 0b1111; + let Inst{7-4} = 0b1111; +} + // Pseudo instruction that will expand into a t2SUBrSPi + a copy. let usesCustomInserter = 1 in { // Expanded after instruction selection. def t2SUBrSPi_ : PseudoInst<(outs GPR:$dst), (ins GPR:$sp, t2_so_imm:$imm), @@ -803,7 +861,7 @@ def t2SUBrSPs_ : PseudoInst<(outs GPR:$dst), (ins GPR:$sp, t2_so_reg:$rhs), // // Load -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in defm t2LDR : T2I_ld<0, 0b10, "ldr", UnOpFrag<(load node:$Src)>>; // Loads with zero extension @@ -926,9 +984,9 @@ def t2LDRSH_POST : T2Iidxldst<1, 0b01, 1, 0, (outs GPR:$dst, GPR:$base_wb), } // Store -defm t2STR : T2I_st<0b10, "str", BinOpFrag<(store node:$LHS, node:$RHS)>>; -defm t2STRB : T2I_st<0b00, "strb", BinOpFrag<(truncstorei8 node:$LHS, node:$RHS)>>; -defm t2STRH : T2I_st<0b01, "strh", BinOpFrag<(truncstorei16 node:$LHS, node:$RHS)>>; +defm t2STR :T2I_st<0b10,"str", BinOpFrag<(store node:$LHS, node:$RHS)>>; +defm t2STRB:T2I_st<0b00,"strb",BinOpFrag<(truncstorei8 node:$LHS, node:$RHS)>>; +defm t2STRH:T2I_st<0b01,"strh",BinOpFrag<(truncstorei16 node:$LHS, node:$RHS)>>; // Store doubleword let mayLoad = 1, hasExtraSrcRegAllocReq = 1 in @@ -989,7 +1047,7 @@ def t2STRB_POST : T2Iidxldst<0, 0b00, 0, 0, (outs GPR:$base_wb), let mayLoad = 1, hasExtraDefRegAllocReq = 1 in def t2LDM : T2XI<(outs), (ins addrmode4:$addr, pred:$p, reglist:$wb, variable_ops), - IIC_iLoadm, "ldm${addr:submode}${p}${addr:wide}\t$addr, $wb", []> { + IIC_iLoadm, "ldm${addr:submode}${p}${addr:wide}\t$addr, $wb", []> { let Inst{31-27} = 0b11101; let Inst{26-25} = 0b00; let Inst{24-23} = {?, ?}; // IA: '01', DB: '10' @@ -1001,7 +1059,7 @@ def t2LDM : T2XI<(outs), let mayStore = 1, hasExtraSrcRegAllocReq = 1 in def t2STM : T2XI<(outs), (ins addrmode4:$addr, pred:$p, reglist:$wb, variable_ops), - IIC_iStorem, "stm${addr:submode}${p}${addr:wide}\t$addr, $wb", []> { + IIC_iStorem, "stm${addr:submode}${p}${addr:wide}\t$addr, $wb", []> { let Inst{31-27} = 0b11101; let Inst{26-25} = 0b00; let Inst{24-23} = {?, ?}; // IA: '01', DB: '10' @@ -1074,13 +1132,15 @@ defm t2SXTB : T2I_unary_rrot<0b100, "sxtb", UnOpFrag<(sext_inreg node:$Src, i8)>>; defm t2SXTH : T2I_unary_rrot<0b000, "sxth", UnOpFrag<(sext_inreg node:$Src, i16)>>; +defm t2SXTB16 : T2I_unary_rrot_DO<0b010, "sxtb16">; defm t2SXTAB : T2I_bin_rrot<0b100, "sxtab", BinOpFrag<(add node:$LHS, (sext_inreg node:$RHS, i8))>>; defm t2SXTAH : T2I_bin_rrot<0b000, "sxtah", BinOpFrag<(add node:$LHS, (sext_inreg node:$RHS,i16))>>; +defm t2SXTAB16 : T2I_bin_rrot_DO<0b010, "sxtab16">; -// TODO: SXT(A){B|H}16 +// TODO: SXT(A){B|H}16 - done for disassembly only // Zero extenders @@ -1101,6 +1161,7 @@ defm t2UXTAB : T2I_bin_rrot<0b101, "uxtab", BinOpFrag<(add node:$LHS, (and node:$RHS, 0x00FF))>>; defm t2UXTAH : T2I_bin_rrot<0b001, "uxtah", BinOpFrag<(add node:$LHS, (and node:$RHS, 0xFFFF))>>; +defm t2UXTAB16 : T2I_bin_rrot_DO<0b011, "uxtab16">; } //===----------------------------------------------------------------------===// @@ -1119,9 +1180,13 @@ defm t2SUBS : T2I_bin_s_irs <0b1101, "sub", BinOpFrag<(subc node:$LHS, node:$RHS)>>; defm t2ADC : T2I_adde_sube_irs<0b1010, "adc", - BinOpFrag<(adde node:$LHS, node:$RHS)>, 1>; + BinOpFrag<(adde_dead_carry node:$LHS, node:$RHS)>, 1>; defm t2SBC : T2I_adde_sube_irs<0b1011, "sbc", - BinOpFrag<(sube node:$LHS, node:$RHS)>>; + BinOpFrag<(sube_dead_carry node:$LHS, node:$RHS)>>; +defm t2ADCS : T2I_adde_sube_s_irs<0b1010, "adc", + BinOpFrag<(adde_live_carry node:$LHS, node:$RHS)>, 1>; +defm t2SBCS : T2I_adde_sube_s_irs<0b1011, "sbc", + BinOpFrag<(sube_live_carry node:$LHS, node:$RHS)>>; // RSB defm t2RSB : T2I_rbin_is <0b1110, "rsb", @@ -1138,6 +1203,155 @@ def : T2Pat<(add GPR:$src, t2_so_imm_neg:$imm), def : T2Pat<(add GPR:$src, imm0_4095_neg:$imm), (t2SUBri12 GPR:$src, imm0_4095_neg:$imm)>; +// Select Bytes -- for disassembly only + +def t2SEL : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b), NoItinerary, "sel", + "\t$dst, $a, $b", []> { + let Inst{31-27} = 0b11111; + let Inst{26-24} = 0b010; + let Inst{23} = 0b1; + let Inst{22-20} = 0b010; + let Inst{15-12} = 0b1111; + let Inst{7} = 0b1; + let Inst{6-4} = 0b000; +} + +// A6.3.13, A6.3.14, A6.3.15 Parallel addition and subtraction (signed/unsigned) +// And Miscellaneous operations -- for disassembly only +class T2I_pam<bits<3> op22_20, bits<4> op7_4, string opc> + : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b), NoItinerary, opc, + "\t$dst, $a, $b", [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11111; + let Inst{26-23} = 0b0101; + let Inst{22-20} = op22_20; + let Inst{15-12} = 0b1111; + let Inst{7-4} = op7_4; +} + +// Saturating add/subtract -- for disassembly only + +def t2QADD : T2I_pam<0b000, 0b1000, "qadd">; +def t2QADD16 : T2I_pam<0b001, 0b0001, "qadd16">; +def t2QADD8 : T2I_pam<0b000, 0b0001, "qadd8">; +def t2QASX : T2I_pam<0b010, 0b0001, "qasx">; +def t2QDADD : T2I_pam<0b000, 0b1001, "qdadd">; +def t2QDSUB : T2I_pam<0b000, 0b1011, "qdsub">; +def t2QSAX : T2I_pam<0b110, 0b0001, "qsax">; +def t2QSUB : T2I_pam<0b000, 0b1010, "qsub">; +def t2QSUB16 : T2I_pam<0b101, 0b0001, "qsub16">; +def t2QSUB8 : T2I_pam<0b100, 0b0001, "qsub8">; +def t2UQADD16 : T2I_pam<0b001, 0b0101, "uqadd16">; +def t2UQADD8 : T2I_pam<0b000, 0b0101, "uqadd8">; +def t2UQASX : T2I_pam<0b010, 0b0101, "uqasx">; +def t2UQSAX : T2I_pam<0b110, 0b0101, "uqsax">; +def t2UQSUB16 : T2I_pam<0b101, 0b0101, "uqsub16">; +def t2UQSUB8 : T2I_pam<0b100, 0b0101, "uqsub8">; + +// Signed/Unsigned add/subtract -- for disassembly only + +def t2SASX : T2I_pam<0b010, 0b0000, "sasx">; +def t2SADD16 : T2I_pam<0b001, 0b0000, "sadd16">; +def t2SADD8 : T2I_pam<0b000, 0b0000, "sadd8">; +def t2SSAX : T2I_pam<0b110, 0b0000, "ssax">; +def t2SSUB16 : T2I_pam<0b101, 0b0000, "ssub16">; +def t2SSUB8 : T2I_pam<0b100, 0b0000, "ssub8">; +def t2UASX : T2I_pam<0b010, 0b0100, "uasx">; +def t2UADD16 : T2I_pam<0b001, 0b0100, "uadd16">; +def t2UADD8 : T2I_pam<0b000, 0b0100, "uadd8">; +def t2USAX : T2I_pam<0b110, 0b0100, "usax">; +def t2USUB16 : T2I_pam<0b101, 0b0100, "usub16">; +def t2USUB8 : T2I_pam<0b100, 0b0100, "usub8">; + +// Signed/Unsigned halving add/subtract -- for disassembly only + +def t2SHASX : T2I_pam<0b010, 0b0010, "shasx">; +def t2SHADD16 : T2I_pam<0b001, 0b0010, "shadd16">; +def t2SHADD8 : T2I_pam<0b000, 0b0010, "shadd8">; +def t2SHSAX : T2I_pam<0b110, 0b0010, "shsax">; +def t2SHSUB16 : T2I_pam<0b101, 0b0010, "shsub16">; +def t2SHSUB8 : T2I_pam<0b100, 0b0010, "shsub8">; +def t2UHASX : T2I_pam<0b010, 0b0110, "uhasx">; +def t2UHADD16 : T2I_pam<0b001, 0b0110, "uhadd16">; +def t2UHADD8 : T2I_pam<0b000, 0b0110, "uhadd8">; +def t2UHSAX : T2I_pam<0b110, 0b0110, "uhsax">; +def t2UHSUB16 : T2I_pam<0b101, 0b0110, "uhsub16">; +def t2UHSUB8 : T2I_pam<0b100, 0b0110, "uhsub8">; + +// Unsigned Sum of Absolute Differences [and Accumulate] -- for disassembly only + +def t2USAD8 : T2I_mac<0, 0b111, 0b0000, (outs GPR:$dst), (ins GPR:$a, GPR:$b), + NoItinerary, "usad8", "\t$dst, $a, $b", []> { + let Inst{15-12} = 0b1111; +} +def t2USADA8 : T2I_mac<0, 0b111, 0b0000, (outs GPR:$dst), + (ins GPR:$a, GPR:$b, GPR:$acc), NoItinerary, "usada8", + "\t$dst, $a, $b, $acc", []>; + +// Signed/Unsigned saturate -- for disassembly only + +def t2SSATlsl : T2I<(outs GPR:$dst), (ins i32imm:$bit_pos,GPR:$a,i32imm:$shamt), + NoItinerary, "ssat", "\t$dst, $bit_pos, $a, lsl $shamt", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{25-22} = 0b1100; + let Inst{20} = 0; + let Inst{15} = 0; + let Inst{21} = 0; // sh = '0' +} + +def t2SSATasr : T2I<(outs GPR:$dst), (ins i32imm:$bit_pos,GPR:$a,i32imm:$shamt), + NoItinerary, "ssat", "\t$dst, $bit_pos, $a, asr $shamt", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{25-22} = 0b1100; + let Inst{20} = 0; + let Inst{15} = 0; + let Inst{21} = 1; // sh = '1' +} + +def t2SSAT16 : T2I<(outs GPR:$dst), (ins i32imm:$bit_pos, GPR:$a), NoItinerary, + "ssat16", "\t$dst, $bit_pos, $a", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{25-22} = 0b1100; + let Inst{20} = 0; + let Inst{15} = 0; + let Inst{21} = 1; // sh = '1' + let Inst{14-12} = 0b000; // imm3 = '000' + let Inst{7-6} = 0b00; // imm2 = '00' +} + +def t2USATlsl : T2I<(outs GPR:$dst), (ins i32imm:$bit_pos,GPR:$a,i32imm:$shamt), + NoItinerary, "usat", "\t$dst, $bit_pos, $a, lsl $shamt", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{25-22} = 0b1110; + let Inst{20} = 0; + let Inst{15} = 0; + let Inst{21} = 0; // sh = '0' +} + +def t2USATasr : T2I<(outs GPR:$dst), (ins i32imm:$bit_pos,GPR:$a,i32imm:$shamt), + NoItinerary, "usat", "\t$dst, $bit_pos, $a, asr $shamt", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{25-22} = 0b1110; + let Inst{20} = 0; + let Inst{15} = 0; + let Inst{21} = 1; // sh = '1' +} + +def t2USAT16 : T2I<(outs GPR:$dst), (ins i32imm:$bit_pos, GPR:$a), NoItinerary, + "usat16", "\t$dst, $bit_pos, $a", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{25-22} = 0b1110; + let Inst{20} = 0; + let Inst{15} = 0; + let Inst{21} = 1; // sh = '1' + let Inst{14-12} = 0b000; // imm3 = '000' + let Inst{7-6} = 0b00; // imm2 = '00' +} //===----------------------------------------------------------------------===// // Shift and rotate Instructions. @@ -1342,6 +1556,8 @@ def t2UMAAL : T2I<(outs GPR:$ldst, GPR:$hdst), (ins GPR:$a, GPR:$b), IIC_iMAC64, } } // neverHasSideEffects +// Rounding variants of the below included for disassembly only + // Most significant word multiply def t2SMMUL : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b), IIC_iMUL32, "smmul", "\t$dst, $a, $b", @@ -1353,6 +1569,15 @@ def t2SMMUL : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b), IIC_iMUL32, let Inst{7-4} = 0b0000; // No Rounding (Inst{4} = 0) } +def t2SMMULR : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b), IIC_iMUL32, + "smmulr", "\t$dst, $a, $b", []> { + let Inst{31-27} = 0b11111; + let Inst{26-23} = 0b0110; + let Inst{22-20} = 0b101; + let Inst{15-12} = 0b1111; // Ra = 0b1111 (no accumulate) + let Inst{7-4} = 0b0001; // Rounding (Inst{4} = 1) +} + def t2SMMLA : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), IIC_iMAC32, "smmla", "\t$dst, $a, $b, $c", [(set GPR:$dst, (add (mulhs GPR:$a, GPR:$b), GPR:$c))]> { @@ -1363,6 +1588,14 @@ def t2SMMLA : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), IIC_iMAC32, let Inst{7-4} = 0b0000; // No Rounding (Inst{4} = 0) } +def t2SMMLAR : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), IIC_iMAC32, + "smmlar", "\t$dst, $a, $b, $c", []> { + let Inst{31-27} = 0b11111; + let Inst{26-23} = 0b0110; + let Inst{22-20} = 0b101; + let Inst{15-12} = {?, ?, ?, ?}; // Ra + let Inst{7-4} = 0b0001; // Rounding (Inst{4} = 1) +} def t2SMMLS : T2I <(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), IIC_iMAC32, "smmls", "\t$dst, $a, $b, $c", @@ -1374,6 +1607,15 @@ def t2SMMLS : T2I <(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), IIC_iMAC32, let Inst{7-4} = 0b0000; // No Rounding (Inst{4} = 0) } +def t2SMMLSR : T2I <(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$c), IIC_iMAC32, + "smmlsr", "\t$dst, $a, $b, $c", []> { + let Inst{31-27} = 0b11111; + let Inst{26-23} = 0b0110; + let Inst{22-20} = 0b110; + let Inst{15-12} = {?, ?, ?, ?}; // Ra + let Inst{7-4} = 0b0001; // Rounding (Inst{4} = 1) +} + multiclass T2I_smul<string opc, PatFrag opnode> { def BB : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b), IIC_iMUL32, !strconcat(opc, "bb"), "\t$dst, $a, $b", @@ -1466,7 +1708,7 @@ multiclass T2I_smla<string opc, PatFrag opnode> { def BT : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$acc), IIC_iMAC16, !strconcat(opc, "bt"), "\t$dst, $a, $b, $acc", [(set GPR:$dst, (add GPR:$acc, (opnode (sext_inreg GPR:$a, i16), - (sra GPR:$b, (i32 16)))))]> { + (sra GPR:$b, (i32 16)))))]> { let Inst{31-27} = 0b11111; let Inst{26-23} = 0b0110; let Inst{22-20} = 0b001; @@ -1490,7 +1732,7 @@ multiclass T2I_smla<string opc, PatFrag opnode> { def TT : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$acc), IIC_iMAC16, !strconcat(opc, "tt"), "\t$dst, $a, $b, $acc", [(set GPR:$dst, (add GPR:$acc, (opnode (sra GPR:$a, (i32 16)), - (sra GPR:$b, (i32 16)))))]> { + (sra GPR:$b, (i32 16)))))]> { let Inst{31-27} = 0b11111; let Inst{26-23} = 0b0110; let Inst{22-20} = 0b001; @@ -1502,7 +1744,7 @@ multiclass T2I_smla<string opc, PatFrag opnode> { def WB : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$acc), IIC_iMAC16, !strconcat(opc, "wb"), "\t$dst, $a, $b, $acc", [(set GPR:$dst, (add GPR:$acc, (sra (opnode GPR:$a, - (sext_inreg GPR:$b, i16)), (i32 16))))]> { + (sext_inreg GPR:$b, i16)), (i32 16))))]> { let Inst{31-27} = 0b11111; let Inst{26-23} = 0b0110; let Inst{22-20} = 0b011; @@ -1514,7 +1756,7 @@ multiclass T2I_smla<string opc, PatFrag opnode> { def WT : T2I<(outs GPR:$dst), (ins GPR:$a, GPR:$b, GPR:$acc), IIC_iMAC16, !strconcat(opc, "wt"), "\t$dst, $a, $b, $acc", [(set GPR:$dst, (add GPR:$acc, (sra (opnode GPR:$a, - (sra GPR:$b, (i32 16))), (i32 16))))]> { + (sra GPR:$b, (i32 16))), (i32 16))))]> { let Inst{31-27} = 0b11111; let Inst{26-23} = 0b0110; let Inst{22-20} = 0b011; @@ -1527,16 +1769,70 @@ multiclass T2I_smla<string opc, PatFrag opnode> { defm t2SMUL : T2I_smul<"smul", BinOpFrag<(mul node:$LHS, node:$RHS)>>; defm t2SMLA : T2I_smla<"smla", BinOpFrag<(mul node:$LHS, node:$RHS)>>; -// TODO: Halfword multiple accumulate long: SMLAL<x><y> -// TODO: Dual halfword multiple: SMUAD, SMUSD, SMLAD, SMLSD, SMLALD, SMLSLD - +// Halfword multiple accumulate long: SMLAL<x><y> -- for disassembly only +def t2SMLALBB : T2I_mac<1, 0b100, 0b1000, (outs GPR:$ldst,GPR:$hdst), + (ins GPR:$a,GPR:$b), IIC_iMAC64, "smlalbb", "\t$ldst, $hdst, $a, $b", + [/* For disassembly only; pattern left blank */]>; +def t2SMLALBT : T2I_mac<1, 0b100, 0b1001, (outs GPR:$ldst,GPR:$hdst), + (ins GPR:$a,GPR:$b), IIC_iMAC64, "smlalbt", "\t$ldst, $hdst, $a, $b", + [/* For disassembly only; pattern left blank */]>; +def t2SMLALTB : T2I_mac<1, 0b100, 0b1010, (outs GPR:$ldst,GPR:$hdst), + (ins GPR:$a,GPR:$b), IIC_iMAC64, "smlaltb", "\t$ldst, $hdst, $a, $b", + [/* For disassembly only; pattern left blank */]>; +def t2SMLALTT : T2I_mac<1, 0b100, 0b1011, (outs GPR:$ldst,GPR:$hdst), + (ins GPR:$a,GPR:$b), IIC_iMAC64, "smlaltt", "\t$ldst, $hdst, $a, $b", + [/* For disassembly only; pattern left blank */]>; + +// Dual halfword multiple: SMUAD, SMUSD, SMLAD, SMLSD, SMLALD, SMLSLD +// These are for disassembly only. + +def t2SMUAD : T2I_mac<0, 0b010, 0b0000, (outs GPR:$dst), (ins GPR:$a, GPR:$b), + IIC_iMAC32, "smuad", "\t$dst, $a, $b", []> { + let Inst{15-12} = 0b1111; +} +def t2SMUADX : T2I_mac<0, 0b010, 0b0001, (outs GPR:$dst), (ins GPR:$a, GPR:$b), + IIC_iMAC32, "smuadx", "\t$dst, $a, $b", []> { + let Inst{15-12} = 0b1111; +} +def t2SMUSD : T2I_mac<0, 0b100, 0b0000, (outs GPR:$dst), (ins GPR:$a, GPR:$b), + IIC_iMAC32, "smusd", "\t$dst, $a, $b", []> { + let Inst{15-12} = 0b1111; +} +def t2SMUSDX : T2I_mac<0, 0b100, 0b0001, (outs GPR:$dst), (ins GPR:$a, GPR:$b), + IIC_iMAC32, "smusdx", "\t$dst, $a, $b", []> { + let Inst{15-12} = 0b1111; +} +def t2SMLAD : T2I_mac<0, 0b010, 0b0000, (outs GPR:$dst), + (ins GPR:$a, GPR:$b, GPR:$acc), IIC_iMAC32, "smlad", + "\t$dst, $a, $b, $acc", []>; +def t2SMLADX : T2I_mac<0, 0b010, 0b0001, (outs GPR:$dst), + (ins GPR:$a, GPR:$b, GPR:$acc), IIC_iMAC32, "smladx", + "\t$dst, $a, $b, $acc", []>; +def t2SMLSD : T2I_mac<0, 0b100, 0b0000, (outs GPR:$dst), + (ins GPR:$a, GPR:$b, GPR:$acc), IIC_iMAC32, "smlsd", + "\t$dst, $a, $b, $acc", []>; +def t2SMLSDX : T2I_mac<0, 0b100, 0b0001, (outs GPR:$dst), + (ins GPR:$a, GPR:$b, GPR:$acc), IIC_iMAC32, "smlsdx", + "\t$dst, $a, $b, $acc", []>; +def t2SMLALD : T2I_mac<1, 0b100, 0b1100, (outs GPR:$ldst,GPR:$hdst), + (ins GPR:$a,GPR:$b), IIC_iMAC64, "smlald", + "\t$ldst, $hdst, $a, $b", []>; +def t2SMLALDX : T2I_mac<1, 0b100, 0b1101, (outs GPR:$ldst,GPR:$hdst), + (ins GPR:$a,GPR:$b), IIC_iMAC64, "smlaldx", + "\t$ldst, $hdst, $a, $b", []>; +def t2SMLSLD : T2I_mac<1, 0b101, 0b1100, (outs GPR:$ldst,GPR:$hdst), + (ins GPR:$a,GPR:$b), IIC_iMAC64, "smlsld", + "\t$ldst, $hdst, $a, $b", []>; +def t2SMLSLDX : T2I_mac<1, 0b101, 0b1101, (outs GPR:$ldst,GPR:$hdst), + (ins GPR:$a,GPR:$b), IIC_iMAC64, "smlsldx", + "\t$ldst, $hdst, $a, $b", []>; //===----------------------------------------------------------------------===// // Misc. Arithmetic Instructions. // -class T2I_misc<bits<2> op1, bits<2> op2, dag oops, dag iops, InstrItinClass itin, - string opc, string asm, list<dag> pattern> +class T2I_misc<bits<2> op1, bits<2> op2, dag oops, dag iops, + InstrItinClass itin, string opc, string asm, list<dag> pattern> : T2I<oops, iops, itin, opc, asm, pattern> { let Inst{31-27} = 0b11111; let Inst{26-22} = 0b01010; @@ -1572,7 +1868,7 @@ def t2REVSH : T2I_misc<0b01, 0b11, (outs GPR:$dst), (ins GPR:$src), IIC_iUNAr, (shl GPR:$src, (i32 8))), i16))]>; def t2PKHBT : T2I<(outs GPR:$dst), (ins GPR:$src1, GPR:$src2, i32imm:$shamt), - IIC_iALUsi, "pkhbt", "\t$dst, $src1, $src2, LSL $shamt", + IIC_iALUsi, "pkhbt", "\t$dst, $src1, $src2, lsl $shamt", [(set GPR:$dst, (or (and GPR:$src1, 0xFFFF), (and (shl GPR:$src2, (i32 imm:$shamt)), 0xFFFF0000)))]> { @@ -1590,7 +1886,7 @@ def : T2Pat<(or (and GPR:$src1, 0xFFFF), (shl GPR:$src2, imm16_31:$shamt)), (t2PKHBT GPR:$src1, GPR:$src2, imm16_31:$shamt)>; def t2PKHTB : T2I<(outs GPR:$dst), (ins GPR:$src1, GPR:$src2, i32imm:$shamt), - IIC_iALUsi, "pkhtb", "\t$dst, $src1, $src2, ASR $shamt", + IIC_iALUsi, "pkhtb", "\t$dst, $src1, $src2, asr $shamt", [(set GPR:$dst, (or (and GPR:$src1, 0xFFFF0000), (and (sra GPR:$src2, imm16_31:$shamt), 0xFFFF)))]> { @@ -1643,7 +1939,7 @@ defm t2TEQ : T2I_cmp_irs<0b0100, "teq", // Conditional moves // FIXME: should be able to write a pattern for ARMcmov, but can't use -// a two-value operand where a dag node expects two operands. :( +// a two-value operand where a dag node expects two operands. :( def t2MOVCCr : T2I<(outs GPR:$dst), (ins GPR:$false, GPR:$true), IIC_iCMOVr, "mov", ".w\t$dst, $true", [/*(set GPR:$dst, (ARMcmov GPR:$false, GPR:$true, imm:$cc, CCR:$ccr))*/]>, @@ -1723,6 +2019,66 @@ def t2Int_SyncBarrierV7 : AInoP<(outs), (ins), } } +// Helper class for multiclass T2MemB -- for disassembly only +class T2I_memb<string opc, string asm> + : T2I<(outs), (ins), NoItinerary, opc, asm, + [/* For disassembly only; pattern left blank */]>, + Requires<[IsThumb2, HasV7]> { + let Inst{31-20} = 0xf3b; + let Inst{15-14} = 0b10; + let Inst{12} = 0; +} + +multiclass T2MemB<bits<4> op7_4, string opc> { + + def st : T2I_memb<opc, "\tst"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b1110; + } + + def ish : T2I_memb<opc, "\tish"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b1011; + } + + def ishst : T2I_memb<opc, "\tishst"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b1010; + } + + def nsh : T2I_memb<opc, "\tnsh"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b0111; + } + + def nshst : T2I_memb<opc, "\tnshst"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b0110; + } + + def osh : T2I_memb<opc, "\tosh"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b0011; + } + + def oshst : T2I_memb<opc, "\toshst"> { + let Inst{7-4} = op7_4; + let Inst{3-0} = 0b0010; + } +} + +// These DMB variants are for disassembly only. +defm t2DMB : T2MemB<0b0101, "dmb">; + +// These DSB variants are for disassembly only. +defm t2DSB : T2MemB<0b0100, "dsb">; + +// ISB has only full system option -- for disassembly only +def t2ISBsy : T2I_memb<"isb", ""> { + let Inst{7-4} = 0b0110; + let Inst{3-0} = 0b1111; +} + class T2I_ldrex<bits<2> opcod, dag oops, dag iops, AddrMode am, SizeFlagVal sz, InstrItinClass itin, string opc, string asm, string cstr, list<dag> pattern, bits<4> rt2 = 0b1111> @@ -1789,6 +2145,16 @@ def t2STREXD : T2I_strex<0b11, (outs GPR:$success), {?, ?, ?, ?}>; } +// Clear-Exclusive is for disassembly only. +def t2CLREX : T2I<(outs), (ins), NoItinerary, "clrex", "", + [/* For disassembly only; pattern left blank */]>, + Requires<[IsARM, HasV7]> { + let Inst{31-20} = 0xf3b; + let Inst{15-14} = 0b10; + let Inst{12} = 0; + let Inst{7-4} = 0b0010; +} + //===----------------------------------------------------------------------===// // TLS Instructions // @@ -1906,6 +2272,24 @@ def t2TBH : let Inst{15-8} = 0b11110000; let Inst{7-4} = 0b0001; // H form } + +// Generic versions of the above two instructions, for disassembly only + +def t2TBBgen : T2I<(outs), (ins GPR:$a, GPR:$b), IIC_Br, + "tbb", "\t[$a, $b]", []>{ + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0001101; + let Inst{15-8} = 0b11110000; + let Inst{7-4} = 0b0000; // B form +} + +def t2TBHgen : T2I<(outs), (ins GPR:$a, GPR:$b), IIC_Br, + "tbh", "\t[$a, $b, lsl #1]", []> { + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0001101; + let Inst{15-8} = 0b11110000; + let Inst{7-4} = 0b0001; // H form +} } // isNotDuplicable, isIndirectBranch } // isBranch, isTerminator, isBarrier @@ -1931,6 +2315,119 @@ def t2IT : Thumb2XI<(outs), (ins it_pred:$cc, it_mask:$mask), let Inst{15-8} = 0b10111111; } +// Branch and Exchange Jazelle -- for disassembly only +// Rm = Inst{19-16} +def t2BXJ : T2I<(outs), (ins GPR:$func), NoItinerary, "bxj", "\t$func", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{26} = 0; + let Inst{25-20} = 0b111100; + let Inst{15-14} = 0b10; + let Inst{12} = 0; +} + +// Change Processor State is a system instruction -- for disassembly only. +// The singleton $opt operand contains the following information: +// opt{4-0} = mode from Inst{4-0} +// opt{5} = changemode from Inst{17} +// opt{8-6} = AIF from Inst{8-6} +// opt{10-9} = imod from Inst{19-18} with 0b10 as enable and 0b11 as disable +def t2CPS : T2XI<(outs),(ins i32imm:$opt), NoItinerary, "cps${opt:cps}", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{26} = 0; + let Inst{25-20} = 0b111010; + let Inst{15-14} = 0b10; + let Inst{12} = 0; +} + +// A6.3.4 Branches and miscellaneous control +// Table A6-14 Change Processor State, and hint instructions +// Helper class for disassembly only. +class T2I_hint<bits<8> op7_0, string opc, string asm> + : T2I<(outs), (ins), NoItinerary, opc, asm, + [/* For disassembly only; pattern left blank */]> { + let Inst{31-20} = 0xf3a; + let Inst{15-14} = 0b10; + let Inst{12} = 0; + let Inst{10-8} = 0b000; + let Inst{7-0} = op7_0; +} + +def t2NOP : T2I_hint<0b00000000, "nop", ".w">; +def t2YIELD : T2I_hint<0b00000001, "yield", ".w">; +def t2WFE : T2I_hint<0b00000010, "wfe", ".w">; +def t2WFI : T2I_hint<0b00000011, "wfi", ".w">; +def t2SEV : T2I_hint<0b00000100, "sev", ".w">; + +def t2DBG : T2I<(outs),(ins i32imm:$opt), NoItinerary, "dbg", "\t$opt", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-20} = 0xf3a; + let Inst{15-14} = 0b10; + let Inst{12} = 0; + let Inst{10-8} = 0b000; + let Inst{7-4} = 0b1111; +} + +// Secure Monitor Call is a system instruction -- for disassembly only +// Option = Inst{19-16} +def t2SMC : T2I<(outs), (ins i32imm:$opt), NoItinerary, "smc", "\t$opt", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{26-20} = 0b1111111; + let Inst{15-12} = 0b1000; +} + +// Store Return State is a system instruction -- for disassembly only +def t2SRSDBW : T2I<(outs),(ins i32imm:$mode),NoItinerary,"srsdb","\tsp!, $mode", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0000010; // W = 1 +} + +def t2SRSDB : T2I<(outs),(ins i32imm:$mode),NoItinerary,"srsdb","\tsp, $mode", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0000000; // W = 0 +} + +def t2SRSIAW : T2I<(outs),(ins i32imm:$mode),NoItinerary,"srsia","\tsp!, $mode", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0011010; // W = 1 +} + +def t2SRSIA : T2I<(outs), (ins i32imm:$mode),NoItinerary,"srsia","\tsp, $mode", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0011000; // W = 0 +} + +// Return From Exception is a system instruction -- for disassembly only +def t2RFEDBW : T2I<(outs), (ins GPR:$base), NoItinerary, "rfedb", "\t$base!", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0000011; // W = 1 +} + +def t2RFEDB : T2I<(outs), (ins GPR:$base), NoItinerary, "rfeab", "\t$base", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0000001; // W = 0 +} + +def t2RFEIAW : T2I<(outs), (ins GPR:$base), NoItinerary, "rfeia", "\t$base!", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0011011; // W = 1 +} + +def t2RFEIA : T2I<(outs), (ins GPR:$base), NoItinerary, "rfeia", "\t$base", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11101; + let Inst{26-20} = 0b0011001; // W = 0 +} + //===----------------------------------------------------------------------===// // Non-Instruction Patterns // @@ -1970,9 +2467,59 @@ def : T2Pat<(ARMWrapperJT tjumptable:$dst, imm:$id), // Pseudo instruction that combines ldr from constpool and add pc. This should // be expanded into two instructions late to allow if-conversion and // scheduling. -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def t2LDRpci_pic : PseudoInst<(outs GPR:$dst), (ins i32imm:$addr, pclabel:$cp), NoItinerary, "@ ldr.w\t$dst, $addr\n$cp:\n\tadd\t$dst, pc", [(set GPR:$dst, (ARMpic_add (load (ARMWrapper tconstpool:$addr)), imm:$cp))]>, Requires<[IsThumb2]>; + +//===----------------------------------------------------------------------===// +// Move between special register and ARM core register -- for disassembly only +// + +// Rd = Instr{11-8} +def t2MRS : T2I<(outs GPR:$dst), (ins), NoItinerary, "mrs", "\t$dst, cpsr", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{26} = 0; + let Inst{25-21} = 0b11111; + let Inst{20} = 0; // The R bit. + let Inst{15-14} = 0b10; + let Inst{12} = 0; +} + +// Rd = Instr{11-8} +def t2MRSsys : T2I<(outs GPR:$dst), (ins), NoItinerary, "mrs", "\t$dst, spsr", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{26} = 0; + let Inst{25-21} = 0b11111; + let Inst{20} = 1; // The R bit. + let Inst{15-14} = 0b10; + let Inst{12} = 0; +} + +// FIXME: mask is ignored for the time being. +// Rn = Inst{19-16} +def t2MSR : T2I<(outs), (ins GPR:$src), NoItinerary, "msr", "\tcpsr, $src", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{26} = 0; + let Inst{25-21} = 0b11100; + let Inst{20} = 0; // The R bit. + let Inst{15-14} = 0b10; + let Inst{12} = 0; +} + +// FIXME: mask is ignored for the time being. +// Rn = Inst{19-16} +def t2MSRsys : T2I<(outs), (ins GPR:$src), NoItinerary, "msr", "\tspsr, $src", + [/* For disassembly only; pattern left blank */]> { + let Inst{31-27} = 0b11110; + let Inst{26} = 0; + let Inst{25-21} = 0b11100; + let Inst{20} = 1; // The R bit. + let Inst{15-14} = 0b10; + let Inst{12} = 0; +} diff --git a/lib/Target/ARM/ARMInstrVFP.td b/lib/Target/ARM/ARMInstrVFP.td index e516593..7c117ed 100644 --- a/lib/Target/ARM/ARMInstrVFP.td +++ b/lib/Target/ARM/ARMInstrVFP.td @@ -54,7 +54,7 @@ def vfp_f64imm : Operand<f64>, // Load / store Instructions. // -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in { +let canFoldAsLoad = 1, isReMaterializable = 1 in { def VLDRD : ADI5<0b1101, 0b01, (outs DPR:$dst), (ins addrmode5:$addr), IIC_fpLoad64, "vldr", ".64\t$dst, $addr", [(set DPR:$dst, (load addrmode5:$addr))]>; diff --git a/lib/Target/ARM/ARMJITInfo.cpp b/lib/Target/ARM/ARMJITInfo.cpp index bef5a06..8c0b720 100644 --- a/lib/Target/ARM/ARMJITInfo.cpp +++ b/lib/Target/ARM/ARMJITInfo.cpp @@ -60,7 +60,7 @@ extern "C" { // whole compilation callback doesn't exist as far as the caller is // concerned, so we can't just preserve the callee saved regs. "stmdb sp!, {r0, r1, r2, r3, lr}\n" -#ifndef __SOFTFP__ +#if (defined(__VFP_FP__) && !defined(__SOFTFP__)) "fstmfdd sp!, {d0, d1, d2, d3, d4, d5, d6, d7}\n" #endif // The LR contains the address of the stub function on entry. @@ -83,7 +83,7 @@ extern "C" { // 6-20 | D0..D7 | Saved VFP registers // +--------+ // -#ifndef __SOFTFP__ +#if (defined(__VFP_FP__) && !defined(__SOFTFP__)) // Restore VFP caller-saved registers. "fldmfdd sp!, {d0, d1, d2, d3, d4, d5, d6, d7}\n" #endif diff --git a/lib/Target/ARM/ARMLoadStoreOptimizer.cpp b/lib/Target/ARM/ARMLoadStoreOptimizer.cpp index 4e2d181..5b4f02d 100644 --- a/lib/Target/ARM/ARMLoadStoreOptimizer.cpp +++ b/lib/Target/ARM/ARMLoadStoreOptimizer.cpp @@ -748,11 +748,19 @@ static bool isMemoryOp(const MachineInstr *MI) { if (MMO->isVolatile()) return false; - // Unaligned ldr/str is emulated by some kernels, but unaligned ldm/stm is not. + // Unaligned ldr/str is emulated by some kernels, but unaligned ldm/stm is + // not. if (MMO->getAlignment() < 4) return false; } + // str <undef> could probably be eliminated entirely, but for now we just want + // to avoid making a mess of it. + // FIXME: Use str <undef> as a wildcard to enable better stm folding. + if (MI->getNumOperands() > 0 && MI->getOperand(0).isReg() && + MI->getOperand(0).isUndef()) + return false; + int Opcode = MI->getOpcode(); switch (Opcode) { default: break; diff --git a/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp b/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp index f60cc33..d6d595c 100644 --- a/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp +++ b/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp @@ -122,6 +122,7 @@ namespace { void printT2AddrModeSoRegOperand(const MachineInstr *MI, int OpNum); void printPredicateOperand(const MachineInstr *MI, int OpNum); + void printMandatoryPredicateOperand(const MachineInstr *MI, int OpNum); void printSBitModifierOperand(const MachineInstr *MI, int OpNum); void printPCLabel(const MachineInstr *MI, int OpNum); void printRegisterList(const MachineInstr *MI, int OpNum); @@ -786,6 +787,12 @@ void ARMAsmPrinter::printPredicateOperand(const MachineInstr *MI, int OpNum) { O << ARMCondCodeToString(CC); } +void ARMAsmPrinter::printMandatoryPredicateOperand(const MachineInstr *MI, + int OpNum) { + ARMCC::CondCodes CC = (ARMCC::CondCodes)MI->getOperand(OpNum).getImm(); + O << ARMCondCodeToString(CC); +} + void ARMAsmPrinter::printSBitModifierOperand(const MachineInstr *MI, int OpNum){ unsigned Reg = MI->getOperand(OpNum).getReg(); if (Reg) { diff --git a/lib/Target/ARM/AsmPrinter/ARMInstPrinter.cpp b/lib/Target/ARM/AsmPrinter/ARMInstPrinter.cpp index d7d8e09..a2084b0 100644 --- a/lib/Target/ARM/AsmPrinter/ARMInstPrinter.cpp +++ b/lib/Target/ARM/AsmPrinter/ARMInstPrinter.cpp @@ -325,6 +325,12 @@ void ARMInstPrinter::printPredicateOperand(const MCInst *MI, unsigned OpNum) { O << ARMCondCodeToString(CC); } +void ARMInstPrinter::printMandatoryPredicateOperand(const MCInst *MI, + unsigned OpNum) { + ARMCC::CondCodes CC = (ARMCC::CondCodes)MI->getOperand(OpNum).getImm(); + O << ARMCondCodeToString(CC); +} + void ARMInstPrinter::printSBitModifierOperand(const MCInst *MI, unsigned OpNum){ if (MI->getOperand(OpNum).getReg()) { assert(MI->getOperand(OpNum).getReg() == ARM::CPSR && diff --git a/lib/Target/ARM/AsmPrinter/ARMInstPrinter.h b/lib/Target/ARM/AsmPrinter/ARMInstPrinter.h index 23a7f05..b7964c9 100644 --- a/lib/Target/ARM/AsmPrinter/ARMInstPrinter.h +++ b/lib/Target/ARM/AsmPrinter/ARMInstPrinter.h @@ -71,6 +71,7 @@ public: void printT2AddrModeSoRegOperand(const MCInst *MI, unsigned OpNum) {} void printPredicateOperand(const MCInst *MI, unsigned OpNum); + void printMandatoryPredicateOperand(const MCInst *MI, unsigned OpNum); void printSBitModifierOperand(const MCInst *MI, unsigned OpNum); void printRegisterList(const MCInst *MI, unsigned OpNum); void printCPInstOperand(const MCInst *MI, unsigned OpNum, diff --git a/lib/Target/ARM/README.txt b/lib/Target/ARM/README.txt index 9efb5a1..57b65cf 100644 --- a/lib/Target/ARM/README.txt +++ b/lib/Target/ARM/README.txt @@ -10,6 +10,8 @@ Reimplement 'select' in terms of 'SEL'. * Implement pre/post increment support. (e.g. PR935) * Implement smarter constant generation for binops with large immediates. +A few ARMv6T2 ops should be pattern matched: BFI, SBFX, and UBFX + //===---------------------------------------------------------------------===// Crazy idea: Consider code that uses lots of 8-bit or 16-bit values. By the diff --git a/lib/Target/ARM/Thumb1RegisterInfo.cpp b/lib/Target/ARM/Thumb1RegisterInfo.cpp index d6630ce..b61ce29 100644 --- a/lib/Target/ARM/Thumb1RegisterInfo.cpp +++ b/lib/Target/ARM/Thumb1RegisterInfo.cpp @@ -450,9 +450,9 @@ Thumb1RegisterInfo::eliminateFrameIndex(MachineBasicBlock::iterator II, Offset -= AFI->getGPRCalleeSavedArea1Offset(); else if (AFI->isGPRCalleeSavedArea2Frame(FrameIndex)) Offset -= AFI->getGPRCalleeSavedArea2Offset(); - else if (hasFP(MF)) { - assert(SPAdj == 0 && "Unexpected"); - // There is alloca()'s in this function, must reference off the frame + else if (MF.getFrameInfo()->hasVarSizedObjects()) { + assert(SPAdj == 0 && hasFP(MF) && "Unexpected"); + // There are alloca()'s in this function, must reference off the frame // pointer instead. FrameReg = getFrameRegister(MF); Offset -= AFI->getFramePtrSpillOffset(); diff --git a/lib/Target/Alpha/AlphaISelDAGToDAG.cpp b/lib/Target/Alpha/AlphaISelDAGToDAG.cpp index d6b17c2..5303d85 100644 --- a/lib/Target/Alpha/AlphaISelDAGToDAG.cpp +++ b/lib/Target/Alpha/AlphaISelDAGToDAG.cpp @@ -159,10 +159,6 @@ namespace { // target-specific node if it hasn't already been changed. SDNode *Select(SDNode *N); - /// InstructionSelect - This callback is invoked by - /// SelectionDAGISel when it has created a SelectionDAG for us to codegen. - virtual void InstructionSelect(); - virtual const char *getPassName() const { return "Alpha DAG->DAG Pattern Instruction Selection"; } @@ -222,20 +218,11 @@ SDNode *AlphaDAGToDAGISel::getGlobalRetAddr() { return CurDAG->getRegister(GlobalRetAddr, TLI.getPointerTy()).getNode(); } -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void AlphaDAGToDAGISel::InstructionSelect() { - // Select target instructions for the DAG. - SelectRoot(*CurDAG); - CurDAG->RemoveDeadNodes(); -} - // Select - Convert the specified operand from a target-independent to a // target-specific node if it hasn't already been changed. SDNode *AlphaDAGToDAGISel::Select(SDNode *N) { - if (N->isMachineOpcode()) { + if (N->isMachineOpcode()) return NULL; // Already selected. - } DebugLoc dl = N->getDebugLoc(); switch (N->getOpcode()) { diff --git a/lib/Target/Alpha/AlphaInstrInfo.td b/lib/Target/Alpha/AlphaInstrInfo.td index 8917e86..341c4a7 100644 --- a/lib/Target/Alpha/AlphaInstrInfo.td +++ b/lib/Target/Alpha/AlphaInstrInfo.td @@ -92,7 +92,7 @@ def immSExt16int : PatLeaf<(imm), [{ //(int)imm fits in a 16 bit sign extended ((int64_t)N->getZExtValue() << 32) >> 32; }], SExt16>; -def zappat : PatFrag<(ops node:$LHS), (and node:$LHS, imm:$L), [{ +def zappat : PatFrag<(ops node:$LHS), (and node:$LHS, imm), [{ ConstantSDNode *RHS = dyn_cast<ConstantSDNode>(N->getOperand(1)); if (!RHS) return 0; uint64_t build = get_zapImm(N->getOperand(0), (uint64_t)RHS->getZExtValue()); @@ -602,9 +602,9 @@ def RPCC : MfcForm<0x18, 0xC000, "rpcc $RA", s_rpcc>; //Read process cycle count def MB : MfcPForm<0x18, 0x4000, "mb", s_imisc>; //memory barrier def WMB : MfcPForm<0x18, 0x4400, "wmb", s_imisc>; //write memory barrier -def : Pat<(membarrier (i64 imm:$ll), (i64 imm:$ls), (i64 imm:$sl), (i64 1), (i64 imm:$dev)), +def : Pat<(membarrier (i64 imm), (i64 imm), (i64 imm), (i64 1), (i64 imm)), (WMB)>; -def : Pat<(membarrier (i64 imm:$ll), (i64 imm:$ls), (i64 imm:$sl), (i64 imm:$ss), (i64 imm:$dev)), +def : Pat<(membarrier (i64 imm), (i64 imm), (i64 imm), (i64 imm), (i64 imm)), (MB)>; //Basic Floating point ops diff --git a/lib/Target/Blackfin/BlackfinISelDAGToDAG.cpp b/lib/Target/Blackfin/BlackfinISelDAGToDAG.cpp index 2c9cc60..c8d71aa 100644 --- a/lib/Target/Blackfin/BlackfinISelDAGToDAG.cpp +++ b/lib/Target/Blackfin/BlackfinISelDAGToDAG.cpp @@ -41,7 +41,7 @@ namespace { BlackfinDAGToDAGISel(BlackfinTargetMachine &TM, CodeGenOpt::Level OptLevel) : SelectionDAGISel(TM, OptLevel) {} - virtual void InstructionSelect(); + virtual void PostprocessISelDAG(); virtual const char *getPassName() const { return "Blackfin DAG->DAG Pattern Instruction Selection"; @@ -72,13 +72,7 @@ FunctionPass *llvm::createBlackfinISelDag(BlackfinTargetMachine &TM, return new BlackfinDAGToDAGISel(TM, OptLevel); } -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void BlackfinDAGToDAGISel::InstructionSelect() { - // Select target instructions for the DAG. - SelectRoot(*CurDAG); - DEBUG(errs() << "Selected selection DAG before regclass fixup:\n"); - DEBUG(CurDAG->dump()); +void BlackfinDAGToDAGISel::PostprocessISelDAG() { FixRegisterClasses(*CurDAG); } diff --git a/lib/Target/CBackend/CBackend.cpp b/lib/Target/CBackend/CBackend.cpp index c1c1d80..10f873f 100644 --- a/lib/Target/CBackend/CBackend.cpp +++ b/lib/Target/CBackend/CBackend.cpp @@ -49,7 +49,6 @@ #include "llvm/System/Host.h" #include "llvm/Config/config.h" #include <algorithm> -#include <sstream> using namespace llvm; extern "C" void LLVMInitializeCBackendTarget() { @@ -153,26 +152,16 @@ namespace { return false; } - raw_ostream &printType(formatted_raw_ostream &Out, - const Type *Ty, + raw_ostream &printType(raw_ostream &Out, const Type *Ty, bool isSigned = false, const std::string &VariableName = "", bool IgnoreName = false, const AttrListPtr &PAL = AttrListPtr()); - std::ostream &printType(std::ostream &Out, const Type *Ty, - bool isSigned = false, - const std::string &VariableName = "", - bool IgnoreName = false, - const AttrListPtr &PAL = AttrListPtr()); - raw_ostream &printSimpleType(formatted_raw_ostream &Out, - const Type *Ty, - bool isSigned, - const std::string &NameSoFar = ""); - std::ostream &printSimpleType(std::ostream &Out, const Type *Ty, - bool isSigned, + raw_ostream &printSimpleType(raw_ostream &Out, const Type *Ty, + bool isSigned, const std::string &NameSoFar = ""); - void printStructReturnPointerFunctionType(formatted_raw_ostream &Out, + void printStructReturnPointerFunctionType(raw_ostream &Out, const AttrListPtr &PAL, const PointerType *Ty); @@ -385,8 +374,8 @@ bool CBackendNameAllUsedStructsAndMergeFunctions::runOnModule(Module &M) { // If this isn't a struct or array type, remove it from our set of types // to name. This simplifies emission later. - if (!isa<StructType>(I->second) && !isa<OpaqueType>(I->second) && - !isa<ArrayType>(I->second)) { + if (!I->second->isStructTy() && !I->second->isOpaqueTy() && + !I->second->isArrayTy()) { TST.remove(I); } else { // If this is not used, remove it from the symbol table. @@ -405,7 +394,7 @@ bool CBackendNameAllUsedStructsAndMergeFunctions::runOnModule(Module &M) { unsigned RenameCounter = 0; for (std::set<const Type *>::const_iterator I = UT.begin(), E = UT.end(); I != E; ++I) - if (isa<StructType>(*I) || isa<ArrayType>(*I)) { + if ((*I)->isStructTy() || (*I)->isArrayTy()) { while (M.addTypeName("unnamed"+utostr(RenameCounter), *I)) ++RenameCounter; Changed = true; @@ -454,11 +443,12 @@ bool CBackendNameAllUsedStructsAndMergeFunctions::runOnModule(Module &M) { /// printStructReturnPointerFunctionType - This is like printType for a struct /// return type, except, instead of printing the type as void (*)(Struct*, ...) /// print it as "Struct (*)(...)", for struct return functions. -void CWriter::printStructReturnPointerFunctionType(formatted_raw_ostream &Out, +void CWriter::printStructReturnPointerFunctionType(raw_ostream &Out, const AttrListPtr &PAL, const PointerType *TheTy) { const FunctionType *FTy = cast<FunctionType>(TheTy->getElementType()); - std::stringstream FunctionInnards; + std::string tstr; + raw_string_ostream FunctionInnards(tstr); FunctionInnards << " (*) ("; bool PrintedType = false; @@ -470,7 +460,7 @@ void CWriter::printStructReturnPointerFunctionType(formatted_raw_ostream &Out, FunctionInnards << ", "; const Type *ArgTy = *I; if (PAL.paramHasAttr(Idx, Attribute::ByVal)) { - assert(isa<PointerType>(ArgTy)); + assert(ArgTy->isPointerTy()); ArgTy = cast<PointerType>(ArgTy)->getElementType(); } printType(FunctionInnards, ArgTy, @@ -484,63 +474,14 @@ void CWriter::printStructReturnPointerFunctionType(formatted_raw_ostream &Out, FunctionInnards << "void"; } FunctionInnards << ')'; - std::string tstr = FunctionInnards.str(); printType(Out, RetTy, - /*isSigned=*/PAL.paramHasAttr(0, Attribute::SExt), tstr); + /*isSigned=*/PAL.paramHasAttr(0, Attribute::SExt), FunctionInnards.str()); } raw_ostream & -CWriter::printSimpleType(formatted_raw_ostream &Out, const Type *Ty, - bool isSigned, +CWriter::printSimpleType(raw_ostream &Out, const Type *Ty, bool isSigned, const std::string &NameSoFar) { - assert((Ty->isPrimitiveType() || Ty->isIntegerTy() || isa<VectorType>(Ty)) && - "Invalid type for printSimpleType"); - switch (Ty->getTypeID()) { - case Type::VoidTyID: return Out << "void " << NameSoFar; - case Type::IntegerTyID: { - unsigned NumBits = cast<IntegerType>(Ty)->getBitWidth(); - if (NumBits == 1) - return Out << "bool " << NameSoFar; - else if (NumBits <= 8) - return Out << (isSigned?"signed":"unsigned") << " char " << NameSoFar; - else if (NumBits <= 16) - return Out << (isSigned?"signed":"unsigned") << " short " << NameSoFar; - else if (NumBits <= 32) - return Out << (isSigned?"signed":"unsigned") << " int " << NameSoFar; - else if (NumBits <= 64) - return Out << (isSigned?"signed":"unsigned") << " long long "<< NameSoFar; - else { - assert(NumBits <= 128 && "Bit widths > 128 not implemented yet"); - return Out << (isSigned?"llvmInt128":"llvmUInt128") << " " << NameSoFar; - } - } - case Type::FloatTyID: return Out << "float " << NameSoFar; - case Type::DoubleTyID: return Out << "double " << NameSoFar; - // Lacking emulation of FP80 on PPC, etc., we assume whichever of these is - // present matches host 'long double'. - case Type::X86_FP80TyID: - case Type::PPC_FP128TyID: - case Type::FP128TyID: return Out << "long double " << NameSoFar; - - case Type::VectorTyID: { - const VectorType *VTy = cast<VectorType>(Ty); - return printSimpleType(Out, VTy->getElementType(), isSigned, - " __attribute__((vector_size(" + - utostr(TD->getTypeAllocSize(VTy)) + " ))) " + NameSoFar); - } - - default: -#ifndef NDEBUG - errs() << "Unknown primitive type: " << *Ty << "\n"; -#endif - llvm_unreachable(0); - } -} - -std::ostream & -CWriter::printSimpleType(std::ostream &Out, const Type *Ty, bool isSigned, - const std::string &NameSoFar) { - assert((Ty->isPrimitiveType() || Ty->isIntegerTy() || isa<VectorType>(Ty)) && + assert((Ty->isPrimitiveType() || Ty->isIntegerTy() || Ty->isVectorTy()) && "Invalid type for printSimpleType"); switch (Ty->getTypeID()) { case Type::VoidTyID: return Out << "void " << NameSoFar; @@ -587,120 +528,16 @@ CWriter::printSimpleType(std::ostream &Out, const Type *Ty, bool isSigned, // Pass the Type* and the variable name and this prints out the variable // declaration. // -raw_ostream &CWriter::printType(formatted_raw_ostream &Out, - const Type *Ty, +raw_ostream &CWriter::printType(raw_ostream &Out, const Type *Ty, bool isSigned, const std::string &NameSoFar, bool IgnoreName, const AttrListPtr &PAL) { - if (Ty->isPrimitiveType() || Ty->isIntegerTy() || isa<VectorType>(Ty)) { - printSimpleType(Out, Ty, isSigned, NameSoFar); - return Out; - } - - // Check to see if the type is named. - if (!IgnoreName || isa<OpaqueType>(Ty)) { - std::map<const Type *, std::string>::iterator I = TypeNames.find(Ty); - if (I != TypeNames.end()) return Out << I->second << ' ' << NameSoFar; - } - - switch (Ty->getTypeID()) { - case Type::FunctionTyID: { - const FunctionType *FTy = cast<FunctionType>(Ty); - std::stringstream FunctionInnards; - FunctionInnards << " (" << NameSoFar << ") ("; - unsigned Idx = 1; - for (FunctionType::param_iterator I = FTy->param_begin(), - E = FTy->param_end(); I != E; ++I) { - const Type *ArgTy = *I; - if (PAL.paramHasAttr(Idx, Attribute::ByVal)) { - assert(isa<PointerType>(ArgTy)); - ArgTy = cast<PointerType>(ArgTy)->getElementType(); - } - if (I != FTy->param_begin()) - FunctionInnards << ", "; - printType(FunctionInnards, ArgTy, - /*isSigned=*/PAL.paramHasAttr(Idx, Attribute::SExt), ""); - ++Idx; - } - if (FTy->isVarArg()) { - if (FTy->getNumParams()) - FunctionInnards << ", ..."; - } else if (!FTy->getNumParams()) { - FunctionInnards << "void"; - } - FunctionInnards << ')'; - std::string tstr = FunctionInnards.str(); - printType(Out, FTy->getReturnType(), - /*isSigned=*/PAL.paramHasAttr(0, Attribute::SExt), tstr); - return Out; - } - case Type::StructTyID: { - const StructType *STy = cast<StructType>(Ty); - Out << NameSoFar + " {\n"; - unsigned Idx = 0; - for (StructType::element_iterator I = STy->element_begin(), - E = STy->element_end(); I != E; ++I) { - Out << " "; - printType(Out, *I, false, "field" + utostr(Idx++)); - Out << ";\n"; - } - Out << '}'; - if (STy->isPacked()) - Out << " __attribute__ ((packed))"; - return Out; - } - - case Type::PointerTyID: { - const PointerType *PTy = cast<PointerType>(Ty); - std::string ptrName = "*" + NameSoFar; - - if (isa<ArrayType>(PTy->getElementType()) || - isa<VectorType>(PTy->getElementType())) - ptrName = "(" + ptrName + ")"; - - if (!PAL.isEmpty()) - // Must be a function ptr cast! - return printType(Out, PTy->getElementType(), false, ptrName, true, PAL); - return printType(Out, PTy->getElementType(), false, ptrName); - } - - case Type::ArrayTyID: { - const ArrayType *ATy = cast<ArrayType>(Ty); - unsigned NumElements = ATy->getNumElements(); - if (NumElements == 0) NumElements = 1; - // Arrays are wrapped in structs to allow them to have normal - // value semantics (avoiding the array "decay"). - Out << NameSoFar << " { "; - printType(Out, ATy->getElementType(), false, - "array[" + utostr(NumElements) + "]"); - return Out << "; }"; - } - - case Type::OpaqueTyID: { - std::string TyName = "struct opaque_" + itostr(OpaqueCounter++); - assert(TypeNames.find(Ty) == TypeNames.end()); - TypeNames[Ty] = TyName; - return Out << TyName << ' ' << NameSoFar; - } - default: - llvm_unreachable("Unhandled case in getTypeProps!"); - } - - return Out; -} - -// Pass the Type* and the variable name and this prints out the variable -// declaration. -// -std::ostream &CWriter::printType(std::ostream &Out, const Type *Ty, - bool isSigned, const std::string &NameSoFar, - bool IgnoreName, const AttrListPtr &PAL) { - if (Ty->isPrimitiveType() || Ty->isIntegerTy() || isa<VectorType>(Ty)) { + if (Ty->isPrimitiveType() || Ty->isIntegerTy() || Ty->isVectorTy()) { printSimpleType(Out, Ty, isSigned, NameSoFar); return Out; } // Check to see if the type is named. - if (!IgnoreName || isa<OpaqueType>(Ty)) { + if (!IgnoreName || Ty->isOpaqueTy()) { std::map<const Type *, std::string>::iterator I = TypeNames.find(Ty); if (I != TypeNames.end()) return Out << I->second << ' ' << NameSoFar; } @@ -708,14 +545,15 @@ std::ostream &CWriter::printType(std::ostream &Out, const Type *Ty, switch (Ty->getTypeID()) { case Type::FunctionTyID: { const FunctionType *FTy = cast<FunctionType>(Ty); - std::stringstream FunctionInnards; + std::string tstr; + raw_string_ostream FunctionInnards(tstr); FunctionInnards << " (" << NameSoFar << ") ("; unsigned Idx = 1; for (FunctionType::param_iterator I = FTy->param_begin(), E = FTy->param_end(); I != E; ++I) { const Type *ArgTy = *I; if (PAL.paramHasAttr(Idx, Attribute::ByVal)) { - assert(isa<PointerType>(ArgTy)); + assert(ArgTy->isPointerTy()); ArgTy = cast<PointerType>(ArgTy)->getElementType(); } if (I != FTy->param_begin()) @@ -731,9 +569,8 @@ std::ostream &CWriter::printType(std::ostream &Out, const Type *Ty, FunctionInnards << "void"; } FunctionInnards << ')'; - std::string tstr = FunctionInnards.str(); printType(Out, FTy->getReturnType(), - /*isSigned=*/PAL.paramHasAttr(0, Attribute::SExt), tstr); + /*isSigned=*/PAL.paramHasAttr(0, Attribute::SExt), FunctionInnards.str()); return Out; } case Type::StructTyID: { @@ -756,8 +593,8 @@ std::ostream &CWriter::printType(std::ostream &Out, const Type *Ty, const PointerType *PTy = cast<PointerType>(Ty); std::string ptrName = "*" + NameSoFar; - if (isa<ArrayType>(PTy->getElementType()) || - isa<VectorType>(PTy->getElementType())) + if (PTy->getElementType()->isArrayTy() || + PTy->getElementType()->isVectorTy()) ptrName = "(" + ptrName + ")"; if (!PAL.isEmpty()) @@ -1144,7 +981,7 @@ void CWriter::printConstant(Constant *CPV, bool Static) { Out << "(("; printType(Out, CPV->getType()); // sign doesn't matter Out << ")/*UNDEF*/"; - if (!isa<VectorType>(CPV->getType())) { + if (!CPV->getType()->isVectorTy()) { Out << "0)"; } else { Out << "{})"; @@ -1660,7 +1497,7 @@ void CWriter::writeOperandWithCast(Value* Operand, const ICmpInst &Cmp) { // If the operand was a pointer, convert to a large integer type. const Type* OpTy = Operand->getType(); - if (isa<PointerType>(OpTy)) + if (OpTy->isPointerTy()) OpTy = TD->getIntPtrType(Operand->getContext()); Out << "(("; @@ -2102,10 +1939,10 @@ bool CWriter::doInitialization(Module &M) { // complete. If the value is an aggregate, print out { 0 }, and let // the compiler figure out the rest of the zeros. Out << " = " ; - if (isa<StructType>(I->getInitializer()->getType()) || - isa<VectorType>(I->getInitializer()->getType())) { + if (I->getInitializer()->getType()->isStructTy() || + I->getInitializer()->getType()->isVectorTy()) { Out << "{ 0 }"; - } else if (isa<ArrayType>(I->getInitializer()->getType())) { + } else if (I->getInitializer()->getType()->isArrayTy()) { // As with structs and vectors, but with an extra set of braces // because arrays are wrapped in structs. Out << "{ { 0 } }"; @@ -2274,7 +2111,7 @@ void CWriter::printModuleTypes(const TypeSymbolTable &TST) { // Out << "/* Structure contents */\n"; for (I = TST.begin(); I != End; ++I) - if (isa<StructType>(I->second) || isa<ArrayType>(I->second)) + if (I->second->isStructTy() || I->second->isArrayTy()) // Only print out used types! printContainedStructs(I->second, StructPrinted); } @@ -2287,7 +2124,7 @@ void CWriter::printModuleTypes(const TypeSymbolTable &TST) { void CWriter::printContainedStructs(const Type *Ty, std::set<const Type*> &StructPrinted) { // Don't walk through pointers. - if (isa<PointerType>(Ty) || Ty->isPrimitiveType() || Ty->isIntegerTy()) + if (Ty->isPointerTy() || Ty->isPrimitiveType() || Ty->isIntegerTy()) return; // Print all contained types first. @@ -2295,7 +2132,7 @@ void CWriter::printContainedStructs(const Type *Ty, E = Ty->subtype_end(); I != E; ++I) printContainedStructs(*I, StructPrinted); - if (isa<StructType>(Ty) || isa<ArrayType>(Ty)) { + if (Ty->isStructTy() || Ty->isArrayTy()) { // Check to see if we have already printed this struct. if (StructPrinted.insert(Ty).second) { // Print structure type out. @@ -2328,7 +2165,8 @@ void CWriter::printFunctionSignature(const Function *F, bool Prototype) { const FunctionType *FT = cast<FunctionType>(F->getFunctionType()); const AttrListPtr &PAL = F->getAttributes(); - std::stringstream FunctionInnards; + std::string tstr; + raw_string_ostream FunctionInnards(tstr); // Print out the name... FunctionInnards << GetValueName(F) << '('; @@ -2383,7 +2221,7 @@ void CWriter::printFunctionSignature(const Function *F, bool Prototype) { if (PrintedArg) FunctionInnards << ", "; const Type *ArgTy = *I; if (PAL.paramHasAttr(Idx, Attribute::ByVal)) { - assert(isa<PointerType>(ArgTy)); + assert(ArgTy->isPointerTy()); ArgTy = cast<PointerType>(ArgTy)->getElementType(); } printType(FunctionInnards, ArgTy, @@ -2714,7 +2552,7 @@ void CWriter::visitPHINode(PHINode &I) { void CWriter::visitBinaryOperator(Instruction &I) { // binary instructions, shift instructions, setCond instructions. - assert(!isa<PointerType>(I.getType())); + assert(!I.getType()->isPointerTy()); // We must cast the results of binary operations which might be promoted. bool needsCast = false; @@ -3490,7 +3328,7 @@ void CWriter::printGEPExpression(Value *Ptr, gep_type_iterator I, // exposed, like a global, avoid emitting (&foo)[0], just emit foo instead. if (isAddressExposed(Ptr)) { writeOperandInternal(Ptr, Static); - } else if (I != E && isa<StructType>(*I)) { + } else if (I != E && (*I)->isStructTy()) { // If we didn't already emit the first operand, see if we can print it as // P->f instead of "P[0].f" writeOperand(Ptr); @@ -3505,13 +3343,13 @@ void CWriter::printGEPExpression(Value *Ptr, gep_type_iterator I, } for (; I != E; ++I) { - if (isa<StructType>(*I)) { + if ((*I)->isStructTy()) { Out << ".field" << cast<ConstantInt>(I.getOperand())->getZExtValue(); - } else if (isa<ArrayType>(*I)) { + } else if ((*I)->isArrayTy()) { Out << ".array["; writeOperandWithCast(I.getOperand(), Instruction::GetElementPtr); Out << ']'; - } else if (!isa<VectorType>(*I)) { + } else if (!(*I)->isVectorTy()) { Out << '['; writeOperandWithCast(I.getOperand(), Instruction::GetElementPtr); Out << ']'; @@ -3669,7 +3507,7 @@ void CWriter::visitInsertValueInst(InsertValueInst &IVI) { i != e; ++i) { const Type *IndexedTy = ExtractValueInst::getIndexedType(IVI.getOperand(0)->getType(), b, i+1); - if (isa<ArrayType>(IndexedTy)) + if (IndexedTy->isArrayTy()) Out << ".array[" << *i << "]"; else Out << ".field" << *i; @@ -3690,7 +3528,7 @@ void CWriter::visitExtractValueInst(ExtractValueInst &EVI) { i != e; ++i) { const Type *IndexedTy = ExtractValueInst::getIndexedType(EVI.getOperand(0)->getType(), b, i+1); - if (isa<ArrayType>(IndexedTy)) + if (IndexedTy->isArrayTy()) Out << ".array[" << *i << "]"; else Out << ".field" << *i; @@ -3706,7 +3544,8 @@ void CWriter::visitExtractValueInst(ExtractValueInst &EVI) { bool CTargetMachine::addPassesToEmitWholeFile(PassManager &PM, formatted_raw_ostream &o, CodeGenFileType FileType, - CodeGenOpt::Level OptLevel) { + CodeGenOpt::Level OptLevel, + bool DisableVerify) { if (FileType != TargetMachine::CGFT_AssemblyFile) return true; PM.add(createGCLoweringPass()); diff --git a/lib/Target/CBackend/CTargetMachine.h b/lib/Target/CBackend/CTargetMachine.h index 715bbda..d178e7f 100644 --- a/lib/Target/CBackend/CTargetMachine.h +++ b/lib/Target/CBackend/CTargetMachine.h @@ -27,7 +27,8 @@ struct CTargetMachine : public TargetMachine { virtual bool addPassesToEmitWholeFile(PassManager &PM, formatted_raw_ostream &Out, CodeGenFileType FileType, - CodeGenOpt::Level OptLevel); + CodeGenOpt::Level OptLevel, + bool DisableVerify); virtual const TargetData *getTargetData() const { return 0; } }; diff --git a/lib/Target/CellSPU/SPU64InstrInfo.td b/lib/Target/CellSPU/SPU64InstrInfo.td index 06eb149..47cb579 100644 --- a/lib/Target/CellSPU/SPU64InstrInfo.td +++ b/lib/Target/CellSPU/SPU64InstrInfo.td @@ -123,8 +123,8 @@ multiclass CompareLogicalGreaterThan64 { defm I64LGT: CompareLogicalGreaterThan64; def : Pat<(setugt R64C:$rA, R64C:$rB), I64LGTr64.Fragment>; -def : Pat<(setugt (v2i64 VECREG:$rA), (v2i64 VECREG:$rB)), - I64LGTv2i64.Fragment>; +//def : Pat<(setugt (v2i64 VECREG:$rA), (v2i64 VECREG:$rB)), +// I64LGTv2i64.Fragment>; // i64 setult: def : I64SETCCNegCond<setule, I64LGTr64>; @@ -201,8 +201,8 @@ multiclass CompareGreaterThan64 { defm I64GT: CompareLogicalGreaterThan64; def : Pat<(setgt R64C:$rA, R64C:$rB), I64GTr64.Fragment>; -def : Pat<(setgt (v2i64 VECREG:$rA), (v2i64 VECREG:$rB)), - I64GTv2i64.Fragment>; +//def : Pat<(setgt (v2i64 VECREG:$rA), (v2i64 VECREG:$rB)), +// I64GTv2i64.Fragment>; // i64 setult: def : I64SETCCNegCond<setle, I64GTr64>; diff --git a/lib/Target/CellSPU/SPUISelDAGToDAG.cpp b/lib/Target/CellSPU/SPUISelDAGToDAG.cpp index 1ed06e3..396a921 100644 --- a/lib/Target/CellSPU/SPUISelDAGToDAG.cpp +++ b/lib/Target/CellSPU/SPUISelDAGToDAG.cpp @@ -294,15 +294,19 @@ namespace { ((vecVT == MVT::v2i64) && ((SPU::get_vec_i16imm(bvNode, *CurDAG, MVT::i64).getNode() != 0) || (SPU::get_ILHUvec_imm(bvNode, *CurDAG, MVT::i64).getNode() != 0) || - (SPU::get_vec_u18imm(bvNode, *CurDAG, MVT::i64).getNode() != 0)))) - return Select(bvNode); + (SPU::get_vec_u18imm(bvNode, *CurDAG, MVT::i64).getNode() != 0)))) { + HandleSDNode Dummy(SDValue(bvNode, 0)); + if (SDNode *N = Select(bvNode)) + return N; + return Dummy.getValue().getNode(); + } // No, need to emit a constant pool spill: std::vector<Constant*> CV; for (size_t i = 0; i < bvNode->getNumOperands(); ++i) { ConstantSDNode *V = dyn_cast<ConstantSDNode > (bvNode->getOperand(i)); - CV.push_back(const_cast<ConstantInt *> (V->getConstantIntValue())); + CV.push_back(const_cast<ConstantInt *>(V->getConstantIntValue())); } Constant *CP = ConstantVector::get(CV); @@ -311,10 +315,15 @@ namespace { SDValue CGPoolOffset = SPU::LowerConstantPool(CPIdx, *CurDAG, SPUtli.getSPUTargetMachine()); - return SelectCode(CurDAG->getLoad(vecVT, dl, - CurDAG->getEntryNode(), CGPoolOffset, - PseudoSourceValue::getConstantPool(), 0, - false, false, Alignment).getNode()); + + HandleSDNode Dummy(CurDAG->getLoad(vecVT, dl, + CurDAG->getEntryNode(), CGPoolOffset, + PseudoSourceValue::getConstantPool(),0, + false, false, Alignment)); + CurDAG->ReplaceAllUsesWith(SDValue(bvNode, 0), Dummy.getValue()); + if (SDNode *N = SelectCode(Dummy.getValue().getNode())) + return N; + return Dummy.getValue().getNode(); } /// Select - Convert the specified operand from a target-independent to a @@ -390,10 +399,6 @@ namespace { return false; } - /// InstructionSelect - This callback is invoked by - /// SelectionDAGISel when it has created a SelectionDAG for us to codegen. - virtual void InstructionSelect(); - virtual const char *getPassName() const { return "Cell SPU DAG->DAG Pattern Instruction Selection"; } @@ -411,16 +416,6 @@ namespace { }; } -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void -SPUDAGToDAGISel::InstructionSelect() -{ - // Select target instructions for the DAG. - SelectRoot(*CurDAG); - CurDAG->RemoveDeadNodes(); -} - /*! \arg Op The ISD instruction operand \arg N The address to be tested @@ -692,9 +687,8 @@ SPUDAGToDAGISel::Select(SDNode *N) { SDValue Ops[8]; DebugLoc dl = N->getDebugLoc(); - if (N->isMachineOpcode()) { + if (N->isMachineOpcode()) return NULL; // Already selected. - } if (Opc == ISD::FrameIndex) { int FI = cast<FrameIndexSDNode>(N)->getIndex(); @@ -759,43 +753,67 @@ SPUDAGToDAGISel::Select(SDNode *N) { } SDNode *shufMaskLoad = emitBuildVector(shufMask.getNode()); - SDNode *PromoteScalar = - SelectCode(CurDAG->getNode(SPUISD::PREFSLOT2VEC, dl, - Op0VecVT, Op0).getNode()); - + + HandleSDNode PromoteScalar(CurDAG->getNode(SPUISD::PREFSLOT2VEC, dl, + Op0VecVT, Op0)); + + SDValue PromScalar; + if (SDNode *N = SelectCode(PromoteScalar.getValue().getNode())) + PromScalar = SDValue(N, 0); + else + PromScalar = PromoteScalar.getValue(); + SDValue zextShuffle = CurDAG->getNode(SPUISD::SHUFB, dl, OpVecVT, - SDValue(PromoteScalar, 0), - SDValue(PromoteScalar, 0), + PromScalar, PromScalar, SDValue(shufMaskLoad, 0)); - // N.B.: BIT_CONVERT replaces and updates the zextShuffle node, so we - // re-use it in the VEC2PREFSLOT selection without needing to explicitly - // call SelectCode (it's already done for us.) - SelectCode(CurDAG->getNode(ISD::BIT_CONVERT, dl, OpVecVT, zextShuffle).getNode()); - return SelectCode(CurDAG->getNode(SPUISD::VEC2PREFSLOT, dl, OpVT, - zextShuffle).getNode()); + HandleSDNode Dummy2(zextShuffle); + if (SDNode *N = SelectCode(Dummy2.getValue().getNode())) + zextShuffle = SDValue(N, 0); + else + zextShuffle = Dummy2.getValue(); + HandleSDNode Dummy(CurDAG->getNode(SPUISD::VEC2PREFSLOT, dl, OpVT, + zextShuffle)); + + CurDAG->ReplaceAllUsesWith(N, Dummy.getValue().getNode()); + SelectCode(Dummy.getValue().getNode()); + return Dummy.getValue().getNode(); } else if (Opc == ISD::ADD && (OpVT == MVT::i64 || OpVT == MVT::v2i64)) { SDNode *CGLoad = emitBuildVector(getCarryGenerateShufMask(*CurDAG, dl).getNode()); - return SelectCode(CurDAG->getNode(SPUISD::ADD64_MARKER, dl, OpVT, - N->getOperand(0), N->getOperand(1), - SDValue(CGLoad, 0)).getNode()); + HandleSDNode Dummy(CurDAG->getNode(SPUISD::ADD64_MARKER, dl, OpVT, + N->getOperand(0), N->getOperand(1), + SDValue(CGLoad, 0))); + + CurDAG->ReplaceAllUsesWith(N, Dummy.getValue().getNode()); + if (SDNode *N = SelectCode(Dummy.getValue().getNode())) + return N; + return Dummy.getValue().getNode(); } else if (Opc == ISD::SUB && (OpVT == MVT::i64 || OpVT == MVT::v2i64)) { SDNode *CGLoad = emitBuildVector(getBorrowGenerateShufMask(*CurDAG, dl).getNode()); - return SelectCode(CurDAG->getNode(SPUISD::SUB64_MARKER, dl, OpVT, - N->getOperand(0), N->getOperand(1), - SDValue(CGLoad, 0)).getNode()); + HandleSDNode Dummy(CurDAG->getNode(SPUISD::SUB64_MARKER, dl, OpVT, + N->getOperand(0), N->getOperand(1), + SDValue(CGLoad, 0))); + + CurDAG->ReplaceAllUsesWith(N, Dummy.getValue().getNode()); + if (SDNode *N = SelectCode(Dummy.getValue().getNode())) + return N; + return Dummy.getValue().getNode(); } else if (Opc == ISD::MUL && (OpVT == MVT::i64 || OpVT == MVT::v2i64)) { SDNode *CGLoad = emitBuildVector(getCarryGenerateShufMask(*CurDAG, dl).getNode()); - return SelectCode(CurDAG->getNode(SPUISD::MUL64_MARKER, dl, OpVT, - N->getOperand(0), N->getOperand(1), - SDValue(CGLoad, 0)).getNode()); + HandleSDNode Dummy(CurDAG->getNode(SPUISD::MUL64_MARKER, dl, OpVT, + N->getOperand(0), N->getOperand(1), + SDValue(CGLoad, 0))); + CurDAG->ReplaceAllUsesWith(N, Dummy.getValue().getNode()); + if (SDNode *N = SelectCode(Dummy.getValue().getNode())) + return N; + return Dummy.getValue().getNode(); } else if (Opc == ISD::TRUNCATE) { SDValue Op0 = N->getOperand(0); if ((Op0.getOpcode() == ISD::SRA || Op0.getOpcode() == ISD::SRL) @@ -832,17 +850,14 @@ SPUDAGToDAGISel::Select(SDNode *N) { } } } else if (Opc == ISD::SHL) { - if (OpVT == MVT::i64) { + if (OpVT == MVT::i64) return SelectSHLi64(N, OpVT); - } } else if (Opc == ISD::SRL) { - if (OpVT == MVT::i64) { + if (OpVT == MVT::i64) return SelectSRLi64(N, OpVT); - } } else if (Opc == ISD::SRA) { - if (OpVT == MVT::i64) { + if (OpVT == MVT::i64) return SelectSRAi64(N, OpVT); - } } else if (Opc == ISD::FNEG && (OpVT == MVT::f64 || OpVT == MVT::v2f64)) { DebugLoc dl = N->getDebugLoc(); @@ -1224,13 +1239,15 @@ SDNode *SPUDAGToDAGISel::SelectI64Constant(uint64_t Value64, EVT OpVT, ? shufmask.getNode() : emitBuildVector(shufmask.getNode())); - SDNode *shufNode = - Select(CurDAG->getNode(SPUISD::SHUFB, dl, OpVecVT, + SDValue shufNode = + CurDAG->getNode(SPUISD::SHUFB, dl, OpVecVT, SDValue(lhsNode, 0), SDValue(rhsNode, 0), - SDValue(shufMaskNode, 0)).getNode()); - - return CurDAG->getMachineNode(SPU::ORi64_v2i64, dl, OpVT, - SDValue(shufNode, 0)); + SDValue(shufMaskNode, 0)); + HandleSDNode Dummy(shufNode); + SDNode *SN = SelectCode(Dummy.getValue().getNode()); + if (SN == 0) SN = Dummy.getValue().getNode(); + + return CurDAG->getMachineNode(SPU::ORi64_v2i64, dl, OpVT, SDValue(SN, 0)); } else if (i64vec.getOpcode() == ISD::BUILD_VECTOR) { return CurDAG->getMachineNode(SPU::ORi64_v2i64, dl, OpVT, SDValue(emitBuildVector(i64vec.getNode()), 0)); diff --git a/lib/Target/CellSPU/SPUISelLowering.cpp b/lib/Target/CellSPU/SPUISelLowering.cpp index b21eb37..e863ee3 100644 --- a/lib/Target/CellSPU/SPUISelLowering.cpp +++ b/lib/Target/CellSPU/SPUISelLowering.cpp @@ -118,8 +118,7 @@ namespace { TLI.LowerCallTo(InChain, RetTy, isSigned, !isSigned, false, false, 0, TLI.getLibcallCallingConv(LC), false, /*isReturnValueUsed=*/true, - Callee, Args, DAG, Op.getDebugLoc(), - DAG.GetOrdering(InChain.getNode())); + Callee, Args, DAG, Op.getDebugLoc()); return CallInfo.first; } diff --git a/lib/Target/CppBackend/CPPBackend.cpp b/lib/Target/CppBackend/CPPBackend.cpp index e3f2e9f..9c5893c 100644 --- a/lib/Target/CppBackend/CPPBackend.cpp +++ b/lib/Target/CppBackend/CPPBackend.cpp @@ -221,7 +221,7 @@ namespace { APFloat APF = APFloat(CFP->getValueAPF()); // copy if (CFP->getType() == Type::getFloatTy(CFP->getContext())) APF.convert(APFloat::IEEEdouble, APFloat::rmNearestTiesToEven, &ignored); - Out << "ConstantFP::get(getGlobalContext(), "; + Out << "ConstantFP::get(mod->getContext(), "; Out << "APFloat("; #if HAVE_PRINTF_A char Buffer[100]; @@ -346,21 +346,21 @@ namespace { // First, handle the primitive types .. easy if (Ty->isPrimitiveType() || Ty->isIntegerTy()) { switch (Ty->getTypeID()) { - case Type::VoidTyID: return "Type::getVoidTy(getGlobalContext())"; + case Type::VoidTyID: return "Type::getVoidTy(mod->getContext())"; case Type::IntegerTyID: { unsigned BitWidth = cast<IntegerType>(Ty)->getBitWidth(); - return "IntegerType::get(getGlobalContext(), " + utostr(BitWidth) + ")"; + return "IntegerType::get(mod->getContext(), " + utostr(BitWidth) + ")"; } - case Type::X86_FP80TyID: return "Type::getX86_FP80Ty(getGlobalContext())"; - case Type::FloatTyID: return "Type::getFloatTy(getGlobalContext())"; - case Type::DoubleTyID: return "Type::getDoubleTy(getGlobalContext())"; - case Type::LabelTyID: return "Type::getLabelTy(getGlobalContext())"; + case Type::X86_FP80TyID: return "Type::getX86_FP80Ty(mod->getContext())"; + case Type::FloatTyID: return "Type::getFloatTy(mod->getContext())"; + case Type::DoubleTyID: return "Type::getDoubleTy(mod->getContext())"; + case Type::LabelTyID: return "Type::getLabelTy(mod->getContext())"; default: error("Invalid primitive type"); break; } // shouldn't be returned, but make it sensible - return "Type::getVoidTy(getGlobalContext())"; + return "Type::getVoidTy(mod->getContext())"; } // Now, see if we've seen the type before and return that @@ -514,7 +514,7 @@ namespace { TypeMap::const_iterator I = UnresolvedTypes.find(Ty); if (I == UnresolvedTypes.end()) { Out << "PATypeHolder " << typeName; - Out << "_fwd = OpaqueType::get(getGlobalContext());"; + Out << "_fwd = OpaqueType::get(mod->getContext());"; nl(Out); UnresolvedTypes[Ty] = typeName; } @@ -615,7 +615,7 @@ namespace { } case Type::OpaqueTyID: { Out << "OpaqueType* " << typeName; - Out << " = OpaqueType::get(getGlobalContext());"; + Out << " = OpaqueType::get(mod->getContext());"; nl(Out); break; } @@ -751,7 +751,7 @@ namespace { if (const ConstantInt *CI = dyn_cast<ConstantInt>(CV)) { std::string constValue = CI->getValue().toString(10, true); Out << "ConstantInt* " << constName - << " = ConstantInt::get(getGlobalContext(), APInt(" + << " = ConstantInt::get(mod->getContext(), APInt(" << cast<IntegerType>(CI->getType())->getBitWidth() << ", StringRef(\"" << constValue << "\"), 10));"; } else if (isa<ConstantAggregateZero>(CV)) { @@ -769,7 +769,7 @@ namespace { CA->getType()->getElementType() == Type::getInt8Ty(CA->getContext())) { Out << "Constant* " << constName << - " = ConstantArray::get(getGlobalContext(), \""; + " = ConstantArray::get(mod->getContext(), \""; std::string tmp = CA->getAsString(); bool nullTerminate = false; if (tmp[tmp.length()-1] == 0) { @@ -995,7 +995,7 @@ namespace { void CppWriter::printVariableHead(const GlobalVariable *GV) { nl(Out) << "GlobalVariable* " << getCppName(GV); if (is_inline) { - Out << " = mod->getGlobalVariable(getGlobalContext(), "; + Out << " = mod->getGlobalVariable(mod->getContext(), "; printEscapedString(GV->getName()); Out << ", " << getCppName(GV->getType()->getElementType()) << ",true)"; nl(Out) << "if (!" << getCppName(GV) << ") {"; @@ -1094,7 +1094,7 @@ namespace { case Instruction::Ret: { const ReturnInst* ret = cast<ReturnInst>(I); - Out << "ReturnInst::Create(getGlobalContext(), " + Out << "ReturnInst::Create(mod->getContext(), " << (ret->getReturnValue() ? opNames[0] + ", " : "") << bbname << ");"; break; } @@ -1171,7 +1171,7 @@ namespace { } case Instruction::Unreachable: { Out << "new UnreachableInst(" - << "getGlobalContext(), " + << "mod->getContext(), " << bbname << ");"; break; } @@ -1673,7 +1673,7 @@ namespace { BI != BE; ++BI) { std::string bbname(getCppName(BI)); Out << "BasicBlock* " << bbname << - " = BasicBlock::Create(getGlobalContext(), \""; + " = BasicBlock::Create(mod->getContext(), \""; if (BI->hasName()) printEscapedString(BI->getName()); Out << "\"," << getCppName(BI->getParent()) << ",0);"; @@ -2009,7 +2009,8 @@ char CppWriter::ID = 0; bool CPPTargetMachine::addPassesToEmitWholeFile(PassManager &PM, formatted_raw_ostream &o, CodeGenFileType FileType, - CodeGenOpt::Level OptLevel) { + CodeGenOpt::Level OptLevel, + bool DisableVerify) { if (FileType != TargetMachine::CGFT_AssemblyFile) return true; PM.add(new CppWriter(o)); return false; diff --git a/lib/Target/CppBackend/CPPTargetMachine.h b/lib/Target/CppBackend/CPPTargetMachine.h index 1f74f76..b7aae91 100644 --- a/lib/Target/CppBackend/CPPTargetMachine.h +++ b/lib/Target/CppBackend/CPPTargetMachine.h @@ -30,7 +30,8 @@ struct CPPTargetMachine : public TargetMachine { virtual bool addPassesToEmitWholeFile(PassManager &PM, formatted_raw_ostream &Out, CodeGenFileType FileType, - CodeGenOpt::Level OptLevel); + CodeGenOpt::Level OptLevel, + bool DisableVerify); virtual const TargetData *getTargetData() const { return 0; } }; diff --git a/lib/Target/MBlaze/AsmPrinter/CMakeLists.txt b/lib/Target/MBlaze/AsmPrinter/CMakeLists.txt new file mode 100644 index 0000000..cfb2fc8 --- /dev/null +++ b/lib/Target/MBlaze/AsmPrinter/CMakeLists.txt @@ -0,0 +1,9 @@ +include_directories(
+ ${CMAKE_CURRENT_BINARY_DIR}/..
+ ${CMAKE_CURRENT_SOURCE_DIR}/..
+ )
+
+add_llvm_library(LLVMMBlazeAsmPrinter
+ MBlazeAsmPrinter.cpp + )
+add_dependencies(LLVMMBlazeAsmPrinter MBlazeCodeGenTable_gen)
\ No newline at end of file diff --git a/lib/Target/MBlaze/AsmPrinter/MBlazeAsmPrinter.cpp b/lib/Target/MBlaze/AsmPrinter/MBlazeAsmPrinter.cpp new file mode 100644 index 0000000..6fe1026 --- /dev/null +++ b/lib/Target/MBlaze/AsmPrinter/MBlazeAsmPrinter.cpp @@ -0,0 +1,302 @@ +//===-- MBlazeAsmPrinter.cpp - MBlaze LLVM assembly writer ----------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains a printer that converts from our internal representation +// of machine-dependent LLVM code to GAS-format MBlaze assembly language. +// +//===----------------------------------------------------------------------===// + +#define DEBUG_TYPE "mblaze-asm-printer" + +#include "MBlaze.h" +#include "MBlazeSubtarget.h" +#include "MBlazeInstrInfo.h" +#include "MBlazeTargetMachine.h" +#include "MBlazeMachineFunction.h" +#include "llvm/Constants.h" +#include "llvm/DerivedTypes.h" +#include "llvm/Module.h" +#include "llvm/CodeGen/AsmPrinter.h" +#include "llvm/CodeGen/DwarfWriter.h" +#include "llvm/CodeGen/MachineFunctionPass.h" +#include "llvm/CodeGen/MachineConstantPool.h" +#include "llvm/CodeGen/MachineFrameInfo.h" +#include "llvm/CodeGen/MachineInstr.h" +#include "llvm/MC/MCStreamer.h" +#include "llvm/MC/MCAsmInfo.h" +#include "llvm/MC/MCSymbol.h" +#include "llvm/Target/TargetData.h" +#include "llvm/Target/TargetLoweringObjectFile.h" +#include "llvm/Target/TargetMachine.h" +#include "llvm/Target/TargetOptions.h" +#include "llvm/Target/TargetRegistry.h" +#include "llvm/Support/ErrorHandling.h" +#include "llvm/ADT/Statistic.h" +#include "llvm/ADT/StringExtras.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/CommandLine.h" +#include "llvm/Support/FormattedStream.h" +#include "llvm/Support/MathExtras.h" +#include <cctype> + +using namespace llvm; + +namespace { + class MBlazeAsmPrinter : public AsmPrinter { + const MBlazeSubtarget *Subtarget; + public: + explicit MBlazeAsmPrinter(formatted_raw_ostream &O, TargetMachine &TM, + MCContext &Ctx, MCStreamer &Streamer, + const MCAsmInfo *T ) + : AsmPrinter(O, TM, Ctx, Streamer, T) { + Subtarget = &TM.getSubtarget<MBlazeSubtarget>(); + } + + virtual const char *getPassName() const { + return "MBlaze Assembly Printer"; + } + + bool PrintAsmOperand(const MachineInstr *MI, unsigned OpNo, + unsigned AsmVariant, const char *ExtraCode); + void printOperand(const MachineInstr *MI, int opNum); + void printUnsignedImm(const MachineInstr *MI, int opNum); + void printFSLImm(const MachineInstr *MI, int opNum); + void printMemOperand(const MachineInstr *MI, int opNum, + const char *Modifier = 0); + void printFCCOperand(const MachineInstr *MI, int opNum, + const char *Modifier = 0); + void printSavedRegsBitmask(); + void printHex32(unsigned int Value); + + const char *emitCurrentABIString(); + void emitFrameDirective(); + + void printInstruction(const MachineInstr *MI); // autogenerated. + void EmitInstruction(const MachineInstr *MI) { + printInstruction(MI); + O << '\n'; + } + virtual void EmitFunctionBodyStart(); + virtual void EmitFunctionBodyEnd(); + static const char *getRegisterName(unsigned RegNo); + + virtual void EmitFunctionEntryLabel(); + void EmitStartOfAsmFile(Module &M); + }; +} // end of anonymous namespace + +#include "MBlazeGenAsmWriter.inc" + +//===----------------------------------------------------------------------===// +// +// MBlaze Asm Directives +// +// -- Frame directive "frame Stackpointer, Stacksize, RARegister" +// Describe the stack frame. +// +// -- Mask directives "mask bitmask, offset" +// Tells the assembler which registers are saved and where. +// bitmask - contain a little endian bitset indicating which registers are +// saved on function prologue (e.g. with a 0x80000000 mask, the +// assembler knows the register 31 (RA) is saved at prologue. +// offset - the position before stack pointer subtraction indicating where +// the first saved register on prologue is located. (e.g. with a +// +// Consider the following function prologue: +// +// .frame R19,48,R15 +// .mask 0xc0000000,-8 +// addiu R1, R1, -48 +// sw R15, 40(R1) +// sw R19, 36(R1) +// +// With a 0xc0000000 mask, the assembler knows the register 15 (R15) and +// 19 (R19) are saved at prologue. As the save order on prologue is from +// left to right, R15 is saved first. A -8 offset means that after the +// stack pointer subtration, the first register in the mask (R15) will be +// saved at address 48-8=40. +// +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// Mask directives +//===----------------------------------------------------------------------===// + +// Create a bitmask with all callee saved registers for CPU or Floating Point +// registers. For CPU registers consider RA, GP and FP for saving if necessary. +void MBlazeAsmPrinter::printSavedRegsBitmask() { + const TargetRegisterInfo &RI = *TM.getRegisterInfo(); + const MBlazeFunctionInfo *MBlazeFI = MF->getInfo<MBlazeFunctionInfo>(); + + // CPU Saved Registers Bitmasks + unsigned int CPUBitmask = 0; + + // Set the CPU Bitmasks + const MachineFrameInfo *MFI = MF->getFrameInfo(); + const std::vector<CalleeSavedInfo> &CSI = MFI->getCalleeSavedInfo(); + for (unsigned i = 0, e = CSI.size(); i != e; ++i) { + unsigned RegNum = MBlazeRegisterInfo::getRegisterNumbering(CSI[i].getReg()); + if (CSI[i].getRegClass() == MBlaze::CPURegsRegisterClass) + CPUBitmask |= (1 << RegNum); + } + + // Return Address and Frame registers must also be set in CPUBitmask. + if (RI.hasFP(*MF)) + CPUBitmask |= (1 << MBlazeRegisterInfo:: + getRegisterNumbering(RI.getFrameRegister(*MF))); + + if (MFI->hasCalls()) + CPUBitmask |= (1 << MBlazeRegisterInfo:: + getRegisterNumbering(RI.getRARegister())); + + // Print CPUBitmask + O << "\t.mask \t"; printHex32(CPUBitmask); O << ',' + << MBlazeFI->getCPUTopSavedRegOff() << '\n'; +} + +// Print a 32 bit hex number with all numbers. +void MBlazeAsmPrinter::printHex32(unsigned int Value) { + O << "0x"; + for (int i = 7; i >= 0; i--) + O << utohexstr( (Value & (0xF << (i*4))) >> (i*4) ); +} + +//===----------------------------------------------------------------------===// +// Frame and Set directives +//===----------------------------------------------------------------------===// + +/// Frame Directive +void MBlazeAsmPrinter::emitFrameDirective() { + const TargetRegisterInfo &RI = *TM.getRegisterInfo(); + + unsigned stackReg = RI.getFrameRegister(*MF); + unsigned returnReg = RI.getRARegister(); + unsigned stackSize = MF->getFrameInfo()->getStackSize(); + + + O << "\t.frame\t" << getRegisterName(stackReg) + << ',' << stackSize << ',' + << getRegisterName(returnReg) + << '\n'; +} + +void MBlazeAsmPrinter::EmitFunctionEntryLabel() { + O << "\t.ent\t" << *CurrentFnSym << '\n'; + OutStreamer.EmitLabel(CurrentFnSym); +} + +/// EmitFunctionBodyStart - Targets can override this to emit stuff before +/// the first basic block in the function. +void MBlazeAsmPrinter::EmitFunctionBodyStart() { + emitFrameDirective(); + printSavedRegsBitmask(); +} + +/// EmitFunctionBodyEnd - Targets can override this to emit stuff after +/// the last basic block in the function. +void MBlazeAsmPrinter::EmitFunctionBodyEnd() { + O << "\t.end\t" << *CurrentFnSym << '\n'; +} + +// Print out an operand for an inline asm expression. +bool MBlazeAsmPrinter:: +PrintAsmOperand(const MachineInstr *MI, unsigned OpNo, + unsigned AsmVariant,const char *ExtraCode){ + // Does this asm operand have a single letter operand modifier? + if (ExtraCode && ExtraCode[0]) + return true; // Unknown modifier. + + printOperand(MI, OpNo); + return false; +} + +void MBlazeAsmPrinter::printOperand(const MachineInstr *MI, int opNum) { + const MachineOperand &MO = MI->getOperand(opNum); + + switch (MO.getType()) { + case MachineOperand::MO_Register: + O << getRegisterName(MO.getReg()); + break; + + case MachineOperand::MO_Immediate: + O << (int)MO.getImm(); + break; + + case MachineOperand::MO_FPImmediate: { + const ConstantFP* fp = MO.getFPImm(); + printHex32(fp->getValueAPF().bitcastToAPInt().getZExtValue()); + O << ";\t# immediate = " << *fp; + break; + } + + case MachineOperand::MO_MachineBasicBlock: + O << *MO.getMBB()->getSymbol(OutContext); + return; + + case MachineOperand::MO_GlobalAddress: + O << *GetGlobalValueSymbol(MO.getGlobal()); + break; + + case MachineOperand::MO_ExternalSymbol: + O << *GetExternalSymbolSymbol(MO.getSymbolName()); + break; + + case MachineOperand::MO_JumpTableIndex: + O << MAI->getPrivateGlobalPrefix() << "JTI" << getFunctionNumber() + << '_' << MO.getIndex(); + break; + + case MachineOperand::MO_ConstantPoolIndex: + O << MAI->getPrivateGlobalPrefix() << "CPI" + << getFunctionNumber() << "_" << MO.getIndex(); + if (MO.getOffset()) + O << "+" << MO.getOffset(); + break; + + default: + llvm_unreachable("<unknown operand type>"); + } +} + +void MBlazeAsmPrinter::printUnsignedImm(const MachineInstr *MI, int opNum) { + const MachineOperand &MO = MI->getOperand(opNum); + if (MO.getType() == MachineOperand::MO_Immediate) + O << (unsigned int)MO.getImm(); + else + printOperand(MI, opNum); +} + +void MBlazeAsmPrinter::printFSLImm(const MachineInstr *MI, int opNum) { + const MachineOperand &MO = MI->getOperand(opNum); + if (MO.getType() == MachineOperand::MO_Immediate) + O << "rfsl" << (unsigned int)MO.getImm(); + else + printOperand(MI, opNum); +} + +void MBlazeAsmPrinter:: +printMemOperand(const MachineInstr *MI, int opNum, const char *Modifier) { + printOperand(MI, opNum+1); + O << ", "; + printOperand(MI, opNum); +} + +void MBlazeAsmPrinter:: +printFCCOperand(const MachineInstr *MI, int opNum, const char *Modifier) { + const MachineOperand& MO = MI->getOperand(opNum); + O << MBlaze::MBlazeFCCToString((MBlaze::CondCode)MO.getImm()); +} + +void MBlazeAsmPrinter::EmitStartOfAsmFile(Module &M) { +} + +// Force static initialization. +extern "C" void LLVMInitializeMBlazeAsmPrinter() { + RegisterAsmPrinter<MBlazeAsmPrinter> X(TheMBlazeTarget); +} diff --git a/lib/Target/MBlaze/AsmPrinter/Makefile b/lib/Target/MBlaze/AsmPrinter/Makefile new file mode 100644 index 0000000..c8e4d8f --- /dev/null +++ b/lib/Target/MBlaze/AsmPrinter/Makefile @@ -0,0 +1,17 @@ +##===- lib/Target/MBlaze/AsmPrinter/Makefile ---------------*- Makefile -*-===## +# +# The LLVM Compiler Infrastructure +# +# This file is distributed under the University of Illinois Open Source +# License. See LICENSE.TXT for details. +# +##===----------------------------------------------------------------------===## + +LEVEL = ../../../.. +LIBRARYNAME = LLVMMBlazeAsmPrinter + +# Hack: we need to include 'main' MBlaze target directory to grab +# private headers +CPPFLAGS = -I$(PROJ_OBJ_DIR)/.. -I$(PROJ_SRC_DIR)/.. + +include $(LEVEL)/Makefile.common diff --git a/lib/Target/MBlaze/CMakeLists.txt b/lib/Target/MBlaze/CMakeLists.txt new file mode 100644 index 0000000..c93e3df --- /dev/null +++ b/lib/Target/MBlaze/CMakeLists.txt @@ -0,0 +1,27 @@ +set(LLVM_TARGET_DEFINITIONS MBlaze.td) + +tablegen(MBlazeGenRegisterInfo.h.inc -gen-register-desc-header) +tablegen(MBlazeGenRegisterNames.inc -gen-register-enums) +tablegen(MBlazeGenRegisterInfo.inc -gen-register-desc) +tablegen(MBlazeGenInstrNames.inc -gen-instr-enums) +tablegen(MBlazeGenInstrInfo.inc -gen-instr-desc) +tablegen(MBlazeGenAsmWriter.inc -gen-asm-writer) +tablegen(MBlazeGenDAGISel.inc -gen-dag-isel) +tablegen(MBlazeGenCallingConv.inc -gen-callingconv) +tablegen(MBlazeGenSubtarget.inc -gen-subtarget) +tablegen(MBlazeGenIntrinsics.inc -gen-tgt-intrinsic) + +add_llvm_target(MBlazeCodeGen + MBlazeDelaySlotFiller.cpp + MBlazeInstrInfo.cpp + MBlazeISelDAGToDAG.cpp + MBlazeISelLowering.cpp + MBlazeMCAsmInfo.cpp + MBlazeRegisterInfo.cpp + MBlazeSubtarget.cpp + MBlazeTargetMachine.cpp + MBlazeTargetObjectFile.cpp + MBlazeIntrinsicInfo.cpp + ) + +target_link_libraries (LLVMMBlazeCodeGen LLVMSelectionDAG) diff --git a/lib/Target/MBlaze/MBlaze.h b/lib/Target/MBlaze/MBlaze.h new file mode 100644 index 0000000..f9d828b --- /dev/null +++ b/lib/Target/MBlaze/MBlaze.h @@ -0,0 +1,39 @@ +//===-- MBlaze.h - Top-level interface for MBlaze ---------------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the entry points for global functions defined in +// the LLVM MBlaze back-end. +// +//===----------------------------------------------------------------------===// + +#ifndef TARGET_MBLAZE_H +#define TARGET_MBLAZE_H + +#include "llvm/Target/TargetMachine.h" + +namespace llvm { + class MBlazeTargetMachine; + class FunctionPass; + class MachineCodeEmitter; + class formatted_raw_ostream; + + FunctionPass *createMBlazeISelDag(MBlazeTargetMachine &TM); + FunctionPass *createMBlazeDelaySlotFillerPass(MBlazeTargetMachine &TM); + + extern Target TheMBlazeTarget; +} // end namespace llvm; + +// Defines symbolic names for MBlaze registers. This defines a mapping from +// register name to register number. +#include "MBlazeGenRegisterNames.inc" + +// Defines symbolic names for the MBlaze instructions. +#include "MBlazeGenInstrNames.inc" + +#endif diff --git a/lib/Target/MBlaze/MBlaze.td b/lib/Target/MBlaze/MBlaze.td new file mode 100644 index 0000000..1679752 --- /dev/null +++ b/lib/Target/MBlaze/MBlaze.td @@ -0,0 +1,85 @@ +//===- MBlaze.td - Describe the MBlaze Target Machine -----------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// This is the top level entry point for the MBlaze target. +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// Target-independent interfaces +//===----------------------------------------------------------------------===// + +include "llvm/Target/Target.td" + +//===----------------------------------------------------------------------===// +// Register File, Calling Conv, Instruction Descriptions +//===----------------------------------------------------------------------===// + +include "MBlazeRegisterInfo.td" +include "MBlazeSchedule.td" +include "MBlazeIntrinsics.td" +include "MBlazeInstrInfo.td" +include "MBlazeCallingConv.td" + +def MBlazeInstrInfo : InstrInfo { + let TSFlagsFields = []; + let TSFlagsShifts = []; +} + + +//===----------------------------------------------------------------------===// +// Microblaze Subtarget features // +//===----------------------------------------------------------------------===// + +def FeaturePipe3 : SubtargetFeature<"pipe3", "HasPipe3", "true", + "Implements 3-stage pipeline.">; +def FeatureBarrel : SubtargetFeature<"barrel", "HasBarrel", "true", + "Implements barrel shifter.">; +def FeatureDiv : SubtargetFeature<"div", "HasDiv", "true", + "Implements hardware divider.">; +def FeatureMul : SubtargetFeature<"mul", "HasMul", "true", + "Implements hardware multiplier.">; +def FeatureFSL : SubtargetFeature<"fsl", "HasFSL", "true", + "Implements FSL instructions.">; +def FeatureEFSL : SubtargetFeature<"efsl", "HasEFSL", "true", + "Implements extended FSL instructions.">; +def FeatureMSRSet : SubtargetFeature<"msrset", "HasMSRSet", "true", + "Implements MSR register set and clear.">; +def FeatureException : SubtargetFeature<"exception", "HasException", "true", + "Implements hardware exception support.">; +def FeaturePatCmp : SubtargetFeature<"patcmp", "HasPatCmp", "true", + "Implements pattern compare instruction.">; +def FeatureFPU : SubtargetFeature<"fpu", "HasFPU", "true", + "Implements floating point unit.">; +def FeatureESR : SubtargetFeature<"esr", "HasESR", "true", + "Implements ESR and EAR registers">; +def FeaturePVR : SubtargetFeature<"pvr", "HasPVR", "true", + "Implements processor version register.">; +def FeatureMul64 : SubtargetFeature<"mul64", "HasMul64", "true", + "Implements multiplier with 64-bit result">; +def FeatureSqrt : SubtargetFeature<"sqrt", "HasSqrt", "true", + "Implements sqrt and floating point convert.">; +def FeatureMMU : SubtargetFeature<"mmu", "HasMMU", "true", + "Implements memory management unit.">; + +//===----------------------------------------------------------------------===// +// MBlaze processors supported. +//===----------------------------------------------------------------------===// + +class Proc<string Name, list<SubtargetFeature> Features> + : Processor<Name, MBlazeGenericItineraries, Features>; + + +def : Proc<"v400", []>; +def : Proc<"v500", []>; +def : Proc<"v600", []>; +def : Proc<"v700", []>; +def : Proc<"v710", []>; + +def MBlaze : Target { + let InstructionSet = MBlazeInstrInfo; +} diff --git a/lib/Target/MBlaze/MBlazeCallingConv.td b/lib/Target/MBlaze/MBlazeCallingConv.td new file mode 100644 index 0000000..dfc87f5 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeCallingConv.td @@ -0,0 +1,41 @@ +//===- MBlazeCallingConv.td - Calling Conventions for MBlaze ----*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// This describes the calling conventions for MBlaze architecture. +//===----------------------------------------------------------------------===// + +/// CCIfSubtarget - Match if the current subtarget has a feature F. +class CCIfSubtarget<string F, CCAction A>: + CCIf<!strconcat("State.getTarget().getSubtarget<MBlazeSubtarget>().", F), A>; + +//===----------------------------------------------------------------------===// +// MBlaze ABI Calling Convention +//===----------------------------------------------------------------------===// + +def CC_MBlaze : CallingConv<[ + // Promote i8/i16 arguments to i32. + CCIfType<[i8, i16], CCPromoteToType<i32>>, + + // Integer arguments are passed in integer registers. + CCIfType<[i32], CCAssignToReg<[R5, R6, R7, R8, R9, R10]>>, + + // Single fp arguments are passed in floating point registers + CCIfType<[f32], CCAssignToReg<[F5, F6, F7, F8, F9, F10]>>, + + // 32-bit values get stored in stack slots that are 4 bytes in + // size and 4-byte aligned. + CCIfType<[i32, f32], CCAssignToStack<4, 4>> +]>; + +def RetCC_MBlaze : CallingConv<[ + // i32 are returned in registers R3, R4 + CCIfType<[i32], CCAssignToReg<[R3, R4]>>, + + // f32 are returned in registers F3, F4 + CCIfType<[f32], CCAssignToReg<[F3, F4]>> +]>; diff --git a/lib/Target/MBlaze/MBlazeDelaySlotFiller.cpp b/lib/Target/MBlaze/MBlazeDelaySlotFiller.cpp new file mode 100644 index 0000000..42fea25 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeDelaySlotFiller.cpp @@ -0,0 +1,75 @@ +//===-- DelaySlotFiller.cpp - MBlaze delay slot filler --------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// Simple pass to fills delay slots with NOPs. +// +//===----------------------------------------------------------------------===// + +#define DEBUG_TYPE "delay-slot-filler" + +#include "MBlaze.h" +#include "MBlazeTargetMachine.h" +#include "llvm/CodeGen/MachineFunctionPass.h" +#include "llvm/CodeGen/MachineInstrBuilder.h" +#include "llvm/Target/TargetInstrInfo.h" +#include "llvm/ADT/Statistic.h" + +using namespace llvm; + +STATISTIC(FilledSlots, "Number of delay slots filled"); + +namespace { + struct Filler : public MachineFunctionPass { + + TargetMachine &TM; + const TargetInstrInfo *TII; + + static char ID; + Filler(TargetMachine &tm) + : MachineFunctionPass(&ID), TM(tm), TII(tm.getInstrInfo()) { } + + virtual const char *getPassName() const { + return "MBlaze Delay Slot Filler"; + } + + bool runOnMachineBasicBlock(MachineBasicBlock &MBB); + bool runOnMachineFunction(MachineFunction &F) { + bool Changed = false; + for (MachineFunction::iterator FI = F.begin(), FE = F.end(); + FI != FE; ++FI) + Changed |= runOnMachineBasicBlock(*FI); + return Changed; + } + + }; + char Filler::ID = 0; +} // end of anonymous namespace + +/// runOnMachineBasicBlock - Fill in delay slots for the given basic block. +/// Currently, we fill delay slots with NOPs. We assume there is only one +/// delay slot per delayed instruction. +bool Filler::runOnMachineBasicBlock(MachineBasicBlock &MBB) { + bool Changed = false; + for (MachineBasicBlock::iterator I = MBB.begin(); I != MBB.end(); ++I) + if (I->getDesc().hasDelaySlot()) { + MachineBasicBlock::iterator J = I; + ++J; + BuildMI(MBB, J, I->getDebugLoc(), TII->get(MBlaze::NOP)); + ++FilledSlots; + Changed = true; + } + return Changed; +} + +/// createMBlazeDelaySlotFillerPass - Returns a pass that fills in delay +/// slots in MBlaze MachineFunctions +FunctionPass *llvm::createMBlazeDelaySlotFillerPass(MBlazeTargetMachine &tm) { + return new Filler(tm); +} + diff --git a/lib/Target/MBlaze/MBlazeISelDAGToDAG.cpp b/lib/Target/MBlaze/MBlazeISelDAGToDAG.cpp new file mode 100644 index 0000000..a0ebea0 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeISelDAGToDAG.cpp @@ -0,0 +1,339 @@ +//===-- MBlazeISelDAGToDAG.cpp - A dag to dag inst selector for MBlaze ----===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file defines an instruction selector for the MBlaze target. +// +//===----------------------------------------------------------------------===// + +#define DEBUG_TYPE "mblaze-isel" +#include "MBlaze.h" +#include "MBlazeISelLowering.h" +#include "MBlazeMachineFunction.h" +#include "MBlazeRegisterInfo.h" +#include "MBlazeSubtarget.h" +#include "MBlazeTargetMachine.h" +#include "llvm/GlobalValue.h" +#include "llvm/Instructions.h" +#include "llvm/Intrinsics.h" +#include "llvm/Support/CFG.h" +#include "llvm/Type.h" +#include "llvm/CodeGen/MachineConstantPool.h" +#include "llvm/CodeGen/MachineFunction.h" +#include "llvm/CodeGen/MachineFrameInfo.h" +#include "llvm/CodeGen/MachineInstrBuilder.h" +#include "llvm/CodeGen/MachineRegisterInfo.h" +#include "llvm/CodeGen/SelectionDAGISel.h" +#include "llvm/Target/TargetMachine.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/ErrorHandling.h" +#include "llvm/Support/raw_ostream.h" +using namespace llvm; + +//===----------------------------------------------------------------------===// +// Instruction Selector Implementation +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// MBlazeDAGToDAGISel - MBlaze specific code to select MBlaze machine +// instructions for SelectionDAG operations. +//===----------------------------------------------------------------------===// +namespace { + +class MBlazeDAGToDAGISel : public SelectionDAGISel { + + /// TM - Keep a reference to MBlazeTargetMachine. + MBlazeTargetMachine &TM; + + /// Subtarget - Keep a pointer to the MBlazeSubtarget around so that we can + /// make the right decision when generating code for different targets. + const MBlazeSubtarget &Subtarget; + +public: + explicit MBlazeDAGToDAGISel(MBlazeTargetMachine &tm) : + SelectionDAGISel(tm), + TM(tm), Subtarget(tm.getSubtarget<MBlazeSubtarget>()) {} + + // Pass Name + virtual const char *getPassName() const { + return "MBlaze DAG->DAG Pattern Instruction Selection"; + } +private: + // Include the pieces autogenerated from the target description. + #include "MBlazeGenDAGISel.inc" + + /// getTargetMachine - Return a reference to the TargetMachine, casted + /// to the target-specific type. + const MBlazeTargetMachine &getTargetMachine() { + return static_cast<const MBlazeTargetMachine &>(TM); + } + + /// getInstrInfo - Return a reference to the TargetInstrInfo, casted + /// to the target-specific type. + const MBlazeInstrInfo *getInstrInfo() { + return getTargetMachine().getInstrInfo(); + } + + SDNode *getGlobalBaseReg(); + SDNode *Select(SDNode *N); + + // Complex Pattern. + bool SelectAddr(SDNode *Op, SDValue N, + SDValue &Base, SDValue &Offset); + + // Address Selection + bool SelectAddrRegReg(SDNode *Op, SDValue N, SDValue &Base, SDValue &Index); + bool SelectAddrRegImm(SDNode *Op, SDValue N, SDValue &Disp, SDValue &Base); + + // getI32Imm - Return a target constant with the specified value, of type i32. + inline SDValue getI32Imm(unsigned Imm) { + return CurDAG->getTargetConstant(Imm, MVT::i32); + } +}; + +} + +/// isIntS32Immediate - This method tests to see if the node is either a 32-bit +/// or 64-bit immediate, and if the value can be accurately represented as a +/// sign extension from a 32-bit value. If so, this returns true and the +/// immediate. +static bool isIntS32Immediate(SDNode *N, int32_t &Imm) { + unsigned Opc = N->getOpcode(); + if (Opc != ISD::Constant) + return false; + + Imm = (int32_t)cast<ConstantSDNode>(N)->getZExtValue(); + if (N->getValueType(0) == MVT::i32) + return Imm == (int32_t)cast<ConstantSDNode>(N)->getZExtValue(); + else + return Imm == (int64_t)cast<ConstantSDNode>(N)->getZExtValue(); +} + +static bool isIntS32Immediate(SDValue Op, int32_t &Imm) { + return isIntS32Immediate(Op.getNode(), Imm); +} + + +/// SelectAddressRegReg - Given the specified addressed, check to see if it +/// can be represented as an indexed [r+r] operation. Returns false if it +/// can be more efficiently represented with [r+imm]. +bool MBlazeDAGToDAGISel:: +SelectAddrRegReg(SDNode *Op, SDValue N, SDValue &Base, SDValue &Index) { + if (N.getOpcode() == ISD::FrameIndex) return false; + if (N.getOpcode() == ISD::TargetExternalSymbol || + N.getOpcode() == ISD::TargetGlobalAddress) + return false; // direct calls. + + if (N.getOperand(0).getOpcode() == ISD::TargetJumpTable || + N.getOperand(1).getOpcode() == ISD::TargetJumpTable) + return false; // jump tables. + + int32_t imm = 0; + if (N.getOpcode() == ISD::ADD || N.getOpcode() == ISD::OR) { + if (isIntS32Immediate(N.getOperand(1), imm)) + return false; // r+i + + Base = N.getOperand(1); + Index = N.getOperand(0); + return true; + } + + return false; +} + +/// Returns true if the address N can be represented by a base register plus +/// a signed 32-bit displacement [r+imm], and if it is not better +/// represented as reg+reg. +bool MBlazeDAGToDAGISel:: +SelectAddrRegImm(SDNode *Op, SDValue N, SDValue &Disp, SDValue &Base) { + // If this can be more profitably realized as r+r, fail. + if (SelectAddrRegReg(Op, N, Disp, Base)) + return false; + + if (N.getOpcode() == ISD::ADD || N.getOpcode() == ISD::OR) { + int32_t imm = 0; + if (isIntS32Immediate(N.getOperand(1), imm)) { + Disp = CurDAG->getTargetConstant(imm, MVT::i32); + if (FrameIndexSDNode *FI = dyn_cast<FrameIndexSDNode>(N.getOperand(0))) { + Base = CurDAG->getTargetFrameIndex(FI->getIndex(), N.getValueType()); + } else { + Base = N.getOperand(0); + } + DEBUG( errs() << "WESLEY: Using Operand Immediate\n" ); + return true; // [r+i] + } + } else if (ConstantSDNode *CN = dyn_cast<ConstantSDNode>(N)) { + // Loading from a constant address. + uint32_t Imm = CN->getZExtValue(); + Disp = CurDAG->getTargetConstant(Imm, CN->getValueType(0)); + Base = CurDAG->getRegister(MBlaze::R0, CN->getValueType(0)); + DEBUG( errs() << "WESLEY: Using Constant Node\n" ); + return true; + } + + Disp = CurDAG->getTargetConstant(0, TM.getTargetLowering()->getPointerTy()); + if (FrameIndexSDNode *FI = dyn_cast<FrameIndexSDNode>(N)) + Base = CurDAG->getTargetFrameIndex(FI->getIndex(), N.getValueType()); + else + Base = N; + return true; // [r+0] +} + +/// getGlobalBaseReg - Output the instructions required to put the +/// GOT address into a register. +SDNode *MBlazeDAGToDAGISel::getGlobalBaseReg() { + unsigned GlobalBaseReg = getInstrInfo()->getGlobalBaseReg(MF); + return CurDAG->getRegister(GlobalBaseReg, TLI.getPointerTy()).getNode(); +} + +/// ComplexPattern used on MBlazeInstrInfo +/// Used on MBlaze Load/Store instructions +bool MBlazeDAGToDAGISel:: +SelectAddr(SDNode *Op, SDValue Addr, SDValue &Offset, SDValue &Base) { + // if Address is FI, get the TargetFrameIndex. + if (FrameIndexSDNode *FIN = dyn_cast<FrameIndexSDNode>(Addr)) { + Base = CurDAG->getTargetFrameIndex(FIN->getIndex(), MVT::i32); + Offset = CurDAG->getTargetConstant(0, MVT::i32); + return true; + } + + // on PIC code Load GA + if (TM.getRelocationModel() == Reloc::PIC_) { + if ((Addr.getOpcode() == ISD::TargetGlobalAddress) || + (Addr.getOpcode() == ISD::TargetConstantPool) || + (Addr.getOpcode() == ISD::TargetJumpTable)){ + Base = CurDAG->getRegister(MBlaze::R15, MVT::i32); + Offset = Addr; + return true; + } + } else { + if ((Addr.getOpcode() == ISD::TargetExternalSymbol || + Addr.getOpcode() == ISD::TargetGlobalAddress)) + return false; + } + + // Operand is a result from an ADD. + if (Addr.getOpcode() == ISD::ADD) { + if (ConstantSDNode *CN = dyn_cast<ConstantSDNode>(Addr.getOperand(1))) { + if (Predicate_immSExt16(CN)) { + + // If the first operand is a FI, get the TargetFI Node + if (FrameIndexSDNode *FIN = dyn_cast<FrameIndexSDNode> + (Addr.getOperand(0))) { + Base = CurDAG->getTargetFrameIndex(FIN->getIndex(), MVT::i32); + } else { + Base = Addr.getOperand(0); + } + + Offset = CurDAG->getTargetConstant(CN->getZExtValue(), MVT::i32); + return true; + } + } + } + + Base = Addr; + Offset = CurDAG->getTargetConstant(0, MVT::i32); + return true; +} + +/// Select instructions not customized! Used for +/// expanded, promoted and normal instructions +SDNode* MBlazeDAGToDAGISel::Select(SDNode *Node) { + unsigned Opcode = Node->getOpcode(); + DebugLoc dl = Node->getDebugLoc(); + + // Dump information about the Node being selected + DEBUG(errs() << "Selecting: "; Node->dump(CurDAG); errs() << "\n"); + + // If we have a custom node, we already have selected! + if (Node->isMachineOpcode()) { + DEBUG(errs() << "== "; Node->dump(CurDAG); errs() << "\n"); + return NULL; + } + + /// + // Instruction Selection not handled by the auto-generated + // tablegen selection should be handled here. + /// + switch(Opcode) { + default: break; + + // Get target GOT address. + case ISD::GLOBAL_OFFSET_TABLE: + return getGlobalBaseReg(); + + case ISD::FrameIndex: { + SDValue imm = CurDAG->getTargetConstant(0, MVT::i32); + int FI = dyn_cast<FrameIndexSDNode>(Node)->getIndex(); + EVT VT = Node->getValueType(0); + SDValue TFI = CurDAG->getTargetFrameIndex(FI, VT); + unsigned Opc = MBlaze::ADDI; + if (Node->hasOneUse()) + return CurDAG->SelectNodeTo(Node, Opc, VT, TFI, imm); + return CurDAG->getMachineNode(Opc, dl, VT, TFI, imm); + } + + + /// Handle direct and indirect calls when using PIC. On PIC, when + /// GOT is smaller than about 64k (small code) the GA target is + /// loaded with only one instruction. Otherwise GA's target must + /// be loaded with 3 instructions. + case MBlazeISD::JmpLink: { + if (TM.getRelocationModel() == Reloc::PIC_) { + SDValue Chain = Node->getOperand(0); + SDValue Callee = Node->getOperand(1); + SDValue R20Reg = CurDAG->getRegister(MBlaze::R20, MVT::i32); + SDValue InFlag(0, 0); + + if ( (isa<GlobalAddressSDNode>(Callee)) || + (isa<ExternalSymbolSDNode>(Callee)) ) + { + /// Direct call for global addresses and external symbols + SDValue GPReg = CurDAG->getRegister(MBlaze::R15, MVT::i32); + + // Use load to get GOT target + SDValue Ops[] = { Callee, GPReg, Chain }; + SDValue Load = SDValue(CurDAG->getMachineNode(MBlaze::LW, dl, + MVT::i32, MVT::Other, Ops, 3), 0); + Chain = Load.getValue(1); + + // Call target must be on T9 + Chain = CurDAG->getCopyToReg(Chain, dl, R20Reg, Load, InFlag); + } else + /// Indirect call + Chain = CurDAG->getCopyToReg(Chain, dl, R20Reg, Callee, InFlag); + + // Emit Jump and Link Register + SDNode *ResNode = CurDAG->getMachineNode(MBlaze::BRLID, dl, MVT::Other, + MVT::Flag, R20Reg, Chain); + Chain = SDValue(ResNode, 0); + InFlag = SDValue(ResNode, 1); + ReplaceUses(SDValue(Node, 0), Chain); + ReplaceUses(SDValue(Node, 1), InFlag); + return ResNode; + } + } + } + + // Select the default instruction + SDNode *ResNode = SelectCode(Node); + + DEBUG(errs() << "=> "); + if (ResNode == NULL || ResNode == Node) + DEBUG(Node->dump(CurDAG)); + else + DEBUG(ResNode->dump(CurDAG)); + DEBUG(errs() << "\n"); + return ResNode; +} + +/// createMBlazeISelDag - This pass converts a legalized DAG into a +/// MBlaze-specific DAG, ready for instruction scheduling. +FunctionPass *llvm::createMBlazeISelDag(MBlazeTargetMachine &TM) { + return new MBlazeDAGToDAGISel(TM); +} diff --git a/lib/Target/MBlaze/MBlazeISelLowering.cpp b/lib/Target/MBlaze/MBlazeISelLowering.cpp new file mode 100644 index 0000000..7790248 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeISelLowering.cpp @@ -0,0 +1,881 @@ +//===-- MBlazeISelLowering.cpp - MBlaze DAG Lowering Implementation -------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file defines the interfaces that MBlaze uses to lower LLVM code into a +// selection DAG. +// +//===----------------------------------------------------------------------===// + +#define DEBUG_TYPE "mblaze-lower" +#include "MBlazeISelLowering.h" +#include "MBlazeMachineFunction.h" +#include "MBlazeTargetMachine.h" +#include "MBlazeTargetObjectFile.h" +#include "MBlazeSubtarget.h" +#include "llvm/DerivedTypes.h" +#include "llvm/Function.h" +#include "llvm/GlobalVariable.h" +#include "llvm/Intrinsics.h" +#include "llvm/CallingConv.h" +#include "llvm/CodeGen/CallingConvLower.h" +#include "llvm/CodeGen/MachineFrameInfo.h" +#include "llvm/CodeGen/MachineFunction.h" +#include "llvm/CodeGen/MachineInstrBuilder.h" +#include "llvm/CodeGen/MachineRegisterInfo.h" +#include "llvm/CodeGen/SelectionDAGISel.h" +#include "llvm/CodeGen/ValueTypes.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/ErrorHandling.h" +#include "llvm/Support/raw_ostream.h" +using namespace llvm; + +const char *MBlazeTargetLowering::getTargetNodeName(unsigned Opcode) const { + switch (Opcode) { + case MBlazeISD::JmpLink : return "MBlazeISD::JmpLink"; + case MBlazeISD::GPRel : return "MBlazeISD::GPRel"; + case MBlazeISD::Wrap : return "MBlazeISD::Wrap"; + case MBlazeISD::ICmp : return "MBlazeISD::ICmp"; + case MBlazeISD::Ret : return "MBlazeISD::Ret"; + case MBlazeISD::Select_CC : return "MBlazeISD::Select_CC"; + default : return NULL; + } +} + +MBlazeTargetLowering::MBlazeTargetLowering(MBlazeTargetMachine &TM) + : TargetLowering(TM, new MBlazeTargetObjectFile()) { + Subtarget = &TM.getSubtarget<MBlazeSubtarget>(); + + // MBlaze does not have i1 type, so use i32 for + // setcc operations results (slt, sgt, ...). + setBooleanContents(ZeroOrOneBooleanContent); + + // Set up the register classes + addRegisterClass(MVT::i32, MBlaze::CPURegsRegisterClass); + if (Subtarget->hasFPU()) { + addRegisterClass(MVT::f32, MBlaze::FGR32RegisterClass); + setOperationAction(ISD::ConstantFP, MVT::f32, Legal); + } + + // Floating point operations which are not supported + setOperationAction(ISD::FREM, MVT::f32, Expand); + setOperationAction(ISD::UINT_TO_FP, MVT::i8, Expand); + setOperationAction(ISD::UINT_TO_FP, MVT::i16, Expand); + setOperationAction(ISD::UINT_TO_FP, MVT::i32, Expand); + setOperationAction(ISD::FP_TO_UINT, MVT::i32, Expand); + setOperationAction(ISD::FP_ROUND, MVT::f32, Expand); + setOperationAction(ISD::FP_ROUND, MVT::f64, Expand); + setOperationAction(ISD::FCOPYSIGN, MVT::f32, Expand); + setOperationAction(ISD::FCOPYSIGN, MVT::f64, Expand); + setOperationAction(ISD::FSIN, MVT::f32, Expand); + setOperationAction(ISD::FCOS, MVT::f32, Expand); + setOperationAction(ISD::FPOWI, MVT::f32, Expand); + setOperationAction(ISD::FPOW, MVT::f32, Expand); + setOperationAction(ISD::FLOG, MVT::f32, Expand); + setOperationAction(ISD::FLOG2, MVT::f32, Expand); + setOperationAction(ISD::FLOG10, MVT::f32, Expand); + setOperationAction(ISD::FEXP, MVT::f32, Expand); + + // Load extented operations for i1 types must be promoted + setLoadExtAction(ISD::EXTLOAD, MVT::i1, Promote); + setLoadExtAction(ISD::ZEXTLOAD, MVT::i1, Promote); + setLoadExtAction(ISD::SEXTLOAD, MVT::i1, Promote); + + // MBlaze has no REM or DIVREM operations. + setOperationAction(ISD::UREM, MVT::i32, Expand); + setOperationAction(ISD::SREM, MVT::i32, Expand); + setOperationAction(ISD::SDIVREM, MVT::i32, Expand); + setOperationAction(ISD::UDIVREM, MVT::i32, Expand); + + // If the processor doesn't support multiply then expand it + if (!Subtarget->hasMul()) { + setOperationAction(ISD::MUL, MVT::i32, Expand); + } + + // If the processor doesn't support 64-bit multiply then expand + if (!Subtarget->hasMul() || !Subtarget->hasMul64()) { + setOperationAction(ISD::MULHS, MVT::i32, Expand); + setOperationAction(ISD::MULHS, MVT::i64, Expand); + setOperationAction(ISD::MULHU, MVT::i32, Expand); + setOperationAction(ISD::MULHU, MVT::i64, Expand); + } + + // If the processor doesn't support division then expand + if (!Subtarget->hasDiv()) { + setOperationAction(ISD::UDIV, MVT::i32, Expand); + setOperationAction(ISD::SDIV, MVT::i32, Expand); + } + + // Expand unsupported conversions + setOperationAction(ISD::BIT_CONVERT, MVT::f32, Expand); + setOperationAction(ISD::BIT_CONVERT, MVT::i32, Expand); + + // Expand SELECT_CC + setOperationAction(ISD::SELECT_CC, MVT::Other, Expand); + + // MBlaze doesn't have MUL_LOHI + setOperationAction(ISD::SMUL_LOHI, MVT::i32, Expand); + setOperationAction(ISD::UMUL_LOHI, MVT::i32, Expand); + setOperationAction(ISD::SMUL_LOHI, MVT::i64, Expand); + setOperationAction(ISD::UMUL_LOHI, MVT::i64, Expand); + + // Used by legalize types to correctly generate the setcc result. + // Without this, every float setcc comes with a AND/OR with the result, + // we don't want this, since the fpcmp result goes to a flag register, + // which is used implicitly by brcond and select operations. + AddPromotedToType(ISD::SETCC, MVT::i1, MVT::i32); + AddPromotedToType(ISD::SELECT, MVT::i1, MVT::i32); + AddPromotedToType(ISD::SELECT_CC, MVT::i1, MVT::i32); + + // MBlaze Custom Operations + setOperationAction(ISD::GlobalAddress, MVT::i32, Custom); + setOperationAction(ISD::GlobalTLSAddress, MVT::i32, Custom); + setOperationAction(ISD::JumpTable, MVT::i32, Custom); + setOperationAction(ISD::ConstantPool, MVT::i32, Custom); + + // Operations not directly supported by MBlaze. + setOperationAction(ISD::DYNAMIC_STACKALLOC, MVT::i32, Expand); + setOperationAction(ISD::BR_JT, MVT::Other, Expand); + setOperationAction(ISD::BR_CC, MVT::Other, Expand); + setOperationAction(ISD::SIGN_EXTEND_INREG, MVT::i1, Expand); + setOperationAction(ISD::ROTL, MVT::i32, Expand); + setOperationAction(ISD::ROTR, MVT::i32, Expand); + setOperationAction(ISD::SHL_PARTS, MVT::i32, Expand); + setOperationAction(ISD::SRA_PARTS, MVT::i32, Expand); + setOperationAction(ISD::SRL_PARTS, MVT::i32, Expand); + setOperationAction(ISD::CTLZ, MVT::i32, Expand); + setOperationAction(ISD::CTTZ, MVT::i32, Expand); + setOperationAction(ISD::CTPOP, MVT::i32, Expand); + setOperationAction(ISD::BSWAP, MVT::i32, Expand); + + // We don't have line number support yet. + setOperationAction(ISD::EH_LABEL, MVT::Other, Expand); + + // Use the default for now + setOperationAction(ISD::STACKSAVE, MVT::Other, Expand); + setOperationAction(ISD::STACKRESTORE, MVT::Other, Expand); + setOperationAction(ISD::MEMBARRIER, MVT::Other, Expand); + + // MBlaze doesn't have extending float->double load/store + setLoadExtAction(ISD::EXTLOAD, MVT::f32, Expand); + setTruncStoreAction(MVT::f64, MVT::f32, Expand); + + setStackPointerRegisterToSaveRestore(MBlaze::R1); + computeRegisterProperties(); +} + +MVT::SimpleValueType MBlazeTargetLowering::getSetCCResultType(EVT VT) const { + return MVT::i32; +} + +/// getFunctionAlignment - Return the Log2 alignment of this function. +unsigned MBlazeTargetLowering::getFunctionAlignment(const Function *) const { + return 2; +} + +SDValue MBlazeTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) { + switch (Op.getOpcode()) + { + case ISD::ConstantPool: return LowerConstantPool(Op, DAG); + case ISD::GlobalAddress: return LowerGlobalAddress(Op, DAG); + case ISD::GlobalTLSAddress: return LowerGlobalTLSAddress(Op, DAG); + case ISD::JumpTable: return LowerJumpTable(Op, DAG); + case ISD::SELECT_CC: return LowerSELECT_CC(Op, DAG); + } + return SDValue(); +} + +//===----------------------------------------------------------------------===// +// Lower helper functions +//===----------------------------------------------------------------------===// +MachineBasicBlock* MBlazeTargetLowering:: +EmitInstrWithCustomInserter(MachineInstr *MI, MachineBasicBlock *BB, + DenseMap<MachineBasicBlock*, + MachineBasicBlock*> *EM) const { + const TargetInstrInfo *TII = getTargetMachine().getInstrInfo(); + DebugLoc dl = MI->getDebugLoc(); + + switch (MI->getOpcode()) { + default: assert(false && "Unexpected instr type to insert"); + case MBlaze::ShiftRL: + case MBlaze::ShiftRA: + case MBlaze::ShiftL: { + // To "insert" a shift left instruction, we actually have to insert a + // simple loop. The incoming instruction knows the destination vreg to + // set, the source vreg to operate over and the shift amount. + const BasicBlock *LLVM_BB = BB->getBasicBlock(); + MachineFunction::iterator It = BB; + ++It; + + // start: + // andi samt, samt, 31 + // beqid samt, finish + // add dst, src, r0 + // loop: + // addik samt, samt, -1 + // sra dst, dst + // bneid samt, loop + // nop + // finish: + MachineFunction *F = BB->getParent(); + MachineRegisterInfo &R = F->getRegInfo(); + MachineBasicBlock *loop = F->CreateMachineBasicBlock(LLVM_BB); + MachineBasicBlock *finish = F->CreateMachineBasicBlock(LLVM_BB); + + unsigned IAMT = R.createVirtualRegister(MBlaze::CPURegsRegisterClass); + BuildMI(BB, dl, TII->get(MBlaze::ANDI), IAMT) + .addReg(MI->getOperand(2).getReg()) + .addImm(31); + + unsigned IVAL = R.createVirtualRegister(MBlaze::CPURegsRegisterClass); + BuildMI(BB, dl, TII->get(MBlaze::ADDI), IVAL) + .addReg(MI->getOperand(1).getReg()) + .addImm(0); + + BuildMI(BB, dl, TII->get(MBlaze::BEQID)) + .addReg(IAMT) + .addMBB(finish); + + F->insert(It, loop); + F->insert(It, finish); + + // Update machine-CFG edges by first adding all successors of the current + // block to the new block which will contain the Phi node for the select. + // Also inform sdisel of the edge changes. + for(MachineBasicBlock::succ_iterator i = BB->succ_begin(), + e = BB->succ_end(); i != e; ++i) { + EM->insert(std::make_pair(*i, finish)); + finish->addSuccessor(*i); + } + + // Next, remove all successors of the current block, and add the true + // and fallthrough blocks as its successors. + while(!BB->succ_empty()) + BB->removeSuccessor(BB->succ_begin()); + BB->addSuccessor(loop); + BB->addSuccessor(finish); + + // Next, add the finish block as a successor of the loop block + loop->addSuccessor(finish); + loop->addSuccessor(loop); + + unsigned DST = R.createVirtualRegister(MBlaze::CPURegsRegisterClass); + unsigned NDST = R.createVirtualRegister(MBlaze::CPURegsRegisterClass); + BuildMI(loop, dl, TII->get(MBlaze::PHI), DST) + .addReg(IVAL).addMBB(BB) + .addReg(NDST).addMBB(loop); + + unsigned SAMT = R.createVirtualRegister(MBlaze::CPURegsRegisterClass); + unsigned NAMT = R.createVirtualRegister(MBlaze::CPURegsRegisterClass); + BuildMI(loop, dl, TII->get(MBlaze::PHI), SAMT) + .addReg(IAMT).addMBB(BB) + .addReg(NAMT).addMBB(loop); + + if (MI->getOpcode() == MBlaze::ShiftL) + BuildMI(loop, dl, TII->get(MBlaze::ADD), NDST).addReg(DST).addReg(DST); + else if (MI->getOpcode() == MBlaze::ShiftRA) + BuildMI(loop, dl, TII->get(MBlaze::SRA), NDST).addReg(DST); + else if (MI->getOpcode() == MBlaze::ShiftRL) + BuildMI(loop, dl, TII->get(MBlaze::SRL), NDST).addReg(DST); + else + llvm_unreachable( "Cannot lower unknown shift instruction" ); + + BuildMI(loop, dl, TII->get(MBlaze::ADDI), NAMT) + .addReg(SAMT) + .addImm(-1); + + BuildMI(loop, dl, TII->get(MBlaze::BNEID)) + .addReg(NAMT) + .addMBB(loop); + + BuildMI(finish, dl, TII->get(MBlaze::PHI), MI->getOperand(0).getReg()) + .addReg(IVAL).addMBB(BB) + .addReg(NDST).addMBB(loop); + + // The pseudo instruction is no longer needed so remove it + F->DeleteMachineInstr(MI); + return finish; + } + + case MBlaze::Select_FCC: + case MBlaze::Select_CC: { + // To "insert" a SELECT_CC instruction, we actually have to insert the + // diamond control-flow pattern. The incoming instruction knows the + // destination vreg to set, the condition code register to branch on, the + // true/false values to select between, and a branch opcode to use. + const BasicBlock *LLVM_BB = BB->getBasicBlock(); + MachineFunction::iterator It = BB; + ++It; + + // thisMBB: + // ... + // TrueVal = ... + // setcc r1, r2, r3 + // bNE r1, r0, copy1MBB + // fallthrough --> copy0MBB + MachineFunction *F = BB->getParent(); + MachineBasicBlock *flsBB = F->CreateMachineBasicBlock(LLVM_BB); + MachineBasicBlock *dneBB = F->CreateMachineBasicBlock(LLVM_BB); + + unsigned Opc; + switch (MI->getOperand(4).getImm()) { + default: llvm_unreachable( "Unknown branch condition" ); + case MBlazeCC::EQ: Opc = MBlaze::BNEID; break; + case MBlazeCC::NE: Opc = MBlaze::BEQID; break; + case MBlazeCC::GT: Opc = MBlaze::BLEID; break; + case MBlazeCC::LT: Opc = MBlaze::BGEID; break; + case MBlazeCC::GE: Opc = MBlaze::BLTID; break; + case MBlazeCC::LE: Opc = MBlaze::BGTID; break; + } + + BuildMI(BB, dl, TII->get(Opc)) + .addReg(MI->getOperand(3).getReg()) + .addMBB(dneBB); + + F->insert(It, flsBB); + F->insert(It, dneBB); + + // Update machine-CFG edges by first adding all successors of the current + // block to the new block which will contain the Phi node for the select. + // Also inform sdisel of the edge changes. + for(MachineBasicBlock::succ_iterator i = BB->succ_begin(), + e = BB->succ_end(); i != e; ++i) { + EM->insert(std::make_pair(*i, dneBB)); + dneBB->addSuccessor(*i); + } + + // Next, remove all successors of the current block, and add the true + // and fallthrough blocks as its successors. + while(!BB->succ_empty()) + BB->removeSuccessor(BB->succ_begin()); + BB->addSuccessor(flsBB); + BB->addSuccessor(dneBB); + flsBB->addSuccessor(dneBB); + + // sinkMBB: + // %Result = phi [ %FalseValue, copy0MBB ], [ %TrueValue, thisMBB ] + // ... + //BuildMI(dneBB, dl, TII->get(MBlaze::PHI), MI->getOperand(0).getReg()) + // .addReg(MI->getOperand(1).getReg()).addMBB(flsBB) + // .addReg(MI->getOperand(2).getReg()).addMBB(BB); + + BuildMI(dneBB, dl, TII->get(MBlaze::PHI), MI->getOperand(0).getReg()) + .addReg(MI->getOperand(2).getReg()).addMBB(flsBB) + .addReg(MI->getOperand(1).getReg()).addMBB(BB); + + F->DeleteMachineInstr(MI); // The pseudo instruction is gone now. + return dneBB; + } + } +} + +//===----------------------------------------------------------------------===// +// Misc Lower Operation implementation +//===----------------------------------------------------------------------===// +// + +SDValue MBlazeTargetLowering::LowerSELECT_CC(SDValue Op, SelectionDAG &DAG) { + SDValue LHS = Op.getOperand(0); + SDValue RHS = Op.getOperand(1); + SDValue TrueVal = Op.getOperand(2); + SDValue FalseVal = Op.getOperand(3); + DebugLoc dl = Op.getDebugLoc(); + unsigned Opc; + + SDValue CompareFlag; + if (LHS.getValueType() == MVT::i32) { + Opc = MBlazeISD::Select_CC; + CompareFlag = DAG.getNode(MBlazeISD::ICmp, dl, MVT::i32, LHS, RHS) + .getValue(1); + } else { + llvm_unreachable( "Cannot lower select_cc with unknown type" ); + } + + return DAG.getNode(Opc, dl, TrueVal.getValueType(), TrueVal, FalseVal, + CompareFlag); +} + +SDValue MBlazeTargetLowering:: +LowerGlobalAddress(SDValue Op, SelectionDAG &DAG) { + // FIXME there isn't actually debug info here + DebugLoc dl = Op.getDebugLoc(); + GlobalValue *GV = cast<GlobalAddressSDNode>(Op)->getGlobal(); + SDValue GA = DAG.getTargetGlobalAddress(GV, MVT::i32); + + return DAG.getNode(MBlazeISD::Wrap, dl, MVT::i32, GA); +} + +SDValue MBlazeTargetLowering:: +LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) { + llvm_unreachable("TLS not implemented for MicroBlaze."); + return SDValue(); // Not reached +} + +SDValue MBlazeTargetLowering:: +LowerJumpTable(SDValue Op, SelectionDAG &DAG) { + SDValue ResNode; + SDValue HiPart; + // FIXME there isn't actually debug info here + DebugLoc dl = Op.getDebugLoc(); + bool IsPIC = getTargetMachine().getRelocationModel() == Reloc::PIC_; + unsigned char OpFlag = IsPIC ? MBlazeII::MO_GOT : MBlazeII::MO_ABS_HILO; + + EVT PtrVT = Op.getValueType(); + JumpTableSDNode *JT = cast<JumpTableSDNode>(Op); + + SDValue JTI = DAG.getTargetJumpTable(JT->getIndex(), PtrVT, OpFlag); + return DAG.getNode(MBlazeISD::Wrap, dl, MVT::i32, JTI); + //return JTI; +} + +SDValue MBlazeTargetLowering:: +LowerConstantPool(SDValue Op, SelectionDAG &DAG) { + SDValue ResNode; + EVT PtrVT = Op.getValueType(); + ConstantPoolSDNode *N = cast<ConstantPoolSDNode>(Op); + Constant *C = N->getConstVal(); + SDValue Zero = DAG.getConstant(0, PtrVT); + // FIXME there isn't actually debug info here + DebugLoc dl = Op.getDebugLoc(); + + SDValue CP = DAG.getTargetConstantPool(C, MVT::i32, N->getAlignment(), + N->getOffset(), MBlazeII::MO_ABS_HILO); + return DAG.getNode(MBlazeISD::Wrap, dl, MVT::i32, CP); +} + +//===----------------------------------------------------------------------===// +// Calling Convention Implementation +//===----------------------------------------------------------------------===// + +#include "MBlazeGenCallingConv.inc" + +//===----------------------------------------------------------------------===// +// Call Calling Convention Implementation +//===----------------------------------------------------------------------===// + +/// LowerCall - functions arguments are copied from virtual regs to +/// (physical regs)/(stack frame), CALLSEQ_START and CALLSEQ_END are emitted. +/// TODO: isVarArg, isTailCall. +SDValue MBlazeTargetLowering:: +LowerCall(SDValue Chain, SDValue Callee, CallingConv::ID CallConv, + bool isVarArg, bool &isTailCall, + const SmallVectorImpl<ISD::OutputArg> &Outs, + const SmallVectorImpl<ISD::InputArg> &Ins, + DebugLoc dl, SelectionDAG &DAG, + SmallVectorImpl<SDValue> &InVals) { + MachineFunction &MF = DAG.getMachineFunction(); + MachineFrameInfo *MFI = MF.getFrameInfo(); + + // Analyze operands of the call, assigning locations to each operand. + SmallVector<CCValAssign, 16> ArgLocs; + CCState CCInfo(CallConv, isVarArg, getTargetMachine(), ArgLocs, + *DAG.getContext()); + CCInfo.AnalyzeCallOperands(Outs, CC_MBlaze); + + // Get a count of how many bytes are to be pushed on the stack. + unsigned NumBytes = CCInfo.getNextStackOffset(); + Chain = DAG.getCALLSEQ_START(Chain, DAG.getIntPtrConstant(NumBytes, true)); + + SmallVector<std::pair<unsigned, SDValue>, 8> RegsToPass; + SmallVector<SDValue, 8> MemOpChains; + + // First/LastArgStackLoc contains the first/last + // "at stack" argument location. + int LastArgStackLoc = 0; + unsigned FirstStackArgLoc = 4; + + // Walk the register/memloc assignments, inserting copies/loads. + for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) { + CCValAssign &VA = ArgLocs[i]; + EVT RegVT = VA.getLocVT(); + SDValue Arg = Outs[i].Val; + + // Promote the value if needed. + switch (VA.getLocInfo()) { + default: llvm_unreachable("Unknown loc info!"); + case CCValAssign::Full: break; + case CCValAssign::SExt: + Arg = DAG.getNode(ISD::SIGN_EXTEND, dl, RegVT, Arg); + break; + case CCValAssign::ZExt: + Arg = DAG.getNode(ISD::ZERO_EXTEND, dl, RegVT, Arg); + break; + case CCValAssign::AExt: + Arg = DAG.getNode(ISD::ANY_EXTEND, dl, RegVT, Arg); + break; + case CCValAssign::BCvt: + Arg = DAG.getNode(ISD::BIT_CONVERT, dl, RegVT, Arg); + break; + } + + // Arguments that can be passed on register must be kept at + // RegsToPass vector + if (VA.isRegLoc()) { + RegsToPass.push_back(std::make_pair(VA.getLocReg(), Arg)); + } else { + // Register can't get to this point... + assert(VA.isMemLoc()); + + // Create the frame index object for this incoming parameter + LastArgStackLoc = (FirstStackArgLoc + VA.getLocMemOffset()); + int FI = MFI->CreateFixedObject(VA.getValVT().getSizeInBits()/8, + LastArgStackLoc, true, false); + + SDValue PtrOff = DAG.getFrameIndex(FI,getPointerTy()); + + // emit ISD::STORE whichs stores the + // parameter value to a stack Location + MemOpChains.push_back(DAG.getStore(Chain, dl, Arg, PtrOff, NULL, 0, + false, false, 0)); + } + } + + // Transform all store nodes into one single node because all store + // nodes are independent of each other. + if (!MemOpChains.empty()) + Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, + &MemOpChains[0], MemOpChains.size()); + + // Build a sequence of copy-to-reg nodes chained together with token + // chain and flag operands which copy the outgoing args into registers. + // The InFlag in necessary since all emited instructions must be + // stuck together. + SDValue InFlag; + for (unsigned i = 0, e = RegsToPass.size(); i != e; ++i) { + Chain = DAG.getCopyToReg(Chain, dl, RegsToPass[i].first, + RegsToPass[i].second, InFlag); + InFlag = Chain.getValue(1); + } + + // If the callee is a GlobalAddress/ExternalSymbol node (quite common, every + // direct call is) turn it into a TargetGlobalAddress/TargetExternalSymbol + // node so that legalize doesn't hack it. + unsigned char OpFlag = MBlazeII::MO_NO_FLAG; + if (GlobalAddressSDNode *G = dyn_cast<GlobalAddressSDNode>(Callee)) + Callee = DAG.getTargetGlobalAddress(G->getGlobal(), + getPointerTy(), 0, OpFlag); + else if (ExternalSymbolSDNode *S = dyn_cast<ExternalSymbolSDNode>(Callee)) + Callee = DAG.getTargetExternalSymbol(S->getSymbol(), + getPointerTy(), OpFlag); + + // MBlazeJmpLink = #chain, #target_address, #opt_in_flags... + // = Chain, Callee, Reg#1, Reg#2, ... + // + // Returns a chain & a flag for retval copy to use. + SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Flag); + SmallVector<SDValue, 8> Ops; + Ops.push_back(Chain); + Ops.push_back(Callee); + + // Add argument registers to the end of the list so that they are + // known live into the call. + for (unsigned i = 0, e = RegsToPass.size(); i != e; ++i) { + Ops.push_back(DAG.getRegister(RegsToPass[i].first, + RegsToPass[i].second.getValueType())); + } + + if (InFlag.getNode()) + Ops.push_back(InFlag); + + Chain = DAG.getNode(MBlazeISD::JmpLink, dl, NodeTys, &Ops[0], Ops.size()); + InFlag = Chain.getValue(1); + + // Create the CALLSEQ_END node. + Chain = DAG.getCALLSEQ_END(Chain, DAG.getIntPtrConstant(NumBytes, true), + DAG.getIntPtrConstant(0, true), InFlag); + if (!Ins.empty()) + InFlag = Chain.getValue(1); + + // Handle result values, copying them out of physregs into vregs that we + // return. + return LowerCallResult(Chain, InFlag, CallConv, isVarArg, + Ins, dl, DAG, InVals); +} + +/// LowerCallResult - Lower the result values of a call into the +/// appropriate copies out of appropriate physical registers. +SDValue MBlazeTargetLowering:: +LowerCallResult(SDValue Chain, SDValue InFlag, CallingConv::ID CallConv, + bool isVarArg, const SmallVectorImpl<ISD::InputArg> &Ins, + DebugLoc dl, SelectionDAG &DAG, + SmallVectorImpl<SDValue> &InVals) { + // Assign locations to each value returned by this call. + SmallVector<CCValAssign, 16> RVLocs; + CCState CCInfo(CallConv, isVarArg, getTargetMachine(), + RVLocs, *DAG.getContext()); + + CCInfo.AnalyzeCallResult(Ins, RetCC_MBlaze); + + // Copy all of the result registers out of their specified physreg. + for (unsigned i = 0; i != RVLocs.size(); ++i) { + Chain = DAG.getCopyFromReg(Chain, dl, RVLocs[i].getLocReg(), + RVLocs[i].getValVT(), InFlag).getValue(1); + InFlag = Chain.getValue(2); + InVals.push_back(Chain.getValue(0)); + } + + return Chain; +} + +//===----------------------------------------------------------------------===// +// Formal Arguments Calling Convention Implementation +//===----------------------------------------------------------------------===// + +/// LowerFormalArguments - transform physical registers into +/// virtual registers and generate load operations for +/// arguments places on the stack. +/// TODO: isVarArg +SDValue MBlazeTargetLowering:: +LowerFormalArguments(SDValue Chain, CallingConv::ID CallConv, bool isVarArg, + const SmallVectorImpl<ISD::InputArg> &Ins, + DebugLoc dl, SelectionDAG &DAG, + SmallVectorImpl<SDValue> &InVals) { + MachineFunction &MF = DAG.getMachineFunction(); + MachineFrameInfo *MFI = MF.getFrameInfo(); + MBlazeFunctionInfo *MBlazeFI = MF.getInfo<MBlazeFunctionInfo>(); + + unsigned StackReg = MF.getTarget().getRegisterInfo()->getFrameRegister(MF); + + // Assign locations to all of the incoming arguments. + SmallVector<CCValAssign, 16> ArgLocs; + CCState CCInfo(CallConv, isVarArg, getTargetMachine(), + ArgLocs, *DAG.getContext()); + + CCInfo.AnalyzeFormalArguments(Ins, CC_MBlaze); + SDValue StackPtr; + + unsigned FirstStackArgLoc = 4; + + for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) { + CCValAssign &VA = ArgLocs[i]; + + // Arguments stored on registers + if (VA.isRegLoc()) { + EVT RegVT = VA.getLocVT(); + TargetRegisterClass *RC = 0; + + if (RegVT == MVT::i32) + RC = MBlaze::CPURegsRegisterClass; + else if (RegVT == MVT::f32) + RC = MBlaze::FGR32RegisterClass; + else + llvm_unreachable("RegVT not supported by LowerFormalArguments"); + + // Transform the arguments stored on + // physical registers into virtual ones + unsigned Reg = MF.addLiveIn(VA.getLocReg(), RC); + SDValue ArgValue = DAG.getCopyFromReg(Chain, dl, Reg, RegVT); + + // If this is an 8 or 16-bit value, it has been passed promoted + // to 32 bits. Insert an assert[sz]ext to capture this, then + // truncate to the right size. + if (VA.getLocInfo() != CCValAssign::Full) { + unsigned Opcode = 0; + if (VA.getLocInfo() == CCValAssign::SExt) + Opcode = ISD::AssertSext; + else if (VA.getLocInfo() == CCValAssign::ZExt) + Opcode = ISD::AssertZext; + if (Opcode) + ArgValue = DAG.getNode(Opcode, dl, RegVT, ArgValue, + DAG.getValueType(VA.getValVT())); + ArgValue = DAG.getNode(ISD::TRUNCATE, dl, VA.getValVT(), ArgValue); + } + + InVals.push_back(ArgValue); + + // To meet ABI, when VARARGS are passed on registers, the registers + // must have their values written to the caller stack frame. + if (isVarArg) { + if (StackPtr.getNode() == 0) + StackPtr = DAG.getRegister(StackReg, getPointerTy()); + + // The stack pointer offset is relative to the caller stack frame. + // Since the real stack size is unknown here, a negative SPOffset + // is used so there's a way to adjust these offsets when the stack + // size get known (on EliminateFrameIndex). A dummy SPOffset is + // used instead of a direct negative address (which is recorded to + // be used on emitPrologue) to avoid mis-calc of the first stack + // offset on PEI::calculateFrameObjectOffsets. + // Arguments are always 32-bit. + int FI = MFI->CreateFixedObject(4, 0, true, false); + MBlazeFI->recordStoreVarArgsFI(FI, -(FirstStackArgLoc+(i*4))); + SDValue PtrOff = DAG.getFrameIndex(FI, getPointerTy()); + + // emit ISD::STORE whichs stores the + // parameter value to a stack Location + InVals.push_back(DAG.getStore(Chain, dl, ArgValue, PtrOff, NULL, 0, + false, false, 0)); + } + + } else { // VA.isRegLoc() + + // sanity check + assert(VA.isMemLoc()); + + // The stack pointer offset is relative to the caller stack frame. + // Since the real stack size is unknown here, a negative SPOffset + // is used so there's a way to adjust these offsets when the stack + // size get known (on EliminateFrameIndex). A dummy SPOffset is + // used instead of a direct negative address (which is recorded to + // be used on emitPrologue) to avoid mis-calc of the first stack + // offset on PEI::calculateFrameObjectOffsets. + // Arguments are always 32-bit. + unsigned ArgSize = VA.getLocVT().getSizeInBits()/8; + int FI = MFI->CreateFixedObject(ArgSize, 0, true, false); + MBlazeFI->recordLoadArgsFI(FI, -(ArgSize+ + (FirstStackArgLoc + VA.getLocMemOffset()))); + + // Create load nodes to retrieve arguments from the stack + SDValue FIN = DAG.getFrameIndex(FI, getPointerTy()); + InVals.push_back(DAG.getLoad(VA.getValVT(), dl, Chain, FIN, NULL, 0, + false, false, 0)); + } + } + + return Chain; +} + +//===----------------------------------------------------------------------===// +// Return Value Calling Convention Implementation +//===----------------------------------------------------------------------===// + +SDValue MBlazeTargetLowering:: +LowerReturn(SDValue Chain, CallingConv::ID CallConv, bool isVarArg, + const SmallVectorImpl<ISD::OutputArg> &Outs, + DebugLoc dl, SelectionDAG &DAG) { + // CCValAssign - represent the assignment of + // the return value to a location + SmallVector<CCValAssign, 16> RVLocs; + + // CCState - Info about the registers and stack slot. + CCState CCInfo(CallConv, isVarArg, getTargetMachine(), + RVLocs, *DAG.getContext()); + + // Analize return values. + CCInfo.AnalyzeReturn(Outs, RetCC_MBlaze); + + // If this is the first return lowered for this function, add + // the regs to the liveout set for the function. + if (DAG.getMachineFunction().getRegInfo().liveout_empty()) { + for (unsigned i = 0; i != RVLocs.size(); ++i) + if (RVLocs[i].isRegLoc()) + DAG.getMachineFunction().getRegInfo().addLiveOut(RVLocs[i].getLocReg()); + } + + SDValue Flag; + + // Copy the result values into the output registers. + for (unsigned i = 0; i != RVLocs.size(); ++i) { + CCValAssign &VA = RVLocs[i]; + assert(VA.isRegLoc() && "Can only return in registers!"); + + Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(), + Outs[i].Val, Flag); + + // guarantee that all emitted copies are + // stuck together, avoiding something bad + Flag = Chain.getValue(1); + } + + // Return on MBlaze is always a "rtsd R15, 8" + if (Flag.getNode()) + return DAG.getNode(MBlazeISD::Ret, dl, MVT::Other, + Chain, DAG.getRegister(MBlaze::R15, MVT::i32), Flag); + else // Return Void + return DAG.getNode(MBlazeISD::Ret, dl, MVT::Other, + Chain, DAG.getRegister(MBlaze::R15, MVT::i32)); +} + +//===----------------------------------------------------------------------===// +// MBlaze Inline Assembly Support +//===----------------------------------------------------------------------===// + +/// getConstraintType - Given a constraint letter, return the type of +/// constraint it is for this target. +MBlazeTargetLowering::ConstraintType MBlazeTargetLowering:: +getConstraintType(const std::string &Constraint) const +{ + // MBlaze specific constrainy + // + // 'd' : An address register. Equivalent to r. + // 'y' : Equivalent to r; retained for + // backwards compatibility. + // 'f' : Floating Point registers. + if (Constraint.size() == 1) { + switch (Constraint[0]) { + default : break; + case 'd': + case 'y': + case 'f': + return C_RegisterClass; + break; + } + } + return TargetLowering::getConstraintType(Constraint); +} + +/// getRegClassForInlineAsmConstraint - Given a constraint letter (e.g. "r"), +/// return a list of registers that can be used to satisfy the constraint. +/// This should only be used for C_RegisterClass constraints. +std::pair<unsigned, const TargetRegisterClass*> MBlazeTargetLowering:: +getRegForInlineAsmConstraint(const std::string &Constraint, EVT VT) const { + if (Constraint.size() == 1) { + switch (Constraint[0]) { + case 'r': + return std::make_pair(0U, MBlaze::CPURegsRegisterClass); + case 'f': + if (VT == MVT::f32) + return std::make_pair(0U, MBlaze::FGR32RegisterClass); + } + } + return TargetLowering::getRegForInlineAsmConstraint(Constraint, VT); +} + +/// Given a register class constraint, like 'r', if this corresponds directly +/// to an LLVM register class, return a register of 0 and the register class +/// pointer. +std::vector<unsigned> MBlazeTargetLowering:: +getRegClassForInlineAsmConstraint(const std::string &Constraint, EVT VT) const { + if (Constraint.size() != 1) + return std::vector<unsigned>(); + + switch (Constraint[0]) { + default : break; + case 'r': + // GCC MBlaze Constraint Letters + case 'd': + case 'y': + return make_vector<unsigned>( + MBlaze::R3, MBlaze::R4, MBlaze::R5, MBlaze::R6, + MBlaze::R7, MBlaze::R9, MBlaze::R10, MBlaze::R11, + MBlaze::R12, MBlaze::R19, MBlaze::R20, MBlaze::R21, + MBlaze::R22, MBlaze::R23, MBlaze::R24, MBlaze::R25, + MBlaze::R26, MBlaze::R27, MBlaze::R28, MBlaze::R29, + MBlaze::R30, MBlaze::R31, 0); + + case 'f': + return make_vector<unsigned>( + MBlaze::F3, MBlaze::F4, MBlaze::F5, MBlaze::F6, + MBlaze::F7, MBlaze::F9, MBlaze::F10, MBlaze::F11, + MBlaze::F12, MBlaze::F19, MBlaze::F20, MBlaze::F21, + MBlaze::F22, MBlaze::F23, MBlaze::F24, MBlaze::F25, + MBlaze::F26, MBlaze::F27, MBlaze::F28, MBlaze::F29, + MBlaze::F30, MBlaze::F31, 0); + } + return std::vector<unsigned>(); +} + +bool MBlazeTargetLowering:: +isOffsetFoldingLegal(const GlobalAddressSDNode *GA) const { + // The MBlaze target isn't yet aware of offsets. + return false; +} + +bool MBlazeTargetLowering::isFPImmLegal(const APFloat &Imm, EVT VT) const { + return VT != MVT::f32; +} diff --git a/lib/Target/MBlaze/MBlazeISelLowering.h b/lib/Target/MBlaze/MBlazeISelLowering.h new file mode 100644 index 0000000..75d2552 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeISelLowering.h @@ -0,0 +1,146 @@ +//===-- MBlazeISelLowering.h - MBlaze DAG Lowering Interface ----*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file defines the interfaces that MBlaze uses to lower LLVM code into a +// selection DAG. +// +//===----------------------------------------------------------------------===// + +#ifndef MBlazeISELLOWERING_H +#define MBlazeISELLOWERING_H + +#include "llvm/CodeGen/SelectionDAG.h" +#include "llvm/Target/TargetLowering.h" +#include "MBlaze.h" +#include "MBlazeSubtarget.h" + +namespace llvm { + namespace MBlazeCC { + enum CC { + FIRST = 0, + EQ, + NE, + GT, + LT, + GE, + LE + }; + } + + namespace MBlazeISD { + enum NodeType { + // Start the numbering from where ISD NodeType finishes. + FIRST_NUMBER = ISD::BUILTIN_OP_END, + + // Jump and link (call) + JmpLink, + + // Handle gp_rel (small data/bss sections) relocation. + GPRel, + + // Select CC Pseudo Instruction + Select_CC, + + // Wrap up multiple types of instructions + Wrap, + + // Integer Compare + ICmp, + + // Return + Ret + }; + } + + //===--------------------------------------------------------------------===// + // TargetLowering Implementation + //===--------------------------------------------------------------------===// + + class MBlazeTargetLowering : public TargetLowering { + public: + + explicit MBlazeTargetLowering(MBlazeTargetMachine &TM); + + /// LowerOperation - Provide custom lowering hooks for some operations. + virtual SDValue LowerOperation(SDValue Op, SelectionDAG &DAG); + + /// getTargetNodeName - This method returns the name of a target specific + // DAG node. + virtual const char *getTargetNodeName(unsigned Opcode) const; + + /// getSetCCResultType - get the ISD::SETCC result ValueType + MVT::SimpleValueType getSetCCResultType(EVT VT) const; + + virtual unsigned getFunctionAlignment(const Function *F) const; + private: + // Subtarget Info + const MBlazeSubtarget *Subtarget; + + + // Lower Operand helpers + SDValue LowerCallResult(SDValue Chain, SDValue InFlag, + CallingConv::ID CallConv, bool isVarArg, + const SmallVectorImpl<ISD::InputArg> &Ins, + DebugLoc dl, SelectionDAG &DAG, + SmallVectorImpl<SDValue> &InVals); + + // Lower Operand specifics + SDValue LowerConstantPool(SDValue Op, SelectionDAG &DAG); + SDValue LowerGlobalAddress(SDValue Op, SelectionDAG &DAG); + SDValue LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG); + SDValue LowerJumpTable(SDValue Op, SelectionDAG &DAG); + SDValue LowerSELECT_CC(SDValue Op, SelectionDAG &DAG); + + virtual SDValue + LowerFormalArguments(SDValue Chain, + CallingConv::ID CallConv, bool isVarArg, + const SmallVectorImpl<ISD::InputArg> &Ins, + DebugLoc dl, SelectionDAG &DAG, + SmallVectorImpl<SDValue> &InVals); + + virtual SDValue + LowerCall(SDValue Chain, SDValue Callee, + CallingConv::ID CallConv, bool isVarArg, + bool &isTailCall, + const SmallVectorImpl<ISD::OutputArg> &Outs, + const SmallVectorImpl<ISD::InputArg> &Ins, + DebugLoc dl, SelectionDAG &DAG, + SmallVectorImpl<SDValue> &InVals); + + virtual SDValue + LowerReturn(SDValue Chain, + CallingConv::ID CallConv, bool isVarArg, + const SmallVectorImpl<ISD::OutputArg> &Outs, + DebugLoc dl, SelectionDAG &DAG); + + virtual MachineBasicBlock *EmitInstrWithCustomInserter(MachineInstr *MI, + MachineBasicBlock *MBB, + DenseMap<MachineBasicBlock*, MachineBasicBlock*> *EM) const; + + // Inline asm support + ConstraintType getConstraintType(const std::string &Constraint) const; + + std::pair<unsigned, const TargetRegisterClass*> + getRegForInlineAsmConstraint(const std::string &Constraint, + EVT VT) const; + + std::vector<unsigned> + getRegClassForInlineAsmConstraint(const std::string &Constraint, + EVT VT) const; + + virtual bool isOffsetFoldingLegal(const GlobalAddressSDNode *GA) const; + + /// isFPImmLegal - Returns true if the target can instruction select the + /// specified FP immediate natively. If false, the legalizer will + /// materialize the FP immediate as a load from a constant pool. + virtual bool isFPImmLegal(const APFloat &Imm, EVT VT) const; + }; +} + +#endif // MBlazeISELLOWERING_H diff --git a/lib/Target/MBlaze/MBlazeInstrFPU.td b/lib/Target/MBlaze/MBlazeInstrFPU.td new file mode 100644 index 0000000..a48a8c9 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeInstrFPU.td @@ -0,0 +1,223 @@ +//===- MBlazeInstrFPU.td - MBlaze FPU Instruction defs ----------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// MBlaze profiles and nodes +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// MBlaze Operand, Complex Patterns and Transformations Definitions. +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// Memory Access Instructions +//===----------------------------------------------------------------------===// +class LoadFM<bits<6> op, string instr_asm, PatFrag OpNode> : + TA<op, 0x000, (outs FGR32:$dst), (ins memrr:$addr), + !strconcat(instr_asm, " $dst, $addr"), + [(set FGR32:$dst, (OpNode xaddr:$addr))], IILoad>; + +class LoadFMI<bits<6> op, string instr_asm, PatFrag OpNode> : + TAI<op, (outs FGR32:$dst), (ins memri:$addr), + !strconcat(instr_asm, " $dst, $addr"), + [(set FGR32:$dst, (OpNode iaddr:$addr))], IILoad>; + +class StoreFM<bits<6> op, string instr_asm, PatFrag OpNode> : + TA<op, 0x000, (outs), (ins FGR32:$dst, memrr:$addr), + !strconcat(instr_asm, " $dst, $addr"), + [(OpNode FGR32:$dst, xaddr:$addr)], IIStore>; + +class StoreFMI<bits<6> op, string instr_asm, PatFrag OpNode> : + TAI<op, (outs), (ins FGR32:$dst, memrr:$addr), + !strconcat(instr_asm, " $dst, $addr"), + [(OpNode FGR32:$dst, iaddr:$addr)], IIStore>; + +class ArithF<bits<6> op, bits<11> flags, string instr_asm, SDNode OpNode, + InstrItinClass itin> : + TA<op, flags, (outs FGR32:$dst), (ins FGR32:$b, FGR32:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [(set FGR32:$dst, (OpNode FGR32:$b, FGR32:$c))], itin>; + +class CmpFN<bits<6> op, bits<11> flags, string instr_asm, + InstrItinClass itin> : + TA<op, flags, (outs CPURegs:$dst), (ins FGR32:$b, FGR32:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [], itin>; + +class ArithFR<bits<6> op, bits<11> flags, string instr_asm, SDNode OpNode, + InstrItinClass itin> : + TA<op, flags, (outs FGR32:$dst), (ins FGR32:$b, FGR32:$c), + !strconcat(instr_asm, " $dst, $c, $b"), + [(set FGR32:$dst, (OpNode FGR32:$b, FGR32:$c))], itin>; + +class ArithF2<bits<6> op, bits<11> flags, string instr_asm, + InstrItinClass itin> : + TF<op, flags, (outs FGR32:$dst), (ins FGR32:$b), + !strconcat(instr_asm, " $dst, $b"), + [], itin>; + +class ArithIF<bits<6> op, bits<11> flags, string instr_asm, + InstrItinClass itin> : + TF<op, flags, (outs FGR32:$dst), (ins CPURegs:$b), + !strconcat(instr_asm, " $dst, $b"), + [], itin>; + +class ArithFI<bits<6> op, bits<11> flags, string instr_asm, + InstrItinClass itin> : + TF<op, flags, (outs CPURegs:$dst), (ins FGR32:$b), + !strconcat(instr_asm, " $dst, $b"), + [], itin>; + +class LogicF<bits<6> op, string instr_asm> : + TAI<op, (outs FGR32:$dst), (ins FGR32:$b, FGR32:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [], + IIAlu>; + +class LogicFI<bits<6> op, string instr_asm> : + TAI<op, (outs FGR32:$dst), (ins FGR32:$b, fimm:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [], + IIAlu>; + +//===----------------------------------------------------------------------===// +// Pseudo instructions +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// FPU Arithmetic Instructions +//===----------------------------------------------------------------------===// +let Predicates=[HasFPU] in { + def FOR : LogicF<0x28, "or ">; + def FORI : LogicFI<0x28, "ori ">; + def FADD : ArithF<0x16, 0x000, "fadd ", fadd, IIAlu>; + def FRSUB : ArithFR<0x16, 0x080, "frsub ", fsub, IIAlu>; + def FMUL : ArithF<0x16, 0x100, "fmul ", fmul, IIAlu>; + def FDIV : ArithF<0x16, 0x180, "fdiv ", fdiv, IIAlu>; + + def LWF : LoadFM<0x32, "lw ", load>; + def LWFI : LoadFMI<0x32, "lwi ", load>; + + def SWF : StoreFM<0x32, "sw ", store>; + def SWFI : StoreFMI<0x32, "swi ", store>; +} + +let Predicates=[HasFPU,HasSqrt] in { + def FLT : ArithIF<0x16, 0x280, "flt ", IIAlu>; + def FINT : ArithFI<0x16, 0x300, "fint ", IIAlu>; + def FSQRT : ArithF2<0x16, 0x300, "fsqrt ", IIAlu>; +} + +let isAsCheapAsAMove = 1 in { + def FCMP_UN : CmpFN<0x16, 0x200, "fcmp.un", IIAlu>; + def FCMP_LT : CmpFN<0x16, 0x210, "fcmp.lt", IIAlu>; + def FCMP_EQ : CmpFN<0x16, 0x220, "fcmp.eq", IIAlu>; + def FCMP_LE : CmpFN<0x16, 0x230, "fcmp.le", IIAlu>; + def FCMP_GT : CmpFN<0x16, 0x240, "fcmp.gt", IIAlu>; + def FCMP_NE : CmpFN<0x16, 0x250, "fcmp.ne", IIAlu>; + def FCMP_GE : CmpFN<0x16, 0x260, "fcmp.ge", IIAlu>; +} + + +let usesCustomInserter = 1 in { + def Select_FCC : MBlazePseudo<(outs FGR32:$dst), + (ins FGR32:$T, FGR32:$F, CPURegs:$CMP, i32imm:$CC), + "; SELECT_FCC PSEUDO!", + []>; +} + +// Floating point conversions +let Predicates=[HasFPU] in { + def : Pat<(sint_to_fp CPURegs:$V), (FLT CPURegs:$V)>; + def : Pat<(fp_to_sint FGR32:$V), (FINT FGR32:$V)>; + def : Pat<(fsqrt FGR32:$V), (FSQRT FGR32:$V)>; +} + +// SET_CC operations +let Predicates=[HasFPU] in { + def : Pat<(setcc FGR32:$L, FGR32:$R, SETEQ), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_EQ FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETNE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_EQ FGR32:$L, FGR32:$R), 1)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETOEQ), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_EQ FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETONE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (XOR (FCMP_UN FGR32:$L, FGR32:$R), + (FCMP_EQ FGR32:$L, FGR32:$R)), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETONE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (OR (FCMP_UN FGR32:$L, FGR32:$R), + (FCMP_EQ FGR32:$L, FGR32:$R)), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETGT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_GT FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETLT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_LT FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETGE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_GE FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETLE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_LE FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETOGT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_GT FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETOLT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_LT FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETOGE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_GE FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETOLE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_LE FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETUEQ), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (OR (FCMP_UN FGR32:$L, FGR32:$R), + (FCMP_EQ FGR32:$L, FGR32:$R)), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETUNE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_NE FGR32:$L, FGR32:$R), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETUGT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (OR (FCMP_UN FGR32:$L, FGR32:$R), + (FCMP_GT FGR32:$L, FGR32:$R)), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETULT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (OR (FCMP_UN FGR32:$L, FGR32:$R), + (FCMP_LT FGR32:$L, FGR32:$R)), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETUGE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (OR (FCMP_UN FGR32:$L, FGR32:$R), + (FCMP_GE FGR32:$L, FGR32:$R)), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETULE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (OR (FCMP_UN FGR32:$L, FGR32:$R), + (FCMP_LE FGR32:$L, FGR32:$R)), 2)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETO), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_UN FGR32:$L, FGR32:$R), 1)>; + def : Pat<(setcc FGR32:$L, FGR32:$R, SETUO), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (FCMP_UN FGR32:$L, FGR32:$R), 2)>; +} + +// SELECT operations +def : Pat<(select CPURegs:$C, FGR32:$T, FGR32:$F), + (Select_FCC FGR32:$T, FGR32:$F, CPURegs:$C, 2)>; + +//===----------------------------------------------------------------------===// +// Patterns for Floating Point Instructions +//===----------------------------------------------------------------------===// +def : Pat<(f32 fpimm:$imm), (FORI F0, fpimm:$imm)>; diff --git a/lib/Target/MBlaze/MBlazeInstrFSL.td b/lib/Target/MBlaze/MBlazeInstrFSL.td new file mode 100644 index 0000000..b59999e --- /dev/null +++ b/lib/Target/MBlaze/MBlazeInstrFSL.td @@ -0,0 +1,153 @@ +//===- MBlazeInstrFSL.td - MBlaze FSL Instruction defs ----------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// FSL Instruction Formats +//===----------------------------------------------------------------------===// +class FSLGetD<bits<6> op, bits<11> flags, string instr_asm, Intrinsic OpNode> : + TA<op, flags, (outs CPURegs:$dst), (ins CPURegs:$b), + !strconcat(instr_asm, " $dst, $b"), + [(set CPURegs:$dst, (OpNode CPURegs:$b))], IIAlu>; + +class FSLGet<bits<6> op, string instr_asm, Intrinsic OpNode> : + TAI<op, (outs CPURegs:$dst), (ins fslimm:$b), + !strconcat(instr_asm, " $dst, $b"), + [(set CPURegs:$dst, (OpNode immZExt4:$b))], IIAlu>; + +class FSLPutD<bits<6> op, bits<11> flags, string instr_asm, Intrinsic OpNode> : + TA<op, flags, (outs), (ins CPURegs:$v, CPURegs:$b), + !strconcat(instr_asm, " $v, $b"), + [(OpNode CPURegs:$v, CPURegs:$b)], IIAlu>; + +class FSLPut<bits<6> op, string instr_asm, Intrinsic OpNode> : + TAI<op, (outs), (ins CPURegs:$v, fslimm:$b), + !strconcat(instr_asm, " $v, $b"), + [(OpNode CPURegs:$v, immZExt4:$b)], IIAlu>; + +class FSLPutTD<bits<6> op, bits<11> flags, string instr_asm, Intrinsic OpNode> : + TA<op, flags, (outs), (ins CPURegs:$b), + !strconcat(instr_asm, " $b"), + [(OpNode CPURegs:$b)], IIAlu>; + +class FSLPutT<bits<6> op, string instr_asm, Intrinsic OpNode> : + TAI<op, (outs), (ins fslimm:$b), + !strconcat(instr_asm, " $b"), + [(OpNode immZExt4:$b)], IIAlu>; + +//===----------------------------------------------------------------------===// +// FSL Get Instructions +//===----------------------------------------------------------------------===// +def GET : FSLGet<0x1B, "get ", int_mblaze_fsl_get>; +def AGET : FSLGet<0x1B, "aget ", int_mblaze_fsl_aget>; +def CGET : FSLGet<0x1B, "cget ", int_mblaze_fsl_cget>; +def CAGET : FSLGet<0x1B, "caget ", int_mblaze_fsl_caget>; +def EGET : FSLGet<0x1B, "eget ", int_mblaze_fsl_eget>; +def EAGET : FSLGet<0x1B, "eaget ", int_mblaze_fsl_eaget>; +def ECGET : FSLGet<0x1B, "ecget ", int_mblaze_fsl_ecget>; +def ECAGET : FSLGet<0x1B, "ecaget ", int_mblaze_fsl_ecaget>; +def NGET : FSLGet<0x1B, "nget ", int_mblaze_fsl_nget>; +def NAGET : FSLGet<0x1B, "naget ", int_mblaze_fsl_naget>; +def NCGET : FSLGet<0x1B, "ncget ", int_mblaze_fsl_ncget>; +def NCAGET : FSLGet<0x1B, "ncaget ", int_mblaze_fsl_ncaget>; +def NEGET : FSLGet<0x1B, "neget ", int_mblaze_fsl_neget>; +def NEAGET : FSLGet<0x1B, "neaget ", int_mblaze_fsl_neaget>; +def NECGET : FSLGet<0x1B, "necget ", int_mblaze_fsl_necget>; +def NECAGET : FSLGet<0x1B, "necaget ", int_mblaze_fsl_necaget>; +def TGET : FSLGet<0x1B, "tget ", int_mblaze_fsl_tget>; +def TAGET : FSLGet<0x1B, "taget ", int_mblaze_fsl_taget>; +def TCGET : FSLGet<0x1B, "tcget ", int_mblaze_fsl_tcget>; +def TCAGET : FSLGet<0x1B, "tcaget ", int_mblaze_fsl_tcaget>; +def TEGET : FSLGet<0x1B, "teget ", int_mblaze_fsl_teget>; +def TEAGET : FSLGet<0x1B, "teaget ", int_mblaze_fsl_teaget>; +def TECGET : FSLGet<0x1B, "tecget ", int_mblaze_fsl_tecget>; +def TECAGET : FSLGet<0x1B, "tecaget ", int_mblaze_fsl_tecaget>; +def TNGET : FSLGet<0x1B, "tnget ", int_mblaze_fsl_tnget>; +def TNAGET : FSLGet<0x1B, "tnaget ", int_mblaze_fsl_tnaget>; +def TNCGET : FSLGet<0x1B, "tncget ", int_mblaze_fsl_tncget>; +def TNCAGET : FSLGet<0x1B, "tncaget ", int_mblaze_fsl_tncaget>; +def TNEGET : FSLGet<0x1B, "tneget ", int_mblaze_fsl_tneget>; +def TNEAGET : FSLGet<0x1B, "tneaget ", int_mblaze_fsl_tneaget>; +def TNECGET : FSLGet<0x1B, "tnecget ", int_mblaze_fsl_tnecget>; +def TNECAGET : FSLGet<0x1B, "tnecaget ", int_mblaze_fsl_tnecaget>; + +//===----------------------------------------------------------------------===// +// FSL Dynamic Get Instructions +//===----------------------------------------------------------------------===// +def GETD : FSLGetD<0x1B, 0x00, "getd ", int_mblaze_fsl_get>; +def AGETD : FSLGetD<0x1B, 0x00, "agetd ", int_mblaze_fsl_aget>; +def CGETD : FSLGetD<0x1B, 0x00, "cgetd ", int_mblaze_fsl_cget>; +def CAGETD : FSLGetD<0x1B, 0x00, "cagetd ", int_mblaze_fsl_caget>; +def EGETD : FSLGetD<0x1B, 0x00, "egetd ", int_mblaze_fsl_eget>; +def EAGETD : FSLGetD<0x1B, 0x00, "eagetd ", int_mblaze_fsl_eaget>; +def ECGETD : FSLGetD<0x1B, 0x00, "ecgetd ", int_mblaze_fsl_ecget>; +def ECAGETD : FSLGetD<0x1B, 0x00, "ecagetd ", int_mblaze_fsl_ecaget>; +def NGETD : FSLGetD<0x1B, 0x00, "ngetd ", int_mblaze_fsl_nget>; +def NAGETD : FSLGetD<0x1B, 0x00, "nagetd ", int_mblaze_fsl_naget>; +def NCGETD : FSLGetD<0x1B, 0x00, "ncgetd ", int_mblaze_fsl_ncget>; +def NCAGETD : FSLGetD<0x1B, 0x00, "ncagetd ", int_mblaze_fsl_ncaget>; +def NEGETD : FSLGetD<0x1B, 0x00, "negetd ", int_mblaze_fsl_neget>; +def NEAGETD : FSLGetD<0x1B, 0x00, "neagetd ", int_mblaze_fsl_neaget>; +def NECGETD : FSLGetD<0x1B, 0x00, "necgetd ", int_mblaze_fsl_necget>; +def NECAGETD : FSLGetD<0x1B, 0x00, "necagetd ", int_mblaze_fsl_necaget>; +def TGETD : FSLGetD<0x1B, 0x00, "tgetd ", int_mblaze_fsl_tget>; +def TAGETD : FSLGetD<0x1B, 0x00, "tagetd ", int_mblaze_fsl_taget>; +def TCGETD : FSLGetD<0x1B, 0x00, "tcgetd ", int_mblaze_fsl_tcget>; +def TCAGETD : FSLGetD<0x1B, 0x00, "tcagetd ", int_mblaze_fsl_tcaget>; +def TEGETD : FSLGetD<0x1B, 0x00, "tegetd ", int_mblaze_fsl_teget>; +def TEAGETD : FSLGetD<0x1B, 0x00, "teagetd ", int_mblaze_fsl_teaget>; +def TECGETD : FSLGetD<0x1B, 0x00, "tecgetd ", int_mblaze_fsl_tecget>; +def TECAGETD : FSLGetD<0x1B, 0x00, "tecagetd ", int_mblaze_fsl_tecaget>; +def TNGETD : FSLGetD<0x1B, 0x00, "tngetd ", int_mblaze_fsl_tnget>; +def TNAGETD : FSLGetD<0x1B, 0x00, "tnagetd ", int_mblaze_fsl_tnaget>; +def TNCGETD : FSLGetD<0x1B, 0x00, "tncgetd ", int_mblaze_fsl_tncget>; +def TNCAGETD : FSLGetD<0x1B, 0x00, "tncagetd ", int_mblaze_fsl_tncaget>; +def TNEGETD : FSLGetD<0x1B, 0x00, "tnegetd ", int_mblaze_fsl_tneget>; +def TNEAGETD : FSLGetD<0x1B, 0x00, "tneagetd ", int_mblaze_fsl_tneaget>; +def TNECGETD : FSLGetD<0x1B, 0x00, "tnecgetd ", int_mblaze_fsl_tnecget>; +def TNECAGETD : FSLGetD<0x1B, 0x00, "tnecagetd", int_mblaze_fsl_tnecaget>; + +//===----------------------------------------------------------------------===// +// FSL Put Instructions +//===----------------------------------------------------------------------===// +def PUT : FSLPut<0x1B, "put ", int_mblaze_fsl_put>; +def APUT : FSLPut<0x1B, "aput ", int_mblaze_fsl_aput>; +def CPUT : FSLPut<0x1B, "cput ", int_mblaze_fsl_cput>; +def CAPUT : FSLPut<0x1B, "caput ", int_mblaze_fsl_caput>; +def NPUT : FSLPut<0x1B, "nput ", int_mblaze_fsl_nput>; +def NAPUT : FSLPut<0x1B, "naput ", int_mblaze_fsl_naput>; +def NCPUT : FSLPut<0x1B, "ncput ", int_mblaze_fsl_ncput>; +def NCAPUT : FSLPut<0x1B, "ncaput ", int_mblaze_fsl_ncaput>; +def TPUT : FSLPutT<0x1B, "tput ", int_mblaze_fsl_tput>; +def TAPUT : FSLPutT<0x1B, "taput ", int_mblaze_fsl_taput>; +def TCPUT : FSLPutT<0x1B, "tcput ", int_mblaze_fsl_tcput>; +def TCAPUT : FSLPutT<0x1B, "tcaput ", int_mblaze_fsl_tcaput>; +def TNPUT : FSLPutT<0x1B, "tnput ", int_mblaze_fsl_tnput>; +def TNAPUT : FSLPutT<0x1B, "tnaput ", int_mblaze_fsl_tnaput>; +def TNCPUT : FSLPutT<0x1B, "tncput ", int_mblaze_fsl_tncput>; +def TNCAPUT : FSLPutT<0x1B, "tncaput ", int_mblaze_fsl_tncaput>; + +//===----------------------------------------------------------------------===// +// FSL Dynamic Put Instructions +//===----------------------------------------------------------------------===// +def PUTD : FSLPutD<0x1B, 0x00, "putd ", int_mblaze_fsl_put>; +def APUTD : FSLPutD<0x1B, 0x00, "aputd ", int_mblaze_fsl_aput>; +def CPUTD : FSLPutD<0x1B, 0x00, "cputd ", int_mblaze_fsl_cput>; +def CAPUTD : FSLPutD<0x1B, 0x00, "caputd ", int_mblaze_fsl_caput>; +def NPUTD : FSLPutD<0x1B, 0x00, "nputd ", int_mblaze_fsl_nput>; +def NAPUTD : FSLPutD<0x1B, 0x00, "naputd ", int_mblaze_fsl_naput>; +def NCPUTD : FSLPutD<0x1B, 0x00, "ncputd ", int_mblaze_fsl_ncput>; +def NCAPUTD : FSLPutD<0x1B, 0x00, "ncaputd ", int_mblaze_fsl_ncaput>; +def TPUTD : FSLPutTD<0x1B, 0x00, "tputd ", int_mblaze_fsl_tput>; +def TAPUTD : FSLPutTD<0x1B, 0x00, "taputd ", int_mblaze_fsl_taput>; +def TCPUTD : FSLPutTD<0x1B, 0x00, "tcputd ", int_mblaze_fsl_tcput>; +def TCAPUTD : FSLPutTD<0x1B, 0x00, "tcaputd ", int_mblaze_fsl_tcaput>; +def TNPUTD : FSLPutTD<0x1B, 0x00, "tnputd ", int_mblaze_fsl_tnput>; +def TNAPUTD : FSLPutTD<0x1B, 0x00, "tnaputd ", int_mblaze_fsl_tnaput>; +def TNCPUTD : FSLPutTD<0x1B, 0x00, "tncputd ", int_mblaze_fsl_tncput>; +def TNCAPUTD : FSLPutTD<0x1B, 0x00, "tncaputd ", int_mblaze_fsl_tncaput>; diff --git a/lib/Target/MBlaze/MBlazeInstrFormats.td b/lib/Target/MBlaze/MBlazeInstrFormats.td new file mode 100644 index 0000000..7d65543 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeInstrFormats.td @@ -0,0 +1,246 @@ +//===- MBlazeInstrFormats.td - MB Instruction defs --------------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// Describe MBlaze instructions format +// +// CPU INSTRUCTION FORMATS +// +// opcode - operation code. +// rd - dst reg. +// ra - first src. reg. +// rb - second src. reg. +// imm16 - 16-bit immediate value. +// +//===----------------------------------------------------------------------===// + +// Generic MBlaze Format +class MBlazeInst<dag outs, dag ins, string asmstr, list<dag> pattern, + InstrItinClass itin> : Instruction +{ + field bits<32> Inst; + + let Namespace = "MBlaze"; + + bits<6> opcode; + + // Top 6 bits are the 'opcode' field + let Inst{0-5} = opcode; + + dag OutOperandList = outs; + dag InOperandList = ins; + + let AsmString = asmstr; + let Pattern = pattern; + let Itinerary = itin; +} + +//===----------------------------------------------------------------------===// +// Pseudo instruction class +//===----------------------------------------------------------------------===// +class MBlazePseudo<dag outs, dag ins, string asmstr, list<dag> pattern>: + MBlazeInst<outs, ins, asmstr, pattern, IIPseudo>; + +//===----------------------------------------------------------------------===// +// Type A instruction class in MBlaze : <|opcode|rd|ra|rb|flags|> +//===----------------------------------------------------------------------===// + +class TA<bits<6> op, bits<11> flags, dag outs, dag ins, string asmstr, + list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> rd; + bits<5> ra; + bits<5> rb; + + let opcode = op; + + let Inst{6-10} = rd; + let Inst{11-15} = ra; + let Inst{16-20} = rb; + let Inst{21-31} = flags; +} + +class TAI<bits<6> op, dag outs, dag ins, string asmstr, + list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> rd; + bits<5> ra; + bits<16> imm16; + + let opcode = op; + + let Inst{6-10} = rd; + let Inst{11-15} = ra; + let Inst{16-31} = imm16; +} + +class TIMM<bits<6> op, dag outs, dag ins, string asmstr, + list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> ra; + bits<16> imm16; + + let opcode = op; + + let Inst{6-15} = 0; + let Inst{16-31} = imm16; +} + +class TADDR<bits<6> op, dag outs, dag ins, string asmstr, + list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<26> addr; + + let opcode = op; + + let Inst{6-31} = addr; +} + +//===----------------------------------------------------------------------===// +// Type B instruction class in MBlaze : <|opcode|rd|ra|immediate|> +//===----------------------------------------------------------------------===// + +class TB<bits<6> op, dag outs, dag ins, string asmstr, list<dag> pattern, + InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> rd; + bits<5> ra; + bits<16> imm16; + + let opcode = op; + + let Inst{6-10} = rd; + let Inst{11-15} = ra; + let Inst{16-31} = imm16; +} + +//===----------------------------------------------------------------------===// +// Float instruction class in MBlaze : <|opcode|rd|ra|flags|> +//===----------------------------------------------------------------------===// + +class TF<bits<6> op, bits<11> flags, dag outs, dag ins, string asmstr, + list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> rd; + bits<5> ra; + + let opcode = op; + + let Inst{6-10} = rd; + let Inst{11-15} = ra; + let Inst{16-20} = 0; + let Inst{21-31} = flags; +} + +//===----------------------------------------------------------------------===// +// Branch instruction class in MBlaze : <|opcode|rd|br|ra|flags|> +//===----------------------------------------------------------------------===// + +class TBR<bits<6> op, bits<5> br, bits<11> flags, dag outs, dag ins, + string asmstr, list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> ra; + + let opcode = op; + + let Inst{6-10} = 0; + let Inst{11-15} = br; + let Inst{16-20} = ra; + let Inst{21-31} = flags; +} + +class TBRC<bits<6> op, bits<5> br, bits<11> flags, dag outs, dag ins, + string asmstr, list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> ra; + bits<5> rb; + + let opcode = op; + + let Inst{6-10} = br; + let Inst{11-15} = ra; + let Inst{16-20} = rb; + let Inst{21-31} = flags; +} + +class TBRL<bits<6> op, bits<5> br, bits<11> flags, dag outs, dag ins, + string asmstr, list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> ra; + + let opcode = op; + + let Inst{6-10} = 0xF; + let Inst{11-15} = br; + let Inst{16-20} = ra; + let Inst{21-31} = flags; +} + +class TBRI<bits<6> op, bits<5> br, dag outs, dag ins, + string asmstr, list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<16> imm16; + + let opcode = op; + + let Inst{6-10} = 0; + let Inst{11-15} = br; + let Inst{16-31} = imm16; +} + +class TBRLI<bits<6> op, bits<5> br, dag outs, dag ins, + string asmstr, list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<16> imm16; + + let opcode = op; + + let Inst{6-10} = 0xF; + let Inst{11-15} = br; + let Inst{16-31} = imm16; +} + +class TBRCI<bits<6> op, bits<5> br, dag outs, dag ins, + string asmstr, list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> ra; + bits<16> imm16; + + let opcode = op; + + let Inst{6-10} = br; + let Inst{11-15} = ra; + let Inst{16-31} = imm16; +} + +class TRET<bits<6> op, dag outs, dag ins, + string asmstr, list<dag> pattern, InstrItinClass itin> : + MBlazeInst<outs, ins, asmstr, pattern, itin> +{ + bits<5> ra; + bits<16> imm16; + + let opcode = op; + + let Inst{6-10} = 0x10; + let Inst{11-15} = ra; + let Inst{16-31} = imm16; +} diff --git a/lib/Target/MBlaze/MBlazeInstrInfo.cpp b/lib/Target/MBlaze/MBlazeInstrInfo.cpp new file mode 100644 index 0000000..a7e8eb7 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeInstrInfo.cpp @@ -0,0 +1,222 @@ +//===- MBlazeInstrInfo.cpp - MBlaze Instruction Information -----*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the MBlaze implementation of the TargetInstrInfo class. +// +//===----------------------------------------------------------------------===// + +#include "MBlazeInstrInfo.h" +#include "MBlazeTargetMachine.h" +#include "MBlazeMachineFunction.h" +#include "llvm/ADT/STLExtras.h" +#include "llvm/CodeGen/MachineInstrBuilder.h" +#include "llvm/CodeGen/MachineRegisterInfo.h" +#include "llvm/Support/ErrorHandling.h" +#include "MBlazeGenInstrInfo.inc" + +using namespace llvm; + +MBlazeInstrInfo::MBlazeInstrInfo(MBlazeTargetMachine &tm) + : TargetInstrInfoImpl(MBlazeInsts, array_lengthof(MBlazeInsts)), + TM(tm), RI(*TM.getSubtargetImpl(), *this) {} + +static bool isZeroImm(const MachineOperand &op) { + return op.isImm() && op.getImm() == 0; +} + +/// Return true if the instruction is a register to register move and +/// leave the source and dest operands in the passed parameters. +bool MBlazeInstrInfo:: +isMoveInstr(const MachineInstr &MI, unsigned &SrcReg, unsigned &DstReg, + unsigned &SrcSubIdx, unsigned &DstSubIdx) const { + SrcSubIdx = DstSubIdx = 0; // No sub-registers. + + // add $dst, $src, $zero || addu $dst, $zero, $src + // or $dst, $src, $zero || or $dst, $zero, $src + if ((MI.getOpcode() == MBlaze::ADD) || (MI.getOpcode() == MBlaze::OR)) { + if (MI.getOperand(1).isReg() && MI.getOperand(1).getReg() == MBlaze::R0) { + DstReg = MI.getOperand(0).getReg(); + SrcReg = MI.getOperand(2).getReg(); + return true; + } else if (MI.getOperand(2).isReg() && + MI.getOperand(2).getReg() == MBlaze::R0) { + DstReg = MI.getOperand(0).getReg(); + SrcReg = MI.getOperand(1).getReg(); + return true; + } + } + + // addi $dst, $src, 0 + // ori $dst, $src, 0 + if ((MI.getOpcode() == MBlaze::ADDI) || (MI.getOpcode() == MBlaze::ORI)) { + if ((MI.getOperand(1).isReg()) && (isZeroImm(MI.getOperand(2)))) { + DstReg = MI.getOperand(0).getReg(); + SrcReg = MI.getOperand(1).getReg(); + return true; + } + } + + return false; +} + +/// isLoadFromStackSlot - If the specified machine instruction is a direct +/// load from a stack slot, return the virtual or physical register number of +/// the destination along with the FrameIndex of the loaded stack slot. If +/// not, return 0. This predicate must return 0 if the instruction has +/// any side effects other than loading from the stack slot. +unsigned MBlazeInstrInfo:: +isLoadFromStackSlot(const MachineInstr *MI, int &FrameIndex) const { + if (MI->getOpcode() == MBlaze::LWI) { + if ((MI->getOperand(2).isFI()) && // is a stack slot + (MI->getOperand(1).isImm()) && // the imm is zero + (isZeroImm(MI->getOperand(1)))) { + FrameIndex = MI->getOperand(2).getIndex(); + return MI->getOperand(0).getReg(); + } + } + + return 0; +} + +/// isStoreToStackSlot - If the specified machine instruction is a direct +/// store to a stack slot, return the virtual or physical register number of +/// the source reg along with the FrameIndex of the loaded stack slot. If +/// not, return 0. This predicate must return 0 if the instruction has +/// any side effects other than storing to the stack slot. +unsigned MBlazeInstrInfo:: +isStoreToStackSlot(const MachineInstr *MI, int &FrameIndex) const { + if (MI->getOpcode() == MBlaze::SWI) { + if ((MI->getOperand(2).isFI()) && // is a stack slot + (MI->getOperand(1).isImm()) && // the imm is zero + (isZeroImm(MI->getOperand(1)))) { + FrameIndex = MI->getOperand(2).getIndex(); + return MI->getOperand(0).getReg(); + } + } + return 0; +} + +/// insertNoop - If data hazard condition is found insert the target nop +/// instruction. +void MBlazeInstrInfo:: +insertNoop(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI) const { + DebugLoc DL = DebugLoc::getUnknownLoc(); + if (MI != MBB.end()) DL = MI->getDebugLoc(); + BuildMI(MBB, MI, DL, get(MBlaze::NOP)); +} + +bool MBlazeInstrInfo:: +copyRegToReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, + unsigned DestReg, unsigned SrcReg, + const TargetRegisterClass *DestRC, + const TargetRegisterClass *SrcRC) const { + DebugLoc dl = DebugLoc::getUnknownLoc(); + llvm::BuildMI(MBB, I, dl, get(MBlaze::ADD), DestReg) + .addReg(SrcReg).addReg(MBlaze::R0); + return true; +} + +void MBlazeInstrInfo:: +storeRegToStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, + unsigned SrcReg, bool isKill, int FI, + const TargetRegisterClass *RC) const { + DebugLoc dl = DebugLoc::getUnknownLoc(); + BuildMI(MBB, I, dl, get(MBlaze::SWI)).addReg(SrcReg,getKillRegState(isKill)) + .addImm(0).addFrameIndex(FI); +} + +void MBlazeInstrInfo:: +loadRegFromStackSlot(MachineBasicBlock &MBB, MachineBasicBlock::iterator I, + unsigned DestReg, int FI, + const TargetRegisterClass *RC) const { + DebugLoc dl = DebugLoc::getUnknownLoc(); + BuildMI(MBB, I, dl, get(MBlaze::LWI), DestReg) + .addImm(0).addFrameIndex(FI); +} + +MachineInstr *MBlazeInstrInfo:: +foldMemoryOperandImpl(MachineFunction &MF, + MachineInstr* MI, + const SmallVectorImpl<unsigned> &Ops, int FI) const { + if (Ops.size() != 1) return NULL; + + MachineInstr *NewMI = NULL; + + switch (MI->getOpcode()) { + case MBlaze::OR: + case MBlaze::ADD: + if ((MI->getOperand(0).isReg()) && + (MI->getOperand(2).isReg()) && + (MI->getOperand(2).getReg() == MBlaze::R0) && + (MI->getOperand(1).isReg())) { + if (Ops[0] == 0) { // COPY -> STORE + unsigned SrcReg = MI->getOperand(1).getReg(); + bool isKill = MI->getOperand(1).isKill(); + bool isUndef = MI->getOperand(1).isUndef(); + NewMI = BuildMI(MF, MI->getDebugLoc(), get(MBlaze::SW)) + .addReg(SrcReg, getKillRegState(isKill) | getUndefRegState(isUndef)) + .addImm(0).addFrameIndex(FI); + } else { // COPY -> LOAD + unsigned DstReg = MI->getOperand(0).getReg(); + bool isDead = MI->getOperand(0).isDead(); + bool isUndef = MI->getOperand(0).isUndef(); + NewMI = BuildMI(MF, MI->getDebugLoc(), get(MBlaze::LW)) + .addReg(DstReg, RegState::Define | getDeadRegState(isDead) | + getUndefRegState(isUndef)) + .addImm(0).addFrameIndex(FI); + } + } + break; + } + + return NewMI; +} + +//===----------------------------------------------------------------------===// +// Branch Analysis +//===----------------------------------------------------------------------===// +unsigned MBlazeInstrInfo:: +InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB, + MachineBasicBlock *FBB, + const SmallVectorImpl<MachineOperand> &Cond) const { + DebugLoc dl = DebugLoc::getUnknownLoc(); + + // Can only insert uncond branches so far. + assert(Cond.empty() && !FBB && TBB && "Can only handle uncond branches!"); + BuildMI(&MBB, dl, get(MBlaze::BRI)).addMBB(TBB); + return 1; +} + +/// getGlobalBaseReg - Return a virtual register initialized with the +/// the global base register value. Output instructions required to +/// initialize the register in the function entry block, if necessary. +/// +unsigned MBlazeInstrInfo::getGlobalBaseReg(MachineFunction *MF) const { + MBlazeFunctionInfo *MBlazeFI = MF->getInfo<MBlazeFunctionInfo>(); + unsigned GlobalBaseReg = MBlazeFI->getGlobalBaseReg(); + if (GlobalBaseReg != 0) + return GlobalBaseReg; + + // Insert the set of GlobalBaseReg into the first MBB of the function + MachineBasicBlock &FirstMBB = MF->front(); + MachineBasicBlock::iterator MBBI = FirstMBB.begin(); + MachineRegisterInfo &RegInfo = MF->getRegInfo(); + const TargetInstrInfo *TII = MF->getTarget().getInstrInfo(); + + GlobalBaseReg = RegInfo.createVirtualRegister(MBlaze::CPURegsRegisterClass); + bool Ok = TII->copyRegToReg(FirstMBB, MBBI, GlobalBaseReg, MBlaze::R20, + MBlaze::CPURegsRegisterClass, + MBlaze::CPURegsRegisterClass); + assert(Ok && "Couldn't assign to global base register!"); + Ok = Ok; // Silence warning when assertions are turned off. + RegInfo.addLiveIn(MBlaze::R20); + + MBlazeFI->setGlobalBaseReg(GlobalBaseReg); + return GlobalBaseReg; +} diff --git a/lib/Target/MBlaze/MBlazeInstrInfo.h b/lib/Target/MBlaze/MBlazeInstrInfo.h new file mode 100644 index 0000000..4f79f1c --- /dev/null +++ b/lib/Target/MBlaze/MBlazeInstrInfo.h @@ -0,0 +1,242 @@ +//===- MBlazeInstrInfo.h - MBlaze Instruction Information -------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the MBlaze implementation of the TargetInstrInfo class. +// +//===----------------------------------------------------------------------===// + +#ifndef MBLAZEINSTRUCTIONINFO_H +#define MBLAZEINSTRUCTIONINFO_H + +#include "MBlaze.h" +#include "llvm/Support/ErrorHandling.h" +#include "llvm/Target/TargetInstrInfo.h" +#include "MBlazeRegisterInfo.h" + +namespace llvm { + +namespace MBlaze { + + // MBlaze Branch Codes + enum FPBranchCode { + BRANCH_F, + BRANCH_T, + BRANCH_FL, + BRANCH_TL, + BRANCH_INVALID + }; + + // MBlaze Condition Codes + enum CondCode { + // To be used with float branch True + FCOND_F, + FCOND_UN, + FCOND_EQ, + FCOND_UEQ, + FCOND_OLT, + FCOND_ULT, + FCOND_OLE, + FCOND_ULE, + FCOND_SF, + FCOND_NGLE, + FCOND_SEQ, + FCOND_NGL, + FCOND_LT, + FCOND_NGE, + FCOND_LE, + FCOND_NGT, + + // To be used with float branch False + // This conditions have the same mnemonic as the + // above ones, but are used with a branch False; + FCOND_T, + FCOND_OR, + FCOND_NEQ, + FCOND_OGL, + FCOND_UGE, + FCOND_OGE, + FCOND_UGT, + FCOND_OGT, + FCOND_ST, + FCOND_GLE, + FCOND_SNE, + FCOND_GL, + FCOND_NLT, + FCOND_GE, + FCOND_NLE, + FCOND_GT, + + // Only integer conditions + COND_E, + COND_GZ, + COND_GEZ, + COND_LZ, + COND_LEZ, + COND_NE, + COND_INVALID + }; + + // Turn condition code into conditional branch opcode. + unsigned GetCondBranchFromCond(CondCode CC); + + /// GetOppositeBranchCondition - Return the inverse of the specified cond, + /// e.g. turning COND_E to COND_NE. + CondCode GetOppositeBranchCondition(MBlaze::CondCode CC); + + /// MBlazeCCToString - Map each FP condition code to its string + inline static const char *MBlazeFCCToString(MBlaze::CondCode CC) + { + switch (CC) { + default: llvm_unreachable("Unknown condition code"); + case FCOND_F: + case FCOND_T: return "f"; + case FCOND_UN: + case FCOND_OR: return "un"; + case FCOND_EQ: + case FCOND_NEQ: return "eq"; + case FCOND_UEQ: + case FCOND_OGL: return "ueq"; + case FCOND_OLT: + case FCOND_UGE: return "olt"; + case FCOND_ULT: + case FCOND_OGE: return "ult"; + case FCOND_OLE: + case FCOND_UGT: return "ole"; + case FCOND_ULE: + case FCOND_OGT: return "ule"; + case FCOND_SF: + case FCOND_ST: return "sf"; + case FCOND_NGLE: + case FCOND_GLE: return "ngle"; + case FCOND_SEQ: + case FCOND_SNE: return "seq"; + case FCOND_NGL: + case FCOND_GL: return "ngl"; + case FCOND_LT: + case FCOND_NLT: return "lt"; + case FCOND_NGE: + case FCOND_GE: return "ge"; + case FCOND_LE: + case FCOND_NLE: return "nle"; + case FCOND_NGT: + case FCOND_GT: return "gt"; + } + } +} + +/// MBlazeII - This namespace holds all of the target specific flags that +/// instruction info tracks. +/// +namespace MBlazeII { + /// Target Operand Flag enum. + enum TOF { + //===------------------------------------------------------------------===// + // MBlaze Specific MachineOperand flags. + MO_NO_FLAG, + + /// MO_GOT - Represents the offset into the global offset table at which + /// the address the relocation entry symbol resides during execution. + MO_GOT, + + /// MO_GOT_CALL - Represents the offset into the global offset table at + /// which the address of a call site relocation entry symbol resides + /// during execution. This is different from the above since this flag + /// can only be present in call instructions. + MO_GOT_CALL, + + /// MO_GPREL - Represents the offset from the current gp value to be used + /// for the relocatable object file being produced. + MO_GPREL, + + /// MO_ABS_HILO - Represents the hi or low part of an absolute symbol + /// address. + MO_ABS_HILO + + }; +} + +class MBlazeInstrInfo : public TargetInstrInfoImpl { + MBlazeTargetMachine &TM; + const MBlazeRegisterInfo RI; +public: + explicit MBlazeInstrInfo(MBlazeTargetMachine &TM); + + /// getRegisterInfo - TargetInstrInfo is a superset of MRegister info. As + /// such, whenever a client has an instance of instruction info, it should + /// always be able to get register info as well (through this method). + /// + virtual const MBlazeRegisterInfo &getRegisterInfo() const { return RI; } + + /// Return true if the instruction is a register to register move and return + /// the source and dest operands and their sub-register indices by reference. + virtual bool isMoveInstr(const MachineInstr &MI, + unsigned &SrcReg, unsigned &DstReg, + unsigned &SrcSubIdx, unsigned &DstSubIdx) const; + + /// isLoadFromStackSlot - If the specified machine instruction is a direct + /// load from a stack slot, return the virtual or physical register number of + /// the destination along with the FrameIndex of the loaded stack slot. If + /// not, return 0. This predicate must return 0 if the instruction has + /// any side effects other than loading from the stack slot. + virtual unsigned isLoadFromStackSlot(const MachineInstr *MI, + int &FrameIndex) const; + + /// isStoreToStackSlot - If the specified machine instruction is a direct + /// store to a stack slot, return the virtual or physical register number of + /// the source reg along with the FrameIndex of the loaded stack slot. If + /// not, return 0. This predicate must return 0 if the instruction has + /// any side effects other than storing to the stack slot. + virtual unsigned isStoreToStackSlot(const MachineInstr *MI, + int &FrameIndex) const; + + /// Branch Analysis + virtual unsigned InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB, + MachineBasicBlock *FBB, + const SmallVectorImpl<MachineOperand> &Cond) const; + virtual bool copyRegToReg(MachineBasicBlock &MBB, + MachineBasicBlock::iterator I, + unsigned DestReg, unsigned SrcReg, + const TargetRegisterClass *DestRC, + const TargetRegisterClass *SrcRC) const; + virtual void storeRegToStackSlot(MachineBasicBlock &MBB, + MachineBasicBlock::iterator MBBI, + unsigned SrcReg, bool isKill, int FrameIndex, + const TargetRegisterClass *RC) const; + + virtual void loadRegFromStackSlot(MachineBasicBlock &MBB, + MachineBasicBlock::iterator MBBI, + unsigned DestReg, int FrameIndex, + const TargetRegisterClass *RC) const; + + virtual MachineInstr* foldMemoryOperandImpl(MachineFunction &MF, + MachineInstr* MI, + const SmallVectorImpl<unsigned> &Ops, + int FrameIndex) const; + + virtual MachineInstr* foldMemoryOperandImpl(MachineFunction &MF, + MachineInstr* MI, + const SmallVectorImpl<unsigned> &Ops, + MachineInstr* LoadMI) const { + return 0; + } + + /// Insert nop instruction when hazard condition is found + virtual void insertNoop(MachineBasicBlock &MBB, + MachineBasicBlock::iterator MI) const; + + /// getGlobalBaseReg - Return a virtual register initialized with the + /// the global base register value. Output instructions required to + /// initialize the register in the function entry block, if necessary. + /// + unsigned getGlobalBaseReg(MachineFunction *MF) const; +}; + +} + +#endif diff --git a/lib/Target/MBlaze/MBlazeInstrInfo.td b/lib/Target/MBlaze/MBlazeInstrInfo.td new file mode 100644 index 0000000..3c406dd --- /dev/null +++ b/lib/Target/MBlaze/MBlazeInstrInfo.td @@ -0,0 +1,672 @@ +//===- MBlazeInstrInfo.td - MBlaze Instruction defs -------------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// Instruction format superclass +//===----------------------------------------------------------------------===// +include "MBlazeInstrFormats.td" + +//===----------------------------------------------------------------------===// +// MBlaze profiles and nodes +//===----------------------------------------------------------------------===// +def SDT_MBlazeRet : SDTypeProfile<0, 1, [SDTCisInt<0>]>; +def SDT_MBlazeJmpLink : SDTypeProfile<0, 1, [SDTCisVT<0, i32>]>; + +// Call +def MBlazeJmpLink : SDNode<"MBlazeISD::JmpLink",SDT_MBlazeJmpLink, + [SDNPHasChain,SDNPOptInFlag,SDNPOutFlag]>; + +// Return +def MBlazeRet : SDNode<"MBlazeISD::Ret", SDT_MBlazeRet, + [SDNPHasChain, SDNPOptInFlag]>; + +// Hi and Lo nodes are used to handle global addresses. Used on +// MBlazeISelLowering to lower stuff like GlobalAddress, ExternalSymbol +// static model. +def MBWrapper : SDNode<"MBlazeISD::Wrap", SDTIntUnaryOp>; +def MBlazeGPRel : SDNode<"MBlazeISD::GPRel", SDTIntUnaryOp>; + +def SDT_MBCallSeqStart : SDCallSeqStart<[SDTCisVT<0, i32>]>; +def SDT_MBCallSeqEnd : SDCallSeqEnd<[SDTCisVT<0, i32>, SDTCisVT<1, i32>]>; + +// These are target-independent nodes, but have target-specific formats. +def callseq_start : SDNode<"ISD::CALLSEQ_START", SDT_MBCallSeqStart, + [SDNPHasChain, SDNPOutFlag]>; +def callseq_end : SDNode<"ISD::CALLSEQ_END", SDT_MBCallSeqEnd, + [SDNPHasChain, SDNPOptInFlag, SDNPOutFlag]>; + +def SDTMBlazeSelectCC : SDTypeProfile<1, 3, [SDTCisSameAs<0, 1>]>; + +//===----------------------------------------------------------------------===// +// MBlaze Instruction Predicate Definitions. +//===----------------------------------------------------------------------===// +def HasPipe3 : Predicate<"Subtarget.hasPipe3()">; +def HasBarrel : Predicate<"Subtarget.hasBarrel()">; +def NoBarrel : Predicate<"!Subtarget.hasBarrel()">; +def HasDiv : Predicate<"Subtarget.hasDiv()">; +def HasMul : Predicate<"Subtarget.hasMul()">; +def HasFSL : Predicate<"Subtarget.hasFSL()">; +def HasEFSL : Predicate<"Subtarget.hasEFSL()">; +def HasMSRSet : Predicate<"Subtarget.hasMSRSet()">; +def HasException : Predicate<"Subtarget.hasException()">; +def HasPatCmp : Predicate<"Subtarget.hasPatCmp()">; +def HasFPU : Predicate<"Subtarget.hasFPU()">; +def HasESR : Predicate<"Subtarget.hasESR()">; +def HasPVR : Predicate<"Subtarget.hasPVR()">; +def HasMul64 : Predicate<"Subtarget.hasMul64()">; +def HasSqrt : Predicate<"Subtarget.hasSqrt()">; +def HasMMU : Predicate<"Subtarget.hasMMU()">; + +//===----------------------------------------------------------------------===// +// MBlaze Operand, Complex Patterns and Transformations Definitions. +//===----------------------------------------------------------------------===// + +// Instruction operand types +def brtarget : Operand<OtherVT>; +def calltarget : Operand<i32>; +def simm16 : Operand<i32>; +def uimm5 : Operand<i32>; +def fimm : Operand<f32>; + +// Unsigned Operand +def uimm16 : Operand<i32> { + let PrintMethod = "printUnsignedImm"; +} + +// FSL Operand +def fslimm : Operand<i32> { + let PrintMethod = "printFSLImm"; +} + +// Address operand +def memri : Operand<i32> { + let PrintMethod = "printMemOperand"; + let MIOperandInfo = (ops simm16, CPURegs); +} + +def memrr : Operand<i32> { + let PrintMethod = "printMemOperand"; + let MIOperandInfo = (ops CPURegs, CPURegs); +} + +// Transformation Function - get the lower 16 bits. +def LO16 : SDNodeXForm<imm, [{ + return getI32Imm((unsigned)N->getZExtValue() & 0xFFFF); +}]>; + +// Transformation Function - get the higher 16 bits. +def HI16 : SDNodeXForm<imm, [{ + return getI32Imm((unsigned)N->getZExtValue() >> 16); +}]>; + +// Node immediate fits as 16-bit sign extended on target immediate. +// e.g. addi, andi +def immSExt16 : PatLeaf<(imm), [{ + return (N->getZExtValue() >> 16) == 0; +}]>; + +// Node immediate fits as 16-bit zero extended on target immediate. +// The LO16 param means that only the lower 16 bits of the node +// immediate are caught. +// e.g. addiu, sltiu +def immZExt16 : PatLeaf<(imm), [{ + return (N->getZExtValue() >> 16) == 0; +}], LO16>; + +// FSL immediate field must fit in 4 bits. +def immZExt4 : PatLeaf<(imm), [{ + return N->getZExtValue() == ((N->getZExtValue()) & 0xf) ; +}]>; + +// shamt field must fit in 5 bits. +def immZExt5 : PatLeaf<(imm), [{ + return N->getZExtValue() == ((N->getZExtValue()) & 0x1f) ; +}]>; + +// MBlaze Address Mode! SDNode frameindex could possibily be a match +// since load and store instructions from stack used it. +def iaddr : ComplexPattern<i32, 2, "SelectAddrRegImm", [frameindex], []>; +def xaddr : ComplexPattern<i32, 2, "SelectAddrRegReg", [], []>; + +//===----------------------------------------------------------------------===// +// Pseudo instructions +//===----------------------------------------------------------------------===// + +// As stack alignment is always done with addiu, we need a 16-bit immediate +let Defs = [R1], Uses = [R1] in { +def ADJCALLSTACKDOWN : MBlazePseudo<(outs), (ins simm16:$amt), + "${:comment} ADJCALLSTACKDOWN $amt", + [(callseq_start timm:$amt)]>; +def ADJCALLSTACKUP : MBlazePseudo<(outs), + (ins uimm16:$amt1, simm16:$amt2), + "${:comment} ADJCALLSTACKUP $amt1", + [(callseq_end timm:$amt1, timm:$amt2)]>; +} + +// Some assembly macros need to avoid pseudoinstructions and assembler +// automatic reodering, we should reorder ourselves. +def MACRO : MBlazePseudo<(outs), (ins), ".set macro", []>; +def REORDER : MBlazePseudo<(outs), (ins), ".set reorder", []>; +def NOMACRO : MBlazePseudo<(outs), (ins), ".set nomacro", []>; +def NOREORDER : MBlazePseudo<(outs), (ins), ".set noreorder", []>; + +// When handling PIC code the assembler needs .cpload and .cprestore +// directives. If the real instructions corresponding these directives +// are used, we have the same behavior, but get also a bunch of warnings +// from the assembler. +def CPLOAD : MBlazePseudo<(outs), (ins CPURegs:$reg), ".cpload $reg", []>; +def CPRESTORE : MBlazePseudo<(outs), (ins uimm16:$l), ".cprestore $l\n", []>; + +//===----------------------------------------------------------------------===// +// Instructions specific format +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// Arithmetic Instructions +//===----------------------------------------------------------------------===// +class Arith<bits<6> op, bits<11> flags, string instr_asm, SDNode OpNode, + InstrItinClass itin> : + TA<op, flags, (outs CPURegs:$dst), (ins CPURegs:$b, CPURegs:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [(set CPURegs:$dst, (OpNode CPURegs:$b, CPURegs:$c))], itin>; + +class ArithI<bits<6> op, string instr_asm, SDNode OpNode, + Operand Od, PatLeaf imm_type> : + TAI<op, (outs CPURegs:$dst), (ins CPURegs:$b, Od:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [(set CPURegs:$dst, (OpNode CPURegs:$b, imm_type:$c))], IIAlu>; + +class ArithR<bits<6> op, bits<11> flags, string instr_asm, SDNode OpNode, + InstrItinClass itin> : + TA<op, flags, (outs CPURegs:$dst), (ins CPURegs:$c, CPURegs:$b), + !strconcat(instr_asm, " $dst, $c, $b"), + [(set CPURegs:$dst, (OpNode CPURegs:$b, CPURegs:$c))], itin>; + +class ArithRI<bits<6> op, string instr_asm, SDNode OpNode, + Operand Od, PatLeaf imm_type> : + TAI<op, (outs CPURegs:$dst), (ins Od:$b, CPURegs:$c), + !strconcat(instr_asm, " $dst, $c, $b"), + [(set CPURegs:$dst, (OpNode imm_type:$b, CPURegs:$c))], IIAlu>; + +class ArithN<bits<6> op, bits<11> flags, string instr_asm, + InstrItinClass itin> : + TA<op, flags, (outs CPURegs:$dst), (ins CPURegs:$b, CPURegs:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [], itin>; + +class ArithNI<bits<6> op, string instr_asm,Operand Od, PatLeaf imm_type> : + TAI<op, (outs CPURegs:$dst), (ins CPURegs:$b, Od:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [], IIAlu>; + +class ArithRN<bits<6> op, bits<11> flags, string instr_asm, + InstrItinClass itin> : + TA<op, flags, (outs CPURegs:$dst), (ins CPURegs:$c, CPURegs:$b), + !strconcat(instr_asm, " $dst, $b, $c"), + [], itin>; + +class ArithRNI<bits<6> op, string instr_asm,Operand Od, PatLeaf imm_type> : + TAI<op, (outs CPURegs:$dst), (ins Od:$c, CPURegs:$b), + !strconcat(instr_asm, " $dst, $b, $c"), + [], IIAlu>; + +//===----------------------------------------------------------------------===// +// Misc Arithmetic Instructions +//===----------------------------------------------------------------------===// + +class Logic<bits<6> op, bits<11> flags, string instr_asm, SDNode OpNode> : + TA<op, flags, (outs CPURegs:$dst), (ins CPURegs:$b, CPURegs:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [(set CPURegs:$dst, (OpNode CPURegs:$b, CPURegs:$c))], IIAlu>; + +class LogicI<bits<6> op, string instr_asm, SDNode OpNode> : + TAI<op, (outs CPURegs:$dst), (ins CPURegs:$b, uimm16:$c), + !strconcat(instr_asm, " $dst, $b, $c"), + [(set CPURegs:$dst, (OpNode CPURegs:$b, immZExt16:$c))], + IIAlu>; + +class EffectiveAddress<string instr_asm> : + TAI<0x08, (outs CPURegs:$dst), (ins memri:$addr), + instr_asm, [(set CPURegs:$dst, iaddr:$addr)], IIAlu>; + +//===----------------------------------------------------------------------===// +// Memory Access Instructions +//===----------------------------------------------------------------------===// +class LoadM<bits<6> op, string instr_asm, PatFrag OpNode> : + TA<op, 0x000, (outs CPURegs:$dst), (ins memrr:$addr), + !strconcat(instr_asm, " $dst, $addr"), + [(set CPURegs:$dst, (OpNode xaddr:$addr))], IILoad>; + +class LoadMI<bits<6> op, string instr_asm, PatFrag OpNode> : + TAI<op, (outs CPURegs:$dst), (ins memri:$addr), + !strconcat(instr_asm, " $dst, $addr"), + [(set CPURegs:$dst, (OpNode iaddr:$addr))], IILoad>; + +class StoreM<bits<6> op, string instr_asm, PatFrag OpNode> : + TA<op, 0x000, (outs), (ins CPURegs:$dst, memrr:$addr), + !strconcat(instr_asm, " $dst, $addr"), + [(OpNode CPURegs:$dst, xaddr:$addr)], IIStore>; + +class StoreMI<bits<6> op, string instr_asm, PatFrag OpNode> : + TAI<op, (outs), (ins CPURegs:$dst, memri:$addr), + !strconcat(instr_asm, " $dst, $addr"), + [(OpNode CPURegs:$dst, iaddr:$addr)], IIStore>; + +//===----------------------------------------------------------------------===// +// Branch Instructions +//===----------------------------------------------------------------------===// +class Branch<bits<6> op, bits<5> br, bits<11> flags, string instr_asm> : + TBR<op, br, flags, (outs), (ins CPURegs:$target), + !strconcat(instr_asm, " $target"), + [(brind CPURegs:$target)], IIBranch>; + +class BranchI<bits<6> op, bits<5> brf, string instr_asm> : + TBRI<op, brf, (outs), (ins brtarget:$target), + !strconcat(instr_asm, " $target"), + [(br bb:$target)], IIBranch>; + +//===----------------------------------------------------------------------===// +// Branch and Link Instructions +//===----------------------------------------------------------------------===// +class BranchL<bits<6> op, bits<5> br, bits<11> flags, string instr_asm> : + TBRL<op, br, flags, (outs), (ins CPURegs:$target), + !strconcat(instr_asm, " r15, $target"), + [], IIBranch>; + +class BranchLI<bits<6> op, bits<5> br, string instr_asm> : + TBRLI<op, br, (outs), (ins calltarget:$target), + !strconcat(instr_asm, " r15, $target"), + [], IIBranch>; + +//===----------------------------------------------------------------------===// +// Conditional Branch Instructions +//===----------------------------------------------------------------------===// +class BranchC<bits<6> op, bits<5> br, bits<11> flags, string instr_asm, + PatFrag cond_op> : + TBRC<op, br, flags, (outs), + (ins CPURegs:$a, CPURegs:$b, brtarget:$offset), + !strconcat(instr_asm, " $a, $b, $offset"), + [], IIBranch>; + //(brcond (cond_op CPURegs:$a, CPURegs:$b), bb:$offset)], + //IIBranch>; + +class BranchCI<bits<6> op, bits<5> br, string instr_asm, PatFrag cond_op> : + TBRCI<op, br, (outs), (ins CPURegs:$a, brtarget:$offset), + !strconcat(instr_asm, " $a, $offset"), + [], IIBranch>; + +//===----------------------------------------------------------------------===// +// MBlaze arithmetic instructions +//===----------------------------------------------------------------------===// + +let isCommutable = 1, isAsCheapAsAMove = 1 in { + def ADD : Arith<0x00, 0x000, "add ", add, IIAlu>; + def ADDC : Arith<0x02, 0x000, "addc ", adde, IIAlu>; + def ADDK : Arith<0x04, 0x000, "addk ", addc, IIAlu>; + def ADDKC : ArithN<0x06, 0x000, "addkc ", IIAlu>; + def AND : Logic<0x21, 0x000, "and ", and>; + def OR : Logic<0x20, 0x000, "or ", or>; + def XOR : Logic<0x22, 0x000, "xor ", xor>; +} + +let isAsCheapAsAMove = 1 in { + def ANDN : ArithN<0x23, 0x000, "andn ", IIAlu>; + def CMP : ArithN<0x05, 0x001, "cmp ", IIAlu>; + def CMPU : ArithN<0x05, 0x003, "cmpu ", IIAlu>; + def RSUB : ArithR<0x01, 0x000, "rsub ", sub, IIAlu>; + def RSUBC : ArithR<0x03, 0x000, "rsubc ", sube, IIAlu>; + def RSUBK : ArithR<0x05, 0x000, "rsubk ", subc, IIAlu>; + def RSUBKC : ArithRN<0x07, 0x000, "rsubkc ", IIAlu>; +} + +let isCommutable = 1, Predicates=[HasMul] in { + def MUL : Arith<0x10, 0x000, "mul ", mul, IIAlu>; +} + +let isCommutable = 1, Predicates=[HasMul,HasMul64] in { + def MULH : Arith<0x10, 0x001, "mulh ", mulhs, IIAlu>; + def MULHU : Arith<0x10, 0x003, "mulhu ", mulhu, IIAlu>; +} + +let Predicates=[HasMul,HasMul64] in { + def MULHSU : ArithN<0x10, 0x002, "mulhsu ", IIAlu>; +} + +let Predicates=[HasBarrel] in { + def BSRL : Arith<0x11, 0x000, "bsrl ", srl, IIAlu>; + def BSRA : Arith<0x11, 0x200, "bsra ", sra, IIAlu>; + def BSLL : Arith<0x11, 0x400, "bsll ", shl, IIAlu>; + def BSRLI : ArithI<0x11, "bsrli ", srl, uimm5, immZExt5>; + def BSRAI : ArithI<0x11, "bsrai ", sra, uimm5, immZExt5>; + def BSLLI : ArithI<0x11, "bslli ", shl, uimm5, immZExt5>; +} + +let Predicates=[HasDiv] in { + def IDIV : Arith<0x12, 0x000, "idiv ", sdiv, IIAlu>; + def IDIVU : Arith<0x12, 0x002, "idivu ", udiv, IIAlu>; +} + +//===----------------------------------------------------------------------===// +// MBlaze immediate mode arithmetic instructions +//===----------------------------------------------------------------------===// + +let isAsCheapAsAMove = 1 in { + def ADDI : ArithI<0x08, "addi ", add, simm16, immSExt16>; + def ADDIC : ArithNI<0x0A, "addic ", simm16, immSExt16>; + def ADDIK : ArithNI<0x0C, "addik ", simm16, immSExt16>; + def ADDIKC : ArithI<0x0E, "addikc ", addc, simm16, immSExt16>; + def RSUBI : ArithRI<0x09, "rsubi ", sub, simm16, immSExt16>; + def RSUBIC : ArithRNI<0x0B, "rsubi ", simm16, immSExt16>; + def RSUBIK : ArithRNI<0x0E, "rsubic ", simm16, immSExt16>; + def RSUBIKC : ArithRI<0x0F, "rsubikc", subc, simm16, immSExt16>; + def ANDNI : ArithNI<0x2B, "andni ", uimm16, immZExt16>; + def ANDI : LogicI<0x29, "andi ", and>; + def ORI : LogicI<0x28, "ori ", or>; + def XORI : LogicI<0x2A, "xori ", xor>; +} + +let Predicates=[HasMul] in { + def MULI : ArithI<0x18, "muli ", mul, simm16, immSExt16>; +} + +//===----------------------------------------------------------------------===// +// MBlaze memory access instructions +//===----------------------------------------------------------------------===// + +let canFoldAsLoad = 1, isReMaterializable = 1 in { + def LBU : LoadM<0x30, "lbu ", zextloadi8>; + def LHU : LoadM<0x31, "lhu ", zextloadi16>; + def LW : LoadM<0x32, "lw ", load>; + + def LBUI : LoadMI<0x30, "lbui ", zextloadi8>; + def LHUI : LoadMI<0x31, "lhui ", zextloadi16>; + def LWI : LoadMI<0x32, "lwi ", load>; +} + + def SB : StoreM<0x34, "sb ", truncstorei8>; + def SH : StoreM<0x35, "sh ", truncstorei16>; + def SW : StoreM<0x36, "sw ", store>; + + def SBI : StoreMI<0x34, "sbi ", truncstorei8>; + def SHI : StoreMI<0x35, "shi ", truncstorei16>; + def SWI : StoreMI<0x36, "swi ", store>; + +//===----------------------------------------------------------------------===// +// MBlaze branch instructions +//===----------------------------------------------------------------------===// + +let isBranch = 1, isTerminator = 1, hasCtrlDep = 1 in { + def BRI : BranchI<0x2E, 0x00, "bri ">; + def BRAI : BranchI<0x2E, 0x08, "brai ">; + def BEQI : BranchCI<0x2F, 0x00, "beqi ", seteq>; + def BNEI : BranchCI<0x2F, 0x01, "bnei ", setne>; + def BLTI : BranchCI<0x2F, 0x02, "blti ", setlt>; + def BLEI : BranchCI<0x2F, 0x03, "blei ", setle>; + def BGTI : BranchCI<0x2F, 0x04, "bgti ", setgt>; + def BGEI : BranchCI<0x2F, 0x05, "bgei ", setge>; +} + +let isBranch = 1, isIndirectBranch = 1, isTerminator = 1, hasCtrlDep = 1 in { + def BR : Branch<0x26, 0x00, 0x000, "br ">; + def BRA : Branch<0x26, 0x08, 0x000, "bra ">; + def BEQ : BranchC<0x27, 0x00, 0x000, "beq ", seteq>; + def BNE : BranchC<0x27, 0x01, 0x000, "bne ", setne>; + def BLT : BranchC<0x27, 0x02, 0x000, "blt ", setlt>; + def BLE : BranchC<0x27, 0x03, 0x000, "ble ", setle>; + def BGT : BranchC<0x27, 0x04, 0x000, "bgt ", setgt>; + def BGE : BranchC<0x27, 0x05, 0x000, "bge ", setge>; +} + +let isBranch = 1, isTerminator = 1, hasDelaySlot = 1, hasCtrlDep = 1 in { + def BRID : BranchI<0x2E, 0x10, "brid ">; + def BRAID : BranchI<0x2E, 0x18, "braid ">; + def BEQID : BranchCI<0x2F, 0x10, "beqid ", seteq>; + def BNEID : BranchCI<0x2F, 0x11, "bneid ", setne>; + def BLTID : BranchCI<0x2F, 0x12, "bltid ", setlt>; + def BLEID : BranchCI<0x2F, 0x13, "bleid ", setle>; + def BGTID : BranchCI<0x2F, 0x14, "bgtid ", setgt>; + def BGEID : BranchCI<0x2F, 0x15, "bgeid ", setge>; +} + +let isBranch = 1, isIndirectBranch = 1, isTerminator = 1, + hasDelaySlot = 1, hasCtrlDep = 1 in { + def BRD : Branch<0x26, 0x10, 0x000, "brd ">; + def BRAD : Branch<0x26, 0x18, 0x000, "brad ">; + def BEQD : BranchC<0x27, 0x10, 0x000, "beqd ", seteq>; + def BNED : BranchC<0x27, 0x11, 0x000, "bned ", setne>; + def BLTD : BranchC<0x27, 0x12, 0x000, "bltd ", setlt>; + def BLED : BranchC<0x27, 0x13, 0x000, "bled ", setle>; + def BGTD : BranchC<0x27, 0x14, 0x000, "bgtd ", setgt>; + def BGED : BranchC<0x27, 0x15, 0x000, "bged ", setge>; +} + +let isCall = 1, hasCtrlDep = 1, isIndirectBranch = 1, + Defs = [R3,R4,R5,R6,R7,R8,R9,R10,R11,R12], + Uses = [R1,R5,R6,R7,R8,R9,R10] in { + def BRL : BranchL<0x26, 0x04, 0x000, "brl ">; + def BRAL : BranchL<0x26, 0x0C, 0x000, "bral ">; +} + +let isCall = 1, hasDelaySlot = 1, hasCtrlDep = 1, + Defs = [R3,R4,R5,R6,R7,R8,R9,R10,R11,R12], + Uses = [R1,R5,R6,R7,R8,R9,R10] in { + def BRLID : BranchLI<0x2E, 0x14, "brlid ">; + def BRALID : BranchLI<0x2E, 0x1C, "bralid ">; +} + +let isCall = 1, hasDelaySlot = 1, hasCtrlDep = 1, isIndirectBranch = 1, + Defs = [R3,R4,R5,R6,R7,R8,R9,R10,R11,R12], + Uses = [R1,R5,R6,R7,R8,R9,R10] in { + def BRLD : BranchL<0x26, 0x14, 0x000, "brld ">; + def BRALD : BranchL<0x26, 0x1C, 0x000, "brald ">; +} + +let isReturn=1, isTerminator=1, hasDelaySlot=1, + isBarrier=1, hasCtrlDep=1, imm16=0x8 in { + def RTSD : TRET<0x2D, (outs), (ins CPURegs:$target), + "rtsd $target, 8", + [(MBlazeRet CPURegs:$target)], + IIBranch>; +} + +//===----------------------------------------------------------------------===// +// MBlaze misc instructions +//===----------------------------------------------------------------------===// + +let addr = 0 in { + def NOP : TADDR<0x00, (outs), (ins), "nop ", [], IIAlu>; +} + +let usesCustomInserter = 1 in { + //class PseudoSelCC<RegisterClass RC, string asmstr>: + // MBlazePseudo<(outs RC:$D), (ins RC:$T, RC:$F, CPURegs:$CMP), asmstr, + // [(set RC:$D, (MBlazeSelectCC RC:$T, RC:$F, CPURegs:$CMP))]>; + //def Select_CC : PseudoSelCC<CPURegs, "# MBlazeSelect_CC">; + + def Select_CC : MBlazePseudo<(outs CPURegs:$dst), + (ins CPURegs:$T, CPURegs:$F, CPURegs:$CMP, i32imm:$CC), + "; SELECT_CC PSEUDO!", + []>; + + def ShiftL : MBlazePseudo<(outs CPURegs:$dst), + (ins CPURegs:$L, CPURegs:$R), + "; ShiftL PSEUDO!", + []>; + + def ShiftRA : MBlazePseudo<(outs CPURegs:$dst), + (ins CPURegs:$L, CPURegs:$R), + "; ShiftRA PSEUDO!", + []>; + + def ShiftRL : MBlazePseudo<(outs CPURegs:$dst), + (ins CPURegs:$L, CPURegs:$R), + "; ShiftRL PSEUDO!", + []>; +} + + +let rb = 0 in { + def SEXT16 : TA<0x24, 0x061, (outs CPURegs:$dst), (ins CPURegs:$src), + "sext16 $dst, $src", [], IIAlu>; + def SEXT8 : TA<0x24, 0x060, (outs CPURegs:$dst), (ins CPURegs:$src), + "sext8 $dst, $src", [], IIAlu>; + def SRL : TA<0x24, 0x041, (outs CPURegs:$dst), (ins CPURegs:$src), + "srl $dst, $src", [], IIAlu>; + def SRA : TA<0x24, 0x001, (outs CPURegs:$dst), (ins CPURegs:$src), + "sra $dst, $src", [], IIAlu>; + def SRC : TA<0x24, 0x021, (outs CPURegs:$dst), (ins CPURegs:$src), + "src $dst, $src", [], IIAlu>; +} + +def LEA_ADDI : EffectiveAddress<"addi $dst, ${addr:stackloc}">; + +//===----------------------------------------------------------------------===// +// Arbitrary patterns that map to one or more instructions +//===----------------------------------------------------------------------===// + +// Small immediates +def : Pat<(i32 0), (ADD R0, R0)>; +def : Pat<(i32 immSExt16:$imm), (ADDI R0, imm:$imm)>; +def : Pat<(i32 immZExt16:$imm), (ORI R0, imm:$imm)>; + +// Arbitrary immediates +def : Pat<(i32 imm:$imm), (ADDI R0, imm:$imm)>; + +// In register sign extension +def : Pat<(sext_inreg CPURegs:$src, i16), (SEXT16 CPURegs:$src)>; +def : Pat<(sext_inreg CPURegs:$src, i8), (SEXT8 CPURegs:$src)>; + +// Call +def : Pat<(MBlazeJmpLink (i32 tglobaladdr:$dst)), (BRLID tglobaladdr:$dst)>; +def : Pat<(MBlazeJmpLink (i32 texternalsym:$dst)),(BRLID texternalsym:$dst)>; +def : Pat<(MBlazeJmpLink CPURegs:$dst), (BRLD CPURegs:$dst)>; + +// Shift Instructions +def : Pat<(shl CPURegs:$L, CPURegs:$R), (ShiftL CPURegs:$L, CPURegs:$R)>; +def : Pat<(sra CPURegs:$L, CPURegs:$R), (ShiftRA CPURegs:$L, CPURegs:$R)>; +def : Pat<(srl CPURegs:$L, CPURegs:$R), (ShiftRL CPURegs:$L, CPURegs:$R)>; + +// SET_CC operations +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETEQ), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMP CPURegs:$L, CPURegs:$R), 1)>; +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETNE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMP CPURegs:$L, CPURegs:$R), 2)>; +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETGT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMP CPURegs:$L, CPURegs:$R), 3)>; +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETLT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMP CPURegs:$L, CPURegs:$R), 4)>; +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETGE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMP CPURegs:$L, CPURegs:$R), 5)>; +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETLE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMP CPURegs:$L, CPURegs:$R), 6)>; +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETUGT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMPU CPURegs:$L, CPURegs:$R), 3)>; +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETULT), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMPU CPURegs:$L, CPURegs:$R), 4)>; +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETUGE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMPU CPURegs:$L, CPURegs:$R), 5)>; +def : Pat<(setcc CPURegs:$L, CPURegs:$R, SETULE), + (Select_CC (ADDI R0, 1), (ADDI R0, 0), + (CMPU CPURegs:$L, CPURegs:$R), 6)>; + +// SELECT operations +def : Pat<(select CPURegs:$C, CPURegs:$T, CPURegs:$F), + (Select_CC CPURegs:$T, CPURegs:$F, CPURegs:$C, 2)>; + +// SELECT_CC +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETEQ), + (Select_CC CPURegs:$T, CPURegs:$F, (CMP CPURegs:$L, CPURegs:$R), 1)>; +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETNE), + (Select_CC CPURegs:$T, CPURegs:$F, (CMP CPURegs:$L, CPURegs:$R), 2)>; +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETGT), + (Select_CC CPURegs:$T, CPURegs:$F, (CMP CPURegs:$L, CPURegs:$R), 3)>; +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETLT), + (Select_CC CPURegs:$T, CPURegs:$F, (CMP CPURegs:$L, CPURegs:$R), 4)>; +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETGE), + (Select_CC CPURegs:$T, CPURegs:$F, (CMP CPURegs:$L, CPURegs:$R), 5)>; +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETLE), + (Select_CC CPURegs:$T, CPURegs:$F, (CMP CPURegs:$L, CPURegs:$R), 6)>; +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETUGT), + (Select_CC CPURegs:$T, CPURegs:$F, (CMPU CPURegs:$L, CPURegs:$R), 3)>; +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETULT), + (Select_CC CPURegs:$T, CPURegs:$F, (CMPU CPURegs:$L, CPURegs:$R), 4)>; +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETUGE), + (Select_CC CPURegs:$T, CPURegs:$F, (CMPU CPURegs:$L, CPURegs:$R), 5)>; +def : Pat<(selectcc CPURegs:$L, CPURegs:$R, CPURegs:$T, CPURegs:$F, SETULE), + (Select_CC CPURegs:$T, CPURegs:$F, (CMPU CPURegs:$L, CPURegs:$R), 6)>; + +// BRCOND instructions +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETEQ), bb:$T), + (BEQID (CMP CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETNE), bb:$T), + (BNEID (CMP CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETGT), bb:$T), + (BGTID (CMP CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETLT), bb:$T), + (BLTID (CMP CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETGE), bb:$T), + (BGEID (CMP CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETLE), bb:$T), + (BLEID (CMP CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETUGT), bb:$T), + (BGTID (CMPU CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETULT), bb:$T), + (BLTID (CMPU CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETUGE), bb:$T), + (BGEID (CMPU CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond (setcc CPURegs:$L, CPURegs:$R, SETULE), bb:$T), + (BLEID (CMPU CPURegs:$R, CPURegs:$L), bb:$T)>; +def : Pat<(brcond CPURegs:$C, bb:$T), + (BNEID CPURegs:$C, bb:$T)>; + +// Jump tables, global addresses, and constant pools +def : Pat<(MBWrapper tglobaladdr:$in), (ORI R0, tglobaladdr:$in)>; +def : Pat<(MBWrapper tjumptable:$in), (ORI R0, tjumptable:$in)>; +def : Pat<(MBWrapper tconstpool:$in), (ORI R0, tconstpool:$in)>; + +// Misc instructions +def : Pat<(and CPURegs:$lh, (not CPURegs:$rh)),(ANDN CPURegs:$lh, CPURegs:$rh)>; + +// Arithmetic with immediates +def : Pat<(add CPURegs:$in, imm:$imm),(ADDI CPURegs:$in, imm:$imm)>; +def : Pat<(or CPURegs:$in, imm:$imm),(ORI CPURegs:$in, imm:$imm)>; +def : Pat<(xor CPURegs:$in, imm:$imm),(XORI CPURegs:$in, imm:$imm)>; + +// extended load and stores +def : Pat<(extloadi1 iaddr:$src), (LBUI iaddr:$src)>; +def : Pat<(extloadi8 iaddr:$src), (LBUI iaddr:$src)>; +def : Pat<(extloadi16 iaddr:$src), (LHUI iaddr:$src)>; +def : Pat<(extloadi1 xaddr:$src), (LBU xaddr:$src)>; +def : Pat<(extloadi8 xaddr:$src), (LBU xaddr:$src)>; +def : Pat<(extloadi16 xaddr:$src), (LHU xaddr:$src)>; + +def : Pat<(sextloadi1 iaddr:$src), (SEXT8 (LBUI iaddr:$src))>; +def : Pat<(sextloadi8 iaddr:$src), (SEXT8 (LBUI iaddr:$src))>; +def : Pat<(sextloadi16 iaddr:$src), (SEXT16 (LHUI iaddr:$src))>; +def : Pat<(sextloadi1 xaddr:$src), (SEXT8 (LBU xaddr:$src))>; +def : Pat<(sextloadi8 xaddr:$src), (SEXT8 (LBU xaddr:$src))>; +def : Pat<(sextloadi16 xaddr:$src), (SEXT16 (LHU xaddr:$src))>; + +// peepholes +def : Pat<(store (i32 0), iaddr:$dst), (SWI R0, iaddr:$dst)>; + +//===----------------------------------------------------------------------===// +// Floating Point Support +//===----------------------------------------------------------------------===// +include "MBlazeInstrFSL.td" +include "MBlazeInstrFPU.td" diff --git a/lib/Target/MBlaze/MBlazeIntrinsicInfo.cpp b/lib/Target/MBlaze/MBlazeIntrinsicInfo.cpp new file mode 100644 index 0000000..c8faffc --- /dev/null +++ b/lib/Target/MBlaze/MBlazeIntrinsicInfo.cpp @@ -0,0 +1,109 @@ +//===- MBlazeIntrinsicInfo.cpp - Intrinsic Information -00-------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the MBlaze implementation of TargetIntrinsicInfo. +// +//===----------------------------------------------------------------------===// + +#include "MBlazeIntrinsicInfo.h" +#include "llvm/DerivedTypes.h" +#include "llvm/Function.h" +#include "llvm/Intrinsics.h" +#include "llvm/Module.h" +#include "llvm/Type.h" +#include "llvm/Support/raw_ostream.h" +#include <cstring> + +using namespace llvm; + +namespace mblazeIntrinsic { + + enum ID { + last_non_mblaze_intrinsic = Intrinsic::num_intrinsics-1, +#define GET_INTRINSIC_ENUM_VALUES +#include "MBlazeGenIntrinsics.inc" +#undef GET_INTRINSIC_ENUM_VALUES + , num_mblaze_intrinsics + }; + +#define GET_LLVM_INTRINSIC_FOR_GCC_BUILTIN +#include "MBlazeGenIntrinsics.inc" +#undef GET_LLVM_INTRINSIC_FOR_GCC_BUILTIN +} + +std::string MBlazeIntrinsicInfo::getName(unsigned IntrID, const Type **Tys, + unsigned numTys) const { + static const char *const names[] = { +#define GET_INTRINSIC_NAME_TABLE +#include "MBlazeGenIntrinsics.inc" +#undef GET_INTRINSIC_NAME_TABLE + }; + + assert(!isOverloaded(IntrID) && "MBlaze intrinsics are not overloaded"); + if (IntrID < Intrinsic::num_intrinsics) + return 0; + assert(IntrID < mblazeIntrinsic::num_mblaze_intrinsics && + "Invalid intrinsic ID"); + + std::string Result(names[IntrID - Intrinsic::num_intrinsics]); + return Result; +} + +unsigned MBlazeIntrinsicInfo:: +lookupName(const char *Name, unsigned Len) const { +#define GET_FUNCTION_RECOGNIZER +#include "MBlazeGenIntrinsics.inc" +#undef GET_FUNCTION_RECOGNIZER + return 0; +} + +unsigned MBlazeIntrinsicInfo:: +lookupGCCName(const char *Name) const { + return mblazeIntrinsic::getIntrinsicForGCCBuiltin("mblaze",Name); +} + +bool MBlazeIntrinsicInfo::isOverloaded(unsigned IntrID) const { + // Overload Table + const bool OTable[] = { +#define GET_INTRINSIC_OVERLOAD_TABLE +#include "MBlazeGenIntrinsics.inc" +#undef GET_INTRINSIC_OVERLOAD_TABLE + }; + if (IntrID == 0) + return false; + else + return OTable[IntrID - Intrinsic::num_intrinsics]; +} + +/// This defines the "getAttributes(ID id)" method. +#define GET_INTRINSIC_ATTRIBUTES +#include "MBlazeGenIntrinsics.inc" +#undef GET_INTRINSIC_ATTRIBUTES + +static const FunctionType *getType(LLVMContext &Context, unsigned id) { + const Type *ResultTy = NULL; + std::vector<const Type*> ArgTys; + bool IsVarArg = false; + +#define GET_INTRINSIC_GENERATOR +#include "MBlazeGenIntrinsics.inc" +#undef GET_INTRINSIC_GENERATOR + + return FunctionType::get(ResultTy, ArgTys, IsVarArg); +} + +Function *MBlazeIntrinsicInfo::getDeclaration(Module *M, unsigned IntrID, + const Type **Tys, + unsigned numTy) const { + assert(!isOverloaded(IntrID) && "MBlaze intrinsics are not overloaded"); + AttrListPtr AList = getAttributes((mblazeIntrinsic::ID) IntrID); + return cast<Function>(M->getOrInsertFunction(getName(IntrID), + getType(M->getContext(), IntrID), + AList)); +} diff --git a/lib/Target/MBlaze/MBlazeIntrinsicInfo.h b/lib/Target/MBlaze/MBlazeIntrinsicInfo.h new file mode 100644 index 0000000..9804c77 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeIntrinsicInfo.h @@ -0,0 +1,33 @@ +//===- MBlazeIntrinsicInfo.h - MBlaze Intrinsic Information -----*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the MBlaze implementation of TargetIntrinsicInfo. +// +//===----------------------------------------------------------------------===// +#ifndef MBLAZEINTRINSICS_H +#define MBLAZEINTRINSICS_H + +#include "llvm/Target/TargetIntrinsicInfo.h" + +namespace llvm { + + class MBlazeIntrinsicInfo : public TargetIntrinsicInfo { + public: + std::string getName(unsigned IntrID, const Type **Tys = 0, + unsigned numTys = 0) const; + unsigned lookupName(const char *Name, unsigned Len) const; + unsigned lookupGCCName(const char *Name) const; + bool isOverloaded(unsigned IID) const; + Function *getDeclaration(Module *M, unsigned ID, const Type **Tys = 0, + unsigned numTys = 0) const; + }; + +} + +#endif diff --git a/lib/Target/MBlaze/MBlazeIntrinsics.td b/lib/Target/MBlaze/MBlazeIntrinsics.td new file mode 100644 index 0000000..76eb563 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeIntrinsics.td @@ -0,0 +1,137 @@ +//===- IntrinsicsMBlaze.td - Defines MBlaze intrinsics -----*- tablegen -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file defines all of the MicroBlaze-specific intrinsics. +// +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// Definitions for all MBlaze intrinsics. +// + +// MBlaze intrinsic classes. +let TargetPrefix = "mblaze", isTarget = 1 in { + class MBFSL_Get_Intrinsic : Intrinsic<[llvm_i32_ty], + [llvm_i32_ty], + [IntrWriteMem]>; + + class MBFSL_Put_Intrinsic : Intrinsic<[llvm_void_ty], + [llvm_i32_ty, llvm_i32_ty], + [IntrWriteMem]>; + + class MBFSL_PutT_Intrinsic : Intrinsic<[llvm_void_ty], + [llvm_i32_ty], + [IntrWriteMem]>; +} + +//===----------------------------------------------------------------------===// +// MicroBlaze FSL Get Intrinsic Definitions. +// + +def int_mblaze_fsl_get : GCCBuiltin<"__builtin_mblaze_fsl_get">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_aget : GCCBuiltin<"__builtin_mblaze_fsl_aget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_cget : GCCBuiltin<"__builtin_mblaze_fsl_cget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_caget : GCCBuiltin<"__builtin_mblaze_fsl_caget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_eget : GCCBuiltin<"__builtin_mblaze_fsl_eget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_eaget : GCCBuiltin<"__builtin_mblaze_fsl_eaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_ecget : GCCBuiltin<"__builtin_mblaze_fsl_ecget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_ecaget : GCCBuiltin<"__builtin_mblaze_fsl_ecaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_nget : GCCBuiltin<"__builtin_mblaze_fsl_nget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_naget : GCCBuiltin<"__builtin_mblaze_fsl_naget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_ncget : GCCBuiltin<"__builtin_mblaze_fsl_ncget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_ncaget : GCCBuiltin<"__builtin_mblaze_fsl_ncaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_neget : GCCBuiltin<"__builtin_mblaze_fsl_neget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_neaget : GCCBuiltin<"__builtin_mblaze_fsl_neaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_necget : GCCBuiltin<"__builtin_mblaze_fsl_necget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_necaget : GCCBuiltin<"__builtin_mblaze_fsl_necaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tget : GCCBuiltin<"__builtin_mblaze_fsl_tget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_taget : GCCBuiltin<"__builtin_mblaze_fsl_taget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tcget : GCCBuiltin<"__builtin_mblaze_fsl_tcget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tcaget : GCCBuiltin<"__builtin_mblaze_fsl_tcaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_teget : GCCBuiltin<"__builtin_mblaze_fsl_teget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_teaget : GCCBuiltin<"__builtin_mblaze_fsl_teaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tecget : GCCBuiltin<"__builtin_mblaze_fsl_tecget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tecaget : GCCBuiltin<"__builtin_mblaze_fsl_tecaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tnget : GCCBuiltin<"__builtin_mblaze_fsl_tnget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tnaget : GCCBuiltin<"__builtin_mblaze_fsl_tnaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tncget : GCCBuiltin<"__builtin_mblaze_fsl_tncget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tncaget : GCCBuiltin<"__builtin_mblaze_fsl_tncaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tneget : GCCBuiltin<"__builtin_mblaze_fsl_tneget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tneaget : GCCBuiltin<"__builtin_mblaze_fsl_tneaget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tnecget : GCCBuiltin<"__builtin_mblaze_fsl_tnecget">, + MBFSL_Get_Intrinsic; +def int_mblaze_fsl_tnecaget : GCCBuiltin<"__builtin_mblaze_fsl_tnecaget">, + MBFSL_Get_Intrinsic; + +//===----------------------------------------------------------------------===// +// MicroBlaze FSL Put Intrinsic Definitions. +// + +def int_mblaze_fsl_put : GCCBuiltin<"__builtin_mblaze_fsl_put">, + MBFSL_Put_Intrinsic; +def int_mblaze_fsl_aput : GCCBuiltin<"__builtin_mblaze_fsl_aput">, + MBFSL_Put_Intrinsic; +def int_mblaze_fsl_cput : GCCBuiltin<"__builtin_mblaze_fsl_cput">, + MBFSL_Put_Intrinsic; +def int_mblaze_fsl_caput : GCCBuiltin<"__builtin_mblaze_fsl_caput">, + MBFSL_Put_Intrinsic; +def int_mblaze_fsl_nput : GCCBuiltin<"__builtin_mblaze_fsl_nput">, + MBFSL_Put_Intrinsic; +def int_mblaze_fsl_naput : GCCBuiltin<"__builtin_mblaze_fsl_naput">, + MBFSL_Put_Intrinsic; +def int_mblaze_fsl_ncput : GCCBuiltin<"__builtin_mblaze_fsl_ncput">, + MBFSL_Put_Intrinsic; +def int_mblaze_fsl_ncaput : GCCBuiltin<"__builtin_mblaze_fsl_ncaput">, + MBFSL_Put_Intrinsic; +def int_mblaze_fsl_tput : GCCBuiltin<"__builtin_mblaze_fsl_tput">, + MBFSL_PutT_Intrinsic; +def int_mblaze_fsl_taput : GCCBuiltin<"__builtin_mblaze_fsl_taput">, + MBFSL_PutT_Intrinsic; +def int_mblaze_fsl_tcput : GCCBuiltin<"__builtin_mblaze_fsl_tcput">, + MBFSL_PutT_Intrinsic; +def int_mblaze_fsl_tcaput : GCCBuiltin<"__builtin_mblaze_fsl_tcaput">, + MBFSL_PutT_Intrinsic; +def int_mblaze_fsl_tnput : GCCBuiltin<"__builtin_mblaze_fsl_tnput">, + MBFSL_PutT_Intrinsic; +def int_mblaze_fsl_tnaput : GCCBuiltin<"__builtin_mblaze_fsl_tnaput">, + MBFSL_PutT_Intrinsic; +def int_mblaze_fsl_tncput : GCCBuiltin<"__builtin_mblaze_fsl_tncput">, + MBFSL_PutT_Intrinsic; +def int_mblaze_fsl_tncaput : GCCBuiltin<"__builtin_mblaze_fsl_tncaput">, + MBFSL_PutT_Intrinsic; diff --git a/lib/Target/MBlaze/MBlazeMCAsmInfo.cpp b/lib/Target/MBlaze/MBlazeMCAsmInfo.cpp new file mode 100644 index 0000000..7ae465d --- /dev/null +++ b/lib/Target/MBlaze/MBlazeMCAsmInfo.cpp @@ -0,0 +1,27 @@ +//===-- MBlazeMCAsmInfo.cpp - MBlaze asm properties -----------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the declarations of the MBlazeMCAsmInfo properties. +// +//===----------------------------------------------------------------------===// + +#include "MBlazeMCAsmInfo.h" +using namespace llvm; + +MBlazeMCAsmInfo::MBlazeMCAsmInfo(const Target &T, const StringRef &TT) { + AlignmentIsInBytes = false; + Data16bitsDirective = "\t.half\t"; + Data32bitsDirective = "\t.word\t"; + Data64bitsDirective = 0; + PrivateGlobalPrefix = "$"; + CommentString = "#"; + ZeroDirective = "\t.space\t"; + GPRel32Directive = "\t.gpword\t"; + HasSetDirective = false; +} diff --git a/lib/Target/MBlaze/MBlazeMCAsmInfo.h b/lib/Target/MBlaze/MBlazeMCAsmInfo.h new file mode 100644 index 0000000..bccb418 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeMCAsmInfo.h @@ -0,0 +1,30 @@ +//=====-- MBlazeMCAsmInfo.h - MBlaze asm properties -----------*- C++ -*--====// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the declaration of the MBlazeMCAsmInfo class. +// +//===----------------------------------------------------------------------===// + +#ifndef MBLAZETARGETASMINFO_H +#define MBLAZETARGETASMINFO_H + +#include "llvm/MC/MCAsmInfo.h" + +namespace llvm { + class Target; + class StringRef; + + class MBlazeMCAsmInfo : public MCAsmInfo { + public: + explicit MBlazeMCAsmInfo(const Target &T, const StringRef &TT); + }; + +} // namespace llvm + +#endif diff --git a/lib/Target/MBlaze/MBlazeMachineFunction.h b/lib/Target/MBlaze/MBlazeMachineFunction.h new file mode 100644 index 0000000..08d4dca --- /dev/null +++ b/lib/Target/MBlaze/MBlazeMachineFunction.h @@ -0,0 +1,136 @@ +//===-- MBlazeMachineFunctionInfo.h - Private data ----------------*- C++ -*-=// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file declares the MBlaze specific subclass of MachineFunctionInfo. +// +//===----------------------------------------------------------------------===// + +#ifndef MBLAZE_MACHINE_FUNCTION_INFO_H +#define MBLAZE_MACHINE_FUNCTION_INFO_H + +#include "llvm/ADT/SmallVector.h" +#include "llvm/ADT/VectorExtras.h" +#include "llvm/CodeGen/MachineFunction.h" +#include "llvm/CodeGen/MachineFrameInfo.h" + +namespace llvm { + +/// MBlazeFunctionInfo - This class is derived from MachineFunction private +/// MBlaze target-specific information for each MachineFunction. +class MBlazeFunctionInfo : public MachineFunctionInfo { + +private: + /// Holds for each function where on the stack the Frame Pointer must be + /// saved. This is used on Prologue and Epilogue to emit FP save/restore + int FPStackOffset; + + /// Holds for each function where on the stack the Return Address must be + /// saved. This is used on Prologue and Epilogue to emit RA save/restore + int RAStackOffset; + + /// At each function entry a special bitmask directive must be emitted + /// to help in debugging CPU callee saved registers. It needs a negative + /// offset from the final stack size and its higher register location on + /// the stack. + int CPUTopSavedRegOff; + + /// MBlazeFIHolder - Holds a FrameIndex and it's Stack Pointer Offset + struct MBlazeFIHolder { + + int FI; + int SPOffset; + + MBlazeFIHolder(int FrameIndex, int StackPointerOffset) + : FI(FrameIndex), SPOffset(StackPointerOffset) {} + }; + + /// When PIC is used the GP must be saved on the stack on the function + /// prologue and must be reloaded from this stack location after every + /// call. A reference to its stack location and frame index must be kept + /// to be used on emitPrologue and processFunctionBeforeFrameFinalized. + MBlazeFIHolder GPHolder; + + /// On LowerFormalArguments the stack size is unknown, so the Stack + /// Pointer Offset calculation of "not in register arguments" must be + /// postponed to emitPrologue. + SmallVector<MBlazeFIHolder, 16> FnLoadArgs; + bool HasLoadArgs; + + // When VarArgs, we must write registers back to caller stack, preserving + // on register arguments. Since the stack size is unknown on + // LowerFormalArguments, the Stack Pointer Offset calculation must be + // postponed to emitPrologue. + SmallVector<MBlazeFIHolder, 4> FnStoreVarArgs; + bool HasStoreVarArgs; + + /// SRetReturnReg - Some subtargets require that sret lowering includes + /// returning the value of the returned struct in a register. This field + /// holds the virtual register into which the sret argument is passed. + unsigned SRetReturnReg; + + /// GlobalBaseReg - keeps track of the virtual register initialized for + /// use as the global base register. This is used for PIC in some PIC + /// relocation models. + unsigned GlobalBaseReg; + +public: + MBlazeFunctionInfo(MachineFunction& MF) + : FPStackOffset(0), RAStackOffset(0), CPUTopSavedRegOff(0), + GPHolder(-1,-1), HasLoadArgs(false), HasStoreVarArgs(false), + SRetReturnReg(0), GlobalBaseReg(0) + {} + + int getFPStackOffset() const { return FPStackOffset; } + void setFPStackOffset(int Off) { FPStackOffset = Off; } + + int getRAStackOffset() const { return RAStackOffset; } + void setRAStackOffset(int Off) { RAStackOffset = Off; } + + int getCPUTopSavedRegOff() const { return CPUTopSavedRegOff; } + void setCPUTopSavedRegOff(int Off) { CPUTopSavedRegOff = Off; } + + int getGPStackOffset() const { return GPHolder.SPOffset; } + int getGPFI() const { return GPHolder.FI; } + void setGPStackOffset(int Off) { GPHolder.SPOffset = Off; } + void setGPFI(int FI) { GPHolder.FI = FI; } + bool needGPSaveRestore() const { return GPHolder.SPOffset != -1; } + + bool hasLoadArgs() const { return HasLoadArgs; } + bool hasStoreVarArgs() const { return HasStoreVarArgs; } + + void recordLoadArgsFI(int FI, int SPOffset) { + if (!HasLoadArgs) HasLoadArgs=true; + FnLoadArgs.push_back(MBlazeFIHolder(FI, SPOffset)); + } + void recordStoreVarArgsFI(int FI, int SPOffset) { + if (!HasStoreVarArgs) HasStoreVarArgs=true; + FnStoreVarArgs.push_back(MBlazeFIHolder(FI, SPOffset)); + } + + void adjustLoadArgsFI(MachineFrameInfo *MFI) const { + if (!hasLoadArgs()) return; + for (unsigned i = 0, e = FnLoadArgs.size(); i != e; ++i) + MFI->setObjectOffset( FnLoadArgs[i].FI, FnLoadArgs[i].SPOffset ); + } + void adjustStoreVarArgsFI(MachineFrameInfo *MFI) const { + if (!hasStoreVarArgs()) return; + for (unsigned i = 0, e = FnStoreVarArgs.size(); i != e; ++i) + MFI->setObjectOffset( FnStoreVarArgs[i].FI, FnStoreVarArgs[i].SPOffset ); + } + + unsigned getSRetReturnReg() const { return SRetReturnReg; } + void setSRetReturnReg(unsigned Reg) { SRetReturnReg = Reg; } + + unsigned getGlobalBaseReg() const { return GlobalBaseReg; } + void setGlobalBaseReg(unsigned Reg) { GlobalBaseReg = Reg; } +}; + +} // end of namespace llvm + +#endif // MBLAZE_MACHINE_FUNCTION_INFO_H diff --git a/lib/Target/MBlaze/MBlazeRegisterInfo.cpp b/lib/Target/MBlaze/MBlazeRegisterInfo.cpp new file mode 100644 index 0000000..9067f8b --- /dev/null +++ b/lib/Target/MBlaze/MBlazeRegisterInfo.cpp @@ -0,0 +1,378 @@ +//===- MBlazeRegisterInfo.cpp - MBlaze Register Information -== -*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the MBlaze implementation of the TargetRegisterInfo +// class. +// +//===----------------------------------------------------------------------===// + +#define DEBUG_TYPE "mblaze-reg-info" + +#include "MBlaze.h" +#include "MBlazeSubtarget.h" +#include "MBlazeRegisterInfo.h" +#include "MBlazeMachineFunction.h" +#include "llvm/Constants.h" +#include "llvm/Type.h" +#include "llvm/Function.h" +#include "llvm/CodeGen/ValueTypes.h" +#include "llvm/CodeGen/MachineInstrBuilder.h" +#include "llvm/CodeGen/MachineFunction.h" +#include "llvm/CodeGen/MachineFrameInfo.h" +#include "llvm/CodeGen/MachineLocation.h" +#include "llvm/Target/TargetFrameInfo.h" +#include "llvm/Target/TargetMachine.h" +#include "llvm/Target/TargetOptions.h" +#include "llvm/Target/TargetInstrInfo.h" +#include "llvm/Support/CommandLine.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/ErrorHandling.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/ADT/BitVector.h" +#include "llvm/ADT/STLExtras.h" + +using namespace llvm; + +MBlazeRegisterInfo:: +MBlazeRegisterInfo(const MBlazeSubtarget &ST, const TargetInstrInfo &tii) + : MBlazeGenRegisterInfo(MBlaze::ADJCALLSTACKDOWN, MBlaze::ADJCALLSTACKUP), + Subtarget(ST), TII(tii) {} + +/// getRegisterNumbering - Given the enum value for some register, e.g. +/// MBlaze::R0, return the number that it corresponds to (e.g. 0). +unsigned MBlazeRegisterInfo::getRegisterNumbering(unsigned RegEnum) { + switch (RegEnum) { + case MBlaze::R0 : case MBlaze::F0 : return 0; + case MBlaze::R1 : case MBlaze::F1 : return 1; + case MBlaze::R2 : case MBlaze::F2 : return 2; + case MBlaze::R3 : case MBlaze::F3 : return 3; + case MBlaze::R4 : case MBlaze::F4 : return 4; + case MBlaze::R5 : case MBlaze::F5 : return 5; + case MBlaze::R6 : case MBlaze::F6 : return 6; + case MBlaze::R7 : case MBlaze::F7 : return 7; + case MBlaze::R8 : case MBlaze::F8 : return 8; + case MBlaze::R9 : case MBlaze::F9 : return 9; + case MBlaze::R10 : case MBlaze::F10 : return 10; + case MBlaze::R11 : case MBlaze::F11 : return 11; + case MBlaze::R12 : case MBlaze::F12 : return 12; + case MBlaze::R13 : case MBlaze::F13 : return 13; + case MBlaze::R14 : case MBlaze::F14 : return 14; + case MBlaze::R15 : case MBlaze::F15 : return 15; + case MBlaze::R16 : case MBlaze::F16 : return 16; + case MBlaze::R17 : case MBlaze::F17 : return 17; + case MBlaze::R18 : case MBlaze::F18 : return 18; + case MBlaze::R19 : case MBlaze::F19 : return 19; + case MBlaze::R20 : case MBlaze::F20 : return 20; + case MBlaze::R21 : case MBlaze::F21 : return 21; + case MBlaze::R22 : case MBlaze::F22 : return 22; + case MBlaze::R23 : case MBlaze::F23 : return 23; + case MBlaze::R24 : case MBlaze::F24 : return 24; + case MBlaze::R25 : case MBlaze::F25 : return 25; + case MBlaze::R26 : case MBlaze::F26 : return 26; + case MBlaze::R27 : case MBlaze::F27 : return 27; + case MBlaze::R28 : case MBlaze::F28 : return 28; + case MBlaze::R29 : case MBlaze::F29 : return 29; + case MBlaze::R30 : case MBlaze::F30 : return 30; + case MBlaze::R31 : case MBlaze::F31 : return 31; + default: llvm_unreachable("Unknown register number!"); + } + return 0; // Not reached +} + +unsigned MBlazeRegisterInfo::getPICCallReg() { + return MBlaze::R20; +} + +//===----------------------------------------------------------------------===// +// Callee Saved Registers methods +//===----------------------------------------------------------------------===// + +/// MBlaze Callee Saved Registers +const unsigned* MBlazeRegisterInfo:: +getCalleeSavedRegs(const MachineFunction *MF) const { + // MBlaze callee-save register range is R20 - R31 + static const unsigned CalleeSavedRegs[] = { + MBlaze::R20, MBlaze::R21, MBlaze::R22, MBlaze::R23, + MBlaze::R24, MBlaze::R25, MBlaze::R26, MBlaze::R27, + MBlaze::R28, MBlaze::R29, MBlaze::R30, MBlaze::R31, + 0 + }; + + return CalleeSavedRegs; +} + +/// MBlaze Callee Saved Register Classes +const TargetRegisterClass* const* MBlazeRegisterInfo:: +getCalleeSavedRegClasses(const MachineFunction *MF) const { + static const TargetRegisterClass * const CalleeSavedRC[] = { + &MBlaze::CPURegsRegClass, &MBlaze::CPURegsRegClass, + &MBlaze::CPURegsRegClass, &MBlaze::CPURegsRegClass, + &MBlaze::CPURegsRegClass, &MBlaze::CPURegsRegClass, + &MBlaze::CPURegsRegClass, &MBlaze::CPURegsRegClass, + &MBlaze::CPURegsRegClass, &MBlaze::CPURegsRegClass, + &MBlaze::CPURegsRegClass, &MBlaze::CPURegsRegClass, + 0 + }; + + return CalleeSavedRC; +} + +BitVector MBlazeRegisterInfo:: +getReservedRegs(const MachineFunction &MF) const { + BitVector Reserved(getNumRegs()); + Reserved.set(MBlaze::R0); + Reserved.set(MBlaze::R1); + Reserved.set(MBlaze::R2); + Reserved.set(MBlaze::R13); + Reserved.set(MBlaze::R14); + Reserved.set(MBlaze::R15); + Reserved.set(MBlaze::R16); + Reserved.set(MBlaze::R17); + Reserved.set(MBlaze::R18); + Reserved.set(MBlaze::R19); + return Reserved; +} + +//===----------------------------------------------------------------------===// +// +// Stack Frame Processing methods +// +----------------------------+ +// +// The stack is allocated decrementing the stack pointer on +// the first instruction of a function prologue. Once decremented, +// all stack references are are done through a positive offset +// from the stack/frame pointer, so the stack is considered +// to grow up. +// +//===----------------------------------------------------------------------===// + +void MBlazeRegisterInfo::adjustMBlazeStackFrame(MachineFunction &MF) const { + MachineFrameInfo *MFI = MF.getFrameInfo(); + MBlazeFunctionInfo *MBlazeFI = MF.getInfo<MBlazeFunctionInfo>(); + + // See the description at MicroBlazeMachineFunction.h + int TopCPUSavedRegOff = -1; + + // Adjust CPU Callee Saved Registers Area. Registers RA and FP must + // be saved in this CPU Area there is the need. This whole Area must + // be aligned to the default Stack Alignment requirements. + unsigned StackOffset = MFI->getStackSize(); + unsigned RegSize = 4; + + // Replace the dummy '0' SPOffset by the negative offsets, as explained on + // LowerFORMAL_ARGUMENTS. Leaving '0' for while is necessary to avoid + // the approach done by calculateFrameObjectOffsets to the stack frame. + MBlazeFI->adjustLoadArgsFI(MFI); + MBlazeFI->adjustStoreVarArgsFI(MFI); + + if (hasFP(MF)) { + MFI->setObjectOffset(MFI->CreateStackObject(RegSize, RegSize, true), + StackOffset); + MBlazeFI->setFPStackOffset(StackOffset); + TopCPUSavedRegOff = StackOffset; + StackOffset += RegSize; + } + + if (MFI->hasCalls()) { + MFI->setObjectOffset(MFI->CreateStackObject(RegSize, RegSize, true), + StackOffset); + MBlazeFI->setRAStackOffset(StackOffset); + TopCPUSavedRegOff = StackOffset; + StackOffset += RegSize; + } + + // Update frame info + MFI->setStackSize(StackOffset); + + // Recalculate the final tops offset. The final values must be '0' + // if there isn't a callee saved register for CPU or FPU, otherwise + // a negative offset is needed. + if (TopCPUSavedRegOff >= 0) + MBlazeFI->setCPUTopSavedRegOff(TopCPUSavedRegOff-StackOffset); +} + +// hasFP - Return true if the specified function should have a dedicated frame +// pointer register. This is true if the function has variable sized allocas or +// if frame pointer elimination is disabled. +bool MBlazeRegisterInfo::hasFP(const MachineFunction &MF) const { + const MachineFrameInfo *MFI = MF.getFrameInfo(); + return NoFramePointerElim || MFI->hasVarSizedObjects(); +} + +// This function eliminate ADJCALLSTACKDOWN, +// ADJCALLSTACKUP pseudo instructions +void MBlazeRegisterInfo:: +eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB, + MachineBasicBlock::iterator I) const { + // Simply discard ADJCALLSTACKDOWN, ADJCALLSTACKUP instructions. + MBB.erase(I); +} + +// FrameIndex represent objects inside a abstract stack. +// We must replace FrameIndex with an stack/frame pointer +// direct reference. +unsigned MBlazeRegisterInfo:: +eliminateFrameIndex(MachineBasicBlock::iterator II, int SPAdj, + int *Value, RegScavenger *RS) const { + MachineInstr &MI = *II; + MachineFunction &MF = *MI.getParent()->getParent(); + + unsigned i = 0; + while (!MI.getOperand(i).isFI()) { + ++i; + assert(i < MI.getNumOperands() && + "Instr doesn't have FrameIndex operand!"); + } + + unsigned oi = i == 2 ? 1 : 2; + + DEBUG(errs() << "\nFunction : " << MF.getFunction()->getName() << "\n"; + errs() << "<--------->\n" << MI); + + int FrameIndex = MI.getOperand(i).getIndex(); + int stackSize = MF.getFrameInfo()->getStackSize(); + int spOffset = MF.getFrameInfo()->getObjectOffset(FrameIndex); + + DEBUG(errs() << "FrameIndex : " << FrameIndex << "\n" + << "spOffset : " << spOffset << "\n" + << "stackSize : " << stackSize << "\n"); + + // as explained on LowerFormalArguments, detect negative offsets + // and adjust SPOffsets considering the final stack size. + int Offset = ((spOffset < 0) ? (stackSize + (-(spOffset+4))) : (spOffset)); + Offset += MI.getOperand(oi).getImm(); + + DEBUG(errs() << "Offset : " << Offset << "\n" << "<--------->\n"); + + MI.getOperand(oi).ChangeToImmediate(Offset); + MI.getOperand(i).ChangeToRegister(getFrameRegister(MF), false); + return 0; +} + +void MBlazeRegisterInfo:: +emitPrologue(MachineFunction &MF) const { + MachineBasicBlock &MBB = MF.front(); + MachineFrameInfo *MFI = MF.getFrameInfo(); + MBlazeFunctionInfo *MBlazeFI = MF.getInfo<MBlazeFunctionInfo>(); + MachineBasicBlock::iterator MBBI = MBB.begin(); + DebugLoc dl = (MBBI != MBB.end() ? + MBBI->getDebugLoc() : DebugLoc::getUnknownLoc()); + + // Get the right frame order for MBlaze. + adjustMBlazeStackFrame(MF); + + // Get the number of bytes to allocate from the FrameInfo. + unsigned StackSize = MFI->getStackSize(); + + // No need to allocate space on the stack. + if (StackSize == 0 && !MFI->hasCalls()) return; + + int FPOffset = MBlazeFI->getFPStackOffset(); + int RAOffset = MBlazeFI->getRAStackOffset(); + + // Adjust stack : addi R1, R1, -imm + BuildMI(MBB, MBBI, dl, TII.get(MBlaze::ADDI), MBlaze::R1) + .addReg(MBlaze::R1).addImm(-StackSize); + + // Save the return address only if the function isnt a leaf one. + // swi R15, R1, stack_loc + if (MFI->hasCalls()) { + BuildMI(MBB, MBBI, dl, TII.get(MBlaze::SWI)) + .addReg(MBlaze::R15).addImm(RAOffset).addReg(MBlaze::R1); + } + + // if framepointer enabled, save it and set it + // to point to the stack pointer + if (hasFP(MF)) { + // swi R19, R1, stack_loc + BuildMI(MBB, MBBI, dl, TII.get(MBlaze::SWI)) + .addReg(MBlaze::R19).addImm(FPOffset).addReg(MBlaze::R1); + + // add R19, R1, R0 + BuildMI(MBB, MBBI, dl, TII.get(MBlaze::ADD), MBlaze::R19) + .addReg(MBlaze::R1).addReg(MBlaze::R0); + } +} + +void MBlazeRegisterInfo:: +emitEpilogue(MachineFunction &MF, MachineBasicBlock &MBB) const { + MachineBasicBlock::iterator MBBI = prior(MBB.end()); + MachineFrameInfo *MFI = MF.getFrameInfo(); + MBlazeFunctionInfo *MBlazeFI = MF.getInfo<MBlazeFunctionInfo>(); + DebugLoc dl = MBBI->getDebugLoc(); + + // Get the number of bytes from FrameInfo + int NumBytes = (int) MFI->getStackSize(); + + // Get the FI's where RA and FP are saved. + int FPOffset = MBlazeFI->getFPStackOffset(); + int RAOffset = MBlazeFI->getRAStackOffset(); + + // if framepointer enabled, restore it and restore the + // stack pointer + if (hasFP(MF)) { + // add R1, R19, R0 + BuildMI(MBB, MBBI, dl, TII.get(MBlaze::ADD), MBlaze::R1) + .addReg(MBlaze::R19).addReg(MBlaze::R0); + + // lwi R19, R1, stack_loc + BuildMI(MBB, MBBI, dl, TII.get(MBlaze::LWI), MBlaze::R19) + .addImm(FPOffset).addReg(MBlaze::R1); + } + + // Restore the return address only if the function isnt a leaf one. + // lwi R15, R1, stack_loc + if (MFI->hasCalls()) { + BuildMI(MBB, MBBI, dl, TII.get(MBlaze::LWI), MBlaze::R15) + .addImm(RAOffset).addReg(MBlaze::R1); + } + + // adjust stack. + // addi R1, R1, imm + if (NumBytes) { + BuildMI(MBB, MBBI, dl, TII.get(MBlaze::ADDI), MBlaze::R1) + .addReg(MBlaze::R1).addImm(NumBytes); + } +} + + +void MBlazeRegisterInfo:: +processFunctionBeforeFrameFinalized(MachineFunction &MF) const { + // Set the stack offset where GP must be saved/loaded from. + MachineFrameInfo *MFI = MF.getFrameInfo(); + MBlazeFunctionInfo *MBlazeFI = MF.getInfo<MBlazeFunctionInfo>(); + if (MBlazeFI->needGPSaveRestore()) + MFI->setObjectOffset(MBlazeFI->getGPFI(), MBlazeFI->getGPStackOffset()); +} + +unsigned MBlazeRegisterInfo::getRARegister() const { + return MBlaze::R15; +} + +unsigned MBlazeRegisterInfo::getFrameRegister(const MachineFunction &MF) const { + return hasFP(MF) ? MBlaze::R19 : MBlaze::R1; +} + +unsigned MBlazeRegisterInfo::getEHExceptionRegister() const { + llvm_unreachable("What is the exception register"); + return 0; +} + +unsigned MBlazeRegisterInfo::getEHHandlerRegister() const { + llvm_unreachable("What is the exception handler register"); + return 0; +} + +int MBlazeRegisterInfo::getDwarfRegNum(unsigned RegNum, bool isEH) const { + llvm_unreachable("What is the dwarf register number"); + return -1; +} + +#include "MBlazeGenRegisterInfo.inc" + diff --git a/lib/Target/MBlaze/MBlazeRegisterInfo.h b/lib/Target/MBlaze/MBlazeRegisterInfo.h new file mode 100644 index 0000000..4847f1e --- /dev/null +++ b/lib/Target/MBlaze/MBlazeRegisterInfo.h @@ -0,0 +1,90 @@ +//===- MBlazeRegisterInfo.h - MBlaze Register Information Impl --*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains the MBlaze implementation of the TargetRegisterInfo +// class. +// +//===----------------------------------------------------------------------===// + +#ifndef MBLAZEREGISTERINFO_H +#define MBLAZEREGISTERINFO_H + +#include "MBlaze.h" +#include "llvm/Target/TargetRegisterInfo.h" +#include "MBlazeGenRegisterInfo.h.inc" + +namespace llvm { +class MBlazeSubtarget; +class TargetInstrInfo; +class Type; + +namespace MBlaze { + /// SubregIndex - The index of various sized subregister classes. Note that + /// these indices must be kept in sync with the class indices in the + /// MBlazeRegisterInfo.td file. + enum SubregIndex { + SUBREG_FPEVEN = 1, SUBREG_FPODD = 2 + }; +} + +struct MBlazeRegisterInfo : public MBlazeGenRegisterInfo { + const MBlazeSubtarget &Subtarget; + const TargetInstrInfo &TII; + + MBlazeRegisterInfo(const MBlazeSubtarget &Subtarget, + const TargetInstrInfo &tii); + + /// getRegisterNumbering - Given the enum value for some register, e.g. + /// MBlaze::RA, return the number that it corresponds to (e.g. 31). + static unsigned getRegisterNumbering(unsigned RegEnum); + + /// Get PIC indirect call register + static unsigned getPICCallReg(); + + /// Adjust the MBlaze stack frame. + void adjustMBlazeStackFrame(MachineFunction &MF) const; + + /// Code Generation virtual methods... + const unsigned *getCalleeSavedRegs(const MachineFunction* MF = 0) const; + + const TargetRegisterClass* const* + getCalleeSavedRegClasses(const MachineFunction* MF = 0) const; + + BitVector getReservedRegs(const MachineFunction &MF) const; + + bool hasFP(const MachineFunction &MF) const; + + void eliminateCallFramePseudoInstr(MachineFunction &MF, + MachineBasicBlock &MBB, + MachineBasicBlock::iterator I) const; + + /// Stack Frame Processing Methods + unsigned eliminateFrameIndex(MachineBasicBlock::iterator II, + int SPAdj, int *Value = NULL, + RegScavenger *RS = NULL) const; + + void processFunctionBeforeFrameFinalized(MachineFunction &MF) const; + + void emitPrologue(MachineFunction &MF) const; + void emitEpilogue(MachineFunction &MF, MachineBasicBlock &MBB) const; + + /// Debug information queries. + unsigned getRARegister() const; + unsigned getFrameRegister(const MachineFunction &MF) const; + + /// Exception handling queries. + unsigned getEHExceptionRegister() const; + unsigned getEHHandlerRegister() const; + + int getDwarfRegNum(unsigned RegNum, bool isEH) const; +}; + +} // end namespace llvm + +#endif diff --git a/lib/Target/MBlaze/MBlazeRegisterInfo.td b/lib/Target/MBlaze/MBlazeRegisterInfo.td new file mode 100644 index 0000000..96a5c98 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeRegisterInfo.td @@ -0,0 +1,186 @@ +//===- MBlazeRegisterInfo.td - MBlaze Register defs -------------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// Declarations that describe the MicroBlaze register file +//===----------------------------------------------------------------------===// + +// We have banks of 32 registers each. +class MBlazeReg<string n> : Register<n> { + field bits<5> Num; + let Namespace = "MBlaze"; +} + +class MBlazeRegWithSubRegs<string n, list<Register> subregs> + : RegisterWithSubRegs<n, subregs> { + field bits<5> Num; + let Namespace = "MBlaze"; +} + +// MBlaze CPU Registers +class MBlazeGPRReg<bits<5> num, string n> : MBlazeReg<n> { + let Num = num; +} + +// MBlaze 32-bit (aliased) FPU Registers +class FPR<bits<5> num, string n, list<Register> subregs> + : MBlazeRegWithSubRegs<n, subregs> { + let Num = num; +} + +//===----------------------------------------------------------------------===// +// Registers +//===----------------------------------------------------------------------===// + +let Namespace = "MBlaze" in { + + // General Purpose Registers + def R0 : MBlazeGPRReg< 0, "r0">, DwarfRegNum<[0]>; + def R1 : MBlazeGPRReg< 1, "r1">, DwarfRegNum<[1]>; + def R2 : MBlazeGPRReg< 2, "r2">, DwarfRegNum<[2]>; + def R3 : MBlazeGPRReg< 3, "r3">, DwarfRegNum<[3]>; + def R4 : MBlazeGPRReg< 4, "r4">, DwarfRegNum<[5]>; + def R5 : MBlazeGPRReg< 5, "r5">, DwarfRegNum<[5]>; + def R6 : MBlazeGPRReg< 6, "r6">, DwarfRegNum<[6]>; + def R7 : MBlazeGPRReg< 7, "r7">, DwarfRegNum<[7]>; + def R8 : MBlazeGPRReg< 8, "r8">, DwarfRegNum<[8]>; + def R9 : MBlazeGPRReg< 9, "r9">, DwarfRegNum<[9]>; + def R10 : MBlazeGPRReg< 10, "r10">, DwarfRegNum<[10]>; + def R11 : MBlazeGPRReg< 11, "r11">, DwarfRegNum<[11]>; + def R12 : MBlazeGPRReg< 12, "r12">, DwarfRegNum<[12]>; + def R13 : MBlazeGPRReg< 13, "r13">, DwarfRegNum<[13]>; + def R14 : MBlazeGPRReg< 14, "r14">, DwarfRegNum<[14]>; + def R15 : MBlazeGPRReg< 15, "r15">, DwarfRegNum<[15]>; + def R16 : MBlazeGPRReg< 16, "r16">, DwarfRegNum<[16]>; + def R17 : MBlazeGPRReg< 17, "r17">, DwarfRegNum<[17]>; + def R18 : MBlazeGPRReg< 18, "r18">, DwarfRegNum<[18]>; + def R19 : MBlazeGPRReg< 19, "r19">, DwarfRegNum<[19]>; + def R20 : MBlazeGPRReg< 20, "r20">, DwarfRegNum<[20]>; + def R21 : MBlazeGPRReg< 21, "r21">, DwarfRegNum<[21]>; + def R22 : MBlazeGPRReg< 22, "r22">, DwarfRegNum<[22]>; + def R23 : MBlazeGPRReg< 23, "r23">, DwarfRegNum<[23]>; + def R24 : MBlazeGPRReg< 24, "r24">, DwarfRegNum<[24]>; + def R25 : MBlazeGPRReg< 25, "r25">, DwarfRegNum<[25]>; + def R26 : MBlazeGPRReg< 26, "r26">, DwarfRegNum<[26]>; + def R27 : MBlazeGPRReg< 27, "r27">, DwarfRegNum<[27]>; + def R28 : MBlazeGPRReg< 28, "r28">, DwarfRegNum<[28]>; + def R29 : MBlazeGPRReg< 29, "r29">, DwarfRegNum<[29]>; + def R30 : MBlazeGPRReg< 30, "r30">, DwarfRegNum<[30]>; + def R31 : MBlazeGPRReg< 31, "r31">, DwarfRegNum<[31]>; + + /// MBlaze Single point precision FPU Registers + def F0 : FPR< 0, "r0", [R0]>, DwarfRegNum<[32]>; + def F1 : FPR< 1, "r1", [R1]>, DwarfRegNum<[33]>; + def F2 : FPR< 2, "r2", [R2]>, DwarfRegNum<[34]>; + def F3 : FPR< 3, "r3", [R3]>, DwarfRegNum<[35]>; + def F4 : FPR< 4, "r4", [R4]>, DwarfRegNum<[36]>; + def F5 : FPR< 5, "r5", [R5]>, DwarfRegNum<[37]>; + def F6 : FPR< 6, "r6", [R6]>, DwarfRegNum<[38]>; + def F7 : FPR< 7, "r7", [R7]>, DwarfRegNum<[39]>; + def F8 : FPR< 8, "r8", [R8]>, DwarfRegNum<[40]>; + def F9 : FPR< 9, "r9", [R9]>, DwarfRegNum<[41]>; + def F10 : FPR<10, "r10", [R10]>, DwarfRegNum<[42]>; + def F11 : FPR<11, "r11", [R11]>, DwarfRegNum<[43]>; + def F12 : FPR<12, "r12", [R12]>, DwarfRegNum<[44]>; + def F13 : FPR<13, "r13", [R13]>, DwarfRegNum<[45]>; + def F14 : FPR<14, "r14", [R14]>, DwarfRegNum<[46]>; + def F15 : FPR<15, "r15", [R15]>, DwarfRegNum<[47]>; + def F16 : FPR<16, "r16", [R16]>, DwarfRegNum<[48]>; + def F17 : FPR<17, "r17", [R17]>, DwarfRegNum<[49]>; + def F18 : FPR<18, "r18", [R18]>, DwarfRegNum<[50]>; + def F19 : FPR<19, "r19", [R19]>, DwarfRegNum<[51]>; + def F20 : FPR<20, "r20", [R20]>, DwarfRegNum<[52]>; + def F21 : FPR<21, "r21", [R21]>, DwarfRegNum<[53]>; + def F22 : FPR<22, "r22", [R22]>, DwarfRegNum<[54]>; + def F23 : FPR<23, "r23", [R23]>, DwarfRegNum<[55]>; + def F24 : FPR<24, "r24", [R24]>, DwarfRegNum<[56]>; + def F25 : FPR<25, "r25", [R25]>, DwarfRegNum<[57]>; + def F26 : FPR<26, "r26", [R26]>, DwarfRegNum<[58]>; + def F27 : FPR<27, "r27", [R27]>, DwarfRegNum<[59]>; + def F28 : FPR<28, "r28", [R28]>, DwarfRegNum<[60]>; + def F29 : FPR<29, "r29", [R29]>, DwarfRegNum<[61]>; + def F30 : FPR<30, "r30", [R30]>, DwarfRegNum<[62]>; + def F31 : FPR<31, "r31", [R31]>, DwarfRegNum<[63]>; +} + +//===----------------------------------------------------------------------===// +// Register Classes +//===----------------------------------------------------------------------===// + +def CPURegs : RegisterClass<"MBlaze", [i32], 32, + [ + // Return Values and Arguments + R3, R4, R5, R6, R7, R8, R9, R10, + + // Not preserved across procedure calls + R11, R12, + + // Callee save + R20, R21, R22, R23, R24, R25, R26, R27, R28, R29, R30, R31, + + // Reserved + R0, // Always zero + R1, // The stack pointer + R2, // Read-only small data area anchor + R13, // Read-write small data area anchor + R14, // Return address for interrupts + R15, // Return address for sub-routines + R16, // Return address for trap + R17, // Return address for exceptions + R18, // Reserved for assembler + R19 // The frame-pointer + ]> +{ + let MethodProtos = [{ + iterator allocation_order_end(const MachineFunction &MF) const; + }]; + let MethodBodies = [{ + CPURegsClass::iterator + CPURegsClass::allocation_order_end(const MachineFunction &MF) const { + // The last 10 registers on the list above are reserved + return end()-10; + } + }]; +} + +def FGR32 : RegisterClass<"MBlaze", [f32], 32, + [ + // Return Values and Arguments + F3, F4, F5, F6, F7, F8, F9, F10, + + // Not preserved across procedure calls + F11, F12, + + // Callee save + F20, F21, F22, F23, F24, F25, F26, F27, F28, F29, F30, F31, + + // Reserved + F0, // Always zero + F1, // The stack pointer + F2, // Read-only small data area anchor + F13, // Read-write small data area anchor + F14, // Return address for interrupts + F15, // Return address for sub-routines + F16, // Return address for trap + F17, // Return address for exceptions + F18, // Reserved for assembler + F19 // The frame pointer + ]> +{ + let MethodProtos = [{ + iterator allocation_order_end(const MachineFunction &MF) const; + }]; + let MethodBodies = [{ + FGR32Class::iterator + FGR32Class::allocation_order_end(const MachineFunction &MF) const { + // The last 10 registers on the list above are reserved + return end()-10; + } + }]; +} diff --git a/lib/Target/MBlaze/MBlazeSchedule.td b/lib/Target/MBlaze/MBlazeSchedule.td new file mode 100644 index 0000000..6a94491 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeSchedule.td @@ -0,0 +1,63 @@ +//===- MBlazeSchedule.td - MBlaze Scheduling Definitions --------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +//===----------------------------------------------------------------------===// +// Functional units across MBlaze chips sets. Based on GCC/MBlaze backend files. +//===----------------------------------------------------------------------===// +def ALU : FuncUnit; +def IMULDIV : FuncUnit; + +//===----------------------------------------------------------------------===// +// Instruction Itinerary classes used for MBlaze +//===----------------------------------------------------------------------===// +def IIAlu : InstrItinClass; +def IILoad : InstrItinClass; +def IIStore : InstrItinClass; +def IIXfer : InstrItinClass; +def IIBranch : InstrItinClass; +def IIHiLo : InstrItinClass; +def IIImul : InstrItinClass; +def IIIdiv : InstrItinClass; +def IIFcvt : InstrItinClass; +def IIFmove : InstrItinClass; +def IIFcmp : InstrItinClass; +def IIFadd : InstrItinClass; +def IIFmulSingle : InstrItinClass; +def IIFmulDouble : InstrItinClass; +def IIFdivSingle : InstrItinClass; +def IIFdivDouble : InstrItinClass; +def IIFsqrtSingle : InstrItinClass; +def IIFsqrtDouble : InstrItinClass; +def IIFrecipFsqrtStep : InstrItinClass; +def IIPseudo : InstrItinClass; + +//===----------------------------------------------------------------------===// +// MBlaze Generic instruction itineraries. +//===----------------------------------------------------------------------===// +def MBlazeGenericItineraries : ProcessorItineraries<[ + InstrItinData<IIAlu , [InstrStage<1, [ALU]>]>, + InstrItinData<IILoad , [InstrStage<3, [ALU]>]>, + InstrItinData<IIStore , [InstrStage<1, [ALU]>]>, + InstrItinData<IIXfer , [InstrStage<2, [ALU]>]>, + InstrItinData<IIBranch , [InstrStage<1, [ALU]>]>, + InstrItinData<IIHiLo , [InstrStage<1, [IMULDIV]>]>, + InstrItinData<IIImul , [InstrStage<17, [IMULDIV]>]>, + InstrItinData<IIIdiv , [InstrStage<38, [IMULDIV]>]>, + InstrItinData<IIFcvt , [InstrStage<1, [ALU]>]>, + InstrItinData<IIFmove , [InstrStage<2, [ALU]>]>, + InstrItinData<IIFcmp , [InstrStage<3, [ALU]>]>, + InstrItinData<IIFadd , [InstrStage<4, [ALU]>]>, + InstrItinData<IIFmulSingle , [InstrStage<7, [ALU]>]>, + InstrItinData<IIFmulDouble , [InstrStage<8, [ALU]>]>, + InstrItinData<IIFdivSingle , [InstrStage<23, [ALU]>]>, + InstrItinData<IIFdivDouble , [InstrStage<36, [ALU]>]>, + InstrItinData<IIFsqrtSingle , [InstrStage<54, [ALU]>]>, + InstrItinData<IIFsqrtDouble , [InstrStage<12, [ALU]>]>, + InstrItinData<IIFrecipFsqrtStep , [InstrStage<5, [ALU]>]> +]>; diff --git a/lib/Target/MBlaze/MBlazeSubtarget.cpp b/lib/Target/MBlaze/MBlazeSubtarget.cpp new file mode 100644 index 0000000..3440521 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeSubtarget.cpp @@ -0,0 +1,31 @@ +//===- MBlazeSubtarget.cpp - MBlaze Subtarget Information -------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the MBlaze specific subclass of TargetSubtarget. +// +//===----------------------------------------------------------------------===// + +#include "MBlazeSubtarget.h" +#include "MBlaze.h" +#include "MBlazeGenSubtarget.inc" +#include "llvm/Support/CommandLine.h" +using namespace llvm; + +MBlazeSubtarget::MBlazeSubtarget(const std::string &TT, const std::string &FS): + HasPipe3(false), HasBarrel(false), HasDiv(false), HasMul(false), + HasFSL(false), HasEFSL(false), HasMSRSet(false), HasException(false), + HasPatCmp(false), HasFPU(false), HasESR(false), HasPVR(false), + HasMul64(false), HasSqrt(false), HasMMU(false) +{ + std::string CPU = "v400"; + MBlazeArchVersion = V400; + + // Parse features string. + ParseSubtargetFeatures(FS, CPU); +} diff --git a/lib/Target/MBlaze/MBlazeSubtarget.h b/lib/Target/MBlaze/MBlazeSubtarget.h new file mode 100644 index 0000000..bebb3f7 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeSubtarget.h @@ -0,0 +1,79 @@ +//=====-- MBlazeSubtarget.h - Define Subtarget for the MBlaze -*- C++ -*--====// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file declares the MBlaze specific subclass of TargetSubtarget. +// +//===----------------------------------------------------------------------===// + +#ifndef MBLAZESUBTARGET_H +#define MBLAZESUBTARGET_H + +#include "llvm/Target/TargetSubtarget.h" +#include "llvm/Target/TargetMachine.h" + +#include <string> + +namespace llvm { + +class MBlazeSubtarget : public TargetSubtarget { + +protected: + + enum MBlazeArchEnum { + V400, V500, V600, V700, V710 + }; + + // MBlaze architecture version + MBlazeArchEnum MBlazeArchVersion; + + bool HasPipe3; + bool HasBarrel; + bool HasDiv; + bool HasMul; + bool HasFSL; + bool HasEFSL; + bool HasMSRSet; + bool HasException; + bool HasPatCmp; + bool HasFPU; + bool HasESR; + bool HasPVR; + bool HasMul64; + bool HasSqrt; + bool HasMMU; + + InstrItineraryData InstrItins; + +public: + + /// This constructor initializes the data members to match that + /// of the specified triple. + MBlazeSubtarget(const std::string &TT, const std::string &FS); + + /// ParseSubtargetFeatures - Parses features string setting specified + /// subtarget options. Definition of function is auto generated by tblgen. + std::string ParseSubtargetFeatures(const std::string &FS, + const std::string &CPU); + + bool hasFPU() const { return HasFPU; } + bool hasSqrt() const { return HasSqrt; } + bool hasMul() const { return HasMul; } + bool hasMul64() const { return HasMul64; } + bool hasDiv() const { return HasDiv; } + bool hasBarrel() const { return HasBarrel; } + + bool isV400() const { return MBlazeArchVersion == V400; } + bool isV500() const { return MBlazeArchVersion == V500; } + bool isV600() const { return MBlazeArchVersion == V600; } + bool isV700() const { return MBlazeArchVersion == V700; } + bool isV710() const { return MBlazeArchVersion == V710; } +}; +} // End llvm namespace + +#endif diff --git a/lib/Target/MBlaze/MBlazeTargetMachine.cpp b/lib/Target/MBlaze/MBlazeTargetMachine.cpp new file mode 100644 index 0000000..9eba2b3 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeTargetMachine.cpp @@ -0,0 +1,66 @@ +//===-- MBlazeTargetMachine.cpp - Define TargetMachine for MBlaze ---------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// Implements the info about MBlaze target spec. +// +//===----------------------------------------------------------------------===// + +#include "MBlaze.h" +#include "MBlazeMCAsmInfo.h" +#include "MBlazeTargetMachine.h" +#include "llvm/PassManager.h" +#include "llvm/Target/TargetRegistry.h" +using namespace llvm; + +extern "C" void LLVMInitializeMBlazeTarget() { + // Register the target. + RegisterTargetMachine<MBlazeTargetMachine> X(TheMBlazeTarget); + RegisterAsmInfo<MBlazeMCAsmInfo> A(TheMBlazeTarget); +} + +// DataLayout --> Big-endian, 32-bit pointer/ABI/alignment +// The stack is always 8 byte aligned +// On function prologue, the stack is created by decrementing +// its pointer. Once decremented, all references are done with positive +// offset from the stack/frame pointer, using StackGrowsUp enables +// an easier handling. +MBlazeTargetMachine:: +MBlazeTargetMachine(const Target &T, const std::string &TT, + const std::string &FS): + LLVMTargetMachine(T, TT), + Subtarget(TT, FS), + DataLayout("E-p:32:32-i8:8:8-i16:16:16-i64:32:32-" + "f64:32:32-v64:32:32-v128:32:32-n32"), + InstrInfo(*this), + FrameInfo(TargetFrameInfo::StackGrowsUp, 8, 0), + TLInfo(*this) { + if (getRelocationModel() == Reloc::Default) { + setRelocationModel(Reloc::Static); + } + + if (getCodeModel() == CodeModel::Default) + setCodeModel(CodeModel::Small); +} + +// Install an instruction selector pass using +// the ISelDag to gen MBlaze code. +bool MBlazeTargetMachine:: +addInstSelector(PassManagerBase &PM, CodeGenOpt::Level OptLevel) { + PM.add(createMBlazeISelDag(*this)); + return false; +} + +// Implemented by targets that want to run passes immediately before +// machine code is emitted. return true if -print-machineinstrs should +// print out the code after the passes. +bool MBlazeTargetMachine:: +addPreEmitPass(PassManagerBase &PM, CodeGenOpt::Level OptLevel) { + PM.add(createMBlazeDelaySlotFillerPass(*this)); + return true; +} diff --git a/lib/Target/MBlaze/MBlazeTargetMachine.h b/lib/Target/MBlaze/MBlazeTargetMachine.h new file mode 100644 index 0000000..85c975c --- /dev/null +++ b/lib/Target/MBlaze/MBlazeTargetMachine.h @@ -0,0 +1,69 @@ +//===-- MBlazeTargetMachine.h - Define TargetMachine for MBlaze --- C++ ---===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file declares the MBlaze specific subclass of TargetMachine. +// +//===----------------------------------------------------------------------===// + +#ifndef MBLAZE_TARGETMACHINE_H +#define MBLAZE_TARGETMACHINE_H + +#include "MBlazeSubtarget.h" +#include "MBlazeInstrInfo.h" +#include "MBlazeISelLowering.h" +#include "MBlazeIntrinsicInfo.h" +#include "llvm/Target/TargetMachine.h" +#include "llvm/Target/TargetData.h" +#include "llvm/Target/TargetFrameInfo.h" + +namespace llvm { + class formatted_raw_ostream; + + class MBlazeTargetMachine : public LLVMTargetMachine { + MBlazeSubtarget Subtarget; + const TargetData DataLayout; // Calculates type size & alignment + MBlazeInstrInfo InstrInfo; + TargetFrameInfo FrameInfo; + MBlazeTargetLowering TLInfo; + MBlazeIntrinsicInfo IntrinsicInfo; + public: + MBlazeTargetMachine(const Target &T, const std::string &TT, + const std::string &FS); + + virtual const MBlazeInstrInfo *getInstrInfo() const + { return &InstrInfo; } + + virtual const TargetFrameInfo *getFrameInfo() const + { return &FrameInfo; } + + virtual const MBlazeSubtarget *getSubtargetImpl() const + { return &Subtarget; } + + virtual const TargetData *getTargetData() const + { return &DataLayout;} + + virtual const MBlazeRegisterInfo *getRegisterInfo() const + { return &InstrInfo.getRegisterInfo(); } + + virtual MBlazeTargetLowering *getTargetLowering() const + { return const_cast<MBlazeTargetLowering*>(&TLInfo); } + + const TargetIntrinsicInfo *getIntrinsicInfo() const + { return &IntrinsicInfo; } + + // Pass Pipeline Configuration + virtual bool addInstSelector(PassManagerBase &PM, + CodeGenOpt::Level OptLevel); + + virtual bool addPreEmitPass(PassManagerBase &PM, + CodeGenOpt::Level OptLevel); + }; +} // End llvm namespace + +#endif diff --git a/lib/Target/MBlaze/MBlazeTargetObjectFile.cpp b/lib/Target/MBlaze/MBlazeTargetObjectFile.cpp new file mode 100644 index 0000000..79c9494 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeTargetObjectFile.cpp @@ -0,0 +1,88 @@ +//===-- MBlazeTargetObjectFile.cpp - MBlaze object files ------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "MBlazeTargetObjectFile.h" +#include "MBlazeSubtarget.h" +#include "llvm/DerivedTypes.h" +#include "llvm/GlobalVariable.h" +#include "llvm/MC/MCSectionELF.h" +#include "llvm/Target/TargetData.h" +#include "llvm/Target/TargetMachine.h" +#include "llvm/Support/CommandLine.h" +using namespace llvm; + +void MBlazeTargetObjectFile:: +Initialize(MCContext &Ctx, const TargetMachine &TM) { + TargetLoweringObjectFileELF::Initialize(Ctx, TM); + + SmallDataSection = + getELFSection(".sdata", MCSectionELF::SHT_PROGBITS, + MCSectionELF::SHF_WRITE | MCSectionELF::SHF_ALLOC, + SectionKind::getDataRel()); + + SmallBSSSection = + getELFSection(".sbss", MCSectionELF::SHT_NOBITS, + MCSectionELF::SHF_WRITE | MCSectionELF::SHF_ALLOC, + SectionKind::getBSS()); + +} + +// A address must be loaded from a small section if its size is less than the +// small section size threshold. Data in this section must be addressed using +// gp_rel operator. +static bool IsInSmallSection(uint64_t Size) { + return Size > 0 && Size <= 8; +} + +bool MBlazeTargetObjectFile:: +IsGlobalInSmallSection(const GlobalValue *GV, const TargetMachine &TM) const { + if (GV->isDeclaration() || GV->hasAvailableExternallyLinkage()) + return false; + + return IsGlobalInSmallSection(GV, TM, getKindForGlobal(GV, TM)); +} + +/// IsGlobalInSmallSection - Return true if this global address should be +/// placed into small data/bss section. +bool MBlazeTargetObjectFile:: +IsGlobalInSmallSection(const GlobalValue *GV, const TargetMachine &TM, + SectionKind Kind) const { + // Only global variables, not functions. + const GlobalVariable *GVA = dyn_cast<GlobalVariable>(GV); + if (!GVA) + return false; + + // We can only do this for datarel or BSS objects for now. + if (!Kind.isBSS() && !Kind.isDataRel()) + return false; + + // If this is a internal constant string, there is a special + // section for it, but not in small data/bss. + if (Kind.isMergeable1ByteCString()) + return false; + + const Type *Ty = GV->getType()->getElementType(); + return IsInSmallSection(TM.getTargetData()->getTypeAllocSize(Ty)); +} + +const MCSection *MBlazeTargetObjectFile:: +SelectSectionForGlobal(const GlobalValue *GV, SectionKind Kind, + Mangler *Mang, const TargetMachine &TM) const { + // TODO: Could also support "weak" symbols as well with ".gnu.linkonce.s.*" + // sections? + + // Handle Small Section classification here. + if (Kind.isBSS() && IsGlobalInSmallSection(GV, TM, Kind)) + return SmallBSSSection; + if (Kind.isDataNoRel() && IsGlobalInSmallSection(GV, TM, Kind)) + return SmallDataSection; + + // Otherwise, we work the same as ELF. + return TargetLoweringObjectFileELF::SelectSectionForGlobal(GV, Kind, Mang,TM); +} diff --git a/lib/Target/MBlaze/MBlazeTargetObjectFile.h b/lib/Target/MBlaze/MBlazeTargetObjectFile.h new file mode 100644 index 0000000..20e7702 --- /dev/null +++ b/lib/Target/MBlaze/MBlazeTargetObjectFile.h @@ -0,0 +1,41 @@ +//===-- llvm/Target/MBlazeTargetObjectFile.h - MBlaze Obj. Info -*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_TARGET_MBLAZE_TARGETOBJECTFILE_H +#define LLVM_TARGET_MBLAZE_TARGETOBJECTFILE_H + +#include "llvm/CodeGen/TargetLoweringObjectFileImpl.h" + +namespace llvm { + + class MBlazeTargetObjectFile : public TargetLoweringObjectFileELF { + const MCSection *SmallDataSection; + const MCSection *SmallBSSSection; + public: + + void Initialize(MCContext &Ctx, const TargetMachine &TM); + + + /// IsGlobalInSmallSection - Return true if this global address should be + /// placed into small data/bss section. + bool IsGlobalInSmallSection(const GlobalValue *GV, + const TargetMachine &TM, + SectionKind Kind) const; + + bool IsGlobalInSmallSection(const GlobalValue *GV, + const TargetMachine &TM) const; + + const MCSection *SelectSectionForGlobal(const GlobalValue *GV, + SectionKind Kind, + Mangler *Mang, + const TargetMachine &TM) const; + }; +} // end namespace llvm + +#endif diff --git a/lib/Target/MBlaze/Makefile b/lib/Target/MBlaze/Makefile new file mode 100644 index 0000000..19e508c --- /dev/null +++ b/lib/Target/MBlaze/Makefile @@ -0,0 +1,23 @@ +##===- lib/Target/MBlaze/Makefile --------------------------*- Makefile -*-===## +# +# The LLVM Compiler Infrastructure +# +# This file is distributed under the University of Illinois Open Source +# License. See LICENSE.TXT for details. +# +##===----------------------------------------------------------------------===## +LEVEL = ../../.. +LIBRARYNAME = LLVMMBlazeCodeGen +TARGET = MBlaze + +# Make sure that tblgen is run, first thing. +BUILT_SOURCES = MBlazeGenRegisterInfo.h.inc MBlazeGenRegisterNames.inc \ + MBlazeGenRegisterInfo.inc MBlazeGenInstrNames.inc \ + MBlazeGenInstrInfo.inc MBlazeGenAsmWriter.inc \ + MBlazeGenDAGISel.inc MBlazeGenCallingConv.inc \ + MBlazeGenSubtarget.inc MBlazeGenIntrinsics.inc + +DIRS = AsmPrinter TargetInfo + +include $(LEVEL)/Makefile.common + diff --git a/lib/Target/MBlaze/TargetInfo/CMakeLists.txt b/lib/Target/MBlaze/TargetInfo/CMakeLists.txt new file mode 100644 index 0000000..5afb14d --- /dev/null +++ b/lib/Target/MBlaze/TargetInfo/CMakeLists.txt @@ -0,0 +1,7 @@ +include_directories( ${CMAKE_CURRENT_BINARY_DIR}/.. ${CMAKE_CURRENT_SOURCE_DIR}/.. ) + +add_llvm_library(LLVMMBlazeInfo + MBlazeTargetInfo.cpp + ) + +add_dependencies(LLVMMBlazeInfo MBlazeCodeGenTable_gen) diff --git a/lib/Target/MBlaze/TargetInfo/MBlazeTargetInfo.cpp b/lib/Target/MBlaze/TargetInfo/MBlazeTargetInfo.cpp new file mode 100644 index 0000000..16e01db --- /dev/null +++ b/lib/Target/MBlaze/TargetInfo/MBlazeTargetInfo.cpp @@ -0,0 +1,19 @@ +//===-- MBlazeTargetInfo.cpp - MBlaze Target Implementation ---------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "MBlaze.h" +#include "llvm/Module.h" +#include "llvm/Target/TargetRegistry.h" +using namespace llvm; + +Target llvm::TheMBlazeTarget; + +extern "C" void LLVMInitializeMBlazeTargetInfo() { + RegisterTarget<Triple::mblaze> X(TheMBlazeTarget, "mblaze", "MBlaze"); +} diff --git a/lib/Target/MBlaze/TargetInfo/Makefile b/lib/Target/MBlaze/TargetInfo/Makefile new file mode 100644 index 0000000..fb7ea11 --- /dev/null +++ b/lib/Target/MBlaze/TargetInfo/Makefile @@ -0,0 +1,15 @@ +##===- lib/Target/MBlaze/TargetInfo/Makefile ---------------*- Makefile -*-===## +# +# The LLVM Compiler Infrastructure +# +# This file is distributed under the University of Illinois Open Source +# License. See LICENSE.TXT for details. +# +##===----------------------------------------------------------------------===## +LEVEL = ../../../.. +LIBRARYNAME = LLVMMBlazeInfo + +# Hack: we need to include 'main' target directory to grab private headers +CPPFLAGS = -I$(PROJ_OBJ_DIR)/.. -I$(PROJ_SRC_DIR)/.. + +include $(LEVEL)/Makefile.common diff --git a/lib/Target/MSIL/MSILWriter.cpp b/lib/Target/MSIL/MSILWriter.cpp index 3330d09..ac41cc8 100644 --- a/lib/Target/MSIL/MSILWriter.cpp +++ b/lib/Target/MSIL/MSILWriter.cpp @@ -38,7 +38,8 @@ namespace llvm { virtual bool addPassesToEmitWholeFile(PassManager &PM, formatted_raw_ostream &Out, CodeGenFileType FileType, - CodeGenOpt::Level OptLevel); + CodeGenOpt::Level OptLevel, + bool DisableVerify); virtual const TargetData *getTargetData() const { return 0; } }; @@ -57,7 +58,7 @@ bool MSILModule::runOnModule(Module &M) { TypeSymbolTable& Table = M.getTypeSymbolTable(); std::set<const Type *> Types = getAnalysis<FindUsedTypes>().getTypes(); for (TypeSymbolTable::iterator I = Table.begin(), E = Table.end(); I!=E; ) { - if (!isa<StructType>(I->second) && !isa<OpaqueType>(I->second)) + if (!I->second->isStructTy() && !I->second->isOpaqueTy()) Table.remove(I++); else { std::set<const Type *>::iterator T = Types.find(I->second); @@ -1459,7 +1460,7 @@ void MSILWriter::printDeclarations(const TypeSymbolTable& ST) { for (std::set<const Type*>::const_iterator UI = UsedTypes->begin(), UE = UsedTypes->end(); UI!=UE; ++UI) { const Type* Ty = *UI; - if (isa<ArrayType>(Ty) || isa<VectorType>(Ty) || isa<StructType>(Ty)) + if (Ty->isArrayTy() || Ty->isVectorTy() || Ty->isStructTy()) Name = getTypeName(Ty, false, true); // Type with no need to declare. else continue; @@ -1688,7 +1689,8 @@ void MSILWriter::printExternals() { bool MSILTarget::addPassesToEmitWholeFile(PassManager &PM, formatted_raw_ostream &o, CodeGenFileType FileType, - CodeGenOpt::Level OptLevel) + CodeGenOpt::Level OptLevel, + bool DisableVerify) { if (FileType != TargetMachine::CGFT_AssemblyFile) return true; MSILWriter* Writer = new MSILWriter(o); diff --git a/lib/Target/MSP430/MSP430ISelDAGToDAG.cpp b/lib/Target/MSP430/MSP430ISelDAGToDAG.cpp index a8c5e0a..911cfcb 100644 --- a/lib/Target/MSP430/MSP430ISelDAGToDAG.cpp +++ b/lib/Target/MSP430/MSP430ISelDAGToDAG.cpp @@ -26,26 +26,12 @@ #include "llvm/CodeGen/SelectionDAG.h" #include "llvm/CodeGen/SelectionDAGISel.h" #include "llvm/Target/TargetLowering.h" -#include "llvm/Support/CommandLine.h" #include "llvm/Support/Compiler.h" #include "llvm/Support/Debug.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Support/raw_ostream.h" -#include "llvm/ADT/Statistic.h" - using namespace llvm; -#ifndef NDEBUG -static cl::opt<bool> -ViewRMWDAGs("view-msp430-rmw-dags", cl::Hidden, - cl::desc("Pop up a window to show isel dags after RMW preprocess")); -#else -static const bool ViewRMWDAGs = false; -#endif - -STATISTIC(NumLoadMoved, "Number of loads moved below TokenFactor"); - - namespace { struct MSP430ISelAddressMode { enum { @@ -123,8 +109,6 @@ namespace { Lowering(*TM.getTargetLowering()), Subtarget(*TM.getSubtargetImpl()) { } - virtual void InstructionSelect(); - virtual const char *getPassName() const { return "MSP430 DAG->DAG Pattern Instruction Selection"; } @@ -133,8 +117,6 @@ namespace { bool MatchWrapper(SDValue N, MSP430ISelAddressMode &AM); bool MatchAddressBase(SDValue N, MSP430ISelAddressMode &AM); - bool IsLegalToFold(SDValue N, SDNode *U, SDNode *Root) const; - virtual bool SelectInlineAsmMemoryOperand(const SDValue &Op, char ConstraintCode, std::vector<SDValue> &OutOps); @@ -143,18 +125,12 @@ namespace { #include "MSP430GenDAGISel.inc" private: - DenseMap<SDNode*, SDNode*> RMWStores; - void PreprocessForRMW(); SDNode *Select(SDNode *N); SDNode *SelectIndexedLoad(SDNode *Op); SDNode *SelectIndexedBinOp(SDNode *Op, SDValue N1, SDValue N2, unsigned Opc8, unsigned Opc16); bool SelectAddr(SDNode *Op, SDValue Addr, SDValue &Base, SDValue &Disp); - - #ifndef NDEBUG - unsigned Indent; - #endif }; } // end anonymous namespace @@ -216,10 +192,7 @@ bool MSP430DAGToDAGISel::MatchAddressBase(SDValue N, MSP430ISelAddressMode &AM) } bool MSP430DAGToDAGISel::MatchAddress(SDValue N, MSP430ISelAddressMode &AM) { - DEBUG({ - errs() << "MatchAddress: "; - AM.dump(); - }); + DEBUG(errs() << "MatchAddress: "; AM.dump()); switch (N.getOpcode()) { default: break; @@ -335,270 +308,6 @@ SelectInlineAsmMemoryOperand(const SDValue &Op, char ConstraintCode, return false; } -bool MSP430DAGToDAGISel::IsLegalToFold(SDValue N, SDNode *U, - SDNode *Root) const { - if (OptLevel == CodeGenOpt::None) return false; - - /// RMW preprocessing creates the following code: - /// [Load1] - /// ^ ^ - /// / | - /// / | - /// [Load2] | - /// ^ ^ | - /// | | | - /// | \-| - /// | | - /// | [Op] - /// | ^ - /// | | - /// \ / - /// \ / - /// [Store] - /// - /// The path Store => Load2 => Load1 is via chain. Note that in general it is - /// not allowed to fold Load1 into Op (and Store) since it will creates a - /// cycle. However, this is perfectly legal for the loads moved below the - /// TokenFactor by PreprocessForRMW. Query the map Store => Load1 (created - /// during preprocessing) to determine whether it's legal to introduce such - /// "cycle" for a moment. - DenseMap<SDNode*, SDNode*>::const_iterator I = RMWStores.find(Root); - if (I != RMWStores.end() && I->second == N.getNode()) - return true; - - // Proceed to 'generic' cycle finder code - return SelectionDAGISel::IsLegalToFold(N, U, Root); -} - - -/// MoveBelowTokenFactor - Replace TokenFactor operand with load's chain operand -/// and move load below the TokenFactor. Replace store's chain operand with -/// load's chain result. -static void MoveBelowTokenFactor(SelectionDAG *CurDAG, SDValue Load, - SDValue Store, SDValue TF) { - SmallVector<SDValue, 4> Ops; - for (unsigned i = 0, e = TF.getNode()->getNumOperands(); i != e; ++i) - if (Load.getNode() == TF.getOperand(i).getNode()) - Ops.push_back(Load.getOperand(0)); - else - Ops.push_back(TF.getOperand(i)); - SDValue NewTF = CurDAG->UpdateNodeOperands(TF, &Ops[0], Ops.size()); - SDValue NewLoad = CurDAG->UpdateNodeOperands(Load, NewTF, - Load.getOperand(1), - Load.getOperand(2)); - CurDAG->UpdateNodeOperands(Store, NewLoad.getValue(1), Store.getOperand(1), - Store.getOperand(2), Store.getOperand(3)); -} - -/// MoveBelowTokenFactor2 - Replace TokenFactor operand with load's chain operand -/// and move load below the TokenFactor. Replace store's chain operand with -/// load's chain result. This a version which sinks two loads below token factor. -/// Look into PreprocessForRMW comments for explanation of transform. -static void MoveBelowTokenFactor2(SelectionDAG *CurDAG, - SDValue Load1, SDValue Load2, - SDValue Store, SDValue TF) { - SmallVector<SDValue, 4> Ops; - for (unsigned i = 0, e = TF.getNode()->getNumOperands(); i != e; ++i) { - SDNode* N = TF.getOperand(i).getNode(); - if (Load2.getNode() == N) - Ops.push_back(Load2.getOperand(0)); - else if (Load1.getNode() != N) - Ops.push_back(TF.getOperand(i)); - } - - SDValue NewTF = SDValue(CurDAG->MorphNodeTo(TF.getNode(), - TF.getOpcode(), - TF.getNode()->getVTList(), - &Ops[0], Ops.size()), TF.getResNo()); - SDValue NewLoad2 = CurDAG->UpdateNodeOperands(Load2, NewTF, - Load2.getOperand(1), - Load2.getOperand(2)); - - SDValue NewLoad1 = CurDAG->UpdateNodeOperands(Load1, NewLoad2.getValue(1), - Load1.getOperand(1), - Load1.getOperand(2)); - - CurDAG->UpdateNodeOperands(Store, - NewLoad1.getValue(1), - Store.getOperand(1), - Store.getOperand(2), Store.getOperand(3)); -} - -/// isAllowedToSink - return true if N a load which can be moved below token -/// factor. Basically, the load should be non-volatile and has single use. -static bool isLoadAllowedToSink(SDValue N, SDValue Chain) { - if (N.getOpcode() == ISD::BIT_CONVERT) - N = N.getOperand(0); - - LoadSDNode *LD = dyn_cast<LoadSDNode>(N); - if (!LD || LD->isVolatile()) - return false; - if (LD->getAddressingMode() != ISD::UNINDEXED) - return false; - - ISD::LoadExtType ExtType = LD->getExtensionType(); - if (ExtType != ISD::NON_EXTLOAD && ExtType != ISD::EXTLOAD) - return false; - - return (N.hasOneUse() && - LD->hasNUsesOfValue(1, 1) && - LD->isOperandOf(Chain.getNode())); -} - - -/// isRMWLoad - Return true if N is a load that's part of RMW sub-DAG. -/// The chain produced by the load must only be used by the store's chain -/// operand, otherwise this may produce a cycle in the DAG. -static bool isRMWLoad(SDValue N, SDValue Chain, SDValue Address, - SDValue &Load) { - if (isLoadAllowedToSink(N, Chain) && - N.getOperand(1) == Address) { - Load = N; - return true; - } - return false; -} - -/// PreprocessForRMW - Preprocess the DAG to make instruction selection better. -/// This is only run if not in -O0 mode. -/// This allows the instruction selector to pick more read-modify-write -/// instructions. This is a common case: -/// -/// [Load chain] -/// ^ -/// | -/// [Load] -/// ^ ^ -/// | | -/// / \- -/// / | -/// [TokenFactor] [Op] -/// ^ ^ -/// | | -/// \ / -/// \ / -/// [Store] -/// -/// The fact the store's chain operand != load's chain will prevent the -/// (store (op (load))) instruction from being selected. We can transform it to: -/// -/// [Load chain] -/// ^ -/// | -/// [TokenFactor] -/// ^ -/// | -/// [Load] -/// ^ ^ -/// | | -/// | \- -/// | | -/// | [Op] -/// | ^ -/// | | -/// \ / -/// \ / -/// [Store] -/// -/// We also recognize the case where second operand of Op is load as well and -/// move it below token factor as well creating DAG as follows: -/// -/// [Load chain] -/// ^ -/// | -/// [TokenFactor] -/// ^ -/// | -/// [Load1] -/// ^ ^ -/// / | -/// / | -/// [Load2] | -/// ^ ^ | -/// | | | -/// | \-| -/// | | -/// | [Op] -/// | ^ -/// | | -/// \ / -/// \ / -/// [Store] -/// -/// This allows selection of mem-mem instructions. Yay! - -void MSP430DAGToDAGISel::PreprocessForRMW() { - for (SelectionDAG::allnodes_iterator I = CurDAG->allnodes_begin(), - E = CurDAG->allnodes_end(); I != E; ++I) { - if (!ISD::isNON_TRUNCStore(I)) - continue; - SDValue Chain = I->getOperand(0); - - if (Chain.getNode()->getOpcode() != ISD::TokenFactor) - continue; - - SDValue N1 = I->getOperand(1); - SDValue N2 = I->getOperand(2); - if ((N1.getValueType().isFloatingPoint() && - !N1.getValueType().isVector()) || - !N1.hasOneUse()) - continue; - - unsigned RModW = 0; - SDValue Load1, Load2; - unsigned Opcode = N1.getNode()->getOpcode(); - switch (Opcode) { - case ISD::ADD: - case ISD::AND: - case ISD::OR: - case ISD::XOR: - case ISD::ADDC: - case ISD::ADDE: { - SDValue N10 = N1.getOperand(0); - SDValue N11 = N1.getOperand(1); - if (isRMWLoad(N10, Chain, N2, Load1)) { - if (isLoadAllowedToSink(N11, Chain)) { - Load2 = N11; - RModW = 2; - } else - RModW = 1; - } else if (isRMWLoad(N11, Chain, N2, Load1)) { - if (isLoadAllowedToSink(N10, Chain)) { - Load2 = N10; - RModW = 2; - } else - RModW = 1; - } - break; - } - case ISD::SUB: - case ISD::SUBC: - case ISD::SUBE: { - SDValue N10 = N1.getOperand(0); - SDValue N11 = N1.getOperand(1); - if (isRMWLoad(N10, Chain, N2, Load1)) { - if (isLoadAllowedToSink(N11, Chain)) { - Load2 = N11; - RModW = 2; - } else - RModW = 1; - } - break; - } - } - - NumLoadMoved += RModW; - if (RModW == 1) - MoveBelowTokenFactor(CurDAG, Load1, SDValue(I, 0), Chain); - else if (RModW == 2) { - MoveBelowTokenFactor2(CurDAG, Load1, Load2, SDValue(I, 0), Chain); - SDNode* Store = I; - RMWStores[Store] = Load2.getNode(); - } - } -} - - static bool isValidIndexedLoad(const LoadSDNode *LD) { ISD::MemIndexedMode AM = LD->getAddressingMode(); if (AM != ISD::POST_INC || LD->getExtensionType() != ISD::NON_EXTLOAD) @@ -681,46 +390,19 @@ SDNode *MSP430DAGToDAGISel::SelectIndexedBinOp(SDNode *Op, } -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void MSP430DAGToDAGISel::InstructionSelect() { - std::string BlockName; - if (ViewRMWDAGs) - BlockName = MF->getFunction()->getNameStr() + ":" + - BB->getBasicBlock()->getNameStr(); - - PreprocessForRMW(); - - if (ViewRMWDAGs) CurDAG->viewGraph("RMW preprocessed:" + BlockName); - - DEBUG(errs() << "Selection DAG after RMW preprocessing:\n"); - DEBUG(CurDAG->dump()); - - // Codegen the basic block. - DEBUG(errs() << "===== Instruction selection begins:\n"); - DEBUG(Indent = 0); - SelectRoot(*CurDAG); - DEBUG(errs() << "===== Instruction selection ends:\n"); - - CurDAG->RemoveDeadNodes(); - RMWStores.clear(); -} - SDNode *MSP430DAGToDAGISel::Select(SDNode *Node) { DebugLoc dl = Node->getDebugLoc(); // Dump information about the Node being selected - DEBUG(errs().indent(Indent) << "Selecting: "); + DEBUG(errs() << "Selecting: "); DEBUG(Node->dump(CurDAG)); DEBUG(errs() << "\n"); - DEBUG(Indent += 2); // If we have a custom node, we already have selected! if (Node->isMachineOpcode()) { - DEBUG(errs().indent(Indent-2) << "== "; + DEBUG(errs() << "== "; Node->dump(CurDAG); errs() << "\n"); - DEBUG(Indent -= 2); return NULL; } @@ -808,13 +490,12 @@ SDNode *MSP430DAGToDAGISel::Select(SDNode *Node) { // Select the default instruction SDNode *ResNode = SelectCode(Node); - DEBUG(errs() << std::string(Indent-2, ' ') << "=> "); + DEBUG(errs() << "=> "); if (ResNode == NULL || ResNode == Node) DEBUG(Node->dump(CurDAG)); else DEBUG(ResNode->dump(CurDAG)); DEBUG(errs() << "\n"); - DEBUG(Indent -= 2); return ResNode; } diff --git a/lib/Target/MSP430/MSP430ISelLowering.cpp b/lib/Target/MSP430/MSP430ISelLowering.cpp index 7281b37..e6c7e1e 100644 --- a/lib/Target/MSP430/MSP430ISelLowering.cpp +++ b/lib/Target/MSP430/MSP430ISelLowering.cpp @@ -795,18 +795,15 @@ SDValue MSP430TargetLowering::LowerSETCC(SDValue Op, SelectionDAG &DAG) { if (andCC) { // C = ~Z, thus Res = SRW & 1, no processing is required } else { - // Res = (SRW >> 1) & 1 + // Res = ~((SRW >> 1) & 1) Shift = true; + Invert = true; } break; case MSP430CC::COND_E: - if (andCC) { - // C = ~Z, thus Res = ~(SRW & 1) - } else { - // Res = ~((SRW >> 1) & 1) - Shift = true; - } - Invert = true; + Shift = true; + // C = ~Z for AND instruction, thus we can put Res = ~(SRW & 1), however, + // Res = (SRW >> 1) & 1 is 1 word shorter. break; } EVT VT = Op.getValueType(); diff --git a/lib/Target/MSP430/MSP430InstrInfo.td b/lib/Target/MSP430/MSP430InstrInfo.td index bb06f7b..144ba26 100644 --- a/lib/Target/MSP430/MSP430InstrInfo.td +++ b/lib/Target/MSP430/MSP430InstrInfo.td @@ -250,7 +250,7 @@ def MOV16ri : I16ri<0x0, [(set GR16:$dst, imm:$src)]>; } -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in { +let canFoldAsLoad = 1, isReMaterializable = 1 in { def MOV8rm : I8rm<0x0, (outs GR8:$dst), (ins memsrc:$src), "mov.b\t{$src, $dst}", diff --git a/lib/Target/Mips/MipsISelDAGToDAG.cpp b/lib/Target/Mips/MipsISelDAGToDAG.cpp index f1d4a67..c4746db 100644 --- a/lib/Target/Mips/MipsISelDAGToDAG.cpp +++ b/lib/Target/Mips/MipsISelDAGToDAG.cpp @@ -59,8 +59,6 @@ public: SelectionDAGISel(tm), TM(tm), Subtarget(tm.getSubtarget<MipsSubtarget>()) {} - virtual void InstructionSelect(); - // Pass Name virtual const char *getPassName() const { return "MIPS DAG->DAG Pattern Instruction Selection"; @@ -98,29 +96,10 @@ private: inline SDValue getI32Imm(unsigned Imm) { return CurDAG->getTargetConstant(Imm, MVT::i32); } - - - #ifndef NDEBUG - unsigned Indent; - #endif }; } -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void MipsDAGToDAGISel::InstructionSelect() { - // Codegen the basic block. - DEBUG(errs() << "===== Instruction selection begins:\n"); - DEBUG(Indent = 0); - - // Select target instructions for the DAG. - SelectRoot(*CurDAG); - - DEBUG(errs() << "===== Instruction selection ends:\n"); - - CurDAG->RemoveDeadNodes(); -} /// getGlobalBaseReg - Output the instructions required to put the /// GOT address into a register. @@ -329,17 +308,11 @@ SDNode* MipsDAGToDAGISel::Select(SDNode *Node) { DebugLoc dl = Node->getDebugLoc(); // Dump information about the Node being selected - DEBUG(errs().indent(Indent) << "Selecting: "; - Node->dump(CurDAG); - errs() << "\n"); - DEBUG(Indent += 2); + DEBUG(errs() << "Selecting: "; Node->dump(CurDAG); errs() << "\n"); // If we have a custom node, we already have selected! if (Node->isMachineOpcode()) { - DEBUG(errs().indent(Indent-2) << "== "; - Node->dump(CurDAG); - errs() << "\n"); - DEBUG(Indent -= 2); + DEBUG(errs() << "== "; Node->dump(CurDAG); errs() << "\n"); return NULL; } @@ -547,14 +520,12 @@ SDNode* MipsDAGToDAGISel::Select(SDNode *Node) { // Select the default instruction SDNode *ResNode = SelectCode(Node); - DEBUG(errs().indent(Indent-2) << "=> "); + DEBUG(errs() << "=> "); if (ResNode == NULL || ResNode == Node) DEBUG(Node->dump(CurDAG)); else DEBUG(ResNode->dump(CurDAG)); DEBUG(errs() << "\n"); - DEBUG(Indent -= 2); - return ResNode; } diff --git a/lib/Target/Mips/MipsInstrInfo.td b/lib/Target/Mips/MipsInstrInfo.td index f16a805..cef3697 100644 --- a/lib/Target/Mips/MipsInstrInfo.td +++ b/lib/Target/Mips/MipsInstrInfo.td @@ -300,9 +300,8 @@ class JumpFR<bits<6> op, bits<6> func, string instr_asm>: // Jump and Link (Call) let isCall=1, hasDelaySlot=1, // All calls clobber the non-callee saved registers... - Defs = [AT, V0, V1, A0, A1, A2, A3, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, - K0, K1, F0, F1, F2, F3, F4, F5, F6, F7, F8, F9, F10, F11, F12, F13, - F14, F15, F16, F17, F18, F19], Uses = [GP] in { + Defs = [AT, V0, V1, A0, A1, A2, A3, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, + K0, K1, D0, D1, D2, D3, D4, D5, D6, D7, D8, D9], Uses = [GP] in { class JumpLink<bits<6> op, string instr_asm>: FJ< op, (outs), @@ -593,8 +592,8 @@ def : Pat<(MipsJmpLink (i32 tglobaladdr:$dst)), (JAL tglobaladdr:$dst)>; def : Pat<(MipsJmpLink (i32 texternalsym:$dst)), (JAL texternalsym:$dst)>; -def : Pat<(MipsJmpLink CPURegs:$dst), - (JALR CPURegs:$dst)>; +//def : Pat<(MipsJmpLink CPURegs:$dst), +// (JALR CPURegs:$dst)>; // hi/lo relocs def : Pat<(MipsHi tglobaladdr:$in), (LUi tglobaladdr:$in)>; diff --git a/lib/Target/PIC16/AsmPrinter/PIC16AsmPrinter.cpp b/lib/Target/PIC16/AsmPrinter/PIC16AsmPrinter.cpp index b015edd..44a6cc0 100644 --- a/lib/Target/PIC16/AsmPrinter/PIC16AsmPrinter.cpp +++ b/lib/Target/PIC16/AsmPrinter/PIC16AsmPrinter.cpp @@ -158,6 +158,7 @@ bool PIC16AsmPrinter::runOnMachineFunction(MachineFunction &MF) { // printOperand - print operand of insn. void PIC16AsmPrinter::printOperand(const MachineInstr *MI, int opNum) { const MachineOperand &MO = MI->getOperand(opNum); + const Function *F = MI->getParent()->getParent()->getFunction(); switch (MO.getType()) { case MachineOperand::MO_Register: @@ -190,19 +191,18 @@ void PIC16AsmPrinter::printOperand(const MachineInstr *MI, int opNum) { } case MachineOperand::MO_ExternalSymbol: { const char *Sname = MO.getSymbolName(); + std::string Printname = Sname; - // If its a libcall name, record it to decls section. - if (PAN::getSymbolTag(Sname) == PAN::LIBCALL) - LibcallDecls.push_back(Sname); - - // Record a call to intrinsic to print the extern declaration for it. - std::string Sym = Sname; - if (PAN::isMemIntrinsic(Sym)) { - Sym = PAN::addPrefix(Sym); - LibcallDecls.push_back(createESName(Sym)); + // Intrinsic stuff needs to be renamed if we are printing IL fn. + if (PAN::isIntrinsicStuff(Printname)) { + if (PAN::isISR(F->getSection())) { + Printname = PAN::Rename(Sname); + } + // Record these decls, we need to print them in asm as extern. + LibcallDecls.push_back(createESName(Printname)); } - O << Sym; + O << Printname; break; } case MachineOperand::MO_MachineBasicBlock: @@ -248,8 +248,6 @@ void PIC16AsmPrinter::printLibcallDecls() { for (std::list<const char*>::const_iterator I = LibcallDecls.begin(); I != LibcallDecls.end(); I++) { O << MAI->getExternDirective() << *I << "\n"; - O << MAI->getExternDirective() << PAN::getArgsLabel(*I) << "\n"; - O << MAI->getExternDirective() << PAN::getRetvalLabel(*I) << "\n"; } O << MAI->getCommentString() << "External decls for libcalls - END." <<"\n"; } diff --git a/lib/Target/PIC16/PIC16ABINames.h b/lib/Target/PIC16/PIC16ABINames.h index b0b9318..4c1a8da 100644 --- a/lib/Target/PIC16/PIC16ABINames.h +++ b/lib/Target/PIC16/PIC16ABINames.h @@ -178,18 +178,21 @@ namespace llvm { return Func1 + tag; } + // Get the retval label for the given function. static std::string getRetvalLabel(const std::string &Func) { std::string Func1 = addPrefix(Func); std::string tag = getTagName(RET_LABEL); return Func1 + tag; } + // Get the argument label for the given function. static std::string getArgsLabel(const std::string &Func) { std::string Func1 = addPrefix(Func); std::string tag = getTagName(ARGS_LABEL); return Func1 + tag; } + // Get the tempdata label for the given function. static std::string getTempdataLabel(const std::string &Func) { std::string Func1 = addPrefix(Func); std::string tag = getTagName(TEMPS_LABEL); @@ -263,6 +266,7 @@ namespace llvm { return false; } + inline static bool isMemIntrinsic (const std::string &Name) { if (Name.compare("@memcpy") == 0 || Name.compare("@memset") == 0 || Name.compare("@memmove") == 0) { @@ -272,6 +276,41 @@ namespace llvm { return false; } + // Currently names of libcalls are assigned during TargetLowering + // object construction. There is no provision to change the when the + // code for a function IL function being generated. + // So we have to change these names while printing assembly. + // We need to do that mainly for names related to intrinsics. This + // function returns true if a name needs to be cloned. + inline static bool isIntrinsicStuff(const std::string &Name) { + // Return true if the name contains LIBCALL marker, or a MemIntrinisc. + // these are mainly ARGS_LABEL, RET_LABEL, and the LIBCALL name itself. + if ((Name.find(getTagName(LIBCALL)) != std::string::npos) + || isMemIntrinsic(Name)) + return true; + + return false; + } + + // Rename the name for IL. + inline static std::string Rename(const std::string &Name) { + std::string Newname; + // If its a label (LIBCALL+Func+LABEL), change it to + // (LIBCALL+Func+IL+LABEL). + TAGS id = getSymbolTag(Name); + if (id == ARGS_LABEL || id == RET_LABEL) { + std::size_t pos = Name.find(getTagName(id)); + Newname = Name.substr(0, pos) + ".IL" + getTagName(id); + return Newname; + } + + // Else, just append IL to name. + return Name + ".IL"; + } + + + + inline static bool isLocalToFunc (std::string &Func, std::string &Var) { if (! isLocalName(Var)) return false; @@ -338,6 +377,22 @@ namespace llvm { inline static std::string getISRAddr(void) { return "0x4"; } + + // Returns the name of clone of a function. + static std::string getCloneFnName(const std::string &Func) { + return (Func + ".IL"); + } + + // Returns the name of clone of a variable. + static std::string getCloneVarName(const std::string &Fn, + const std::string &Var) { + std::string cloneVarName = Var; + // These vars are named like fun.auto.var. + // Just replace the function name, with clone function name. + std::string cloneFnName = getCloneFnName(Fn); + cloneVarName.replace(cloneVarName.find(Fn), Fn.length(), cloneFnName); + return cloneVarName; + } }; // class PAN. } // end namespace llvm; diff --git a/lib/Target/PIC16/PIC16DebugInfo.cpp b/lib/Target/PIC16/PIC16DebugInfo.cpp index c517b1b..877e4ff 100644 --- a/lib/Target/PIC16/PIC16DebugInfo.cpp +++ b/lib/Target/PIC16/PIC16DebugInfo.cpp @@ -419,7 +419,7 @@ void PIC16DbgInfo::EmitAuxEntry(const std::string VarName, int Aux[], int Num, if (TagName != "") O << ", " << TagName; for (int i = 0; i<Num; i++) - O << "," << Aux[i]; + O << "," << (Aux[i] && 0xff); } /// EmitSymbol - Emit .def for a symbol. Value is offset for the member. diff --git a/lib/Target/PIC16/PIC16ISelDAGToDAG.cpp b/lib/Target/PIC16/PIC16ISelDAGToDAG.cpp index 82197ae..6cbd002 100644 --- a/lib/Target/PIC16/PIC16ISelDAGToDAG.cpp +++ b/lib/Target/PIC16/PIC16ISelDAGToDAG.cpp @@ -14,10 +14,7 @@ #define DEBUG_TYPE "pic16-isel" #include "llvm/Support/ErrorHandling.h" -#include "llvm/Support/raw_ostream.h" #include "PIC16ISelDAGToDAG.h" -#include "llvm/Support/Debug.h" - using namespace llvm; /// createPIC16ISelDag - This pass converts a legalized DAG into a @@ -27,13 +24,6 @@ FunctionPass *llvm::createPIC16ISelDag(PIC16TargetMachine &TM) { } -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void PIC16DAGToDAGISel::InstructionSelect() { - SelectRoot(*CurDAG); - CurDAG->RemoveDeadNodes(); -} - /// Select - Select instructions not customized! Used for /// expanded, promoted and normal instructions. SDNode* PIC16DAGToDAGISel::Select(SDNode *N) { diff --git a/lib/Target/PIC16/PIC16ISelDAGToDAG.h b/lib/Target/PIC16/PIC16ISelDAGToDAG.h index 813a540..8ed5bf7 100644 --- a/lib/Target/PIC16/PIC16ISelDAGToDAG.h +++ b/lib/Target/PIC16/PIC16ISelDAGToDAG.h @@ -19,6 +19,8 @@ #include "PIC16TargetMachine.h" #include "llvm/CodeGen/SelectionDAGISel.h" #include "llvm/Support/Compiler.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Support/Debug.h" #include "llvm/Intrinsics.h" using namespace llvm; @@ -46,8 +48,6 @@ public: return "PIC16 DAG->DAG Pattern Instruction Selection"; } - virtual void InstructionSelect(); - private: // Include the pieces autogenerated from the target description. #include "PIC16GenDAGISel.inc" diff --git a/lib/Target/PIC16/PIC16ISelLowering.cpp b/lib/Target/PIC16/PIC16ISelLowering.cpp index d2fc8db..d17abb9 100644 --- a/lib/Target/PIC16/PIC16ISelLowering.cpp +++ b/lib/Target/PIC16/PIC16ISelLowering.cpp @@ -419,8 +419,7 @@ PIC16TargetLowering::MakePIC16Libcall(PIC16ISD::PIC16Libcall Call, LowerCallTo(DAG.getEntryNode(), RetTy, isSigned, !isSigned, false, false, 0, CallingConv::C, false, /*isReturnValueUsed=*/true, - Callee, Args, DAG, dl, - DAG.GetOrdering(DAG.getEntryNode().getNode())); + Callee, Args, DAG, dl); return CallInfo.first; } @@ -1527,10 +1526,24 @@ bool PIC16TargetLowering::NeedToConvertToMemOp(SDValue Op, unsigned &MemOp, return true; if (isDirectLoad(Op.getOperand(1))) { - if (Op.getOperand(1).hasOneUse()) - return false; - else - MemOp = 1; + if (Op.getOperand(1).hasOneUse()) { + // Legal and profitable folding check uses the NodeId of DAG nodes. + // This NodeId is assigned by topological order. Therefore first + // assign topological order then perform legal and profitable check. + // Note:- Though this ordering is done before begining with legalization, + // newly added node during legalization process have NodeId=-1 (NewNode) + // therefore before performing any check proper ordering of the node is + // required. + DAG.AssignTopologicalOrder(); + + // Direct load operands are folded in binary operations. But before folding + // verify if this folding is legal. Fold only if it is legal otherwise + // convert this direct load to a separate memory operation. + if(ISel->IsLegalToFold(Op.getOperand(1), Op.getNode(), Op.getNode())) + return false; + else + MemOp = 1; + } } return true; } diff --git a/lib/Target/PIC16/PIC16Passes/PIC16Cloner.cpp b/lib/Target/PIC16/PIC16Passes/PIC16Cloner.cpp new file mode 100644 index 0000000..865da35 --- /dev/null +++ b/lib/Target/PIC16/PIC16Passes/PIC16Cloner.cpp @@ -0,0 +1,299 @@ +//===-- PIC16Cloner.cpp - PIC16 LLVM Cloner for shared functions -*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains code to clone all functions that are shared between +// the main line code (ML) and interrupt line code (IL). It clones all such +// shared functions and their automatic global vars by adding the .IL suffix. +// +// This pass is supposed to be run on the linked .bc module. +// It traveses the module call graph twice. Once starting from the main function +// and marking each reached function as "ML". Again, starting from the ISR +// and cloning any reachable function that was marked as "ML". After cloning +// the function, it remaps all the call sites in IL functions to call the +// cloned functions. +//===----------------------------------------------------------------------===// + +#include "llvm/Analysis/CallGraph.h" +#include "llvm/Pass.h" +#include "llvm/Module.h" +#include "llvm/Support/ErrorHandling.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Transforms/Utils/Cloning.h" +#include "PIC16Cloner.h" +#include "../PIC16ABINames.h" +#include <vector> + +using namespace llvm; +using std::vector; +using std::string; +using std::map; + +namespace llvm { + char PIC16Cloner::ID = 0; + + ModulePass *createPIC16ClonerPass() { return new PIC16Cloner(); } +} + +// We currently intend to run these passes in opt, which does not have any +// diagnostic support. So use these functions for now. In future +// we will probably write our own driver tool. +// +void PIC16Cloner::reportError(string ErrorString) { + errs() << "ERROR : " << ErrorString << "\n"; + exit(1); +} + +void PIC16Cloner:: +reportError (string ErrorString, vector<string> &Values) { + unsigned ValCount = Values.size(); + string TargetString; + for (unsigned i=0; i<ValCount; ++i) { + TargetString = "%"; + TargetString += ((char)i + '0'); + ErrorString.replace(ErrorString.find(TargetString), TargetString.length(), + Values[i]); + } + errs() << "ERROR : " << ErrorString << "\n"; + exit(1); +} + + +// Entry point +// +bool PIC16Cloner::runOnModule(Module &M) { + CallGraph &CG = getAnalysis<CallGraph>(); + + // Search for the "main" and "ISR" functions. + CallGraphNode *mainCGN = NULL, *isrCGN = NULL; + for (CallGraph::iterator it = CG.begin() ; it != CG.end(); it++) + { + // External calling node doesn't have any function associated with it. + if (! it->first) + continue; + + if (it->first->getName().str() == "main") { + mainCGN = it->second; + } + + if (PAN::isISR(it->first->getSection())) { + isrCGN = it->second; + } + + // Don't search further if we've found both. + if (mainCGN && isrCGN) + break; + } + + // We have nothing to do if any of the main or ISR is missing. + if (! mainCGN || ! isrCGN) return false; + + // Time for some diagnostics. + // See if the main itself is interrupt function then report an error. + if (PAN::isISR(mainCGN->getFunction()->getSection())) { + reportError("Function 'main' can't be interrupt function"); + } + + + // Mark all reachable functions from main as ML. + markCallGraph(mainCGN, "ML"); + + // And then all the functions reachable from ISR will be cloned. + cloneSharedFunctions(isrCGN); + + return true; +} + +// Mark all reachable functions from the given node, with the given mark. +// +void PIC16Cloner::markCallGraph(CallGraphNode *CGN, string StringMark) { + // Mark the top node first. + Function *thisF = CGN->getFunction(); + + thisF->setSection(StringMark); + + // Mark all the called functions + for(CallGraphNode::iterator cgn_it = CGN->begin(); + cgn_it != CGN->end(); ++cgn_it) { + Function *CalledF = cgn_it->second->getFunction(); + + // If calling an external function then CallGraphNode + // will not be associated with any function. + if (! CalledF) + continue; + + // Issue diagnostic if interrupt function is being called. + if (PAN::isISR(CalledF->getSection())) { + vector<string> Values; + Values.push_back(CalledF->getName().str()); + reportError("Interrupt function (%0) can't be called", Values); + } + + // Has already been mark + if (CalledF->getSection().find(StringMark) != string::npos) { + // Should we do anything here? + } else { + // Mark now + CalledF->setSection(StringMark); + } + + // Before going any further mark all the called function by current + // function. + markCallGraph(cgn_it->second ,StringMark); + } // end of loop of all called functions. +} + + +// For PIC16, automatic variables of a function are emitted as globals. +// Clone the auto variables of a function and put them in ValueMap, +// this ValueMap will be used while +// Cloning the code of function itself. +// +void PIC16Cloner::CloneAutos(Function *F) { + // We'll need to update module's globals list as well. So keep a reference + // handy. + Module *M = F->getParent(); + Module::GlobalListType &Globals = M->getGlobalList(); + + // Clear the leftovers in ValueMap by any previous cloning. + ValueMap.clear(); + + // Find the auto globls for this function and clone them, and put them + // in ValueMap. + std::string FnName = F->getName().str(); + std::string VarName, ClonedVarName; + for (Module::global_iterator I = M->global_begin(), E = M->global_end(); + I != E; ++I) { + VarName = I->getName().str(); + if (PAN::isLocalToFunc(FnName, VarName)) { + // Auto variable for current function found. Clone it. + GlobalVariable *GV = I; + + const Type *InitTy = GV->getInitializer()->getType(); + GlobalVariable *ClonedGV = + new GlobalVariable(InitTy, false, GV->getLinkage(), + GV->getInitializer()); + ClonedGV->setName(PAN::getCloneVarName(FnName, VarName)); + // Add these new globals to module's globals list. + Globals.push_back(ClonedGV); + + // Update ValueMap. + ValueMap[GV] = ClonedGV; + } + } +} + + +// Clone all functions that are reachable from ISR and are already +// marked as ML. +// +void PIC16Cloner::cloneSharedFunctions(CallGraphNode *CGN) { + + // Check all the called functions from ISR. + for(CallGraphNode::iterator cgn_it = CGN->begin(); + cgn_it != CGN->end(); ++cgn_it) { + Function *CalledF = cgn_it->second->getFunction(); + + // If calling an external function then CallGraphNode + // will not be associated with any function. + if (!CalledF) + continue; + + // Issue diagnostic if interrupt function is being called. + if (PAN::isISR(CalledF->getSection())) { + vector<string> Values; + Values.push_back(CalledF->getName().str()); + reportError("Interrupt function (%0) can't be called", Values); + } + + if (CalledF->getSection().find("ML") != string::npos) { + // Function is alternatively marked. It should be a shared one. + // Create IL copy. Passing called function as first argument + // and the caller as the second argument. + + // Before making IL copy, first ensure that this function has a + // body. If the function does have a body. It can't be cloned. + // Such a case may occur when the function has been declarated + // in the C source code but its body exists in assembly file. + if (!CalledF->isDeclaration()) { + Function *cf = cloneFunction(CalledF); + remapAllSites(CGN->getFunction(), CalledF, cf); + } else { + // It is called only from ISR. Still mark it as we need this info + // in code gen while calling intrinsics.Function is not marked. + CalledF->setSection("IL"); + } + } + // Before going any further clone all the shared function reachaable + // by current function. + cloneSharedFunctions(cgn_it->second); + } // end of loop of all called functions. +} + +// Clone the given function and return it. +// Note: it uses the ValueMap member of the class, which is already populated +// by cloneAutos by the time we reach here. +// FIXME: Should we just pass ValueMap's ref as a parameter here? rather +// than keeping the ValueMap as a member. +Function * +PIC16Cloner::cloneFunction(Function *OrgF) { + Function *ClonedF; + + // See if we already cloned it. Return that. + cloned_map_iterator cm_it = ClonedFunctionMap.find(OrgF); + if(cm_it != ClonedFunctionMap.end()) { + ClonedF = cm_it->second; + return ClonedF; + } + + // Clone does not exist. + // First clone the autos, and populate ValueMap. + CloneAutos(OrgF); + + // Now create the clone. + ClonedF = CloneFunction(OrgF, ValueMap); + + // The new function should be for interrupt line. Therefore should have + // the name suffixed with IL and section attribute marked with IL. + ClonedF->setName(PAN::getCloneFnName(OrgF->getName())); + ClonedF->setSection("IL"); + + // Add the newly created function to the module. + OrgF->getParent()->getFunctionList().push_back(ClonedF); + + // Update the ClonedFunctionMap to record this cloning activity. + ClonedFunctionMap[OrgF] = ClonedF; + + return ClonedF; +} + + +// Remap the call sites of shared functions, that are in IL. +// Change the IL call site of a shared function to its clone. +// +void PIC16Cloner:: +remapAllSites(Function *Caller, Function *OrgF, Function *Clone) { + // First find the caller to update. If the caller itself is cloned + // then use the cloned caller. Otherwise use it. + cloned_map_iterator cm_it = ClonedFunctionMap.find(Caller); + if (cm_it != ClonedFunctionMap.end()) + Caller = cm_it->second; + + // For the lack of a better call site finding mechanism, iterate over + // all insns to find the uses of original fn. + for (Function::iterator BI = Caller->begin(); BI != Caller->end(); ++BI) { + BasicBlock &BB = *BI; + for (BasicBlock::iterator II = BB.begin(); II != BB.end(); ++II) { + if (II->getNumOperands() > 0 && II->getOperand(0) == OrgF) + II->setOperand(0, Clone); + } + } +} + + + diff --git a/lib/Target/PIC16/PIC16Passes/PIC16Cloner.h b/lib/Target/PIC16/PIC16Passes/PIC16Cloner.h new file mode 100644 index 0000000..24c1152 --- /dev/null +++ b/lib/Target/PIC16/PIC16Passes/PIC16Cloner.h @@ -0,0 +1,83 @@ +//===-- PIC16Cloner.h - PIC16 LLVM Cloner for shared functions --*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file contains declaration of a cloner class clone all functions that +// are shared between the main line code (ML) and interrupt line code (IL). +// +//===----------------------------------------------------------------------===// + +#ifndef PIC16CLONER_H +#define PIC16CLONER_H + +#include "llvm/ADT/DenseMap.h" + +using namespace llvm; +using std::vector; +using std::string; +using std::map; + +namespace llvm { + // forward classes. + class Value; + class Function; + class Module; + class ModulePass; + class CallGraph; + class CallGraphNode; + class AnalysisUsage; + + class PIC16Cloner : public ModulePass { + public: + static char ID; // Class identification + PIC16Cloner() : ModulePass(&ID) {} + + virtual void getAnalysisUsage(AnalysisUsage &AU) const { + AU.addRequired<CallGraph>(); + } + virtual bool runOnModule(Module &M); + + private: // Functions + // Mark reachable functions for the MainLine or InterruptLine. + void markCallGraph(CallGraphNode *CGN, string StringMark); + + // Clone auto variables of function specified. + void CloneAutos(Function *F); + + // Clone the body of a function. + Function *cloneFunction(Function *F); + + // Clone all shared functions. + void cloneSharedFunctions(CallGraphNode *isrCGN); + + // Remap all call sites to the shared function. + void remapAllSites(Function *Caller, Function *OrgF, Function *Clone); + + // Error reporting for PIC16Pass + void reportError(string ErrorString, vector<string> &Values); + void reportError(string ErrorString); + + private: //data + // Records if the interrupt function has already been found. + // If more than one interrupt function is found then an error + // should be thrown. + bool foundISR; + + // This ValueMap maps the auto variables of the original functions with + // the corresponding cloned auto variable of the cloned function. + // This value map is passed during the function cloning so that all the + // uses of auto variables be updated properly. + DenseMap<const Value*, Value*> ValueMap; + + // Map of a already cloned functions. + map<Function *, Function *> ClonedFunctionMap; + typedef map<Function *, Function *>::iterator cloned_map_iterator; + }; +} // End of anonymous namespace + +#endif diff --git a/lib/Target/PIC16/PIC16Passes/PIC16Overlay.cpp b/lib/Target/PIC16/PIC16Passes/PIC16Overlay.cpp index 197c398..5ecb6aa 100644 --- a/lib/Target/PIC16/PIC16Passes/PIC16Overlay.cpp +++ b/lib/Target/PIC16/PIC16Passes/PIC16Overlay.cpp @@ -24,27 +24,27 @@ using namespace llvm; namespace llvm { - char PIC16FrameOverlay::ID = 0; - ModulePass *createPIC16OverlayPass() { return new PIC16FrameOverlay(); } + char PIC16Overlay::ID = 0; + ModulePass *createPIC16OverlayPass() { return new PIC16Overlay(); } } -void PIC16FrameOverlay::getAnalysisUsage(AnalysisUsage &AU) const { +void PIC16Overlay::getAnalysisUsage(AnalysisUsage &AU) const { AU.setPreservesAll(); AU.addRequired<CallGraph>(); } -void PIC16FrameOverlay::DFSTraverse(CallGraphNode *CGN, unsigned Depth) { +void PIC16Overlay::DFSTraverse(CallGraphNode *CGN, unsigned Depth) { // Do not set any color for external calling node. if (Depth != 0 && CGN->getFunction()) { unsigned Color = getColor(CGN->getFunction()); // Handle indirectly called functions - if (Color >= PIC16Overlay::StartIndirectCallColor || - Depth >= PIC16Overlay::StartIndirectCallColor) { + if (Color >= PIC16OVERLAY::StartIndirectCallColor || + Depth >= PIC16OVERLAY::StartIndirectCallColor) { // All functions called from an indirectly called function are given // an unique color. - if (Color < PIC16Overlay::StartIndirectCallColor && - Depth >= PIC16Overlay::StartIndirectCallColor) + if (Color < PIC16OVERLAY::StartIndirectCallColor && + Depth >= PIC16OVERLAY::StartIndirectCallColor) setColor(CGN->getFunction(), Depth); for (unsigned int i = 0; i < CGN->size(); i++) @@ -65,7 +65,7 @@ void PIC16FrameOverlay::DFSTraverse(CallGraphNode *CGN, unsigned Depth) { DFSTraverse((*CGN)[i], Depth+1); } -unsigned PIC16FrameOverlay::ModifyDepthForInterrupt(CallGraphNode *CGN, +unsigned PIC16Overlay::ModifyDepthForInterrupt(CallGraphNode *CGN, unsigned Depth) { Function *Fn = CGN->getFunction(); @@ -81,7 +81,7 @@ unsigned PIC16FrameOverlay::ModifyDepthForInterrupt(CallGraphNode *CGN, return Depth; } -void PIC16FrameOverlay::setColor(Function *Fn, unsigned Color) { +void PIC16Overlay::setColor(Function *Fn, unsigned Color) { std::string Section = ""; if (Fn->hasSection()) Section = Fn->getSection(); @@ -119,7 +119,7 @@ void PIC16FrameOverlay::setColor(Function *Fn, unsigned Color) { Fn->setSection(Section); } -unsigned PIC16FrameOverlay::getColor(Function *Fn) { +unsigned PIC16Overlay::getColor(Function *Fn) { int Color = 0; if (!Fn->hasSection()) return 0; @@ -150,7 +150,7 @@ unsigned PIC16FrameOverlay::getColor(Function *Fn) { return Color; } -bool PIC16FrameOverlay::runOnModule(Module &M) { +bool PIC16Overlay::runOnModule(Module &M) { CallGraph &CG = getAnalysis<CallGraph>(); CallGraphNode *ECN = CG.getExternalCallingNode(); @@ -164,7 +164,7 @@ bool PIC16FrameOverlay::runOnModule(Module &M) { return false; } -void PIC16FrameOverlay::MarkIndirectlyCalledFunctions(Module &M) { +void PIC16Overlay::MarkIndirectlyCalledFunctions(Module &M) { // If the use of a function is not a call instruction then this // function might be called indirectly. In that case give it // an unique color. diff --git a/lib/Target/PIC16/PIC16Passes/PIC16Overlay.h b/lib/Target/PIC16/PIC16Passes/PIC16Overlay.h index d70c4e7..5a2551f 100644 --- a/lib/Target/PIC16/PIC16Passes/PIC16Overlay.h +++ b/lib/Target/PIC16/PIC16Passes/PIC16Overlay.h @@ -1,4 +1,4 @@ -//===-- PIC16FrameOverlay.h - Interface for PIC16 Frame Overlay -*- C++ -*-===// +//===-- PIC16Overlay.h - Interface for PIC16 Frame Overlay -*- C++ -*-===// // // The LLVM Compiler Infrastructure // @@ -14,30 +14,35 @@ #ifndef PIC16FRAMEOVERLAY_H #define PIC16FRAMEOVERLAY_H -#include "llvm/Analysis/CallGraph.h" -#include "llvm/Pass.h" -#include "llvm/CallGraphSCCPass.h" using std::string; using namespace llvm; namespace llvm { - namespace PIC16Overlay { + // Forward declarations. + class Function; + class Module; + class ModulePass; + class AnalysisUsage; + class CallGraphNode; + class CallGraph; + + namespace PIC16OVERLAY { enum OverlayConsts { StartInterruptColor = 200, StartIndirectCallColor = 300 }; } - class PIC16FrameOverlay : public ModulePass { + class PIC16Overlay : public ModulePass { std::string OverlayStr; unsigned InterruptDepth; unsigned IndirectCallColor; public: static char ID; // Class identification - PIC16FrameOverlay() : ModulePass(&ID) { + PIC16Overlay() : ModulePass(&ID) { OverlayStr = "Overlay="; - InterruptDepth = PIC16Overlay::StartInterruptColor; - IndirectCallColor = PIC16Overlay::StartIndirectCallColor; + InterruptDepth = PIC16OVERLAY::StartInterruptColor; + IndirectCallColor = PIC16OVERLAY::StartIndirectCallColor; } virtual void getAnalysisUsage(AnalysisUsage &AU) const; diff --git a/lib/Target/PowerPC/PPCHazardRecognizers.cpp b/lib/Target/PowerPC/PPCHazardRecognizers.cpp index 3a15f7e..66dfd4b 100644 --- a/lib/Target/PowerPC/PPCHazardRecognizers.cpp +++ b/lib/Target/PowerPC/PPCHazardRecognizers.cpp @@ -257,7 +257,7 @@ void PPCHazardRecognizer970::EmitInstruction(SUnit *SU) { case PPC::STWX: case PPC::STWX8: case PPC::STWUX: case PPC::STW: case PPC::STW8: - case PPC::STWU: case PPC::STWU8: + case PPC::STWU: case PPC::STVEWX: case PPC::STFIWX: case PPC::STWBRX: diff --git a/lib/Target/PowerPC/PPCISelDAGToDAG.cpp b/lib/Target/PowerPC/PPCISelDAGToDAG.cpp index 004997f..9d79c0d 100644 --- a/lib/Target/PowerPC/PPCISelDAGToDAG.cpp +++ b/lib/Target/PowerPC/PPCISelDAGToDAG.cpp @@ -156,10 +156,6 @@ namespace { SDValue BuildSDIVSequence(SDNode *N); SDValue BuildUDIVSequence(SDNode *N); - /// InstructionSelect - This callback is invoked by - /// SelectionDAGISel when it has created a SelectionDAG for us to codegen. - virtual void InstructionSelect(); - void InsertVRSaveCode(MachineFunction &MF); virtual const char *getPassName() const { @@ -184,14 +180,6 @@ private: }; } -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void PPCDAGToDAGISel::InstructionSelect() { - // Select target instructions for the DAG. - SelectRoot(*CurDAG); - CurDAG->RemoveDeadNodes(); -} - /// InsertVRSaveCode - Once the entire function has been instruction selected, /// all virtual registers are created and all machine instructions are built, /// check to see if we need to save/restore VRSAVE. If so, do it. diff --git a/lib/Target/PowerPC/PPCISelLowering.cpp b/lib/Target/PowerPC/PPCISelLowering.cpp index e73af56..3d81afa 100644 --- a/lib/Target/PowerPC/PPCISelLowering.cpp +++ b/lib/Target/PowerPC/PPCISelLowering.cpp @@ -1334,7 +1334,7 @@ SDValue PPCTargetLowering::LowerTRAMPOLINE(SDValue Op, SelectionDAG &DAG) { false, false, false, false, 0, CallingConv::C, false, /*isReturnValueUsed=*/true, DAG.getExternalSymbol("__trampoline_setup", PtrVT), - Args, DAG, dl, DAG.GetOrdering(Chain.getNode())); + Args, DAG, dl); SDValue Ops[] = { CallResult.first, CallResult.second }; diff --git a/lib/Target/PowerPC/PPCInstr64Bit.td b/lib/Target/PowerPC/PPCInstr64Bit.td index 219efb9..a0781b9 100644 --- a/lib/Target/PowerPC/PPCInstr64Bit.td +++ b/lib/Target/PowerPC/PPCInstr64Bit.td @@ -366,7 +366,7 @@ def ADDE8 : XOForm_1<31, 138, 0, (outs G8RC:$rT), (ins G8RC:$rA, G8RC:$rB), [(set G8RC:$rT, (adde G8RC:$rA, G8RC:$rB))]>; def ADDME8 : XOForm_3<31, 234, 0, (outs G8RC:$rT), (ins G8RC:$rA), "addme $rT, $rA", IntGeneral, - [(set G8RC:$rT, (adde G8RC:$rA, immAllOnes))]>; + [(set G8RC:$rT, (adde G8RC:$rA, -1))]>; def ADDZE8 : XOForm_3<31, 202, 0, (outs G8RC:$rT), (ins G8RC:$rA), "addze $rT, $rA", IntGeneral, [(set G8RC:$rT, (adde G8RC:$rA, 0))]>; @@ -375,7 +375,7 @@ def SUBFE8 : XOForm_1<31, 136, 0, (outs G8RC:$rT), (ins G8RC:$rA, G8RC:$rB), [(set G8RC:$rT, (sube G8RC:$rB, G8RC:$rA))]>; def SUBFME8 : XOForm_3<31, 232, 0, (outs G8RC:$rT), (ins G8RC:$rA), "subfme $rT, $rA", IntGeneral, - [(set G8RC:$rT, (sube immAllOnes, G8RC:$rA))]>; + [(set G8RC:$rT, (sube -1, G8RC:$rA))]>; def SUBFZE8 : XOForm_3<31, 200, 0, (outs G8RC:$rT), (ins G8RC:$rA), "subfze $rT, $rA", IntGeneral, [(set G8RC:$rT, (sube 0, G8RC:$rA))]>; @@ -635,13 +635,6 @@ def STHU8 : DForm_1<45, (outs ptr_rc:$ea_res), (ins G8RC:$rS, (pre_truncsti16 G8RC:$rS, ptr_rc:$ptrreg, iaddroff:$ptroff))]>, RegConstraint<"$ptrreg = $ea_res">, NoEncode<"$ea_res">; -def STWU8 : DForm_1<37, (outs ptr_rc:$ea_res), (ins G8RC:$rS, - symbolLo:$ptroff, ptr_rc:$ptrreg), - "stwu $rS, $ptroff($ptrreg)", LdStGeneral, - [(set ptr_rc:$ea_res, (pre_store G8RC:$rS, ptr_rc:$ptrreg, - iaddroff:$ptroff))]>, - RegConstraint<"$ptrreg = $ea_res">, NoEncode<"$ea_res">; - def STDU : DSForm_1<62, 1, (outs ptr_rc:$ea_res), (ins G8RC:$rS, s16immX4:$ptroff, ptr_rc:$ptrreg), diff --git a/lib/Target/PowerPC/PPCInstrInfo.cpp b/lib/Target/PowerPC/PPCInstrInfo.cpp index 3db623a..9895bea 100644 --- a/lib/Target/PowerPC/PPCInstrInfo.cpp +++ b/lib/Target/PowerPC/PPCInstrInfo.cpp @@ -19,6 +19,7 @@ #include "PPCTargetMachine.h" #include "llvm/ADT/STLExtras.h" #include "llvm/CodeGen/MachineInstrBuilder.h" +#include "llvm/CodeGen/MachineRegisterInfo.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Support/raw_ostream.h" @@ -73,8 +74,7 @@ bool PPCInstrInfo::isMoveInstr(const MachineInstr& MI, destReg = MI.getOperand(0).getReg(); return true; } - } else if (oc == PPC::FMRS || oc == PPC::FMRD || - oc == PPC::FMRSD) { // fmr r1, r2 + } else if (oc == PPC::FMR || oc == PPC::FMRSD) { // fmr r1, r2 assert(MI.getNumOperands() >= 2 && MI.getOperand(0).isReg() && MI.getOperand(1).isReg() && @@ -344,10 +344,9 @@ bool PPCInstrInfo::copyRegToReg(MachineBasicBlock &MBB, BuildMI(MBB, MI, DL, get(PPC::OR), DestReg).addReg(SrcReg).addReg(SrcReg); } else if (DestRC == PPC::G8RCRegisterClass) { BuildMI(MBB, MI, DL, get(PPC::OR8), DestReg).addReg(SrcReg).addReg(SrcReg); - } else if (DestRC == PPC::F4RCRegisterClass) { - BuildMI(MBB, MI, DL, get(PPC::FMRS), DestReg).addReg(SrcReg); - } else if (DestRC == PPC::F8RCRegisterClass) { - BuildMI(MBB, MI, DL, get(PPC::FMRD), DestReg).addReg(SrcReg); + } else if (DestRC == PPC::F4RCRegisterClass || + DestRC == PPC::F8RCRegisterClass) { + BuildMI(MBB, MI, DL, get(PPC::FMR), DestReg).addReg(SrcReg); } else if (DestRC == PPC::CRRCRegisterClass) { BuildMI(MBB, MI, DL, get(PPC::MCRF), DestReg).addReg(SrcReg); } else if (DestRC == PPC::VRRCRegisterClass) { @@ -688,33 +687,21 @@ MachineInstr *PPCInstrInfo::foldMemoryOperandImpl(MachineFunction &MF, getUndefRegState(isUndef)), FrameIndex); } - } else if (Opc == PPC::FMRD) { - if (OpNum == 0) { // move -> store - unsigned InReg = MI->getOperand(1).getReg(); - bool isKill = MI->getOperand(1).isKill(); - bool isUndef = MI->getOperand(1).isUndef(); - NewMI = addFrameReference(BuildMI(MF, MI->getDebugLoc(), get(PPC::STFD)) - .addReg(InReg, - getKillRegState(isKill) | - getUndefRegState(isUndef)), - FrameIndex); - } else { // move -> load - unsigned OutReg = MI->getOperand(0).getReg(); - bool isDead = MI->getOperand(0).isDead(); - bool isUndef = MI->getOperand(0).isUndef(); - NewMI = addFrameReference(BuildMI(MF, MI->getDebugLoc(), get(PPC::LFD)) - .addReg(OutReg, - RegState::Define | - getDeadRegState(isDead) | - getUndefRegState(isUndef)), - FrameIndex); - } - } else if (Opc == PPC::FMRS) { + } else if (Opc == PPC::FMR || Opc == PPC::FMRSD) { + // The register may be F4RC or F8RC, and that determines the memory op. + unsigned OrigReg = MI->getOperand(OpNum).getReg(); + // We cannot tell the register class from a physreg alone. + if (TargetRegisterInfo::isPhysicalRegister(OrigReg)) + return NULL; + const TargetRegisterClass *RC = MF.getRegInfo().getRegClass(OrigReg); + const bool is64 = RC == PPC::F8RCRegisterClass; + if (OpNum == 0) { // move -> store unsigned InReg = MI->getOperand(1).getReg(); bool isKill = MI->getOperand(1).isKill(); bool isUndef = MI->getOperand(1).isUndef(); - NewMI = addFrameReference(BuildMI(MF, MI->getDebugLoc(), get(PPC::STFS)) + NewMI = addFrameReference(BuildMI(MF, MI->getDebugLoc(), + get(is64 ? PPC::STFD : PPC::STFS)) .addReg(InReg, getKillRegState(isKill) | getUndefRegState(isUndef)), @@ -723,7 +710,8 @@ MachineInstr *PPCInstrInfo::foldMemoryOperandImpl(MachineFunction &MF, unsigned OutReg = MI->getOperand(0).getReg(); bool isDead = MI->getOperand(0).isDead(); bool isUndef = MI->getOperand(0).isUndef(); - NewMI = addFrameReference(BuildMI(MF, MI->getDebugLoc(), get(PPC::LFS)) + NewMI = addFrameReference(BuildMI(MF, MI->getDebugLoc(), + get(is64 ? PPC::LFD : PPC::LFS)) .addReg(OutReg, RegState::Define | getDeadRegState(isDead) | @@ -749,7 +737,7 @@ bool PPCInstrInfo::canFoldMemoryOperand(const MachineInstr *MI, else if ((Opc == PPC::OR8 && MI->getOperand(1).getReg() == MI->getOperand(2).getReg())) return true; - else if (Opc == PPC::FMRD || Opc == PPC::FMRS) + else if (Opc == PPC::FMR || Opc == PPC::FMRSD) return true; return false; diff --git a/lib/Target/PowerPC/PPCInstrInfo.td b/lib/Target/PowerPC/PPCInstrInfo.td index 842f8ee..845cd8f 100644 --- a/lib/Target/PowerPC/PPCInstrInfo.td +++ b/lib/Target/PowerPC/PPCInstrInfo.td @@ -1019,20 +1019,16 @@ let Uses = [RM] in { } } -/// FMR is split into 3 versions, one for 4/8 byte FP, and one for extending. +/// FMR is split into 2 versions, one for 4/8 byte FP, and one for extending. /// /// Note that these are defined as pseudo-ops on the PPC970 because they are /// often coalesced away and we don't want the dispatch group builder to think /// that they will fill slots (which could cause the load of a LSU reject to /// sneak into a d-group with a store). -def FMRS : XForm_26<63, 72, (outs F4RC:$frD), (ins F4RC:$frB), - "fmr $frD, $frB", FPGeneral, - []>, // (set F4RC:$frD, F4RC:$frB) - PPC970_Unit_Pseudo; -def FMRD : XForm_26<63, 72, (outs F8RC:$frD), (ins F8RC:$frB), - "fmr $frD, $frB", FPGeneral, - []>, // (set F8RC:$frD, F8RC:$frB) - PPC970_Unit_Pseudo; +def FMR : XForm_26<63, 72, (outs F4RC:$frD), (ins F4RC:$frB), + "fmr $frD, $frB", FPGeneral, + []>, // (set F4RC:$frD, F4RC:$frB) + PPC970_Unit_Pseudo; def FMRSD : XForm_26<63, 72, (outs F8RC:$frD), (ins F4RC:$frB), "fmr $frD, $frB", FPGeneral, [(set F8RC:$frD, (fextend F4RC:$frB))]>, @@ -1215,7 +1211,7 @@ def ADDE : XOForm_1<31, 138, 0, (outs GPRC:$rT), (ins GPRC:$rA, GPRC:$rB), [(set GPRC:$rT, (adde GPRC:$rA, GPRC:$rB))]>; def ADDME : XOForm_3<31, 234, 0, (outs GPRC:$rT), (ins GPRC:$rA), "addme $rT, $rA", IntGeneral, - [(set GPRC:$rT, (adde GPRC:$rA, immAllOnes))]>; + [(set GPRC:$rT, (adde GPRC:$rA, -1))]>; def ADDZE : XOForm_3<31, 202, 0, (outs GPRC:$rT), (ins GPRC:$rA), "addze $rT, $rA", IntGeneral, [(set GPRC:$rT, (adde GPRC:$rA, 0))]>; @@ -1224,7 +1220,7 @@ def SUBFE : XOForm_1<31, 136, 0, (outs GPRC:$rT), (ins GPRC:$rA, GPRC:$rB), [(set GPRC:$rT, (sube GPRC:$rB, GPRC:$rA))]>; def SUBFME : XOForm_3<31, 232, 0, (outs GPRC:$rT), (ins GPRC:$rA), "subfme $rT, $rA", IntGeneral, - [(set GPRC:$rT, (sube immAllOnes, GPRC:$rA))]>; + [(set GPRC:$rT, (sube -1, GPRC:$rA))]>; def SUBFZE : XOForm_3<31, 200, 0, (outs GPRC:$rT), (ins GPRC:$rA), "subfze $rT, $rA", IntGeneral, [(set GPRC:$rT, (sube 0, GPRC:$rA))]>; @@ -1480,11 +1476,11 @@ def : Pat<(extloadf32 xaddr:$src), (FMRSD (LFSX xaddr:$src))>; // Memory barriers -def : Pat<(membarrier (i32 imm:$ll), - (i32 imm:$ls), - (i32 imm:$sl), - (i32 imm:$ss), - (i32 imm:$device)), +def : Pat<(membarrier (i32 imm /*ll*/), + (i32 imm /*ls*/), + (i32 imm /*sl*/), + (i32 imm /*ss*/), + (i32 imm /*device*/)), (SYNC)>; include "PPCInstrAltivec.td" diff --git a/lib/Target/PowerPC/PPCRegisterInfo.td b/lib/Target/PowerPC/PPCRegisterInfo.td index 049e893..1cb7340 100644 --- a/lib/Target/PowerPC/PPCRegisterInfo.td +++ b/lib/Target/PowerPC/PPCRegisterInfo.td @@ -287,10 +287,8 @@ def GPRC : RegisterClass<"PPC", [i32], 32, GPRCClass::allocation_order_begin(const MachineFunction &MF) const { // 32-bit SVR4 ABI: r2 is reserved for the OS. // 64-bit SVR4 ABI: r2 is reserved for the TOC pointer. - if (!MF.getTarget().getSubtarget<PPCSubtarget>().isDarwin()) - return begin()+1; - - return begin(); + // Darwin: R2 is reserved for CR save/restore sequence. + return begin()+1; } GPRCClass::iterator GPRCClass::allocation_order_end(const MachineFunction &MF) const { @@ -325,10 +323,8 @@ def G8RC : RegisterClass<"PPC", [i64], 64, G8RCClass::iterator G8RCClass::allocation_order_begin(const MachineFunction &MF) const { // 64-bit SVR4 ABI: r2 is reserved for the TOC pointer. - if (!MF.getTarget().getSubtarget<PPCSubtarget>().isDarwin()) - return begin()+1; - - return begin(); + // Darwin: r2 is reserved for CR save/restore sequence. + return begin()+1; } G8RCClass::iterator G8RCClass::allocation_order_end(const MachineFunction &MF) const { diff --git a/lib/Target/PowerPC/README.txt b/lib/Target/PowerPC/README.txt index e49bda0..3465779 100644 --- a/lib/Target/PowerPC/README.txt +++ b/lib/Target/PowerPC/README.txt @@ -889,7 +889,7 @@ entry: ; recognize a more elaborate tree than a simple SETxx. define double @test_FNEG_sel(double %A, double %B, double %C) { - %D = sub double -0.000000e+00, %A ; <double> [#uses=1] + %D = fsub double -0.000000e+00, %A ; <double> [#uses=1] %Cond = fcmp ugt double %D, -0.000000e+00 ; <i1> [#uses=1] %E = select i1 %Cond, double %B, double %C ; <double> [#uses=1] ret double %E diff --git a/lib/Target/Sparc/AsmPrinter/SparcAsmPrinter.cpp b/lib/Target/Sparc/AsmPrinter/SparcAsmPrinter.cpp index 9a2ce6b..d6b45be 100644 --- a/lib/Target/Sparc/AsmPrinter/SparcAsmPrinter.cpp +++ b/lib/Target/Sparc/AsmPrinter/SparcAsmPrinter.cpp @@ -56,6 +56,9 @@ namespace { unsigned AsmVariant, const char *ExtraCode); bool printGetPCX(const MachineInstr *MI, unsigned OpNo); + + virtual bool isBlockOnlyReachableByFallthrough(const MachineBasicBlock *MBB) + const; }; } // end of anonymous namespace @@ -140,18 +143,19 @@ bool SparcAsmPrinter::printGetPCX(const MachineInstr *MI, unsigned opNum) { break; } + unsigned mfNum = MI->getParent()->getParent()->getFunctionNumber(); unsigned bbNum = MI->getParent()->getNumber(); - O << '\n' << ".LLGETPCH" << bbNum << ":\n"; - O << "\tcall\t.LLGETPC" << bbNum << '\n' ; + O << '\n' << ".LLGETPCH" << mfNum << '_' << bbNum << ":\n"; + O << "\tcall\t.LLGETPC" << mfNum << '_' << bbNum << '\n' ; O << "\t sethi\t" - << "%hi(_GLOBAL_OFFSET_TABLE_+(.-.LLGETPCH" << bbNum << ")), " + << "%hi(_GLOBAL_OFFSET_TABLE_+(.-.LLGETPCH" << mfNum << '_' << bbNum << ")), " << operand << '\n' ; - O << ".LLGETPC" << bbNum << ":\n" ; + O << ".LLGETPC" << mfNum << '_' << bbNum << ":\n" ; O << "\tor\t" << operand - << ", %lo(_GLOBAL_OFFSET_TABLE_+(.-.LLGETPCH" << bbNum << ")), " + << ", %lo(_GLOBAL_OFFSET_TABLE_+(.-.LLGETPCH" << mfNum << '_' << bbNum << ")), " << operand << '\n'; O << "\tadd\t" << operand << ", %o7, " << operand << '\n'; @@ -197,6 +201,38 @@ bool SparcAsmPrinter::PrintAsmMemoryOperand(const MachineInstr *MI, return false; } +/// isBlockOnlyReachableByFallthough - Return true if the basic block has +/// exactly one predecessor and the control transfer mechanism between +/// the predecessor and this block is a fall-through. +/// Override AsmPrinter implementation to handle delay slots +bool SparcAsmPrinter::isBlockOnlyReachableByFallthrough(const MachineBasicBlock *MBB) + const { + // If this is a landing pad, it isn't a fall through. If it has no preds, + // then nothing falls through to it. + if (MBB->isLandingPad() || MBB->pred_empty()) + return false; + + // If there isn't exactly one predecessor, it can't be a fall through. + MachineBasicBlock::const_pred_iterator PI = MBB->pred_begin(), PI2 = PI; + ++PI2; + if (PI2 != MBB->pred_end()) + return false; + + // The predecessor has to be immediately before this block. + const MachineBasicBlock *Pred = *PI; + + if (!Pred->isLayoutSuccessor(MBB)) + return false; + + // Check if the last terminator is an unconditional branch + MachineBasicBlock::const_iterator I = Pred->end(); + while( I != Pred->begin() && !(--I)->getDesc().isTerminator() ) + ; /* Noop */ + return I == Pred->end() || !I->getDesc().isBarrier(); +} + + + // Force static initialization. extern "C" void LLVMInitializeSparcAsmPrinter() { RegisterAsmPrinter<SparcAsmPrinter> X(TheSparcTarget); diff --git a/lib/Target/Sparc/README.txt b/lib/Target/Sparc/README.txt index cc24abf..b4991fe 100644 --- a/lib/Target/Sparc/README.txt +++ b/lib/Target/Sparc/README.txt @@ -56,3 +56,4 @@ int %t1(int %a, int %b) { leaf fns. * Fill delay slots +* Implement JIT support diff --git a/lib/Target/Sparc/SparcISelDAGToDAG.cpp b/lib/Target/Sparc/SparcISelDAGToDAG.cpp index e1b3299..a7d1805 100644 --- a/lib/Target/Sparc/SparcISelDAGToDAG.cpp +++ b/lib/Target/Sparc/SparcISelDAGToDAG.cpp @@ -35,7 +35,6 @@ class SparcDAGToDAGISel : public SelectionDAGISel { /// make the right decision when generating code for different targets. const SparcSubtarget &Subtarget; SparcTargetMachine& TM; - MachineBasicBlock *CurBB; public: explicit SparcDAGToDAGISel(SparcTargetMachine &tm) : SelectionDAGISel(tm), @@ -56,10 +55,6 @@ public: char ConstraintCode, std::vector<SDValue> &OutOps); - /// InstructionSelect - This callback is invoked by - /// SelectionDAGISel when it has created a SelectionDAG for us to codegen. - virtual void InstructionSelect(); - virtual const char *getPassName() const { return "SPARC DAG->DAG Pattern Instruction Selection"; } @@ -72,17 +67,8 @@ private: }; } // end anonymous namespace -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void SparcDAGToDAGISel::InstructionSelect() { - CurBB = BB; - // Select target instructions for the DAG. - SelectRoot(*CurDAG); - CurDAG->RemoveDeadNodes(); -} - SDNode* SparcDAGToDAGISel::getGlobalBaseReg() { - MachineFunction *MF = CurBB->getParent(); + MachineFunction *MF = BB->getParent(); unsigned GlobalBaseReg = TM.getInstrInfo()->getGlobalBaseReg(MF); return CurDAG->getRegister(GlobalBaseReg, TLI.getPointerTy()).getNode(); } diff --git a/lib/Target/Sparc/SparcMachineFunctionInfo.h b/lib/Target/Sparc/SparcMachineFunctionInfo.h index e457235..56d8708 100644 --- a/lib/Target/Sparc/SparcMachineFunctionInfo.h +++ b/lib/Target/Sparc/SparcMachineFunctionInfo.h @@ -22,7 +22,7 @@ namespace llvm { unsigned GlobalBaseReg; public: SparcMachineFunctionInfo() : GlobalBaseReg(0) {} - SparcMachineFunctionInfo(MachineFunction &MF) : GlobalBaseReg(0) {} + explicit SparcMachineFunctionInfo(MachineFunction &MF) : GlobalBaseReg(0) {} unsigned getGlobalBaseReg() const { return GlobalBaseReg; } void setGlobalBaseReg(unsigned Reg) { GlobalBaseReg = Reg; } diff --git a/lib/Target/SystemZ/SystemZISelDAGToDAG.cpp b/lib/Target/SystemZ/SystemZISelDAGToDAG.cpp index 7f0d9fb..8152e1d 100644 --- a/lib/Target/SystemZ/SystemZISelDAGToDAG.cpp +++ b/lib/Target/SystemZ/SystemZISelDAGToDAG.cpp @@ -100,8 +100,6 @@ namespace { Lowering(*TM.getTargetLowering()), Subtarget(*TM.getSubtargetImpl()) { } - virtual void InstructionSelect(); - virtual const char *getPassName() const { return "SystemZ DAG->DAG Pattern Instruction Selection"; } @@ -152,10 +150,6 @@ namespace { bool MatchAddressBase(SDValue N, SystemZRRIAddressMode &AM); bool MatchAddressRI(SDValue N, SystemZRRIAddressMode &AM, bool is12Bit); - - #ifndef NDEBUG - unsigned Indent; - #endif }; } // end anonymous namespace @@ -599,35 +593,17 @@ bool SystemZDAGToDAGISel::TryFoldLoad(SDNode *P, SDValue N, return false; } -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void SystemZDAGToDAGISel::InstructionSelect() { - // Codegen the basic block. - DEBUG(errs() << "===== Instruction selection begins:\n"); - DEBUG(Indent = 0); - SelectRoot(*CurDAG); - DEBUG(errs() << "===== Instruction selection ends:\n"); - - CurDAG->RemoveDeadNodes(); -} - SDNode *SystemZDAGToDAGISel::Select(SDNode *Node) { EVT NVT = Node->getValueType(0); DebugLoc dl = Node->getDebugLoc(); unsigned Opcode = Node->getOpcode(); // Dump information about the Node being selected - DEBUG(errs().indent(Indent) << "Selecting: "; - Node->dump(CurDAG); - errs() << "\n"); - DEBUG(Indent += 2); + DEBUG(errs() << "Selecting: "; Node->dump(CurDAG); errs() << "\n"); // If we have a custom node, we already have selected! if (Node->isMachineOpcode()) { - DEBUG(errs().indent(Indent-2) << "== "; - Node->dump(CurDAG); - errs() << "\n"); - DEBUG(Indent -= 2); + DEBUG(errs() << "== "; Node->dump(CurDAG); errs() << "\n"); return NULL; // Already selected. } @@ -693,9 +669,7 @@ SDNode *SystemZDAGToDAGISel::Select(SDNode *Node) { MVT::i32)); ReplaceUses(SDValue(Node, 0), SDValue(Div, 0)); - DEBUG(errs().indent(Indent-2) << "=> "; - Result->dump(CurDAG); - errs() << "\n"); + DEBUG(errs() << "=> "; Result->dump(CurDAG); errs() << "\n"); } // Copy the remainder (even subreg) result, if it is needed. @@ -708,15 +682,9 @@ SDNode *SystemZDAGToDAGISel::Select(SDNode *Node) { MVT::i32)); ReplaceUses(SDValue(Node, 1), SDValue(Rem, 0)); - DEBUG(errs().indent(Indent-2) << "=> "; - Result->dump(CurDAG); - errs() << "\n"); + DEBUG(errs() << "=> "; Result->dump(CurDAG); errs() << "\n"); } -#ifndef NDEBUG - Indent -= 2; -#endif - return NULL; } case ISD::UDIVREM: { @@ -782,9 +750,7 @@ SDNode *SystemZDAGToDAGISel::Select(SDNode *Node) { CurDAG->getTargetConstant(SubRegIdx, MVT::i32)); ReplaceUses(SDValue(Node, 0), SDValue(Div, 0)); - DEBUG(errs().indent(Indent-2) << "=> "; - Result->dump(CurDAG); - errs() << "\n"); + DEBUG(errs() << "=> "; Result->dump(CurDAG); errs() << "\n"); } // Copy the remainder (even subreg) result, if it is needed. @@ -796,15 +762,9 @@ SDNode *SystemZDAGToDAGISel::Select(SDNode *Node) { CurDAG->getTargetConstant(SubRegIdx, MVT::i32)); ReplaceUses(SDValue(Node, 1), SDValue(Rem, 0)); - DEBUG(errs().indent(Indent-2) << "=> "; - Result->dump(CurDAG); - errs() << "\n"); + DEBUG(errs() << "=> "; Result->dump(CurDAG); errs() << "\n"); } -#ifndef NDEBUG - Indent -= 2; -#endif - return NULL; } } @@ -812,14 +772,12 @@ SDNode *SystemZDAGToDAGISel::Select(SDNode *Node) { // Select the default instruction SDNode *ResNode = SelectCode(Node); - DEBUG(errs().indent(Indent-2) << "=> "; + DEBUG(errs() << "=> "; if (ResNode == NULL || ResNode == Node) Node->dump(CurDAG); else ResNode->dump(CurDAG); errs() << "\n"; ); - DEBUG(Indent -= 2); - return ResNode; } diff --git a/lib/Target/SystemZ/SystemZInstrFP.td b/lib/Target/SystemZ/SystemZInstrFP.td index 336e20e..f46840c 100644 --- a/lib/Target/SystemZ/SystemZInstrFP.td +++ b/lib/Target/SystemZ/SystemZInstrFP.td @@ -58,7 +58,7 @@ def FMOV64rr : Pseudo<(outs FP64:$dst), (ins FP64:$src), []>; } -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in { +let canFoldAsLoad = 1, isReMaterializable = 1 in { def FMOV32rm : Pseudo<(outs FP32:$dst), (ins rriaddr12:$src), "le\t{$dst, $src}", [(set FP32:$dst, (load rriaddr12:$src))]>; diff --git a/lib/Target/SystemZ/SystemZInstrInfo.td b/lib/Target/SystemZ/SystemZInstrInfo.td index 1891bba..a44f6d9 100644 --- a/lib/Target/SystemZ/SystemZInstrInfo.td +++ b/lib/Target/SystemZ/SystemZInstrInfo.td @@ -257,7 +257,7 @@ def MOV64rihi32 : RILI<0xEC0, (outs GR64:$dst), (ins i64imm:$src), [(set GR64:$dst, i64hi32:$src)]>; } -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in { +let canFoldAsLoad = 1, isReMaterializable = 1 in { def MOV32rm : RXI<0x58, (outs GR32:$dst), (ins rriaddr12:$src), "l\t{$dst, $src}", diff --git a/lib/Target/SystemZ/SystemZMachineFunctionInfo.h b/lib/Target/SystemZ/SystemZMachineFunctionInfo.h index e47d419..fd6e330 100644 --- a/lib/Target/SystemZ/SystemZMachineFunctionInfo.h +++ b/lib/Target/SystemZ/SystemZMachineFunctionInfo.h @@ -33,7 +33,8 @@ class SystemZMachineFunctionInfo : public MachineFunctionInfo { public: SystemZMachineFunctionInfo() : CalleeSavedFrameSize(0) {} - SystemZMachineFunctionInfo(MachineFunction &MF) : CalleeSavedFrameSize(0) {} + explicit SystemZMachineFunctionInfo(MachineFunction &MF) + : CalleeSavedFrameSize(0) {} unsigned getCalleeSavedFrameSize() const { return CalleeSavedFrameSize; } void setCalleeSavedFrameSize(unsigned bytes) { CalleeSavedFrameSize = bytes; } diff --git a/lib/Target/TargetData.cpp b/lib/Target/TargetData.cpp index 295b30f..9a16808 100644 --- a/lib/Target/TargetData.cpp +++ b/lib/Target/TargetData.cpp @@ -580,7 +580,7 @@ const IntegerType *TargetData::getIntPtrType(LLVMContext &C) const { uint64_t TargetData::getIndexedOffset(const Type *ptrTy, Value* const* Indices, unsigned NumIndices) const { const Type *Ty = ptrTy; - assert(isa<PointerType>(Ty) && "Illegal argument for getIndexedOffset()"); + assert(Ty->isPointerTy() && "Illegal argument for getIndexedOffset()"); uint64_t Result = 0; generic_gep_type_iterator<Value* const*> diff --git a/lib/Target/TargetLoweringObjectFile.cpp b/lib/Target/TargetLoweringObjectFile.cpp index 0c105e9..82619c7 100644 --- a/lib/Target/TargetLoweringObjectFile.cpp +++ b/lib/Target/TargetLoweringObjectFile.cpp @@ -307,7 +307,7 @@ getSymbolForDwarfReference(const MCSymbol *Sym, MachineModuleInfo *MMI, switch (Encoding & 0xF0) { default: - llvm_report_error("Do not support this DWARF encoding yet!"); + llvm_report_error("We do not support this DWARF encoding yet!"); break; case dwarf::DW_EH_PE_absptr: // Do nothing special diff --git a/lib/Target/X86/AsmPrinter/X86ATTInstPrinter.cpp b/lib/Target/X86/AsmPrinter/X86ATTInstPrinter.cpp index 1a35a49..734a545 100644 --- a/lib/Target/X86/AsmPrinter/X86ATTInstPrinter.cpp +++ b/lib/Target/X86/AsmPrinter/X86ATTInstPrinter.cpp @@ -73,7 +73,7 @@ void X86ATTInstPrinter::printOperand(const MCInst *MI, unsigned OpNo) { O << '$' << Op.getImm(); if (CommentStream && (Op.getImm() > 255 || Op.getImm() < -256)) - *CommentStream << format("imm = 0x%X\n", Op.getImm()); + *CommentStream << format("imm = 0x%llX\n", (long long)Op.getImm()); } else { assert(Op.isExpr() && "unknown operand kind in printOperand"); diff --git a/lib/Target/X86/CMakeLists.txt b/lib/Target/X86/CMakeLists.txt index 61f26a7..eed3b45 100644 --- a/lib/Target/X86/CMakeLists.txt +++ b/lib/Target/X86/CMakeLists.txt @@ -15,6 +15,7 @@ tablegen(X86GenCallingConv.inc -gen-callingconv) tablegen(X86GenSubtarget.inc -gen-subtarget) set(sources + X86AsmBackend.cpp X86CodeEmitter.cpp X86COFFMachineModuleInfo.cpp X86ELFWriterInfo.cpp diff --git a/lib/Target/X86/README-SSE.txt b/lib/Target/X86/README-SSE.txt index 19eb05e..e5f84e8 100644 --- a/lib/Target/X86/README-SSE.txt +++ b/lib/Target/X86/README-SSE.txt @@ -67,8 +67,8 @@ no_exit.i7: ; preds = %no_exit.i7, %build_tree.exit [ %tmp.34.i18, %no_exit.i7 ] %tmp.0.0.0.i10 = phi double [ 0.000000e+00, %build_tree.exit ], [ %tmp.28.i16, %no_exit.i7 ] - %tmp.28.i16 = add double %tmp.0.0.0.i10, 0.000000e+00 - %tmp.34.i18 = add double %tmp.0.1.0.i9, 0.000000e+00 + %tmp.28.i16 = fadd double %tmp.0.0.0.i10, 0.000000e+00 + %tmp.34.i18 = fadd double %tmp.0.1.0.i9, 0.000000e+00 br i1 false, label %Compute_Tree.exit23, label %no_exit.i7 Compute_Tree.exit23: ; preds = %no_exit.i7 @@ -97,7 +97,7 @@ pcmp/pand/pandn/por to do a selection instead of a conditional branch: double %X(double %Y, double %Z, double %A, double %B) { %C = setlt double %A, %B - %z = add double %Z, 0.0 ;; select operand is not a load + %z = fadd double %Z, 0.0 ;; select operand is not a load %D = select bool %C, double %Y, double %z ret double %D } @@ -545,7 +545,7 @@ eliminates a constant pool load. For example, consider: define i64 @ccosf(float %z.0, float %z.1) nounwind readonly { entry: - %tmp6 = sub float -0.000000e+00, %z.1 ; <float> [#uses=1] + %tmp6 = fsub float -0.000000e+00, %z.1 ; <float> [#uses=1] %tmp20 = tail call i64 @ccoshf( float %tmp6, float %z.0 ) nounwind readonly ret i64 %tmp20 } diff --git a/lib/Target/X86/README.txt b/lib/Target/X86/README.txt index 3c6138b..d4545a6 100644 --- a/lib/Target/X86/README.txt +++ b/lib/Target/X86/README.txt @@ -227,11 +227,6 @@ lambda, siod, optimizer-eval, ackermann, hash2, nestedloop, strcat, and Treesor. //===---------------------------------------------------------------------===// -Teach the coalescer to coalesce vregs of different register classes. e.g. FR32 / -FR64 to VR128. - -//===---------------------------------------------------------------------===// - Adding to the list of cmp / test poor codegen issues: int test(__m128 *A, __m128 *B) { diff --git a/lib/Target/X86/X86.h b/lib/Target/X86/X86.h index 1a1e447..ba0ee6c 100644 --- a/lib/Target/X86/X86.h +++ b/lib/Target/X86/X86.h @@ -19,13 +19,15 @@ namespace llvm { -class X86TargetMachine; class FunctionPass; -class MachineCodeEmitter; +class JITCodeEmitter; +class MCAssembler; class MCCodeEmitter; class MCContext; -class JITCodeEmitter; +class MachineCodeEmitter; class Target; +class TargetAsmBackend; +class X86TargetMachine; class formatted_raw_ostream; /// createX86ISelDag - This pass converts a legalized DAG into a @@ -55,6 +57,9 @@ MCCodeEmitter *createX86_32MCCodeEmitter(const Target &, TargetMachine &TM, MCCodeEmitter *createX86_64MCCodeEmitter(const Target &, TargetMachine &TM, MCContext &Ctx); +TargetAsmBackend *createX86_32AsmBackend(const Target &, MCAssembler &); +TargetAsmBackend *createX86_64AsmBackend(const Target &, MCAssembler &); + /// createX86EmitCodeToMemory - Returns a pass that converts a register /// allocated function into raw machine code in a dynamically /// allocated chunk of memory. diff --git a/lib/Target/X86/X86AsmBackend.cpp b/lib/Target/X86/X86AsmBackend.cpp new file mode 100644 index 0000000..e6654ef --- /dev/null +++ b/lib/Target/X86/X86AsmBackend.cpp @@ -0,0 +1,34 @@ +//===-- X86AsmBackend.cpp - X86 Assembler Backend -------------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "llvm/Target/TargetAsmBackend.h" +#include "X86.h" +#include "llvm/Target/TargetRegistry.h" +#include "llvm/Target/TargetAsmBackend.h" +using namespace llvm; + +namespace { + +class X86AsmBackend : public TargetAsmBackend { +public: + X86AsmBackend(const Target &T, MCAssembler &A) + : TargetAsmBackend(T) {} +}; + +} + +TargetAsmBackend *llvm::createX86_32AsmBackend(const Target &T, + MCAssembler &A) { + return new X86AsmBackend(T, A); +} + +TargetAsmBackend *llvm::createX86_64AsmBackend(const Target &T, + MCAssembler &A) { + return new X86AsmBackend(T, A); +} diff --git a/lib/Target/X86/X86FastISel.cpp b/lib/Target/X86/X86FastISel.cpp index 69a9d60..17366ee 100644 --- a/lib/Target/X86/X86FastISel.cpp +++ b/lib/Target/X86/X86FastISel.cpp @@ -1161,6 +1161,8 @@ bool X86FastISel::X86VisitIntrinsicCall(IntrinsicInst &I) { if (!X86SelectAddress(DI->getAddress(), AM)) return false; const TargetInstrDesc &II = TII.get(TargetOpcode::DBG_VALUE); + // FIXME may need to add RegState::Debug to any registers produced, + // although ESP/EBP should be the only ones at the moment. addFullAddress(BuildMI(MBB, DL, II), AM).addImm(0). addMetadata(DI->getVariable()); return true; diff --git a/lib/Target/X86/X86ISelDAGToDAG.cpp b/lib/Target/X86/X86ISelDAGToDAG.cpp index 7b349f6..08030e0 100644 --- a/lib/Target/X86/X86ISelDAGToDAG.cpp +++ b/lib/Target/X86/X86ISelDAGToDAG.cpp @@ -12,15 +12,6 @@ // //===----------------------------------------------------------------------===// -// Force NDEBUG on in any optimized build on Darwin. -// -// FIXME: This is a huge hack, to work around ridiculously awful compile times -// on this file with gcc-4.2 on Darwin, in Release mode. -#if (!defined(__llvm__) && defined(__APPLE__) && \ - defined(__OPTIMIZE__) && !defined(NDEBUG)) -#define NDEBUG -#endif - #define DEBUG_TYPE "x86-isel" #include "X86.h" #include "X86InstrBuilder.h" @@ -177,15 +168,11 @@ namespace { return "X86 DAG->DAG Instruction Selection"; } - /// InstructionSelect - This callback is invoked by - /// SelectionDAGISel when it has created a SelectionDAG for us to codegen. - virtual void InstructionSelect(); - virtual void EmitFunctionEntryCode(Function &Fn, MachineFunction &MF); virtual bool IsProfitableToFold(SDValue N, SDNode *U, SDNode *Root) const; - virtual bool IsLegalToFold(SDValue N, SDNode *U, SDNode *Root) const; + virtual void PreprocessISelDAG(); // Include the pieces autogenerated from the target description. #include "X86GenDAGISel.inc" @@ -209,18 +196,17 @@ namespace { SDValue &Scale, SDValue &Index, SDValue &Disp); bool SelectTLSADDRAddr(SDNode *Op, SDValue N, SDValue &Base, SDValue &Scale, SDValue &Index, SDValue &Disp); - bool SelectScalarSSELoad(SDNode *Op, SDValue Pred, - SDValue N, SDValue &Base, SDValue &Scale, + bool SelectScalarSSELoad(SDNode *Root, SDValue N, + SDValue &Base, SDValue &Scale, SDValue &Index, SDValue &Disp, SDValue &Segment, - SDValue &InChain, SDValue &OutChain); + SDValue &NodeWithChain); + bool TryFoldLoad(SDNode *P, SDValue N, SDValue &Base, SDValue &Scale, SDValue &Index, SDValue &Disp, SDValue &Segment); - void PreprocessForRMW(); - void PreprocessForFPConvert(); - + /// SelectInlineAsmMemoryOperand - Implement addressing mode selection for /// inline asm expressions. virtual bool SelectInlineAsmMemoryOperand(const SDValue &Op, @@ -296,10 +282,6 @@ namespace { const X86InstrInfo *getInstrInfo() { return getTargetMachine().getInstrInfo(); } - -#ifndef NDEBUG - unsigned Indent; -#endif }; } @@ -367,65 +349,6 @@ X86DAGToDAGISel::IsProfitableToFold(SDValue N, SDNode *U, SDNode *Root) const { return true; } - -bool X86DAGToDAGISel::IsLegalToFold(SDValue N, SDNode *U, SDNode *Root) const { - if (OptLevel == CodeGenOpt::None) return false; - - // Proceed to 'generic' cycle finder code - return SelectionDAGISel::IsLegalToFold(N, U, Root); -} - -/// MoveBelowTokenFactor - Replace TokenFactor operand with load's chain operand -/// and move load below the TokenFactor. Replace store's chain operand with -/// load's chain result. -static void MoveBelowTokenFactor(SelectionDAG *CurDAG, SDValue Load, - SDValue Store, SDValue TF) { - SmallVector<SDValue, 4> Ops; - for (unsigned i = 0, e = TF.getNode()->getNumOperands(); i != e; ++i) - if (Load.getNode() == TF.getOperand(i).getNode()) - Ops.push_back(Load.getOperand(0)); - else - Ops.push_back(TF.getOperand(i)); - SDValue NewTF = CurDAG->UpdateNodeOperands(TF, &Ops[0], Ops.size()); - SDValue NewLoad = CurDAG->UpdateNodeOperands(Load, NewTF, - Load.getOperand(1), - Load.getOperand(2)); - CurDAG->UpdateNodeOperands(Store, NewLoad.getValue(1), Store.getOperand(1), - Store.getOperand(2), Store.getOperand(3)); -} - -/// isRMWLoad - Return true if N is a load that's part of RMW sub-DAG. The -/// chain produced by the load must only be used by the store's chain operand, -/// otherwise this may produce a cycle in the DAG. -/// -static bool isRMWLoad(SDValue N, SDValue Chain, SDValue Address, - SDValue &Load) { - if (N.getOpcode() == ISD::BIT_CONVERT) { - if (!N.hasOneUse()) - return false; - N = N.getOperand(0); - } - - LoadSDNode *LD = dyn_cast<LoadSDNode>(N); - if (!LD || LD->isVolatile()) - return false; - if (LD->getAddressingMode() != ISD::UNINDEXED) - return false; - - ISD::LoadExtType ExtType = LD->getExtensionType(); - if (ExtType != ISD::NON_EXTLOAD && ExtType != ISD::EXTLOAD) - return false; - - if (N.hasOneUse() && - LD->hasNUsesOfValue(1, 1) && - N.getOperand(1) == Address && - LD->isOperandOf(Chain.getNode())) { - Load = N; - return true; - } - return false; -} - /// MoveBelowCallSeqStart - Replace CALLSEQ_START operand with load's chain /// operand and move load below the call's chain operand. static void MoveBelowCallSeqStart(SelectionDAG *CurDAG, SDValue Load, @@ -489,51 +412,14 @@ static bool isCalleeLoad(SDValue Callee, SDValue &Chain) { return false; } - -/// PreprocessForRMW - Preprocess the DAG to make instruction selection better. -/// This is only run if not in -O0 mode. -/// This allows the instruction selector to pick more read-modify-write -/// instructions. This is a common case: -/// -/// [Load chain] -/// ^ -/// | -/// [Load] -/// ^ ^ -/// | | -/// / \- -/// / | -/// [TokenFactor] [Op] -/// ^ ^ -/// | | -/// \ / -/// \ / -/// [Store] -/// -/// The fact the store's chain operand != load's chain will prevent the -/// (store (op (load))) instruction from being selected. We can transform it to: -/// -/// [Load chain] -/// ^ -/// | -/// [TokenFactor] -/// ^ -/// | -/// [Load] -/// ^ ^ -/// | | -/// | \- -/// | | -/// | [Op] -/// | ^ -/// | | -/// \ / -/// \ / -/// [Store] -void X86DAGToDAGISel::PreprocessForRMW() { +void X86DAGToDAGISel::PreprocessISelDAG() { + OptForSize = MF->getFunction()->hasFnAttr(Attribute::OptimizeForSize); + for (SelectionDAG::allnodes_iterator I = CurDAG->allnodes_begin(), - E = CurDAG->allnodes_end(); I != E; ++I) { - if (I->getOpcode() == X86ISD::CALL) { + E = CurDAG->allnodes_end(); I != E; ) { + SDNode *N = I++; // Preincrement iterator to avoid invalidation issues. + + if (OptLevel != CodeGenOpt::None && N->getOpcode() == X86ISD::CALL) { /// Also try moving call address load from outside callseq_start to just /// before the call to allow it to be folded. /// @@ -553,85 +439,23 @@ void X86DAGToDAGISel::PreprocessForRMW() { /// \ / /// \ / /// [CALL] - SDValue Chain = I->getOperand(0); - SDValue Load = I->getOperand(1); + SDValue Chain = N->getOperand(0); + SDValue Load = N->getOperand(1); if (!isCalleeLoad(Load, Chain)) continue; - MoveBelowCallSeqStart(CurDAG, Load, SDValue(I, 0), Chain); + MoveBelowCallSeqStart(CurDAG, Load, SDValue(N, 0), Chain); ++NumLoadMoved; continue; } - - if (!ISD::isNON_TRUNCStore(I)) - continue; - SDValue Chain = I->getOperand(0); - - if (Chain.getNode()->getOpcode() != ISD::TokenFactor) - continue; - - SDValue N1 = I->getOperand(1); - SDValue N2 = I->getOperand(2); - if ((N1.getValueType().isFloatingPoint() && - !N1.getValueType().isVector()) || - !N1.hasOneUse()) - continue; - - bool RModW = false; - SDValue Load; - unsigned Opcode = N1.getNode()->getOpcode(); - switch (Opcode) { - case ISD::ADD: - case ISD::MUL: - case ISD::AND: - case ISD::OR: - case ISD::XOR: - case ISD::ADDC: - case ISD::ADDE: - case ISD::VECTOR_SHUFFLE: { - SDValue N10 = N1.getOperand(0); - SDValue N11 = N1.getOperand(1); - RModW = isRMWLoad(N10, Chain, N2, Load); - if (!RModW) - RModW = isRMWLoad(N11, Chain, N2, Load); - break; - } - case ISD::SUB: - case ISD::SHL: - case ISD::SRA: - case ISD::SRL: - case ISD::ROTL: - case ISD::ROTR: - case ISD::SUBC: - case ISD::SUBE: - case X86ISD::SHLD: - case X86ISD::SHRD: { - SDValue N10 = N1.getOperand(0); - RModW = isRMWLoad(N10, Chain, N2, Load); - break; - } - } - - if (RModW) { - MoveBelowTokenFactor(CurDAG, Load, SDValue(I, 0), Chain); - ++NumLoadMoved; - checkForCycles(I); - } - } -} - - -/// PreprocessForFPConvert - Walk over the dag lowering fpround and fpextend -/// nodes that target the FP stack to be store and load to the stack. This is a -/// gross hack. We would like to simply mark these as being illegal, but when -/// we do that, legalize produces these when it expands calls, then expands -/// these in the same legalize pass. We would like dag combine to be able to -/// hack on these between the call expansion and the node legalization. As such -/// this pass basically does "really late" legalization of these inline with the -/// X86 isel pass. -void X86DAGToDAGISel::PreprocessForFPConvert() { - for (SelectionDAG::allnodes_iterator I = CurDAG->allnodes_begin(), - E = CurDAG->allnodes_end(); I != E; ) { - SDNode *N = I++; // Preincrement iterator to avoid invalidation issues. + + // Lower fpround and fpextend nodes that target the FP stack to be store and + // load to the stack. This is a gross hack. We would like to simply mark + // these as being illegal, but when we do that, legalize produces these when + // it expands calls, then expands these in the same legalize pass. We would + // like dag combine to be able to hack on these between the call expansion + // and the node legalization. As such this pass basically does "really + // late" legalization of these inline with the X86 isel pass. + // FIXME: This should only happen when not compiled with -O0. if (N->getOpcode() != ISD::FP_ROUND && N->getOpcode() != ISD::FP_EXTEND) continue; @@ -687,30 +511,6 @@ void X86DAGToDAGISel::PreprocessForFPConvert() { } } -/// InstructionSelectBasicBlock - This callback is invoked by SelectionDAGISel -/// when it has created a SelectionDAG for us to codegen. -void X86DAGToDAGISel::InstructionSelect() { - const Function *F = MF->getFunction(); - OptForSize = F->hasFnAttr(Attribute::OptimizeForSize); - - if (OptLevel != CodeGenOpt::None) - PreprocessForRMW(); - - // FIXME: This should only happen when not compiled with -O0. - PreprocessForFPConvert(); - - // Codegen the basic block. -#ifndef NDEBUG - DEBUG(dbgs() << "===== Instruction selection begins:\n"); - Indent = 0; -#endif - SelectRoot(*CurDAG); -#ifndef NDEBUG - DEBUG(dbgs() << "===== Instruction selection ends:\n"); -#endif - - CurDAG->RemoveDeadNodes(); -} /// EmitSpecialCodeForMain - Emit any code that needs to be executed only in /// the main function. @@ -1317,22 +1117,24 @@ bool X86DAGToDAGISel::SelectAddr(SDNode *Op, SDValue N, SDValue &Base, /// SelectScalarSSELoad - Match a scalar SSE load. In particular, we want to /// match a load whose top elements are either undef or zeros. The load flavor /// is derived from the type of N, which is either v4f32 or v2f64. -bool X86DAGToDAGISel::SelectScalarSSELoad(SDNode *Op, SDValue Pred, +/// +/// We also return: +/// PatternChainNode: this is the matched node that has a chain input and +/// output. +bool X86DAGToDAGISel::SelectScalarSSELoad(SDNode *Root, SDValue N, SDValue &Base, SDValue &Scale, SDValue &Index, SDValue &Disp, SDValue &Segment, - SDValue &InChain, - SDValue &OutChain) { + SDValue &PatternNodeWithChain) { if (N.getOpcode() == ISD::SCALAR_TO_VECTOR) { - InChain = N.getOperand(0).getValue(1); - if (ISD::isNON_EXTLoad(InChain.getNode()) && - InChain.getValue(0).hasOneUse() && - IsProfitableToFold(N, Pred.getNode(), Op) && - IsLegalToFold(N, Pred.getNode(), Op)) { - LoadSDNode *LD = cast<LoadSDNode>(InChain); - if (!SelectAddr(Op, LD->getBasePtr(), Base, Scale, Index, Disp, Segment)) + PatternNodeWithChain = N.getOperand(0); + if (ISD::isNON_EXTLoad(PatternNodeWithChain.getNode()) && + PatternNodeWithChain.hasOneUse() && + IsProfitableToFold(N.getOperand(0), N.getNode(), Root) && + IsLegalToFold(N.getOperand(0), N.getNode(), Root)) { + LoadSDNode *LD = cast<LoadSDNode>(PatternNodeWithChain); + if (!SelectAddr(Root, LD->getBasePtr(), Base, Scale, Index, Disp,Segment)) return false; - OutChain = LD->getChain(); return true; } } @@ -1344,13 +1146,14 @@ bool X86DAGToDAGISel::SelectScalarSSELoad(SDNode *Op, SDValue Pred, N.getOperand(0).getOpcode() == ISD::SCALAR_TO_VECTOR && N.getOperand(0).getNode()->hasOneUse() && ISD::isNON_EXTLoad(N.getOperand(0).getOperand(0).getNode()) && - N.getOperand(0).getOperand(0).hasOneUse()) { + N.getOperand(0).getOperand(0).hasOneUse() && + IsProfitableToFold(N.getOperand(0), N.getNode(), Root) && + IsLegalToFold(N.getOperand(0), N.getNode(), Root)) { // Okay, this is a zero extending load. Fold it. LoadSDNode *LD = cast<LoadSDNode>(N.getOperand(0).getOperand(0)); - if (!SelectAddr(Op, LD->getBasePtr(), Base, Scale, Index, Disp, Segment)) + if (!SelectAddr(Root, LD->getBasePtr(), Base, Scale, Index, Disp, Segment)) return false; - OutChain = LD->getChain(); - InChain = SDValue(LD, 1); + PatternNodeWithChain = SDValue(LD, 0); return true; } return false; @@ -1424,7 +1227,6 @@ bool X86DAGToDAGISel::SelectLEAAddr(SDNode *Op, SDValue N, bool X86DAGToDAGISel::SelectTLSADDRAddr(SDNode *Op, SDValue N, SDValue &Base, SDValue &Scale, SDValue &Index, SDValue &Disp) { - assert(Op->getOpcode() == X86ISD::TLSADDR); assert(N.getOpcode() == ISD::TargetGlobalTLSAddress); const GlobalAddressSDNode *GA = cast<GlobalAddressSDNode>(N); @@ -1451,11 +1253,12 @@ bool X86DAGToDAGISel::TryFoldLoad(SDNode *P, SDValue N, SDValue &Base, SDValue &Scale, SDValue &Index, SDValue &Disp, SDValue &Segment) { - if (ISD::isNON_EXTLoad(N.getNode()) && - IsProfitableToFold(N, P, P) && - IsLegalToFold(N, P, P)) - return SelectAddr(P, N.getOperand(1), Base, Scale, Index, Disp, Segment); - return false; + if (!ISD::isNON_EXTLoad(N.getNode()) || + !IsProfitableToFold(N, P, P) || + !IsLegalToFold(N, P, P)) + return false; + + return SelectAddr(P, N.getOperand(1), Base, Scale, Index, Disp, Segment); } /// getGlobalBaseReg - Return an SDNode that returns the value of @@ -1558,7 +1361,7 @@ SDNode *X86DAGToDAGISel::SelectAtomicLoadAdd(SDNode *Node, EVT NVT) { Opc = X86::LOCK_DEC16m; else if (isSub) { if (isCN) { - if (Predicate_i16immSExt8(Val.getNode())) + if (Predicate_immSext8(Val.getNode())) Opc = X86::LOCK_SUB16mi8; else Opc = X86::LOCK_SUB16mi; @@ -1566,7 +1369,7 @@ SDNode *X86DAGToDAGISel::SelectAtomicLoadAdd(SDNode *Node, EVT NVT) { Opc = X86::LOCK_SUB16mr; } else { if (isCN) { - if (Predicate_i16immSExt8(Val.getNode())) + if (Predicate_immSext8(Val.getNode())) Opc = X86::LOCK_ADD16mi8; else Opc = X86::LOCK_ADD16mi; @@ -1581,7 +1384,7 @@ SDNode *X86DAGToDAGISel::SelectAtomicLoadAdd(SDNode *Node, EVT NVT) { Opc = X86::LOCK_DEC32m; else if (isSub) { if (isCN) { - if (Predicate_i32immSExt8(Val.getNode())) + if (Predicate_immSext8(Val.getNode())) Opc = X86::LOCK_SUB32mi8; else Opc = X86::LOCK_SUB32mi; @@ -1589,7 +1392,7 @@ SDNode *X86DAGToDAGISel::SelectAtomicLoadAdd(SDNode *Node, EVT NVT) { Opc = X86::LOCK_SUB32mr; } else { if (isCN) { - if (Predicate_i32immSExt8(Val.getNode())) + if (Predicate_immSext8(Val.getNode())) Opc = X86::LOCK_ADD32mi8; else Opc = X86::LOCK_ADD32mi; @@ -1605,7 +1408,7 @@ SDNode *X86DAGToDAGISel::SelectAtomicLoadAdd(SDNode *Node, EVT NVT) { else if (isSub) { Opc = X86::LOCK_SUB64mr; if (isCN) { - if (Predicate_i64immSExt8(Val.getNode())) + if (Predicate_immSext8(Val.getNode())) Opc = X86::LOCK_SUB64mi8; else if (Predicate_i64immSExt32(Val.getNode())) Opc = X86::LOCK_SUB64mi32; @@ -1613,7 +1416,7 @@ SDNode *X86DAGToDAGISel::SelectAtomicLoadAdd(SDNode *Node, EVT NVT) { } else { Opc = X86::LOCK_ADD64mr; if (isCN) { - if (Predicate_i64immSExt8(Val.getNode())) + if (Predicate_immSext8(Val.getNode())) Opc = X86::LOCK_ADD64mi8; else if (Predicate_i64immSExt32(Val.getNode())) Opc = X86::LOCK_ADD64mi32; @@ -1710,24 +1513,10 @@ SDNode *X86DAGToDAGISel::Select(SDNode *Node) { unsigned Opcode = Node->getOpcode(); DebugLoc dl = Node->getDebugLoc(); -#ifndef NDEBUG - DEBUG({ - dbgs() << std::string(Indent, ' ') << "Selecting: "; - Node->dump(CurDAG); - dbgs() << '\n'; - }); - Indent += 2; -#endif + DEBUG(dbgs() << "Selecting: "; Node->dump(CurDAG); dbgs() << '\n'); if (Node->isMachineOpcode()) { -#ifndef NDEBUG - DEBUG({ - dbgs() << std::string(Indent-2, ' ') << "== "; - Node->dump(CurDAG); - dbgs() << '\n'; - }); - Indent -= 2; -#endif + DEBUG(dbgs() << "== "; Node->dump(CurDAG); dbgs() << '\n'); return NULL; // Already selected. } @@ -1823,13 +1612,7 @@ SDNode *X86DAGToDAGISel::Select(SDNode *Node) { LoReg, NVT, InFlag); InFlag = Result.getValue(2); ReplaceUses(SDValue(Node, 0), Result); -#ifndef NDEBUG - DEBUG({ - dbgs() << std::string(Indent-2, ' ') << "=> "; - Result.getNode()->dump(CurDAG); - dbgs() << '\n'; - }); -#endif + DEBUG(dbgs() << "=> "; Result.getNode()->dump(CurDAG); dbgs() << '\n'); } // Copy the high half of the result, if it is needed. if (!SDValue(Node, 1).use_empty()) { @@ -1852,19 +1635,9 @@ SDNode *X86DAGToDAGISel::Select(SDNode *Node) { InFlag = Result.getValue(2); } ReplaceUses(SDValue(Node, 1), Result); -#ifndef NDEBUG - DEBUG({ - dbgs() << std::string(Indent-2, ' ') << "=> "; - Result.getNode()->dump(CurDAG); - dbgs() << '\n'; - }); -#endif + DEBUG(dbgs() << "=> "; Result.getNode()->dump(CurDAG); dbgs() << '\n'); } -#ifndef NDEBUG - Indent -= 2; -#endif - return NULL; } @@ -1979,13 +1752,7 @@ SDNode *X86DAGToDAGISel::Select(SDNode *Node) { LoReg, NVT, InFlag); InFlag = Result.getValue(2); ReplaceUses(SDValue(Node, 0), Result); -#ifndef NDEBUG - DEBUG({ - dbgs() << std::string(Indent-2, ' ') << "=> "; - Result.getNode()->dump(CurDAG); - dbgs() << '\n'; - }); -#endif + DEBUG(dbgs() << "=> "; Result.getNode()->dump(CurDAG); dbgs() << '\n'); } // Copy the remainder (high) result, if it is needed. if (!SDValue(Node, 1).use_empty()) { @@ -2009,19 +1776,8 @@ SDNode *X86DAGToDAGISel::Select(SDNode *Node) { InFlag = Result.getValue(2); } ReplaceUses(SDValue(Node, 1), Result); -#ifndef NDEBUG - DEBUG({ - dbgs() << std::string(Indent-2, ' ') << "=> "; - Result.getNode()->dump(CurDAG); - dbgs() << '\n'; - }); -#endif + DEBUG(dbgs() << "=> "; Result.getNode()->dump(CurDAG); dbgs() << '\n'); } - -#ifndef NDEBUG - Indent -= 2; -#endif - return NULL; } @@ -2134,17 +1890,12 @@ SDNode *X86DAGToDAGISel::Select(SDNode *Node) { SDNode *ResNode = SelectCode(Node); -#ifndef NDEBUG - DEBUG({ - dbgs() << std::string(Indent-2, ' ') << "=> "; - if (ResNode == NULL || ResNode == Node) - Node->dump(CurDAG); - else - ResNode->dump(CurDAG); - dbgs() << '\n'; - }); - Indent -= 2; -#endif + DEBUG(dbgs() << "=> "; + if (ResNode == NULL || ResNode == Node) + Node->dump(CurDAG); + else + ResNode->dump(CurDAG); + dbgs() << '\n'); return ResNode; } diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp index 9974d8c..e2b8193 100644 --- a/lib/Target/X86/X86ISelLowering.cpp +++ b/lib/Target/X86/X86ISelLowering.cpp @@ -73,7 +73,7 @@ static TargetLoweringObjectFile *createTLOF(X86TargetMachine &TM) { case X86Subtarget::isDarwin: if (TM.getSubtarget<X86Subtarget>().is64Bit()) return new X8664_MachoTargetObjectFile(); - return new X8632_MachoTargetObjectFile(); + return new TargetLoweringObjectFileMachO(); case X86Subtarget::isELF: if (TM.getSubtarget<X86Subtarget>().is64Bit()) return new X8664_ELFTargetObjectFile(TM); @@ -990,6 +990,7 @@ X86TargetLowering::X86TargetLowering(X86TargetMachine &TM) setTargetDAGCombine(ISD::VECTOR_SHUFFLE); setTargetDAGCombine(ISD::BUILD_VECTOR); setTargetDAGCombine(ISD::SELECT); + setTargetDAGCombine(ISD::AND); setTargetDAGCombine(ISD::SHL); setTargetDAGCombine(ISD::SRA); setTargetDAGCombine(ISD::SRL); @@ -1743,7 +1744,7 @@ EmitTailCallStoreRetAddr(SelectionDAG & DAG, MachineFunction &MF, // Calculate the new stack slot for the return address. int SlotSize = Is64Bit ? 8 : 4; int NewReturnAddrFI = - MF.getFrameInfo()->CreateFixedObject(SlotSize, FPDiff-SlotSize, true,false); + MF.getFrameInfo()->CreateFixedObject(SlotSize, FPDiff-SlotSize, false, false); EVT VT = Is64Bit ? MVT::i64 : MVT::i32; SDValue NewRetAddrFrIdx = DAG.getFrameIndex(NewReturnAddrFI, VT); Chain = DAG.getStore(Chain, dl, RetAddrFrIdx, NewRetAddrFrIdx, @@ -2376,7 +2377,7 @@ SDValue X86TargetLowering::getReturnAddressFrameIndex(SelectionDAG &DAG) { // Set up a frame object for the return address. uint64_t SlotSize = TD->getPointerSize(); ReturnAddrIndex = MF.getFrameInfo()->CreateFixedObject(SlotSize, -SlotSize, - true, false); + false, false); FuncInfo->setRAIndex(ReturnAddrIndex); } @@ -4816,8 +4817,16 @@ X86TargetLowering::LowerINSERT_VECTOR_ELT_SSE4(SDValue Op, SelectionDAG &DAG){ if ((EltVT.getSizeInBits() == 8 || EltVT.getSizeInBits() == 16) && isa<ConstantSDNode>(N2)) { - unsigned Opc = (EltVT.getSizeInBits() == 8) ? X86ISD::PINSRB - : X86ISD::PINSRW; + unsigned Opc; + if (VT == MVT::v8i16) + Opc = X86ISD::PINSRW; + else if (VT == MVT::v4i16) + Opc = X86ISD::MMX_PINSRW; + else if (VT == MVT::v16i8) + Opc = X86ISD::PINSRB; + else + Opc = X86ISD::PINSRB; + // Transform it so it match pinsr{b,w} which expects a GR32 as its second // argument. if (N1.getValueType() != MVT::i32) @@ -4868,7 +4877,8 @@ X86TargetLowering::LowerINSERT_VECTOR_ELT(SDValue Op, SelectionDAG &DAG) { N1 = DAG.getNode(ISD::ANY_EXTEND, dl, MVT::i32, N1); if (N2.getValueType() != MVT::i32) N2 = DAG.getIntPtrConstant(cast<ConstantSDNode>(N2)->getZExtValue()); - return DAG.getNode(X86ISD::PINSRW, dl, VT, N0, N1, N2); + return DAG.getNode(VT == MVT::v8i16 ? X86ISD::PINSRW : X86ISD::MMX_PINSRW, + dl, VT, N0, N1, N2); } return SDValue(); } @@ -5244,7 +5254,7 @@ SDValue X86TargetLowering::LowerShift(SDValue Op, SelectionDAG &DAG) { SDValue AndNode = DAG.getNode(ISD::AND, dl, MVT::i8, ShAmt, DAG.getConstant(VTBits, MVT::i8)); - SDValue Cond = DAG.getNode(X86ISD::CMP, dl, VT, + SDValue Cond = DAG.getNode(X86ISD::CMP, dl, MVT::i32, AndNode, DAG.getConstant(0, MVT::i8)); SDValue Hi, Lo; @@ -5873,26 +5883,31 @@ SDValue X86TargetLowering::EmitCmp(SDValue Op0, SDValue Op1, unsigned X86CC, /// LowerToBT - Result of 'and' is compared against zero. Turn it into a BT node /// if it's possible. -static SDValue LowerToBT(SDValue Op0, ISD::CondCode CC, +static SDValue LowerToBT(SDValue And, ISD::CondCode CC, DebugLoc dl, SelectionDAG &DAG) { + SDValue Op0 = And.getOperand(0); + SDValue Op1 = And.getOperand(1); + if (Op0.getOpcode() == ISD::TRUNCATE) + Op0 = Op0.getOperand(0); + if (Op1.getOpcode() == ISD::TRUNCATE) + Op1 = Op1.getOperand(0); + SDValue LHS, RHS; - if (Op0.getOperand(1).getOpcode() == ISD::SHL) { - if (ConstantSDNode *Op010C = - dyn_cast<ConstantSDNode>(Op0.getOperand(1).getOperand(0))) - if (Op010C->getZExtValue() == 1) { - LHS = Op0.getOperand(0); - RHS = Op0.getOperand(1).getOperand(1); + if (Op1.getOpcode() == ISD::SHL) { + if (ConstantSDNode *And10C = dyn_cast<ConstantSDNode>(Op1.getOperand(0))) + if (And10C->getZExtValue() == 1) { + LHS = Op0; + RHS = Op1.getOperand(1); } - } else if (Op0.getOperand(0).getOpcode() == ISD::SHL) { - if (ConstantSDNode *Op000C = - dyn_cast<ConstantSDNode>(Op0.getOperand(0).getOperand(0))) - if (Op000C->getZExtValue() == 1) { - LHS = Op0.getOperand(1); - RHS = Op0.getOperand(0).getOperand(1); + } else if (Op0.getOpcode() == ISD::SHL) { + if (ConstantSDNode *And00C = dyn_cast<ConstantSDNode>(Op0.getOperand(0))) + if (And00C->getZExtValue() == 1) { + LHS = Op1; + RHS = Op0.getOperand(1); } - } else if (Op0.getOperand(1).getOpcode() == ISD::Constant) { - ConstantSDNode *AndRHS = cast<ConstantSDNode>(Op0.getOperand(1)); - SDValue AndLHS = Op0.getOperand(0); + } else if (Op1.getOpcode() == ISD::Constant) { + ConstantSDNode *AndRHS = cast<ConstantSDNode>(Op1); + SDValue AndLHS = Op0; if (AndRHS->getZExtValue() == 1 && AndLHS.getOpcode() == ISD::SRL) { LHS = AndLHS.getOperand(0); RHS = AndLHS.getOperand(1); @@ -5942,6 +5957,21 @@ SDValue X86TargetLowering::LowerSETCC(SDValue Op, SelectionDAG &DAG) { return NewSetCC; } + // Look for "(setcc) == / != 1" to avoid unncessary setcc. + if (Op0.getOpcode() == X86ISD::SETCC && + Op1.getOpcode() == ISD::Constant && + (cast<ConstantSDNode>(Op1)->getZExtValue() == 1 || + cast<ConstantSDNode>(Op1)->isNullValue()) && + (CC == ISD::SETEQ || CC == ISD::SETNE)) { + X86::CondCode CCode = (X86::CondCode)Op0.getConstantOperandVal(0); + bool Invert = (CC == ISD::SETNE) ^ + cast<ConstantSDNode>(Op1)->isNullValue(); + if (Invert) + CCode = X86::GetOppositeBranchCondition(CCode); + return DAG.getNode(X86ISD::SETCC, dl, MVT::i8, + DAG.getConstant(CCode, MVT::i8), Op0.getOperand(1)); + } + bool isFP = Op.getOperand(1).getValueType().isFloatingPoint(); unsigned X86CC = TranslateX86CC(CC, isFP, Op0, Op1, DAG); if (X86CC == X86::COND_INVALID) @@ -6444,8 +6474,7 @@ X86TargetLowering::EmitTargetCodeForMemset(SelectionDAG &DAG, DebugLoc dl, LowerCallTo(Chain, Type::getVoidTy(*DAG.getContext()), false, false, false, false, 0, CallingConv::C, false, /*isReturnValueUsed=*/false, - DAG.getExternalSymbol(bzeroEntry, IntPtr), Args, DAG, dl, - DAG.GetOrdering(Chain.getNode())); + DAG.getExternalSymbol(bzeroEntry, IntPtr), Args, DAG, dl); return CallResult.second; } @@ -7662,6 +7691,7 @@ const char *X86TargetLowering::getTargetNodeName(unsigned Opcode) const { case X86ISD::INSERTPS: return "X86ISD::INSERTPS"; case X86ISD::PINSRB: return "X86ISD::PINSRB"; case X86ISD::PINSRW: return "X86ISD::PINSRW"; + case X86ISD::MMX_PINSRW: return "X86ISD::MMX_PINSRW"; case X86ISD::PSHUFB: return "X86ISD::PSHUFB"; case X86ISD::FMAX: return "X86ISD::FMAX"; case X86ISD::FMIN: return "X86ISD::FMIN"; @@ -7769,7 +7799,7 @@ bool X86TargetLowering::isTruncateFree(const Type *Ty1, const Type *Ty2) const { unsigned NumBits2 = Ty2->getPrimitiveSizeInBits(); if (NumBits1 <= NumBits2) return false; - return Subtarget->is64Bit() || NumBits1 < 64; + return true; } bool X86TargetLowering::isTruncateFree(EVT VT1, EVT VT2) const { @@ -7779,7 +7809,7 @@ bool X86TargetLowering::isTruncateFree(EVT VT1, EVT VT2) const { unsigned NumBits2 = VT2.getSizeInBits(); if (NumBits1 <= NumBits2) return false; - return Subtarget->is64Bit() || NumBits1 < 64; + return true; } bool X86TargetLowering::isZExtFree(const Type *Ty1, const Type *Ty2) const { @@ -8792,10 +8822,9 @@ static SDValue PerformSELECTCombine(SDNode *N, SelectionDAG &DAG, SDValue RHS = N->getOperand(2); // If we have SSE[12] support, try to form min/max nodes. SSE min/max - // instructions have the peculiarity that if either operand is a NaN, - // they chose what we call the RHS operand (and as such are not symmetric). - // It happens that this matches the semantics of the common C idiom - // x<y?x:y and related forms, so we can recognize these cases. + // instructions match the semantics of the common C idiom x<y?x:y but not + // x<=y?x:y, because of how they handle negative zero (which can be + // ignored in unsafe-math mode). if (Subtarget->hasSSE2() && (LHS.getValueType() == MVT::f32 || LHS.getValueType() == MVT::f64) && Cond.getOpcode() == ISD::SETCC) { @@ -8803,36 +8832,34 @@ static SDValue PerformSELECTCombine(SDNode *N, SelectionDAG &DAG, unsigned Opcode = 0; // Check for x CC y ? x : y. - if (LHS == Cond.getOperand(0) && RHS == Cond.getOperand(1)) { + if (DAG.isEqualTo(LHS, Cond.getOperand(0)) && + DAG.isEqualTo(RHS, Cond.getOperand(1))) { switch (CC) { default: break; case ISD::SETULT: - // This can be a min if we can prove that at least one of the operands - // is not a nan. - if (!FiniteOnlyFPMath()) { - if (DAG.isKnownNeverNaN(RHS)) { - // Put the potential NaN in the RHS so that SSE will preserve it. - std::swap(LHS, RHS); - } else if (!DAG.isKnownNeverNaN(LHS)) + // Converting this to a min would handle NaNs incorrectly, and swapping + // the operands would cause it to handle comparisons between positive + // and negative zero incorrectly. + if (!FiniteOnlyFPMath() && + (!DAG.isKnownNeverNaN(LHS) || !DAG.isKnownNeverNaN(RHS))) { + if (!UnsafeFPMath && + !(DAG.isKnownNeverZero(LHS) || DAG.isKnownNeverZero(RHS))) break; + std::swap(LHS, RHS); } Opcode = X86ISD::FMIN; break; case ISD::SETOLE: - // This can be a min if we can prove that at least one of the operands - // is not a nan. - if (!FiniteOnlyFPMath()) { - if (DAG.isKnownNeverNaN(LHS)) { - // Put the potential NaN in the RHS so that SSE will preserve it. - std::swap(LHS, RHS); - } else if (!DAG.isKnownNeverNaN(RHS)) - break; - } + // Converting this to a min would handle comparisons between positive + // and negative zero incorrectly. + if (!UnsafeFPMath && + !DAG.isKnownNeverZero(LHS) && !DAG.isKnownNeverZero(RHS)) + break; Opcode = X86ISD::FMIN; break; case ISD::SETULE: - // This can be a min, but if either operand is a NaN we need it to - // preserve the original LHS. + // Converting this to a min would handle both negative zeros and NaNs + // incorrectly, but we can swap the operands to fix both. std::swap(LHS, RHS); case ISD::SETOLT: case ISD::SETLT: @@ -8841,32 +8868,29 @@ static SDValue PerformSELECTCombine(SDNode *N, SelectionDAG &DAG, break; case ISD::SETOGE: - // This can be a max if we can prove that at least one of the operands - // is not a nan. - if (!FiniteOnlyFPMath()) { - if (DAG.isKnownNeverNaN(LHS)) { - // Put the potential NaN in the RHS so that SSE will preserve it. - std::swap(LHS, RHS); - } else if (!DAG.isKnownNeverNaN(RHS)) - break; - } + // Converting this to a max would handle comparisons between positive + // and negative zero incorrectly. + if (!UnsafeFPMath && + !DAG.isKnownNeverZero(LHS) && !DAG.isKnownNeverZero(LHS)) + break; Opcode = X86ISD::FMAX; break; case ISD::SETUGT: - // This can be a max if we can prove that at least one of the operands - // is not a nan. - if (!FiniteOnlyFPMath()) { - if (DAG.isKnownNeverNaN(RHS)) { - // Put the potential NaN in the RHS so that SSE will preserve it. - std::swap(LHS, RHS); - } else if (!DAG.isKnownNeverNaN(LHS)) + // Converting this to a max would handle NaNs incorrectly, and swapping + // the operands would cause it to handle comparisons between positive + // and negative zero incorrectly. + if (!FiniteOnlyFPMath() && + (!DAG.isKnownNeverNaN(LHS) || !DAG.isKnownNeverNaN(RHS))) { + if (!UnsafeFPMath && + !(DAG.isKnownNeverZero(LHS) || DAG.isKnownNeverZero(RHS))) break; + std::swap(LHS, RHS); } Opcode = X86ISD::FMAX; break; case ISD::SETUGE: - // This can be a max, but if either operand is a NaN we need it to - // preserve the original LHS. + // Converting this to a max would handle both negative zeros and NaNs + // incorrectly, but we can swap the operands to fix both. std::swap(LHS, RHS); case ISD::SETOGT: case ISD::SETGT: @@ -8875,36 +8899,33 @@ static SDValue PerformSELECTCombine(SDNode *N, SelectionDAG &DAG, break; } // Check for x CC y ? y : x -- a min/max with reversed arms. - } else if (LHS == Cond.getOperand(1) && RHS == Cond.getOperand(0)) { + } else if (DAG.isEqualTo(LHS, Cond.getOperand(1)) && + DAG.isEqualTo(RHS, Cond.getOperand(0))) { switch (CC) { default: break; case ISD::SETOGE: - // This can be a min if we can prove that at least one of the operands - // is not a nan. - if (!FiniteOnlyFPMath()) { - if (DAG.isKnownNeverNaN(RHS)) { - // Put the potential NaN in the RHS so that SSE will preserve it. - std::swap(LHS, RHS); - } else if (!DAG.isKnownNeverNaN(LHS)) + // Converting this to a min would handle comparisons between positive + // and negative zero incorrectly, and swapping the operands would + // cause it to handle NaNs incorrectly. + if (!UnsafeFPMath && + !(DAG.isKnownNeverZero(LHS) || DAG.isKnownNeverZero(RHS))) { + if (!FiniteOnlyFPMath() && + (!DAG.isKnownNeverNaN(LHS) || !DAG.isKnownNeverNaN(RHS))) break; + std::swap(LHS, RHS); } Opcode = X86ISD::FMIN; break; case ISD::SETUGT: - // This can be a min if we can prove that at least one of the operands - // is not a nan. - if (!FiniteOnlyFPMath()) { - if (DAG.isKnownNeverNaN(LHS)) { - // Put the potential NaN in the RHS so that SSE will preserve it. - std::swap(LHS, RHS); - } else if (!DAG.isKnownNeverNaN(RHS)) - break; - } + // Converting this to a min would handle NaNs incorrectly. + if (!UnsafeFPMath && + (!DAG.isKnownNeverNaN(LHS) || !DAG.isKnownNeverNaN(RHS))) + break; Opcode = X86ISD::FMIN; break; case ISD::SETUGE: - // This can be a min, but if either operand is a NaN we need it to - // preserve the original LHS. + // Converting this to a min would handle both negative zeros and NaNs + // incorrectly, but we can swap the operands to fix both. std::swap(LHS, RHS); case ISD::SETOGT: case ISD::SETGT: @@ -8913,32 +8934,28 @@ static SDValue PerformSELECTCombine(SDNode *N, SelectionDAG &DAG, break; case ISD::SETULT: - // This can be a max if we can prove that at least one of the operands - // is not a nan. - if (!FiniteOnlyFPMath()) { - if (DAG.isKnownNeverNaN(LHS)) { - // Put the potential NaN in the RHS so that SSE will preserve it. - std::swap(LHS, RHS); - } else if (!DAG.isKnownNeverNaN(RHS)) - break; - } + // Converting this to a max would handle NaNs incorrectly. + if (!FiniteOnlyFPMath() && + (!DAG.isKnownNeverNaN(LHS) || !DAG.isKnownNeverNaN(RHS))) + break; Opcode = X86ISD::FMAX; break; case ISD::SETOLE: - // This can be a max if we can prove that at least one of the operands - // is not a nan. - if (!FiniteOnlyFPMath()) { - if (DAG.isKnownNeverNaN(RHS)) { - // Put the potential NaN in the RHS so that SSE will preserve it. - std::swap(LHS, RHS); - } else if (!DAG.isKnownNeverNaN(LHS)) + // Converting this to a max would handle comparisons between positive + // and negative zero incorrectly, and swapping the operands would + // cause it to handle NaNs incorrectly. + if (!UnsafeFPMath && + !DAG.isKnownNeverZero(LHS) && !DAG.isKnownNeverZero(RHS)) { + if (!FiniteOnlyFPMath() && + (!DAG.isKnownNeverNaN(LHS) || !DAG.isKnownNeverNaN(RHS))) break; + std::swap(LHS, RHS); } Opcode = X86ISD::FMAX; break; case ISD::SETULE: - // This can be a max, but if either operand is a NaN we need it to - // preserve the original LHS. + // Converting this to a max would handle both negative zeros and NaNs + // incorrectly, but we can swap the operands to fix both. std::swap(LHS, RHS); case ISD::SETOLT: case ISD::SETLT: @@ -9157,16 +9174,64 @@ static SDValue PerformCMOVCombine(SDNode *N, SelectionDAG &DAG, return SDValue(); } +/// PerformANDCombine - Look for SSE and instructions of this form: +/// (and x, (build_vector signbit,signbit,signbit,signbit)). If there +/// exists a use of a build_vector that's the bitwise complement of the mask, +/// then transform the node to +/// (and (xor x, (build_vector -1,-1,-1,-1)), (build_vector ~sb,~sb,~sb,~sb)). +static SDValue PerformANDCombine(SDNode *N, SelectionDAG &DAG, + TargetLowering::DAGCombinerInfo &DCI) { + EVT VT = N->getValueType(0); + if (!VT.isVector() || !VT.isInteger()) + return SDValue(); + + SDValue N0 = N->getOperand(0); + SDValue N1 = N->getOperand(1); + if (N0.getOpcode() == ISD::XOR || !N1.hasOneUse()) + return SDValue(); + + if (N1.getOpcode() == ISD::BUILD_VECTOR) { + unsigned NumElts = VT.getVectorNumElements(); + EVT EltVT = VT.getVectorElementType(); + SmallVector<SDValue, 8> Mask; + Mask.reserve(NumElts); + for (unsigned i = 0; i != NumElts; ++i) { + SDValue Arg = N1.getOperand(i); + if (Arg.getOpcode() == ISD::UNDEF) { + Mask.push_back(Arg); + continue; + } + ConstantSDNode *C = dyn_cast<ConstantSDNode>(Arg); + if (!C) + return SDValue(); + if (!C->getAPIntValue().isSignBit() && + !C->getAPIntValue().isMaxSignedValue()) + return SDValue(); + Mask.push_back(DAG.getConstant(~C->getAPIntValue(), EltVT)); + } + N1 = DAG.getNode(ISD::BUILD_VECTOR, N1.getDebugLoc(), VT, + &Mask[0], NumElts); + if (!N1.use_empty()) { + unsigned Bits = EltVT.getSizeInBits(); + Mask.clear(); + for (unsigned i = 0; i != NumElts; ++i) + Mask.push_back(DAG.getConstant(APInt::getAllOnesValue(Bits), EltVT)); + SDValue NewMask = DAG.getNode(ISD::BUILD_VECTOR, N->getDebugLoc(), + VT, &Mask[0], NumElts); + return DAG.getNode(ISD::AND, N->getDebugLoc(), VT, + DAG.getNode(ISD::XOR, N->getDebugLoc(), VT, + N0, NewMask), N1); + } + } + + return SDValue(); +} /// PerformMulCombine - Optimize a single multiply with constant into two /// in order to implement it with two cheaper instructions, e.g. /// LEA + SHL, LEA + LEA. static SDValue PerformMulCombine(SDNode *N, SelectionDAG &DAG, TargetLowering::DAGCombinerInfo &DCI) { - if (DAG.getMachineFunction(). - getFunction()->hasFnAttr(Attribute::OptimizeForSize)) - return SDValue(); - if (DCI.isBeforeLegalize() || DCI.isCalledByLegalizer()) return SDValue(); @@ -9305,7 +9370,7 @@ static SDValue PerformShiftCombine(SDNode* N, SelectionDAG &DAG, } } else if (InVec.getOpcode() == ISD::INSERT_VECTOR_ELT) { if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(InVec.getOperand(2))) { - unsigned SplatIdx = cast<ShuffleVectorSDNode>(ShAmtOp)->getSplatIndex(); + unsigned SplatIdx= cast<ShuffleVectorSDNode>(ShAmtOp)->getSplatIndex(); if (C->getZExtValue() == SplatIdx) BaseShAmt = InVec.getOperand(1); } @@ -9690,6 +9755,7 @@ SDValue X86TargetLowering::PerformDAGCombine(SDNode *N, case ISD::VECTOR_SHUFFLE: return PerformShuffleCombine(N, DAG, *this); case ISD::SELECT: return PerformSELECTCombine(N, DAG, Subtarget); case X86ISD::CMOV: return PerformCMOVCombine(N, DAG, DCI); + case ISD::AND: return PerformANDCombine(N, DAG, DCI); case ISD::MUL: return PerformMulCombine(N, DAG, DCI); case ISD::SHL: case ISD::SRA: diff --git a/lib/Target/X86/X86ISelLowering.h b/lib/Target/X86/X86ISelLowering.h index cf0eb40..ffaf1cf 100644 --- a/lib/Target/X86/X86ISelLowering.h +++ b/lib/Target/X86/X86ISelLowering.h @@ -180,7 +180,7 @@ namespace llvm { /// PINSRW - Insert the lower 16-bits of a 32-bit value to a vector, /// corresponds to X86::PINSRW. - PINSRW, + PINSRW, MMX_PINSRW, /// PSHUFB - Shuffle 16 8-bit values within a vector. PSHUFB, diff --git a/lib/Target/X86/X86Instr64bit.td b/lib/Target/X86/X86Instr64bit.td index 4ea3739..8462255 100644 --- a/lib/Target/X86/X86Instr64bit.td +++ b/lib/Target/X86/X86Instr64bit.td @@ -59,10 +59,11 @@ def tls64addr : ComplexPattern<i64, 4, "SelectTLSADDRAddr", // Pattern fragments. // -def i64immSExt8 : PatLeaf<(i64 imm), [{ - // i64immSExt8 predicate - True if the 64-bit immediate fits in a 8-bit - // sign extended field. - return (int64_t)N->getZExtValue() == (int8_t)N->getZExtValue(); +def i64immSExt8 : PatLeaf<(i64 immSext8)>; + +def GetLo32XForm : SDNodeXForm<imm, [{ + // Transformation function: get the low 32 bits. + return getI32Imm((unsigned)N->getZExtValue()); }]>; def i64immSExt32 : PatLeaf<(i64 imm), [{ @@ -71,6 +72,7 @@ def i64immSExt32 : PatLeaf<(i64 imm), [{ return (int64_t)N->getZExtValue() == (int32_t)N->getZExtValue(); }]>; + def i64immZExt32 : PatLeaf<(i64 imm), [{ // i64immZExt32 predicate - True if the 64-bit immediate fits in a 32-bit // unsignedsign extended field. @@ -325,7 +327,7 @@ def MOV64ri32 : RIi32<0xC7, MRM0r, (outs GR64:$dst), (ins i64i32imm:$src), def MOV64rr_REV : RI<0x8B, MRMSrcReg, (outs GR64:$dst), (ins GR64:$src), "mov{q}\t{$src, $dst|$dst, $src}", []>; -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def MOV64rm : RI<0x8B, MRMSrcMem, (outs GR64:$dst), (ins i64mem:$src), "mov{q}\t{$src, $dst|$dst, $src}", [(set GR64:$dst, (load addr:$src))]>; @@ -556,7 +558,7 @@ def ADC64mi8 : RIi8<0x83, MRM2m, (outs), (ins i64mem:$dst, i64i8imm :$src2), addr:$dst)]>; def ADC64mi32 : RIi32<0x81, MRM2m, (outs), (ins i64mem:$dst, i64i32imm:$src2), "adc{q}\t{$src2, $dst|$dst, $src2}", - [(store (adde (load addr:$dst), i64immSExt8:$src2), + [(store (adde (load addr:$dst), i64immSExt32:$src2), addr:$dst)]>; } // Uses = [EFLAGS] @@ -1981,7 +1983,7 @@ def : Pat<(and GR64:$src, i64immZExt32:$imm), (i64 0), (AND32ri (EXTRACT_SUBREG GR64:$src, x86_subreg_32bit), - imm:$imm), + (i32 (GetLo32XForm imm:$imm))), x86_subreg_32bit)>; // r & (2^32-1) ==> movz @@ -2105,34 +2107,34 @@ def : Pat<(store (i8 (trunc_su (srl_su GR16:$src, (i8 8)))), addr:$dst), def : Pat<(shl GR64:$src1, (i8 1)), (ADD64rr GR64:$src1, GR64:$src1)>; // (shl x (and y, 63)) ==> (shl x, y) -def : Pat<(shl GR64:$src1, (and CL:$amt, 63)), +def : Pat<(shl GR64:$src1, (and CL, 63)), (SHL64rCL GR64:$src1)>; -def : Pat<(store (shl (loadi64 addr:$dst), (and CL:$amt, 63)), addr:$dst), +def : Pat<(store (shl (loadi64 addr:$dst), (and CL, 63)), addr:$dst), (SHL64mCL addr:$dst)>; -def : Pat<(srl GR64:$src1, (and CL:$amt, 63)), +def : Pat<(srl GR64:$src1, (and CL, 63)), (SHR64rCL GR64:$src1)>; -def : Pat<(store (srl (loadi64 addr:$dst), (and CL:$amt, 63)), addr:$dst), +def : Pat<(store (srl (loadi64 addr:$dst), (and CL, 63)), addr:$dst), (SHR64mCL addr:$dst)>; -def : Pat<(sra GR64:$src1, (and CL:$amt, 63)), +def : Pat<(sra GR64:$src1, (and CL, 63)), (SAR64rCL GR64:$src1)>; -def : Pat<(store (sra (loadi64 addr:$dst), (and CL:$amt, 63)), addr:$dst), +def : Pat<(store (sra (loadi64 addr:$dst), (and CL, 63)), addr:$dst), (SAR64mCL addr:$dst)>; // Double shift patterns -def : Pat<(shrd GR64:$src1, (i8 imm:$amt1), GR64:$src2, (i8 imm:$amt2)), +def : Pat<(shrd GR64:$src1, (i8 imm:$amt1), GR64:$src2, (i8 imm)), (SHRD64rri8 GR64:$src1, GR64:$src2, (i8 imm:$amt1))>; def : Pat<(store (shrd (loadi64 addr:$dst), (i8 imm:$amt1), - GR64:$src2, (i8 imm:$amt2)), addr:$dst), + GR64:$src2, (i8 imm)), addr:$dst), (SHRD64mri8 addr:$dst, GR64:$src2, (i8 imm:$amt1))>; -def : Pat<(shld GR64:$src1, (i8 imm:$amt1), GR64:$src2, (i8 imm:$amt2)), +def : Pat<(shld GR64:$src1, (i8 imm:$amt1), GR64:$src2, (i8 imm)), (SHLD64rri8 GR64:$src1, GR64:$src2, (i8 imm:$amt1))>; def : Pat<(store (shld (loadi64 addr:$dst), (i8 imm:$amt1), - GR64:$src2, (i8 imm:$amt2)), addr:$dst), + GR64:$src2, (i8 imm)), addr:$dst), (SHLD64mri8 addr:$dst, GR64:$src2, (i8 imm:$amt1))>; // (or x1, x2) -> (add x1, x2) if two operands are known not to share bits. diff --git a/lib/Target/X86/X86InstrFPStack.td b/lib/Target/X86/X86InstrFPStack.td index e22a903..ae24bfb 100644 --- a/lib/Target/X86/X86InstrFPStack.td +++ b/lib/Target/X86/X86InstrFPStack.td @@ -397,7 +397,7 @@ def CMOVNP_F : FPI<0xD8, AddRegFrm, (outs RST:$op), (ins), let canFoldAsLoad = 1 in { def LD_Fp32m : FpIf32<(outs RFP32:$dst), (ins f32mem:$src), ZeroArgFP, [(set RFP32:$dst, (loadf32 addr:$src))]>; -let isReMaterializable = 1, mayHaveSideEffects = 1 in +let isReMaterializable = 1 in def LD_Fp64m : FpIf64<(outs RFP64:$dst), (ins f64mem:$src), ZeroArgFP, [(set RFP64:$dst, (loadf64 addr:$src))]>; def LD_Fp80m : FpI_<(outs RFP80:$dst), (ins f80mem:$src), ZeroArgFP, diff --git a/lib/Target/X86/X86InstrInfo.cpp b/lib/Target/X86/X86InstrInfo.cpp index a0d0312..39bda04 100644 --- a/lib/Target/X86/X86InstrInfo.cpp +++ b/lib/Target/X86/X86InstrInfo.cpp @@ -276,11 +276,8 @@ X86InstrInfo::X86InstrInfo(X86TargetMachine &tm) { X86::MOVDQArr, X86::MOVDQAmr, 0, 16 }, { X86::MOVPDI2DIrr, X86::MOVPDI2DImr, 0, 0 }, { X86::MOVPQIto64rr,X86::MOVPQI2QImr, 0, 0 }, - { X86::MOVPS2SSrr, X86::MOVPS2SSmr, 0, 0 }, - { X86::MOVSDrr, X86::MOVSDmr, 0, 0 }, { X86::MOVSDto64rr, X86::MOVSDto64mr, 0, 0 }, { X86::MOVSS2DIrr, X86::MOVSS2DImr, 0, 0 }, - { X86::MOVSSrr, X86::MOVSSmr, 0, 0 }, { X86::MOVUPDrr, X86::MOVUPDmr, 0, 0 }, { X86::MOVUPSrr, X86::MOVUPSmr, 0, 0 }, { X86::MUL16r, X86::MUL16m, 1, 0 }, @@ -389,12 +386,8 @@ X86InstrInfo::X86InstrInfo(X86TargetMachine &tm) { X86::MOVDI2PDIrr, X86::MOVDI2PDIrm, 0 }, { X86::MOVDI2SSrr, X86::MOVDI2SSrm, 0 }, { X86::MOVDQArr, X86::MOVDQArm, 16 }, - { X86::MOVSD2PDrr, X86::MOVSD2PDrm, 0 }, - { X86::MOVSDrr, X86::MOVSDrm, 0 }, { X86::MOVSHDUPrr, X86::MOVSHDUPrm, 16 }, { X86::MOVSLDUPrr, X86::MOVSLDUPrm, 16 }, - { X86::MOVSS2PSrr, X86::MOVSS2PSrm, 0 }, - { X86::MOVSSrr, X86::MOVSSrm, 0 }, { X86::MOVSX16rr8, X86::MOVSX16rm8, 0 }, { X86::MOVSX32rr16, X86::MOVSX32rm16, 0 }, { X86::MOVSX32rr8, X86::MOVSX32rm8, 0 }, @@ -682,23 +675,20 @@ bool X86InstrInfo::isMoveInstr(const MachineInstr& MI, case X86::MOV16rr: case X86::MOV32rr: case X86::MOV64rr: - case X86::MOVSSrr: - case X86::MOVSDrr: // FP Stack register class copies case X86::MOV_Fp3232: case X86::MOV_Fp6464: case X86::MOV_Fp8080: case X86::MOV_Fp3264: case X86::MOV_Fp3280: case X86::MOV_Fp6432: case X86::MOV_Fp8032: - + + // Note that MOVSSrr and MOVSDrr are not considered copies. FR32 and FR64 + // copies are done with FsMOVAPSrr and FsMOVAPDrr. + case X86::FsMOVAPSrr: case X86::FsMOVAPDrr: case X86::MOVAPSrr: case X86::MOVAPDrr: case X86::MOVDQArr: - case X86::MOVSS2PSrr: - case X86::MOVSD2PDrr: - case X86::MOVPS2SSrr: - case X86::MOVPD2SDrr: case X86::MMX_MOVQ64rr: assert(MI.getNumOperands() >= 2 && MI.getOperand(0).isReg() && @@ -1083,7 +1073,7 @@ void X86InstrInfo::reMaterialize(MachineBasicBlock &MBB, case X86::MOV8r0: Opc = X86::MOV8ri; break; case X86::MOV16r0: Opc = X86::MOV16ri; break; case X86::MOV32r0: Opc = X86::MOV32ri; break; - case X86::MOV64r0: Opc = X86::MOV64ri; break; + case X86::MOV64r0: Opc = X86::MOV64ri64i32; break; } Clone = false; } @@ -1860,7 +1850,7 @@ bool X86InstrInfo::copyRegToReg(MachineBasicBlock &MBB, CommonRC = SrcRC; else if (!DestRC->hasSubClass(SrcRC)) { // Neither of GR64_NOREX or GR64_NOSP is a superclass of the other, - // but we want to copy then as GR64. Similarly, for GR32_NOREX and + // but we want to copy them as GR64. Similarly, for GR32_NOREX and // GR32_NOSP, copy as GR32. if (SrcRC->hasSuperClass(&X86::GR64RegClass) && DestRC->hasSuperClass(&X86::GR64RegClass)) diff --git a/lib/Target/X86/X86InstrInfo.td b/lib/Target/X86/X86InstrInfo.td index 25cd297..cfe71a5 100644 --- a/lib/Target/X86/X86InstrInfo.td +++ b/lib/Target/X86/X86InstrInfo.td @@ -343,18 +343,37 @@ def X86_COND_O : PatLeaf<(i8 13)>; def X86_COND_P : PatLeaf<(i8 14)>; // alt. COND_PE def X86_COND_S : PatLeaf<(i8 15)>; -def i16immSExt8 : PatLeaf<(i16 imm), [{ - // i16immSExt8 predicate - True if the 16-bit immediate fits in a 8-bit - // sign extended field. - return (int16_t)N->getZExtValue() == (int8_t)N->getZExtValue(); +def immSext8 : PatLeaf<(imm), [{ + return N->getSExtValue() == (int8_t)N->getSExtValue(); }]>; -def i32immSExt8 : PatLeaf<(i32 imm), [{ - // i32immSExt8 predicate - True if the 32-bit immediate fits in a 8-bit - // sign extended field. - return (int32_t)N->getZExtValue() == (int8_t)N->getZExtValue(); +def i16immSExt8 : PatLeaf<(i16 immSext8)>; +def i32immSExt8 : PatLeaf<(i32 immSext8)>; + +/// Load patterns: these constraint the match to the right address space. +def dsload : PatFrag<(ops node:$ptr), (load node:$ptr), [{ + if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) + if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) + if (PT->getAddressSpace() > 255) + return false; + return true; +}]>; + +def gsload : PatFrag<(ops node:$ptr), (load node:$ptr), [{ + if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) + if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) + return PT->getAddressSpace() == 256; + return false; +}]>; + +def fsload : PatFrag<(ops node:$ptr), (load node:$ptr), [{ + if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) + if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) + return PT->getAddressSpace() == 257; + return false; }]>; + // Helper fragments for loads. // It's always safe to treat a anyext i16 load as a i32 load if the i16 is // known to be 32-bit aligned or better. Ditto for i8 to i16. @@ -372,8 +391,7 @@ def loadi16 : PatFrag<(ops node:$ptr), (i16 (unindexedload node:$ptr)), [{ return false; }]>; -def loadi16_anyext : PatFrag<(ops node:$ptr), (i32 (unindexedload node:$ptr)), -[{ +def loadi16_anyext : PatFrag<(ops node:$ptr), (i32 (unindexedload node:$ptr)),[{ LoadSDNode *LD = cast<LoadSDNode>(N); if (const Value *Src = LD->getSrcValue()) if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) @@ -399,72 +417,11 @@ def loadi32 : PatFrag<(ops node:$ptr), (i32 (unindexedload node:$ptr)), [{ return false; }]>; -def nvloadi32 : PatFrag<(ops node:$ptr), (i32 (unindexedload node:$ptr)), [{ - LoadSDNode *LD = cast<LoadSDNode>(N); - if (const Value *Src = LD->getSrcValue()) - if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) - if (PT->getAddressSpace() > 255) - return false; - if (LD->isVolatile()) - return false; - ISD::LoadExtType ExtType = LD->getExtensionType(); - if (ExtType == ISD::NON_EXTLOAD) - return true; - if (ExtType == ISD::EXTLOAD) - return LD->getAlignment() >= 4; - return false; -}]>; - -def gsload : PatFrag<(ops node:$ptr), (load node:$ptr), [{ - if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) - if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) - return PT->getAddressSpace() == 256; - return false; -}]>; - -def fsload : PatFrag<(ops node:$ptr), (load node:$ptr), [{ - if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) - if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) - return PT->getAddressSpace() == 257; - return false; -}]>; - -def loadi8 : PatFrag<(ops node:$ptr), (i8 (load node:$ptr)), [{ - if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) - if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) - if (PT->getAddressSpace() > 255) - return false; - return true; -}]>; -def loadi64 : PatFrag<(ops node:$ptr), (i64 (load node:$ptr)), [{ - if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) - if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) - if (PT->getAddressSpace() > 255) - return false; - return true; -}]>; - -def loadf32 : PatFrag<(ops node:$ptr), (f32 (load node:$ptr)), [{ - if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) - if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) - if (PT->getAddressSpace() > 255) - return false; - return true; -}]>; -def loadf64 : PatFrag<(ops node:$ptr), (f64 (load node:$ptr)), [{ - if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) - if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) - if (PT->getAddressSpace() > 255) - return false; - return true; -}]>; -def loadf80 : PatFrag<(ops node:$ptr), (f80 (load node:$ptr)), [{ - if (const Value *Src = cast<LoadSDNode>(N)->getSrcValue()) - if (const PointerType *PT = dyn_cast<PointerType>(Src->getType())) - if (PT->getAddressSpace() > 255) - return false; - return true; -}]>; +def loadi8 : PatFrag<(ops node:$ptr), (i8 (dsload node:$ptr))>; +def loadi64 : PatFrag<(ops node:$ptr), (i64 (dsload node:$ptr))>; +def loadf32 : PatFrag<(ops node:$ptr), (f32 (dsload node:$ptr))>; +def loadf64 : PatFrag<(ops node:$ptr), (f64 (dsload node:$ptr))>; +def loadf80 : PatFrag<(ops node:$ptr), (f80 (dsload node:$ptr))>; def sextloadi16i8 : PatFrag<(ops node:$ptr), (i16 (sextloadi8 node:$ptr))>; def sextloadi32i8 : PatFrag<(ops node:$ptr), (i32 (sextloadi8 node:$ptr))>; @@ -1037,7 +994,7 @@ def MOV16rr_REV : I<0x8B, MRMSrcReg, (outs GR16:$dst), (ins GR16:$src), def MOV32rr_REV : I<0x8B, MRMSrcReg, (outs GR32:$dst), (ins GR32:$src), "mov{l}\t{$src, $dst|$dst, $src}", []>; -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in { +let canFoldAsLoad = 1, isReMaterializable = 1 in { def MOV8rm : I<0x8A, MRMSrcMem, (outs GR8 :$dst), (ins i8mem :$src), "mov{b}\t{$src, $dst|$dst, $src}", [(set GR8:$dst, (loadi8 addr:$src))]>; @@ -1071,7 +1028,7 @@ def MOV8mr_NOREX : I<0x88, MRMDestMem, (outs), (ins i8mem_NOREX:$dst, GR8_NOREX:$src), "mov{b}\t{$src, $dst|$dst, $src} # NOREX", []>; let mayLoad = 1, - canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in + canFoldAsLoad = 1, isReMaterializable = 1 in def MOV8rm_NOREX : I<0x8A, MRMSrcMem, (outs GR8_NOREX:$dst), (ins i8mem_NOREX:$src), "mov{b}\t{$src, $dst|$dst, $src} # NOREX", []>; @@ -1156,7 +1113,7 @@ def IMUL32m : I<0xF7, MRM5m, (outs), (ins i32mem:$src), } // neverHasSideEffects // unsigned division/remainder -let Defs = [AL,AH,EFLAGS], Uses = [AX] in +let Defs = [AX,EFLAGS], Uses = [AX] in def DIV8r : I<0xF6, MRM6r, (outs), (ins GR8:$src), // AX/r8 = AL,AH "div{b}\t$src", []>; let Defs = [AX,DX,EFLAGS], Uses = [AX,DX] in @@ -4442,12 +4399,6 @@ def : Pat<(i16 (anyext GR8 :$src)), (MOVZX16rr8 GR8 :$src)>; def : Pat<(i32 (anyext GR8 :$src)), (MOVZX32rr8 GR8 :$src)>; def : Pat<(i32 (anyext GR16:$src)), (MOVZX32rr16 GR16:$src)>; -// (and (i32 load), 255) -> (zextload i8) -def : Pat<(i32 (and (nvloadi32 addr:$src), (i32 255))), - (MOVZX32rm8 addr:$src)>; -def : Pat<(i32 (and (nvloadi32 addr:$src), (i32 65535))), - (MOVZX32rm16 addr:$src)>; - //===----------------------------------------------------------------------===// // Some peepholes //===----------------------------------------------------------------------===// @@ -4543,43 +4494,43 @@ def : Pat<(shl GR16:$src1, (i8 1)), (ADD16rr GR16:$src1, GR16:$src1)>; def : Pat<(shl GR32:$src1, (i8 1)), (ADD32rr GR32:$src1, GR32:$src1)>; // (shl x (and y, 31)) ==> (shl x, y) -def : Pat<(shl GR8:$src1, (and CL:$amt, 31)), +def : Pat<(shl GR8:$src1, (and CL, 31)), (SHL8rCL GR8:$src1)>; -def : Pat<(shl GR16:$src1, (and CL:$amt, 31)), +def : Pat<(shl GR16:$src1, (and CL, 31)), (SHL16rCL GR16:$src1)>; -def : Pat<(shl GR32:$src1, (and CL:$amt, 31)), +def : Pat<(shl GR32:$src1, (and CL, 31)), (SHL32rCL GR32:$src1)>; -def : Pat<(store (shl (loadi8 addr:$dst), (and CL:$amt, 31)), addr:$dst), +def : Pat<(store (shl (loadi8 addr:$dst), (and CL, 31)), addr:$dst), (SHL8mCL addr:$dst)>; -def : Pat<(store (shl (loadi16 addr:$dst), (and CL:$amt, 31)), addr:$dst), +def : Pat<(store (shl (loadi16 addr:$dst), (and CL, 31)), addr:$dst), (SHL16mCL addr:$dst)>; -def : Pat<(store (shl (loadi32 addr:$dst), (and CL:$amt, 31)), addr:$dst), +def : Pat<(store (shl (loadi32 addr:$dst), (and CL, 31)), addr:$dst), (SHL32mCL addr:$dst)>; -def : Pat<(srl GR8:$src1, (and CL:$amt, 31)), +def : Pat<(srl GR8:$src1, (and CL, 31)), (SHR8rCL GR8:$src1)>; -def : Pat<(srl GR16:$src1, (and CL:$amt, 31)), +def : Pat<(srl GR16:$src1, (and CL, 31)), (SHR16rCL GR16:$src1)>; -def : Pat<(srl GR32:$src1, (and CL:$amt, 31)), +def : Pat<(srl GR32:$src1, (and CL, 31)), (SHR32rCL GR32:$src1)>; -def : Pat<(store (srl (loadi8 addr:$dst), (and CL:$amt, 31)), addr:$dst), +def : Pat<(store (srl (loadi8 addr:$dst), (and CL, 31)), addr:$dst), (SHR8mCL addr:$dst)>; -def : Pat<(store (srl (loadi16 addr:$dst), (and CL:$amt, 31)), addr:$dst), +def : Pat<(store (srl (loadi16 addr:$dst), (and CL, 31)), addr:$dst), (SHR16mCL addr:$dst)>; -def : Pat<(store (srl (loadi32 addr:$dst), (and CL:$amt, 31)), addr:$dst), +def : Pat<(store (srl (loadi32 addr:$dst), (and CL, 31)), addr:$dst), (SHR32mCL addr:$dst)>; -def : Pat<(sra GR8:$src1, (and CL:$amt, 31)), +def : Pat<(sra GR8:$src1, (and CL, 31)), (SAR8rCL GR8:$src1)>; -def : Pat<(sra GR16:$src1, (and CL:$amt, 31)), +def : Pat<(sra GR16:$src1, (and CL, 31)), (SAR16rCL GR16:$src1)>; -def : Pat<(sra GR32:$src1, (and CL:$amt, 31)), +def : Pat<(sra GR32:$src1, (and CL, 31)), (SAR32rCL GR32:$src1)>; -def : Pat<(store (sra (loadi8 addr:$dst), (and CL:$amt, 31)), addr:$dst), +def : Pat<(store (sra (loadi8 addr:$dst), (and CL, 31)), addr:$dst), (SAR8mCL addr:$dst)>; -def : Pat<(store (sra (loadi16 addr:$dst), (and CL:$amt, 31)), addr:$dst), +def : Pat<(store (sra (loadi16 addr:$dst), (and CL, 31)), addr:$dst), (SAR16mCL addr:$dst)>; -def : Pat<(store (sra (loadi32 addr:$dst), (and CL:$amt, 31)), addr:$dst), +def : Pat<(store (sra (loadi32 addr:$dst), (and CL, 31)), addr:$dst), (SAR32mCL addr:$dst)>; // (or (x >> c) | (y << (32 - c))) ==> (shrd32 x, y, c) @@ -4600,11 +4551,11 @@ def : Pat<(store (or (srl (loadi32 addr:$dst), (i8 (trunc ECX:$amt))), addr:$dst), (SHRD32mrCL addr:$dst, GR32:$src2)>; -def : Pat<(shrd GR32:$src1, (i8 imm:$amt1), GR32:$src2, (i8 imm:$amt2)), +def : Pat<(shrd GR32:$src1, (i8 imm:$amt1), GR32:$src2, (i8 imm/*:$amt2*/)), (SHRD32rri8 GR32:$src1, GR32:$src2, (i8 imm:$amt1))>; def : Pat<(store (shrd (loadi32 addr:$dst), (i8 imm:$amt1), - GR32:$src2, (i8 imm:$amt2)), addr:$dst), + GR32:$src2, (i8 imm/*:$amt2*/)), addr:$dst), (SHRD32mri8 addr:$dst, GR32:$src2, (i8 imm:$amt1))>; // (or (x << c) | (y >> (32 - c))) ==> (shld32 x, y, c) @@ -4625,11 +4576,11 @@ def : Pat<(store (or (shl (loadi32 addr:$dst), (i8 (trunc ECX:$amt))), addr:$dst), (SHLD32mrCL addr:$dst, GR32:$src2)>; -def : Pat<(shld GR32:$src1, (i8 imm:$amt1), GR32:$src2, (i8 imm:$amt2)), +def : Pat<(shld GR32:$src1, (i8 imm:$amt1), GR32:$src2, (i8 imm/*:$amt2*/)), (SHLD32rri8 GR32:$src1, GR32:$src2, (i8 imm:$amt1))>; def : Pat<(store (shld (loadi32 addr:$dst), (i8 imm:$amt1), - GR32:$src2, (i8 imm:$amt2)), addr:$dst), + GR32:$src2, (i8 imm/*:$amt2*/)), addr:$dst), (SHLD32mri8 addr:$dst, GR32:$src2, (i8 imm:$amt1))>; // (or (x >> c) | (y << (16 - c))) ==> (shrd16 x, y, c) @@ -4650,11 +4601,11 @@ def : Pat<(store (or (srl (loadi16 addr:$dst), (i8 (trunc CX:$amt))), addr:$dst), (SHRD16mrCL addr:$dst, GR16:$src2)>; -def : Pat<(shrd GR16:$src1, (i8 imm:$amt1), GR16:$src2, (i8 imm:$amt2)), +def : Pat<(shrd GR16:$src1, (i8 imm:$amt1), GR16:$src2, (i8 imm/*:$amt2*/)), (SHRD16rri8 GR16:$src1, GR16:$src2, (i8 imm:$amt1))>; def : Pat<(store (shrd (loadi16 addr:$dst), (i8 imm:$amt1), - GR16:$src2, (i8 imm:$amt2)), addr:$dst), + GR16:$src2, (i8 imm/*:$amt2*/)), addr:$dst), (SHRD16mri8 addr:$dst, GR16:$src2, (i8 imm:$amt1))>; // (or (x << c) | (y >> (16 - c))) ==> (shld16 x, y, c) @@ -4675,11 +4626,11 @@ def : Pat<(store (or (shl (loadi16 addr:$dst), (i8 (trunc CX:$amt))), addr:$dst), (SHLD16mrCL addr:$dst, GR16:$src2)>; -def : Pat<(shld GR16:$src1, (i8 imm:$amt1), GR16:$src2, (i8 imm:$amt2)), +def : Pat<(shld GR16:$src1, (i8 imm:$amt1), GR16:$src2, (i8 imm/*:$amt2*/)), (SHLD16rri8 GR16:$src1, GR16:$src2, (i8 imm:$amt1))>; def : Pat<(store (shld (loadi16 addr:$dst), (i8 imm:$amt1), - GR16:$src2, (i8 imm:$amt2)), addr:$dst), + GR16:$src2, (i8 imm/*:$amt2*/)), addr:$dst), (SHLD16mri8 addr:$dst, GR16:$src2, (i8 imm:$amt1))>; // (anyext (setcc_carry)) -> (setcc_carry) diff --git a/lib/Target/X86/X86InstrMMX.td b/lib/Target/X86/X86InstrMMX.td index 89f020c..c8e0723 100644 --- a/lib/Target/X86/X86InstrMMX.td +++ b/lib/Target/X86/X86InstrMMX.td @@ -141,7 +141,7 @@ def MMX_MOVD64rrv164 : MMXI<0x6E, MRMSrcReg, (outs VR64:$dst), (ins GR64:$src), let neverHasSideEffects = 1 in def MMX_MOVQ64rr : MMXI<0x6F, MRMSrcReg, (outs VR64:$dst), (ins VR64:$src), "movq\t{$src, $dst|$dst, $src}", []>; -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def MMX_MOVQ64rm : MMXI<0x6F, MRMSrcMem, (outs VR64:$dst), (ins i64mem:$src), "movq\t{$src, $dst|$dst, $src}", [(set VR64:$dst, (load_mmx addr:$src))]>; @@ -426,13 +426,15 @@ def MMX_CVTTPS2PIrm : MMXI<0x2C, MRMSrcMem, (outs VR64:$dst), (ins f64mem:$src), // Extract / Insert -def MMX_X86pextrw : SDNode<"X86ISD::PEXTRW", SDTypeProfile<1, 2, []>, []>; -def MMX_X86pinsrw : SDNode<"X86ISD::PINSRW", SDTypeProfile<1, 3, []>, []>; +def MMX_X86pinsrw : SDNode<"X86ISD::MMX_PINSRW", + SDTypeProfile<1, 3, [SDTCisVT<0, v4i16>, SDTCisSameAs<0,1>, + SDTCisVT<2, i32>, SDTCisPtrTy<3>]>>; + def MMX_PEXTRWri : MMXIi8<0xC5, MRMSrcReg, (outs GR32:$dst), (ins VR64:$src1, i16i8imm:$src2), "pextrw\t{$src2, $src1, $dst|$dst, $src1, $src2}", - [(set GR32:$dst, (MMX_X86pextrw (v4i16 VR64:$src1), + [(set GR32:$dst, (X86pextrw (v4i16 VR64:$src1), (iPTR imm:$src2)))]>; let Constraints = "$src1 = $dst" in { def MMX_PINSRWrri : MMXIi8<0xC4, MRMSrcReg, @@ -597,13 +599,6 @@ let AddedComplexity = 10 in { (MMX_PUNPCKHDQrr VR64:$src, VR64:$src)>; } -// Patterns to perform vector shuffling with a zeroed out vector. -let AddedComplexity = 20 in { - def : Pat<(bc_v2i32 (mmx_unpckl immAllZerosV, - (v2i32 (scalar_to_vector (load_mmx addr:$src))))), - (MMX_PUNPCKLDQrm VR64:$src, VR64:$src)>; -} - // Some special case PANDN patterns. // FIXME: Get rid of these. def : Pat<(v1i64 (and (xor VR64:$src1, (bc_v1i64 (v2i32 immAllOnesV))), diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td index 9b2140f..2743dba 100644 --- a/lib/Target/X86/X86InstrSSE.td +++ b/lib/Target/X86/X86InstrSSE.td @@ -160,6 +160,32 @@ def memopv4i16 : PatFrag<(ops node:$ptr), (v4i16 (memop64 node:$ptr))>; def memopv8i16 : PatFrag<(ops node:$ptr), (v8i16 (memop64 node:$ptr))>; def memopv2i32 : PatFrag<(ops node:$ptr), (v2i32 (memop64 node:$ptr))>; +// MOVNT Support +// Like 'store', but requires the non-temporal bit to be set +def nontemporalstore : PatFrag<(ops node:$val, node:$ptr), + (st node:$val, node:$ptr), [{ + if (StoreSDNode *ST = dyn_cast<StoreSDNode>(N)) + return ST->isNonTemporal(); + return false; +}]>; + +def alignednontemporalstore : PatFrag<(ops node:$val, node:$ptr), + (st node:$val, node:$ptr), [{ + if (StoreSDNode *ST = dyn_cast<StoreSDNode>(N)) + return ST->isNonTemporal() && !ST->isTruncatingStore() && + ST->getAddressingMode() == ISD::UNINDEXED && + ST->getAlignment() >= 16; + return false; +}]>; + +def unalignednontemporalstore : PatFrag<(ops node:$val, node:$ptr), + (st node:$val, node:$ptr), [{ + if (StoreSDNode *ST = dyn_cast<StoreSDNode>(N)) + return ST->isNonTemporal() && + ST->getAlignment() < 16; + return false; +}]>; + def bc_v4f32 : PatFrag<(ops node:$in), (v4f32 (bitconvert node:$in))>; def bc_v2f64 : PatFrag<(ops node:$in), (v2f64 (bitconvert node:$in))>; def bc_v16i8 : PatFrag<(ops node:$in), (v16i8 (bitconvert node:$in))>; @@ -344,18 +370,56 @@ let Uses = [EFLAGS], usesCustomInserter = 1 in { // SSE1 Instructions //===----------------------------------------------------------------------===// -// Move Instructions -let neverHasSideEffects = 1 in -def MOVSSrr : SSI<0x10, MRMSrcReg, (outs FR32:$dst), (ins FR32:$src), - "movss\t{$src, $dst|$dst, $src}", []>; -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +// Move Instructions. Register-to-register movss is not used for FR32 +// register copies because it's a partial register update; FsMOVAPSrr is +// used instead. Register-to-register movss is not modeled as an INSERT_SUBREG +// because INSERT_SUBREG requires that the insert be implementable in terms of +// a copy, and just mentioned, we don't use movss for copies. +let Constraints = "$src1 = $dst" in +def MOVSSrr : SSI<0x10, MRMSrcReg, + (outs VR128:$dst), (ins VR128:$src1, FR32:$src2), + "movss\t{$src2, $dst|$dst, $src2}", + [(set VR128:$dst, + (movl VR128:$src1, (scalar_to_vector FR32:$src2)))]>; + +// Extract the low 32-bit value from one vector and insert it into another. +let AddedComplexity = 15 in +def : Pat<(v4f32 (movl VR128:$src1, VR128:$src2)), + (MOVSSrr VR128:$src1, + (EXTRACT_SUBREG (v4f32 VR128:$src2), x86_subreg_ss))>; + +// Implicitly promote a 32-bit scalar to a vector. +def : Pat<(v4f32 (scalar_to_vector FR32:$src)), + (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), FR32:$src, x86_subreg_ss)>; + +// Loading from memory automatically zeroing upper bits. +let canFoldAsLoad = 1, isReMaterializable = 1 in def MOVSSrm : SSI<0x10, MRMSrcMem, (outs FR32:$dst), (ins f32mem:$src), "movss\t{$src, $dst|$dst, $src}", [(set FR32:$dst, (loadf32 addr:$src))]>; + +// MOVSSrm zeros the high parts of the register; represent this +// with SUBREG_TO_REG. +let AddedComplexity = 20 in { +def : Pat<(v4f32 (X86vzmovl (v4f32 (scalar_to_vector (loadf32 addr:$src))))), + (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), x86_subreg_ss)>; +def : Pat<(v4f32 (scalar_to_vector (loadf32 addr:$src))), + (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), x86_subreg_ss)>; +def : Pat<(v4f32 (X86vzmovl (loadv4f32 addr:$src))), + (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), x86_subreg_ss)>; +} + +// Store scalar value to memory. def MOVSSmr : SSI<0x11, MRMDestMem, (outs), (ins f32mem:$dst, FR32:$src), "movss\t{$src, $dst|$dst, $src}", [(store FR32:$src, addr:$dst)]>; +// Extract and store. +def : Pat<(store (f32 (vector_extract (v4f32 VR128:$src), (iPTR 0))), + addr:$dst), + (MOVSSmr addr:$dst, + (EXTRACT_SUBREG (v4f32 VR128:$src), x86_subreg_ss))>; + // Conversion instructions def CVTTSS2SIrr : SSI<0x2C, MRMSrcReg, (outs GR32:$dst), (ins FR32:$src), "cvttss2si\t{$src, $dst|$dst, $src}", @@ -518,7 +582,7 @@ def FsMOVAPSrr : PSI<0x28, MRMSrcReg, (outs FR32:$dst), (ins FR32:$src), // Alias instruction to load FR32 from f128mem using movaps. Upper bits are // disregarded. -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def FsMOVAPSrm : PSI<0x28, MRMSrcMem, (outs FR32:$dst), (ins f128mem:$src), "movaps\t{$src, $dst|$dst, $src}", [(set FR32:$dst, (alignedloadfsf32 addr:$src))]>; @@ -715,7 +779,7 @@ defm MIN : sse1_fp_binop_rm<0x5D, "min", X86fmin, let neverHasSideEffects = 1 in def MOVAPSrr : PSI<0x28, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src), "movaps\t{$src, $dst|$dst, $src}", []>; -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def MOVAPSrm : PSI<0x28, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src), "movaps\t{$src, $dst|$dst, $src}", [(set VR128:$dst, (alignedloadv4f32 addr:$src))]>; @@ -727,7 +791,7 @@ def MOVAPSmr : PSI<0x29, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src), let neverHasSideEffects = 1 in def MOVUPSrr : PSI<0x10, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src), "movups\t{$src, $dst|$dst, $src}", []>; -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def MOVUPSrm : PSI<0x10, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src), "movups\t{$src, $dst|$dst, $src}", [(set VR128:$dst, (loadv4f32 addr:$src))]>; @@ -736,7 +800,7 @@ def MOVUPSmr : PSI<0x11, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src), [(store (v4f32 VR128:$src), addr:$dst)]>; // Intrinsic forms of MOVUPS load and store -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def MOVUPSrm_Int : PSI<0x10, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src), "movups\t{$src, $dst|$dst, $src}", [(set VR128:$dst, (int_x86_sse_loadu_ps addr:$src))]>; @@ -796,9 +860,9 @@ def MOVHLPSrr : PSI<0x12, MRMSrcReg, (outs VR128:$dst), let AddedComplexity = 20 in { def : Pat<(v4f32 (movddup VR128:$src, (undef))), - (MOVLHPSrr VR128:$src, VR128:$src)>, Requires<[HasSSE1]>; + (MOVLHPSrr VR128:$src, VR128:$src)>; def : Pat<(v2i64 (movddup VR128:$src, (undef))), - (MOVLHPSrr VR128:$src, VR128:$src)>, Requires<[HasSSE1]>; + (MOVLHPSrr VR128:$src, VR128:$src)>; } @@ -1013,10 +1077,33 @@ def PREFETCHNTA : PSI<0x18, MRM0m, (outs), (ins i8mem:$src), "prefetchnta\t$src", [(prefetch addr:$src, imm, (i32 0))]>; // Non-temporal stores -def MOVNTPSmr : PSI<0x2B, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src), +def MOVNTPSmr_Int : PSI<0x2B, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src), "movntps\t{$src, $dst|$dst, $src}", [(int_x86_sse_movnt_ps addr:$dst, VR128:$src)]>; +let AddedComplexity = 400 in { // Prefer non-temporal versions +def MOVNTPSmr : PSI<0x2B, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src), + "movntps\t{$src, $dst|$dst, $src}", + [(alignednontemporalstore (v4f32 VR128:$src), addr:$dst)]>; + +def MOVNTDQ_64mr : PDI<0xE7, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src), + "movntdq\t{$src, $dst|$dst, $src}", + [(alignednontemporalstore (v2f64 VR128:$src), addr:$dst)]>; + +def : Pat<(alignednontemporalstore (v2i64 VR128:$src), addr:$dst), + (MOVNTDQ_64mr VR128:$src, addr:$dst)>; + +def MOVNTImr : I<0xC3, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src), + "movnti\t{$src, $dst|$dst, $src}", + [(nontemporalstore (i32 GR32:$src), addr:$dst)]>, + TB, Requires<[HasSSE2]>; + +def MOVNTI_64mr : RI<0xC3, MRMDestMem, (outs), (ins i64mem:$dst, GR64:$src), + "movnti\t{$src, $dst|$dst, $src}", + [(nontemporalstore (i64 GR64:$src), addr:$dst)]>, + TB, Requires<[HasSSE2]>; +} + // Load, store, and memory fence def SFENCE : PSI<0xAE, MRM7r, (outs), (ins), "sfence", [(int_x86_sse_sfence)]>; @@ -1035,84 +1122,73 @@ let isReMaterializable = 1, isAsCheapAsAMove = 1, canFoldAsLoad = 1, def V_SET0 : PSI<0x57, MRMInitReg, (outs VR128:$dst), (ins), "", [(set VR128:$dst, (v4i32 immAllZerosV))]>; -let Predicates = [HasSSE1] in { - def : Pat<(v2i64 immAllZerosV), (V_SET0)>; - def : Pat<(v8i16 immAllZerosV), (V_SET0)>; - def : Pat<(v16i8 immAllZerosV), (V_SET0)>; - def : Pat<(v2f64 immAllZerosV), (V_SET0)>; - def : Pat<(v4f32 immAllZerosV), (V_SET0)>; -} - -// FR32 to 128-bit vector conversion. -let isAsCheapAsAMove = 1 in -def MOVSS2PSrr : SSI<0x10, MRMSrcReg, (outs VR128:$dst), (ins FR32:$src), - "movss\t{$src, $dst|$dst, $src}", - [(set VR128:$dst, - (v4f32 (scalar_to_vector FR32:$src)))]>; -def MOVSS2PSrm : SSI<0x10, MRMSrcMem, (outs VR128:$dst), (ins f32mem:$src), - "movss\t{$src, $dst|$dst, $src}", - [(set VR128:$dst, - (v4f32 (scalar_to_vector (loadf32 addr:$src))))]>; - -// FIXME: may not be able to eliminate this movss with coalescing the src and -// dest register classes are different. We really want to write this pattern -// like this: -// def : Pat<(f32 (vector_extract (v4f32 VR128:$src), (iPTR 0))), -// (f32 FR32:$src)>; -let isAsCheapAsAMove = 1 in -def MOVPS2SSrr : SSI<0x10, MRMSrcReg, (outs FR32:$dst), (ins VR128:$src), - "movss\t{$src, $dst|$dst, $src}", - [(set FR32:$dst, (vector_extract (v4f32 VR128:$src), - (iPTR 0)))]>; -def MOVPS2SSmr : SSI<0x11, MRMDestMem, (outs), (ins f32mem:$dst, VR128:$src), - "movss\t{$src, $dst|$dst, $src}", - [(store (f32 (vector_extract (v4f32 VR128:$src), - (iPTR 0))), addr:$dst)]>; - - -// Move to lower bits of a VR128, leaving upper bits alone. -// Three operand (but two address) aliases. -let Constraints = "$src1 = $dst" in { -let neverHasSideEffects = 1 in - def MOVLSS2PSrr : SSI<0x10, MRMSrcReg, - (outs VR128:$dst), (ins VR128:$src1, FR32:$src2), - "movss\t{$src2, $dst|$dst, $src2}", []>; - - let AddedComplexity = 15 in - def MOVLPSrr : SSI<0x10, MRMSrcReg, - (outs VR128:$dst), (ins VR128:$src1, VR128:$src2), - "movss\t{$src2, $dst|$dst, $src2}", - [(set VR128:$dst, - (v4f32 (movl VR128:$src1, VR128:$src2)))]>; -} +def : Pat<(v2i64 immAllZerosV), (V_SET0)>; +def : Pat<(v8i16 immAllZerosV), (V_SET0)>; +def : Pat<(v16i8 immAllZerosV), (V_SET0)>; +def : Pat<(v2f64 immAllZerosV), (V_SET0)>; +def : Pat<(v4f32 immAllZerosV), (V_SET0)>; -// Move to lower bits of a VR128 and zeroing upper bits. -// Loading from memory automatically zeroing upper bits. -let AddedComplexity = 20 in -def MOVZSS2PSrm : SSI<0x10, MRMSrcMem, (outs VR128:$dst), (ins f32mem:$src), - "movss\t{$src, $dst|$dst, $src}", - [(set VR128:$dst, (v4f32 (X86vzmovl (v4f32 (scalar_to_vector - (loadf32 addr:$src))))))]>; - -def : Pat<(v4f32 (X86vzmovl (loadv4f32 addr:$src))), - (MOVZSS2PSrm addr:$src)>; +def : Pat<(f32 (vector_extract (v4f32 VR128:$src), (iPTR 0))), + (f32 (EXTRACT_SUBREG (v4f32 VR128:$src), x86_subreg_ss))>; //===---------------------------------------------------------------------===// // SSE2 Instructions //===---------------------------------------------------------------------===// -// Move Instructions -let neverHasSideEffects = 1 in -def MOVSDrr : SDI<0x10, MRMSrcReg, (outs FR64:$dst), (ins FR64:$src), - "movsd\t{$src, $dst|$dst, $src}", []>; -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +// Move Instructions. Register-to-register movsd is not used for FR64 +// register copies because it's a partial register update; FsMOVAPDrr is +// used instead. Register-to-register movsd is not modeled as an INSERT_SUBREG +// because INSERT_SUBREG requires that the insert be implementable in terms of +// a copy, and just mentioned, we don't use movsd for copies. +let Constraints = "$src1 = $dst" in +def MOVSDrr : SDI<0x10, MRMSrcReg, + (outs VR128:$dst), (ins VR128:$src1, FR64:$src2), + "movsd\t{$src2, $dst|$dst, $src2}", + [(set VR128:$dst, + (movl VR128:$src1, (scalar_to_vector FR64:$src2)))]>; + +// Extract the low 64-bit value from one vector and insert it into another. +let AddedComplexity = 15 in +def : Pat<(v2f64 (movl VR128:$src1, VR128:$src2)), + (MOVSDrr VR128:$src1, + (EXTRACT_SUBREG (v2f64 VR128:$src2), x86_subreg_sd))>; + +// Implicitly promote a 64-bit scalar to a vector. +def : Pat<(v2f64 (scalar_to_vector FR64:$src)), + (INSERT_SUBREG (v2f64 (IMPLICIT_DEF)), FR64:$src, x86_subreg_sd)>; + +// Loading from memory automatically zeroing upper bits. +let canFoldAsLoad = 1, isReMaterializable = 1, AddedComplexity = 20 in def MOVSDrm : SDI<0x10, MRMSrcMem, (outs FR64:$dst), (ins f64mem:$src), "movsd\t{$src, $dst|$dst, $src}", [(set FR64:$dst, (loadf64 addr:$src))]>; + +// MOVSDrm zeros the high parts of the register; represent this +// with SUBREG_TO_REG. +let AddedComplexity = 20 in { +def : Pat<(v2f64 (X86vzmovl (v2f64 (scalar_to_vector (loadf64 addr:$src))))), + (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>; +def : Pat<(v2f64 (scalar_to_vector (loadf64 addr:$src))), + (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>; +def : Pat<(v2f64 (X86vzmovl (loadv2f64 addr:$src))), + (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>; +def : Pat<(v2f64 (X86vzmovl (bc_v2f64 (loadv4f32 addr:$src)))), + (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>; +def : Pat<(v2f64 (X86vzload addr:$src)), + (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), x86_subreg_sd)>; +} + +// Store scalar value to memory. def MOVSDmr : SDI<0x11, MRMDestMem, (outs), (ins f64mem:$dst, FR64:$src), "movsd\t{$src, $dst|$dst, $src}", [(store FR64:$src, addr:$dst)]>; +// Extract and store. +def : Pat<(store (f64 (vector_extract (v2f64 VR128:$src), (iPTR 0))), + addr:$dst), + (MOVSDmr addr:$dst, + (EXTRACT_SUBREG (v2f64 VR128:$src), x86_subreg_sd))>; + // Conversion instructions def CVTTSD2SIrr : SDI<0x2C, MRMSrcReg, (outs GR32:$dst), (ins FR64:$src), "cvttsd2si\t{$src, $dst|$dst, $src}", @@ -1166,7 +1242,8 @@ def CVTSS2SDrm : I<0x5A, MRMSrcMem, (outs FR64:$dst), (ins f32mem:$src), Requires<[HasSSE2, OptForSize]>; def : Pat<(extloadf32 addr:$src), - (CVTSS2SDrr (MOVSSrm addr:$src))>, Requires<[HasSSE2, OptForSpeed]>; + (CVTSS2SDrr (MOVSSrm addr:$src))>, + Requires<[HasSSE2, OptForSpeed]>; // Match intrinsics which expect XMM operand(s). def Int_CVTSD2SIrr : SDI<0x2D, MRMSrcReg, (outs GR32:$dst), (ins VR128:$src), @@ -1285,7 +1362,7 @@ def FsMOVAPDrr : PDI<0x28, MRMSrcReg, (outs FR64:$dst), (ins FR64:$src), // Alias instruction to load FR64 from f128mem using movapd. Upper bits are // disregarded. -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def FsMOVAPDrm : PDI<0x28, MRMSrcMem, (outs FR64:$dst), (ins f128mem:$src), "movapd\t{$src, $dst|$dst, $src}", [(set FR64:$dst, (alignedloadfsf64 addr:$src))]>; @@ -1483,7 +1560,7 @@ defm MIN : sse2_fp_binop_rm<0x5D, "min", X86fmin, let neverHasSideEffects = 1 in def MOVAPDrr : PDI<0x28, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src), "movapd\t{$src, $dst|$dst, $src}", []>; -let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in +let canFoldAsLoad = 1, isReMaterializable = 1 in def MOVAPDrm : PDI<0x28, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src), "movapd\t{$src, $dst|$dst, $src}", [(set VR128:$dst, (alignedloadv2f64 addr:$src))]>; @@ -2298,17 +2375,30 @@ def MASKMOVDQU64 : PDI<0xF7, MRMSrcReg, (outs), (ins VR128:$src, VR128:$mask), [(int_x86_sse2_maskmov_dqu VR128:$src, VR128:$mask, RDI)]>; // Non-temporal stores -def MOVNTPDmr : PDI<0x2B, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src), - "movntpd\t{$src, $dst|$dst, $src}", - [(int_x86_sse2_movnt_pd addr:$dst, VR128:$src)]>; -def MOVNTDQmr : PDI<0xE7, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src), - "movntdq\t{$src, $dst|$dst, $src}", - [(int_x86_sse2_movnt_dq addr:$dst, VR128:$src)]>; -def MOVNTImr : I<0xC3, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src), +def MOVNTPDmr_Int : PDI<0x2B, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src), + "movntpd\t{$src, $dst|$dst, $src}", + [(int_x86_sse2_movnt_pd addr:$dst, VR128:$src)]>; +def MOVNTDQmr_Int : PDI<0xE7, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src), + "movntdq\t{$src, $dst|$dst, $src}", + [(int_x86_sse2_movnt_dq addr:$dst, VR128:$src)]>; +def MOVNTImr_Int : I<0xC3, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src), "movnti\t{$src, $dst|$dst, $src}", [(int_x86_sse2_movnt_i addr:$dst, GR32:$src)]>, TB, Requires<[HasSSE2]>; +let AddedComplexity = 400 in { // Prefer non-temporal versions +def MOVNTPDmr : PDI<0x2B, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src), + "movntpd\t{$src, $dst|$dst, $src}", + [(alignednontemporalstore(v2f64 VR128:$src), addr:$dst)]>; + +def MOVNTDQmr : PDI<0xE7, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src), + "movntdq\t{$src, $dst|$dst, $src}", + [(alignednontemporalstore (v4f32 VR128:$src), addr:$dst)]>; + +def : Pat<(alignednontemporalstore (v4i32 VR128:$src), addr:$dst), + (MOVNTDQmr VR128:$src, addr:$dst)>; +} + // Flush cache def CLFLUSH : I<0xAE, MRM7m, (outs), (ins i8mem:$src), "clflush\t$src", [(int_x86_sse2_clflush addr:$src)]>, @@ -2321,11 +2411,11 @@ def MFENCE : I<0xAE, MRM_F0, (outs), (ins), "mfence", [(int_x86_sse2_mfence)]>, TB, Requires<[HasSSE2]>; //TODO: custom lower this so as to never even generate the noop -def : Pat<(membarrier (i8 imm:$ll), (i8 imm:$ls), (i8 imm:$sl), (i8 imm:$ss), +def : Pat<(membarrier (i8 imm), (i8 imm), (i8 imm), (i8 imm), (i8 0)), (NOOP)>; def : Pat<(membarrier (i8 0), (i8 0), (i8 0), (i8 1), (i8 1)), (SFENCE)>; def : Pat<(membarrier (i8 1), (i8 0), (i8 0), (i8 0), (i8 1)), (LFENCE)>; -def : Pat<(membarrier (i8 imm:$ll), (i8 imm:$ls), (i8 imm:$sl), (i8 imm:$ss), +def : Pat<(membarrier (i8 imm), (i8 imm), (i8 imm), (i8 imm), (i8 1)), (MFENCE)>; // Alias instructions that map zero vector to pxor / xorp* for sse. @@ -2337,17 +2427,6 @@ let isReMaterializable = 1, isAsCheapAsAMove = 1, canFoldAsLoad = 1, def V_SETALLONES : PDI<0x76, MRMInitReg, (outs VR128:$dst), (ins), "", [(set VR128:$dst, (v4i32 immAllOnesV))]>; -// FR64 to 128-bit vector conversion. -let isAsCheapAsAMove = 1 in -def MOVSD2PDrr : SDI<0x10, MRMSrcReg, (outs VR128:$dst), (ins FR64:$src), - "movsd\t{$src, $dst|$dst, $src}", - [(set VR128:$dst, - (v2f64 (scalar_to_vector FR64:$src)))]>; -def MOVSD2PDrm : SDI<0x10, MRMSrcMem, (outs VR128:$dst), (ins f64mem:$src), - "movsd\t{$src, $dst|$dst, $src}", - [(set VR128:$dst, - (v2f64 (scalar_to_vector (loadf64 addr:$src))))]>; - def MOVDI2PDIrr : PDI<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR32:$src), "movd\t{$src, $dst|$dst, $src}", [(set VR128:$dst, @@ -2376,20 +2455,9 @@ def MOVPQI2QImr : PDI<0xD6, MRMDestMem, (outs), (ins i64mem:$dst, VR128:$src), [(store (i64 (vector_extract (v2i64 VR128:$src), (iPTR 0))), addr:$dst)]>; -// FIXME: may not be able to eliminate this movss with coalescing the src and -// dest register classes are different. We really want to write this pattern -// like this: -// def : Pat<(f32 (vector_extract (v4f32 VR128:$src), (iPTR 0))), -// (f32 FR32:$src)>; -let isAsCheapAsAMove = 1 in -def MOVPD2SDrr : SDI<0x10, MRMSrcReg, (outs FR64:$dst), (ins VR128:$src), - "movsd\t{$src, $dst|$dst, $src}", - [(set FR64:$dst, (vector_extract (v2f64 VR128:$src), - (iPTR 0)))]>; -def MOVPD2SDmr : SDI<0x11, MRMDestMem, (outs), (ins f64mem:$dst, VR128:$src), - "movsd\t{$src, $dst|$dst, $src}", - [(store (f64 (vector_extract (v2f64 VR128:$src), - (iPTR 0))), addr:$dst)]>; +def : Pat<(f64 (vector_extract (v2f64 VR128:$src), (iPTR 0))), + (f64 (EXTRACT_SUBREG (v2f64 VR128:$src), x86_subreg_sd))>; + def MOVPDI2DIrr : PDI<0x7E, MRMDestReg, (outs GR32:$dst), (ins VR128:$src), "movd\t{$src, $dst|$dst, $src}", [(set GR32:$dst, (vector_extract (v4i32 VR128:$src), @@ -2406,44 +2474,11 @@ def MOVSS2DImr : PDI<0x7E, MRMDestMem, (outs), (ins i32mem:$dst, FR32:$src), "movd\t{$src, $dst|$dst, $src}", [(store (i32 (bitconvert FR32:$src)), addr:$dst)]>; - -// Move to lower bits of a VR128, leaving upper bits alone. -// Three operand (but two address) aliases. -let Constraints = "$src1 = $dst" in { - let neverHasSideEffects = 1 in - def MOVLSD2PDrr : SDI<0x10, MRMSrcReg, - (outs VR128:$dst), (ins VR128:$src1, FR64:$src2), - "movsd\t{$src2, $dst|$dst, $src2}", []>; - - let AddedComplexity = 15 in - def MOVLPDrr : SDI<0x10, MRMSrcReg, - (outs VR128:$dst), (ins VR128:$src1, VR128:$src2), - "movsd\t{$src2, $dst|$dst, $src2}", - [(set VR128:$dst, - (v2f64 (movl VR128:$src1, VR128:$src2)))]>; -} - // Store / copy lower 64-bits of a XMM register. def MOVLQ128mr : PDI<0xD6, MRMDestMem, (outs), (ins i64mem:$dst, VR128:$src), "movq\t{$src, $dst|$dst, $src}", [(int_x86_sse2_storel_dq addr:$dst, VR128:$src)]>; -// Move to lower bits of a VR128 and zeroing upper bits. -// Loading from memory automatically zeroing upper bits. -let AddedComplexity = 20 in { -def MOVZSD2PDrm : SDI<0x10, MRMSrcMem, (outs VR128:$dst), (ins f64mem:$src), - "movsd\t{$src, $dst|$dst, $src}", - [(set VR128:$dst, - (v2f64 (X86vzmovl (v2f64 (scalar_to_vector - (loadf64 addr:$src))))))]>; - -def : Pat<(v2f64 (X86vzmovl (loadv2f64 addr:$src))), - (MOVZSD2PDrm addr:$src)>; -def : Pat<(v2f64 (X86vzmovl (bc_v2f64 (loadv4f32 addr:$src)))), - (MOVZSD2PDrm addr:$src)>; -def : Pat<(v2f64 (X86vzload addr:$src)), (MOVZSD2PDrm addr:$src)>; -} - // movd / movq to XMM register zero-extends let AddedComplexity = 15 in { def MOVZDI2PDIrr : PDI<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR32:$src), @@ -2989,13 +3024,15 @@ let Predicates = [HasSSE2] in { let AddedComplexity = 15 in { // Zeroing a VR128 then do a MOVS{S|D} to the lower bits. def : Pat<(v2f64 (X86vzmovl (v2f64 (scalar_to_vector FR64:$src)))), - (MOVLSD2PDrr (V_SET0), FR64:$src)>, Requires<[HasSSE2]>; + (MOVSDrr (v2f64 (V_SET0)), FR64:$src)>; def : Pat<(v4f32 (X86vzmovl (v4f32 (scalar_to_vector FR32:$src)))), - (MOVLSS2PSrr (V_SET0), FR32:$src)>, Requires<[HasSSE1]>; + (MOVSSrr (v4f32 (V_SET0)), FR32:$src)>; def : Pat<(v4f32 (X86vzmovl (v4f32 VR128:$src))), - (MOVLPSrr (V_SET0), VR128:$src)>, Requires<[HasSSE1]>; + (MOVSSrr (v4f32 (V_SET0)), + (f32 (EXTRACT_SUBREG (v4f32 VR128:$src), x86_subreg_ss)))>; def : Pat<(v4i32 (X86vzmovl (v4i32 VR128:$src))), - (MOVLPSrr (V_SET0), VR128:$src)>, Requires<[HasSSE1]>; + (MOVSSrr (v4i32 (V_SET0)), + (EXTRACT_SUBREG (v4i32 VR128:$src), x86_subreg_ss))>; } // Splat v2f64 / v2i64 @@ -3013,8 +3050,7 @@ def : Pat<(unpckh (v2i64 VR128:$src), (undef)), // Special unary SHUFPSrri case. def : Pat<(v4f32 (pshufd:$src3 VR128:$src1, (undef))), (SHUFPSrri VR128:$src1, VR128:$src1, - (SHUFFLE_get_shuf_imm VR128:$src3))>, - Requires<[HasSSE1]>; + (SHUFFLE_get_shuf_imm VR128:$src3))>; let AddedComplexity = 5 in def : Pat<(v4f32 (pshufd:$src2 VR128:$src1, (undef))), (PSHUFDri VR128:$src1, (SHUFFLE_get_shuf_imm VR128:$src2))>, @@ -3060,13 +3096,13 @@ def : Pat<(v4f32 (unpckl_undef:$src2 VR128:$src, (undef))), } let AddedComplexity = 10 in { def : Pat<(v4f32 (unpckl_undef VR128:$src, (undef))), - (UNPCKLPSrr VR128:$src, VR128:$src)>, Requires<[HasSSE1]>; + (UNPCKLPSrr VR128:$src, VR128:$src)>; def : Pat<(v16i8 (unpckl_undef VR128:$src, (undef))), - (PUNPCKLBWrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>; + (PUNPCKLBWrr VR128:$src, VR128:$src)>; def : Pat<(v8i16 (unpckl_undef VR128:$src, (undef))), - (PUNPCKLWDrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>; + (PUNPCKLWDrr VR128:$src, VR128:$src)>; def : Pat<(v4i32 (unpckl_undef VR128:$src, (undef))), - (PUNPCKLDQrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>; + (PUNPCKLDQrr VR128:$src, VR128:$src)>; } // vector_shuffle v1, <undef>, <2, 2, 3, 3, ...> @@ -3080,13 +3116,13 @@ def : Pat<(v4f32 (unpckh_undef:$src2 VR128:$src, (undef))), } let AddedComplexity = 10 in { def : Pat<(v4f32 (unpckh_undef VR128:$src, (undef))), - (UNPCKHPSrr VR128:$src, VR128:$src)>, Requires<[HasSSE1]>; + (UNPCKHPSrr VR128:$src, VR128:$src)>; def : Pat<(v16i8 (unpckh_undef VR128:$src, (undef))), - (PUNPCKHBWrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>; + (PUNPCKHBWrr VR128:$src, VR128:$src)>; def : Pat<(v8i16 (unpckh_undef VR128:$src, (undef))), - (PUNPCKHWDrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>; + (PUNPCKHWDrr VR128:$src, VR128:$src)>; def : Pat<(v4i32 (unpckh_undef VR128:$src, (undef))), - (PUNPCKHDQrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>; + (PUNPCKHDQrr VR128:$src, VR128:$src)>; } let AddedComplexity = 20 in { @@ -3108,45 +3144,49 @@ def : Pat<(v4i32 (movhlps_undef VR128:$src1, (undef))), let AddedComplexity = 20 in { // vector_shuffle v1, (load v2) <4, 5, 2, 3> using MOVLPS def : Pat<(v4f32 (movlp VR128:$src1, (load addr:$src2))), - (MOVLPSrm VR128:$src1, addr:$src2)>, Requires<[HasSSE1]>; + (MOVLPSrm VR128:$src1, addr:$src2)>; def : Pat<(v2f64 (movlp VR128:$src1, (load addr:$src2))), - (MOVLPDrm VR128:$src1, addr:$src2)>, Requires<[HasSSE2]>; + (MOVLPDrm VR128:$src1, addr:$src2)>; def : Pat<(v4i32 (movlp VR128:$src1, (load addr:$src2))), - (MOVLPSrm VR128:$src1, addr:$src2)>, Requires<[HasSSE2]>; + (MOVLPSrm VR128:$src1, addr:$src2)>; def : Pat<(v2i64 (movlp VR128:$src1, (load addr:$src2))), - (MOVLPDrm VR128:$src1, addr:$src2)>, Requires<[HasSSE2]>; + (MOVLPDrm VR128:$src1, addr:$src2)>; } // (store (vector_shuffle (load addr), v2, <4, 5, 2, 3>), addr) using MOVLPS def : Pat<(store (v4f32 (movlp (load addr:$src1), VR128:$src2)), addr:$src1), - (MOVLPSmr addr:$src1, VR128:$src2)>, Requires<[HasSSE1]>; + (MOVLPSmr addr:$src1, VR128:$src2)>; def : Pat<(store (v2f64 (movlp (load addr:$src1), VR128:$src2)), addr:$src1), - (MOVLPDmr addr:$src1, VR128:$src2)>, Requires<[HasSSE2]>; + (MOVLPDmr addr:$src1, VR128:$src2)>; def : Pat<(store (v4i32 (movlp (bc_v4i32 (loadv2i64 addr:$src1)), VR128:$src2)), addr:$src1), - (MOVLPSmr addr:$src1, VR128:$src2)>, Requires<[HasSSE1]>; + (MOVLPSmr addr:$src1, VR128:$src2)>; def : Pat<(store (v2i64 (movlp (load addr:$src1), VR128:$src2)), addr:$src1), - (MOVLPDmr addr:$src1, VR128:$src2)>, Requires<[HasSSE2]>; + (MOVLPDmr addr:$src1, VR128:$src2)>; let AddedComplexity = 15 in { // Setting the lowest element in the vector. def : Pat<(v4i32 (movl VR128:$src1, VR128:$src2)), - (MOVLPSrr VR128:$src1, VR128:$src2)>, Requires<[HasSSE2]>; + (MOVSSrr (v4i32 VR128:$src1), + (EXTRACT_SUBREG (v4i32 VR128:$src2), x86_subreg_ss))>; def : Pat<(v2i64 (movl VR128:$src1, VR128:$src2)), - (MOVLPDrr VR128:$src1, VR128:$src2)>, Requires<[HasSSE2]>; + (MOVSDrr (v2i64 VR128:$src1), + (EXTRACT_SUBREG (v2i64 VR128:$src2), x86_subreg_sd))>; -// vector_shuffle v1, v2 <4, 5, 2, 3> using MOVLPDrr (movsd) +// vector_shuffle v1, v2 <4, 5, 2, 3> using movsd def : Pat<(v4f32 (movlp VR128:$src1, VR128:$src2)), - (MOVLPDrr VR128:$src1, VR128:$src2)>, Requires<[HasSSE2]>; + (MOVSDrr VR128:$src1, (EXTRACT_SUBREG VR128:$src2, x86_subreg_sd))>, + Requires<[HasSSE2]>; def : Pat<(v4i32 (movlp VR128:$src1, VR128:$src2)), - (MOVLPDrr VR128:$src1, VR128:$src2)>, Requires<[HasSSE2]>; + (MOVSDrr VR128:$src1, (EXTRACT_SUBREG VR128:$src2, x86_subreg_sd))>, + Requires<[HasSSE2]>; } // vector_shuffle v1, v2 <4, 5, 2, 3> using SHUFPSrri (we prefer movsd, but // fall back to this for SSE1) def : Pat<(v4f32 (movlp:$src3 VR128:$src1, (v4f32 VR128:$src2))), (SHUFPSrri VR128:$src2, VR128:$src1, - (SHUFFLE_get_shuf_imm VR128:$src3))>, Requires<[HasSSE1]>; + (SHUFFLE_get_shuf_imm VR128:$src3))>; // Set lowest element and zero upper elements. let AddedComplexity = 15 in @@ -3188,30 +3228,30 @@ def : Pat<(v2i32 (fp_to_sint (v2f64 VR128:$src))), // Use movaps / movups for SSE integer load / store (one byte shorter). def : Pat<(alignedloadv4i32 addr:$src), - (MOVAPSrm addr:$src)>, Requires<[HasSSE1]>; + (MOVAPSrm addr:$src)>; def : Pat<(loadv4i32 addr:$src), - (MOVUPSrm addr:$src)>, Requires<[HasSSE1]>; + (MOVUPSrm addr:$src)>; def : Pat<(alignedloadv2i64 addr:$src), - (MOVAPSrm addr:$src)>, Requires<[HasSSE2]>; + (MOVAPSrm addr:$src)>; def : Pat<(loadv2i64 addr:$src), - (MOVUPSrm addr:$src)>, Requires<[HasSSE2]>; + (MOVUPSrm addr:$src)>; def : Pat<(alignedstore (v2i64 VR128:$src), addr:$dst), - (MOVAPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; + (MOVAPSmr addr:$dst, VR128:$src)>; def : Pat<(alignedstore (v4i32 VR128:$src), addr:$dst), - (MOVAPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; + (MOVAPSmr addr:$dst, VR128:$src)>; def : Pat<(alignedstore (v8i16 VR128:$src), addr:$dst), - (MOVAPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; + (MOVAPSmr addr:$dst, VR128:$src)>; def : Pat<(alignedstore (v16i8 VR128:$src), addr:$dst), - (MOVAPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; + (MOVAPSmr addr:$dst, VR128:$src)>; def : Pat<(store (v2i64 VR128:$src), addr:$dst), - (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; + (MOVUPSmr addr:$dst, VR128:$src)>; def : Pat<(store (v4i32 VR128:$src), addr:$dst), - (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; + (MOVUPSmr addr:$dst, VR128:$src)>; def : Pat<(store (v8i16 VR128:$src), addr:$dst), - (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; + (MOVUPSmr addr:$dst, VR128:$src)>; def : Pat<(store (v16i8 VR128:$src), addr:$dst), - (MOVUPSmr addr:$dst, VR128:$src)>, Requires<[HasSSE2]>; + (MOVUPSmr addr:$dst, VR128:$src)>; //===----------------------------------------------------------------------===// // SSE4.1 Instructions @@ -3400,7 +3440,7 @@ let Constraints = "$src1 = $dst" in { (ins VR128:$src1, i128mem:$src2), !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"), [(set VR128:$dst, - (OpNode VR128:$src1, (memop addr:$src2)))]>, OpSize; + (OpVT (OpNode VR128:$src1, (memop addr:$src2))))]>, OpSize; def rm_int : SS48I<opc, MRMSrcMem, (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2), !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"), diff --git a/lib/Target/X86/X86MCAsmInfo.cpp b/lib/Target/X86/X86MCAsmInfo.cpp index 91c0fbb..250634f 100644 --- a/lib/Target/X86/X86MCAsmInfo.cpp +++ b/lib/Target/X86/X86MCAsmInfo.cpp @@ -55,6 +55,11 @@ X86MCAsmInfoDarwin::X86MCAsmInfoDarwin(const Triple &Triple) { if (!is64Bit) Data64bitsDirective = 0; // we can't emit a 64-bit unit + // Use ## as a comment string so that .s files generated by llvm can go + // through the GCC preprocessor without causing an error. This is needed + // because "clang foo.s" runs the C preprocessor, which is usually reserved + // for .S files on other systems. Perhaps this is because the file system + // wasn't always case preserving or something. CommentString = "##"; PCSymbol = "."; @@ -70,6 +75,8 @@ X86ELFMCAsmInfo::X86ELFMCAsmInfo(const Triple &Triple) { AsmTransCBE = x86_asm_table; AssemblerDialect = AsmWriterFlavor; + TextAlignFillValue = 0x90; + PrivateGlobalPrefix = ".L"; WeakRefDirective = "\t.weak\t"; PCSymbol = "."; @@ -94,4 +101,6 @@ MCSection *X86ELFMCAsmInfo::getNonexecutableStackSection(MCContext &Ctx) const { X86MCAsmInfoCOFF::X86MCAsmInfoCOFF(const Triple &Triple) { AsmTransCBE = x86_asm_table; AssemblerDialect = AsmWriterFlavor; -}
\ No newline at end of file + + TextAlignFillValue = 0x90; +} diff --git a/lib/Target/X86/X86RegisterInfo.cpp b/lib/Target/X86/X86RegisterInfo.cpp index 8524236..946d6b2 100644 --- a/lib/Target/X86/X86RegisterInfo.cpp +++ b/lib/Target/X86/X86RegisterInfo.cpp @@ -191,6 +191,8 @@ X86RegisterInfo::getMatchingSuperRegClass(const TargetRegisterClass *A, return &X86::GR16_NOREXRegClass; else if (A == &X86::GR16_ABCDRegClass) return &X86::GR16_ABCDRegClass; + } else if (B == &X86::FR32RegClass) { + return A; } break; case 2: @@ -207,6 +209,8 @@ X86RegisterInfo::getMatchingSuperRegClass(const TargetRegisterClass *A, else if (A == &X86::GR16RegClass || A == &X86::GR16_ABCDRegClass || A == &X86::GR16_NOREXRegClass) return &X86::GR16_ABCDRegClass; + } else if (B == &X86::FR64RegClass) { + return A; } break; case 3: @@ -234,6 +238,8 @@ X86RegisterInfo::getMatchingSuperRegClass(const TargetRegisterClass *A, return &X86::GR32_NOREXRegClass; else if (A == &X86::GR32_ABCDRegClass) return &X86::GR64_ABCDRegClass; + } else if (B == &X86::VR128RegClass) { + return A; } break; case 4: @@ -446,8 +452,10 @@ bool X86RegisterInfo::canRealignStack(const MachineFunction &MF) const { bool X86RegisterInfo::needsStackRealignment(const MachineFunction &MF) const { const MachineFrameInfo *MFI = MF.getFrameInfo(); + const Function *F = MF.getFunction(); bool requiresRealignment = - RealignStack && (MFI->getMaxAlignment() > StackAlign); + RealignStack && ((MFI->getMaxAlignment() > StackAlign) || + F->hasFnAttr(Attribute::StackAlignment)); // FIXME: Currently we don't support stack realignment for functions with // variable-sized allocas. diff --git a/lib/Target/X86/X86RegisterInfo.h b/lib/Target/X86/X86RegisterInfo.h index 8fb5e92..e4bdb4e 100644 --- a/lib/Target/X86/X86RegisterInfo.h +++ b/lib/Target/X86/X86RegisterInfo.h @@ -35,7 +35,8 @@ namespace X86 { /// these indices must be kept in sync with the class indices in the /// X86RegisterInfo.td file. enum SubregIndex { - SUBREG_8BIT = 1, SUBREG_8BIT_HI = 2, SUBREG_16BIT = 3, SUBREG_32BIT = 4 + SUBREG_8BIT = 1, SUBREG_8BIT_HI = 2, SUBREG_16BIT = 3, SUBREG_32BIT = 4, + SUBREG_SS = 1, SUBREG_SD = 2, SUBREG_XMM = 3 }; } diff --git a/lib/Target/X86/X86RegisterInfo.td b/lib/Target/X86/X86RegisterInfo.td index 1559bf7..ed2ce6c 100644 --- a/lib/Target/X86/X86RegisterInfo.td +++ b/lib/Target/X86/X86RegisterInfo.td @@ -158,22 +158,22 @@ let Namespace = "X86" in { def XMM15: Register<"xmm15">, DwarfRegNum<[32, -2, -2]>; // YMM Registers, used by AVX instructions - def YMM0: Register<"ymm0">, DwarfRegNum<[17, 21, 21]>; - def YMM1: Register<"ymm1">, DwarfRegNum<[18, 22, 22]>; - def YMM2: Register<"ymm2">, DwarfRegNum<[19, 23, 23]>; - def YMM3: Register<"ymm3">, DwarfRegNum<[20, 24, 24]>; - def YMM4: Register<"ymm4">, DwarfRegNum<[21, 25, 25]>; - def YMM5: Register<"ymm5">, DwarfRegNum<[22, 26, 26]>; - def YMM6: Register<"ymm6">, DwarfRegNum<[23, 27, 27]>; - def YMM7: Register<"ymm7">, DwarfRegNum<[24, 28, 28]>; - def YMM8: Register<"ymm8">, DwarfRegNum<[25, -2, -2]>; - def YMM9: Register<"ymm9">, DwarfRegNum<[26, -2, -2]>; - def YMM10: Register<"ymm10">, DwarfRegNum<[27, -2, -2]>; - def YMM11: Register<"ymm11">, DwarfRegNum<[28, -2, -2]>; - def YMM12: Register<"ymm12">, DwarfRegNum<[29, -2, -2]>; - def YMM13: Register<"ymm13">, DwarfRegNum<[30, -2, -2]>; - def YMM14: Register<"ymm14">, DwarfRegNum<[31, -2, -2]>; - def YMM15: Register<"ymm15">, DwarfRegNum<[32, -2, -2]>; + def YMM0: RegisterWithSubRegs<"ymm0", [XMM0]>, DwarfRegNum<[17, 21, 21]>; + def YMM1: RegisterWithSubRegs<"ymm1", [XMM1]>, DwarfRegNum<[18, 22, 22]>; + def YMM2: RegisterWithSubRegs<"ymm2", [XMM2]>, DwarfRegNum<[19, 23, 23]>; + def YMM3: RegisterWithSubRegs<"ymm3", [XMM3]>, DwarfRegNum<[20, 24, 24]>; + def YMM4: RegisterWithSubRegs<"ymm4", [XMM4]>, DwarfRegNum<[21, 25, 25]>; + def YMM5: RegisterWithSubRegs<"ymm5", [XMM5]>, DwarfRegNum<[22, 26, 26]>; + def YMM6: RegisterWithSubRegs<"ymm6", [XMM6]>, DwarfRegNum<[23, 27, 27]>; + def YMM7: RegisterWithSubRegs<"ymm7", [XMM7]>, DwarfRegNum<[24, 28, 28]>; + def YMM8: RegisterWithSubRegs<"ymm8", [XMM8]>, DwarfRegNum<[25, -2, -2]>; + def YMM9: RegisterWithSubRegs<"ymm9", [XMM9]>, DwarfRegNum<[26, -2, -2]>; + def YMM10: RegisterWithSubRegs<"ymm10", [XMM10]>, DwarfRegNum<[27, -2, -2]>; + def YMM11: RegisterWithSubRegs<"ymm11", [XMM11]>, DwarfRegNum<[28, -2, -2]>; + def YMM12: RegisterWithSubRegs<"ymm12", [XMM12]>, DwarfRegNum<[29, -2, -2]>; + def YMM13: RegisterWithSubRegs<"ymm13", [XMM13]>, DwarfRegNum<[30, -2, -2]>; + def YMM14: RegisterWithSubRegs<"ymm14", [XMM14]>, DwarfRegNum<[31, -2, -2]>; + def YMM15: RegisterWithSubRegs<"ymm15", [XMM15]>, DwarfRegNum<[32, -2, -2]>; // Floating point stack registers def ST0 : Register<"st(0)">, DwarfRegNum<[33, 12, 11]>; @@ -238,6 +238,10 @@ def x86_subreg_8bit_hi : PatLeaf<(i32 2)>; def x86_subreg_16bit : PatLeaf<(i32 3)>; def x86_subreg_32bit : PatLeaf<(i32 4)>; +def x86_subreg_ss : PatLeaf<(i32 1)>; +def x86_subreg_sd : PatLeaf<(i32 2)>; +def x86_subreg_xmm : PatLeaf<(i32 3)>; + def : SubRegSet<1, [AX, CX, DX, BX, SP, BP, SI, DI, R8W, R9W, R10W, R11W, R12W, R13W, R14W, R15W], [AL, CL, DL, BL, SPL, BPL, SIL, DIL, @@ -277,11 +281,31 @@ def : SubRegSet<4, [RAX, RCX, RDX, RBX, RSP, RBP, RSI, RDI, [EAX, ECX, EDX, EBX, ESP, EBP, ESI, EDI, R8D, R9D, R10D, R11D, R12D, R13D, R14D, R15D]>; -def : SubRegSet<1, [YMM0, YMM1, YMM2, YMM3, YMM4, YMM5, YMM6, YMM7, +def : SubRegSet<1, [YMM0, YMM1, YMM2, YMM3, YMM4, YMM5, YMM6, YMM7, + YMM8, YMM9, YMM10, YMM11, YMM12, YMM13, YMM14, YMM15], + [XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7, + XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15]>; + +def : SubRegSet<2, [YMM0, YMM1, YMM2, YMM3, YMM4, YMM5, YMM6, YMM7, + YMM8, YMM9, YMM10, YMM11, YMM12, YMM13, YMM14, YMM15], + [XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7, + XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15]>; + +def : SubRegSet<3, [YMM0, YMM1, YMM2, YMM3, YMM4, YMM5, YMM6, YMM7, YMM8, YMM9, YMM10, YMM11, YMM12, YMM13, YMM14, YMM15], [XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7, XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15]>; +def : SubRegSet<1, [XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7, + XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15], + [XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7, + XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15]>; + +def : SubRegSet<2, [XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7, + XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15], + [XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7, + XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15]>; + //===----------------------------------------------------------------------===// // Register Class Definitions... now that we have all of the pieces, define the // top-level register classes. The order specified in the register list is @@ -793,6 +817,7 @@ def VR128 : RegisterClass<"X86", [v16i8, v8i16, v4i32, v2i64, v4f32, v2f64],128, [XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7, XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15]> { + let SubRegClassList = [FR32, FR64]; let MethodProtos = [{ iterator allocation_order_end(const MachineFunction &MF) const; }]; @@ -811,7 +836,9 @@ def VR128 : RegisterClass<"X86", [v16i8, v8i16, v4i32, v2i64, v4f32, v2f64],128, def VR256 : RegisterClass<"X86", [ v8i32, v4i64, v8f32, v4f64],256, [YMM0, YMM1, YMM2, YMM3, YMM4, YMM5, YMM6, YMM7, YMM8, YMM9, YMM10, YMM11, - YMM12, YMM13, YMM14, YMM15]>; + YMM12, YMM13, YMM14, YMM15]> { + let SubRegClassList = [FR32, FR64, VR128]; +} // Status flags registers. def CCR : RegisterClass<"X86", [i32], 32, [EFLAGS]> { diff --git a/lib/Target/X86/X86Subtarget.h b/lib/Target/X86/X86Subtarget.h index 5e05c2f..594a470 100644 --- a/lib/Target/X86/X86Subtarget.h +++ b/lib/Target/X86/X86Subtarget.h @@ -20,9 +20,9 @@ namespace llvm { class GlobalValue; class TargetMachine; - + /// PICStyles - The X86 backend supports a number of different styles of PIC. -/// +/// namespace PICStyles { enum Style { StubPIC, // Used on i386-darwin in -fPIC mode. @@ -46,7 +46,7 @@ protected: /// PICStyle - Which PIC style to use /// PICStyles::Style PICStyle; - + /// X86SSELevel - MMX, SSE1, SSE2, SSE3, SSSE3, SSE41, SSE42, or /// none supported. X86SSEEnum X86SSELevel; @@ -58,7 +58,7 @@ protected: /// HasCMov - True if this processor has conditional move instructions /// (generally pentium pro+). bool HasCMov; - + /// HasX86_64 - True if the processor supports X86-64 instructions. /// bool HasX86_64; @@ -78,8 +78,9 @@ protected: /// IsBTMemSlow - True if BT (bit test) of memory instructions are slow. bool IsBTMemSlow; - /// HasVectorUAMem - True if SIMD operations can have unaligned memory operands. - /// This may require setting a feature bit in the processor. + /// HasVectorUAMem - True if SIMD operations can have unaligned memory + /// operands. This may require setting a feature bit in the + /// processor. bool HasVectorUAMem; /// DarwinVers - Nonzero if this is a darwin platform: the numeric @@ -150,20 +151,20 @@ public: bool isTargetDarwin() const { return TargetType == isDarwin; } bool isTargetELF() const { return TargetType == isELF; } - + bool isTargetWindows() const { return TargetType == isWindows; } bool isTargetMingw() const { return TargetType == isMingw; } bool isTargetCygwin() const { return TargetType == isCygwin; } bool isTargetCygMing() const { return TargetType == isMingw || TargetType == isCygwin; } - + /// isTargetCOFF - Return true if this is any COFF/Windows target variant. bool isTargetCOFF() const { return TargetType == isMingw || TargetType == isCygwin || TargetType == isWindows; } - + bool isTargetWin64() const { return Is64Bit && (TargetType == isMingw || TargetType == isWindows); } @@ -196,11 +197,11 @@ public: bool isPICStyleStubAny() const { return PICStyle == PICStyles::StubDynamicNoPIC || PICStyle == PICStyles::StubPIC; } - + /// getDarwinVers - Return the darwin version number, 8 = Tiger, 9 = Leopard, /// 10 = Snow Leopard, etc. unsigned getDarwinVers() const { return DarwinVers; } - + /// ClassifyGlobalReference - Classify a global variable reference for the /// current subtarget according to how we should reference it in a non-pcrel /// context. diff --git a/lib/Target/X86/X86TargetMachine.cpp b/lib/Target/X86/X86TargetMachine.cpp index 7802f98..56ddaf8 100644 --- a/lib/Target/X86/X86TargetMachine.cpp +++ b/lib/Target/X86/X86TargetMachine.cpp @@ -51,6 +51,12 @@ extern "C" void LLVMInitializeX86Target() { createX86_32MCCodeEmitter); TargetRegistry::RegisterCodeEmitter(TheX86_64Target, createX86_64MCCodeEmitter); + + // Register the asm backend. + TargetRegistry::RegisterAsmBackend(TheX86_32Target, + createX86_32AsmBackend); + TargetRegistry::RegisterAsmBackend(TheX86_64Target, + createX86_64AsmBackend); } diff --git a/lib/Target/X86/X86TargetObjectFile.cpp b/lib/Target/X86/X86TargetObjectFile.cpp index d1ee3fc..29a0be5 100644 --- a/lib/Target/X86/X86TargetObjectFile.cpp +++ b/lib/Target/X86/X86TargetObjectFile.cpp @@ -18,38 +18,6 @@ using namespace llvm; using namespace dwarf; -const MCExpr *X8632_MachoTargetObjectFile:: -getSymbolForDwarfGlobalReference(const GlobalValue *GV, Mangler *Mang, - MachineModuleInfo *MMI, unsigned Encoding) const { - // The mach-o version of this method defaults to returning a stub reference. - - if (Encoding & DW_EH_PE_indirect) { - MachineModuleInfoMachO &MachOMMI = - MMI->getObjFileInfo<MachineModuleInfoMachO>(); - - SmallString<128> Name; - Mang->getNameWithPrefix(Name, GV, true); - Name += "$non_lazy_ptr"; - - // Add information about the stub reference to MachOMMI so that the stub - // gets emitted by the asmprinter. - MCSymbol *Sym = getContext().GetOrCreateSymbol(Name.str()); - MCSymbol *&StubSym = MachOMMI.getGVStubEntry(Sym); - if (StubSym == 0) { - Name.clear(); - Mang->getNameWithPrefix(Name, GV, false); - StubSym = getContext().GetOrCreateSymbol(Name.str()); - } - - return TargetLoweringObjectFile:: - getSymbolForDwarfReference(Sym, MMI, - Encoding & ~dwarf::DW_EH_PE_indirect); - } - - return TargetLoweringObjectFileMachO:: - getSymbolForDwarfGlobalReference(GV, Mang, MMI, Encoding); -} - const MCExpr *X8664_MachoTargetObjectFile:: getSymbolForDwarfGlobalReference(const GlobalValue *GV, Mangler *Mang, MachineModuleInfo *MMI, unsigned Encoding) const { @@ -148,35 +116,3 @@ unsigned X8664_ELFTargetObjectFile::getTTypeEncoding() const { return DW_EH_PE_absptr; } - -unsigned X8632_MachoTargetObjectFile::getPersonalityEncoding() const { - return DW_EH_PE_indirect | DW_EH_PE_pcrel | DW_EH_PE_sdata4; -} - -unsigned X8632_MachoTargetObjectFile::getLSDAEncoding() const { - return DW_EH_PE_pcrel | DW_EH_PE_sdata4; -} - -unsigned X8632_MachoTargetObjectFile::getFDEEncoding() const { - return DW_EH_PE_pcrel | DW_EH_PE_sdata4; -} - -unsigned X8632_MachoTargetObjectFile::getTTypeEncoding() const { - return DW_EH_PE_indirect | DW_EH_PE_pcrel | DW_EH_PE_sdata4; -} - -unsigned X8664_MachoTargetObjectFile::getPersonalityEncoding() const { - return DW_EH_PE_indirect | DW_EH_PE_pcrel | DW_EH_PE_sdata4; -} - -unsigned X8664_MachoTargetObjectFile::getLSDAEncoding() const { - return DW_EH_PE_pcrel | DW_EH_PE_sdata4; -} - -unsigned X8664_MachoTargetObjectFile::getFDEEncoding() const { - return DW_EH_PE_pcrel | DW_EH_PE_sdata4; -} - -unsigned X8664_MachoTargetObjectFile::getTTypeEncoding() const { - return DW_EH_PE_indirect | DW_EH_PE_pcrel | DW_EH_PE_sdata4; -} diff --git a/lib/Target/X86/X86TargetObjectFile.h b/lib/Target/X86/X86TargetObjectFile.h index 0fff194..0444417 100644 --- a/lib/Target/X86/X86TargetObjectFile.h +++ b/lib/Target/X86/X86TargetObjectFile.h @@ -17,20 +17,6 @@ namespace llvm { class X86TargetMachine; - /// X8632_MachoTargetObjectFile - This TLOF implementation is used for - /// Darwin/x86-32. - class X8632_MachoTargetObjectFile : public TargetLoweringObjectFileMachO { - public: - - virtual const MCExpr * - getSymbolForDwarfGlobalReference(const GlobalValue *GV, Mangler *Mang, - MachineModuleInfo *MMI, unsigned Encoding) const; - virtual unsigned getPersonalityEncoding() const; - virtual unsigned getLSDAEncoding() const; - virtual unsigned getFDEEncoding() const; - virtual unsigned getTTypeEncoding() const; - }; - /// X8664_MachoTargetObjectFile - This TLOF implementation is used for /// Darwin/x86-64. class X8664_MachoTargetObjectFile : public TargetLoweringObjectFileMachO { @@ -39,17 +25,13 @@ namespace llvm { virtual const MCExpr * getSymbolForDwarfGlobalReference(const GlobalValue *GV, Mangler *Mang, MachineModuleInfo *MMI, unsigned Encoding) const; - virtual unsigned getPersonalityEncoding() const; - virtual unsigned getLSDAEncoding() const; - virtual unsigned getFDEEncoding() const; - virtual unsigned getTTypeEncoding() const; }; class X8632_ELFTargetObjectFile : public TargetLoweringObjectFileELF { const X86TargetMachine &TM; public: X8632_ELFTargetObjectFile(const X86TargetMachine &tm) - :TM(tm) { }; + :TM(tm) { } virtual unsigned getPersonalityEncoding() const; virtual unsigned getLSDAEncoding() const; virtual unsigned getFDEEncoding() const; @@ -60,7 +42,7 @@ namespace llvm { const X86TargetMachine &TM; public: X8664_ELFTargetObjectFile(const X86TargetMachine &tm) - :TM(tm) { }; + :TM(tm) { } virtual unsigned getPersonalityEncoding() const; virtual unsigned getLSDAEncoding() const; virtual unsigned getFDEEncoding() const; diff --git a/lib/Target/XCore/AsmPrinter/XCoreAsmPrinter.cpp b/lib/Target/XCore/AsmPrinter/XCoreAsmPrinter.cpp index d18f55d..82e23a1 100644 --- a/lib/Target/XCore/AsmPrinter/XCoreAsmPrinter.cpp +++ b/lib/Target/XCore/AsmPrinter/XCoreAsmPrinter.cpp @@ -27,6 +27,7 @@ #include "llvm/CodeGen/MachineFunctionPass.h" #include "llvm/CodeGen/MachineConstantPool.h" #include "llvm/CodeGen/MachineInstr.h" +#include "llvm/CodeGen/MachineJumpTableInfo.h" #include "llvm/MC/MCStreamer.h" #include "llvm/MC/MCSymbol.h" #include "llvm/Target/TargetData.h" @@ -62,6 +63,11 @@ namespace { } void printMemOperand(const MachineInstr *MI, int opNum); + void printInlineJT(const MachineInstr *MI, int opNum, + const std::string &directive = ".jmptable"); + void printInlineJT32(const MachineInstr *MI, int opNum) { + printInlineJT(MI, opNum, ".jmptable32"); + } void printOperand(const MachineInstr *MI, int opNum); bool PrintAsmOperand(const MachineInstr *MI, unsigned OpNo, unsigned AsmVariant, const char *ExtraCode); @@ -257,6 +263,23 @@ void XCoreAsmPrinter::printMemOperand(const MachineInstr *MI, int opNum) printOperand(MI, opNum+1); } +void XCoreAsmPrinter:: +printInlineJT(const MachineInstr *MI, int opNum, const std::string &directive) +{ + unsigned JTI = MI->getOperand(opNum).getIndex(); + const MachineFunction *MF = MI->getParent()->getParent(); + const MachineJumpTableInfo *MJTI = MF->getJumpTableInfo(); + const std::vector<MachineJumpTableEntry> &JT = MJTI->getJumpTables(); + const std::vector<MachineBasicBlock*> &JTBBs = JT[JTI].MBBs; + O << "\t" << directive << " "; + for (unsigned i = 0, e = JTBBs.size(); i != e; ++i) { + MachineBasicBlock *MBB = JTBBs[i]; + if (i > 0) + O << ","; + O << *MBB->getSymbol(OutContext); + } +} + void XCoreAsmPrinter::printOperand(const MachineInstr *MI, int opNum) { const MachineOperand &MO = MI->getOperand(opNum); switch (MO.getType()) { diff --git a/lib/Target/XCore/XCoreISelDAGToDAG.cpp b/lib/Target/XCore/XCoreISelDAGToDAG.cpp index 383fd91..b1ab132 100644 --- a/lib/Target/XCore/XCoreISelDAGToDAG.cpp +++ b/lib/Target/XCore/XCoreISelDAGToDAG.cpp @@ -65,8 +65,6 @@ namespace { bool SelectADDRcpii(SDNode *Op, SDValue Addr, SDValue &Base, SDValue &Offset); - virtual void InstructionSelect(); - virtual const char *getPassName() const { return "XCore DAG->DAG Pattern Instruction Selection"; } @@ -147,15 +145,6 @@ bool XCoreDAGToDAGISel::SelectADDRcpii(SDNode *Op, SDValue Addr, return false; } -/// InstructionSelect - This callback is invoked by -/// SelectionDAGISel when it has created a SelectionDAG for us to codegen. -void XCoreDAGToDAGISel::InstructionSelect() { - // Select target instructions for the DAG. - SelectRoot(*CurDAG); - - CurDAG->RemoveDeadNodes(); -} - SDNode *XCoreDAGToDAGISel::Select(SDNode *N) { DebugLoc dl = N->getDebugLoc(); EVT NVT = N->getValueType(0); @@ -164,7 +153,11 @@ SDNode *XCoreDAGToDAGISel::Select(SDNode *N) { default: break; case ISD::Constant: { if (Predicate_immMskBitp(N)) { - SDValue MskSize = Transform_msksize_xform(N); + // Transformation function: get the size of a mask + int64_t MaskVal = cast<ConstantSDNode>(N)->getZExtValue(); + assert(isMask_32(MaskVal)); + // Look for the first non-zero bit + SDValue MskSize = getI32Imm(32 - CountLeadingZeros_32(MaskVal)); return CurDAG->getMachineNode(XCore::MKMSK_rus, dl, MVT::i32, MskSize); } diff --git a/lib/Target/XCore/XCoreISelLowering.cpp b/lib/Target/XCore/XCoreISelLowering.cpp index 57fd43b..e6515d8 100644 --- a/lib/Target/XCore/XCoreISelLowering.cpp +++ b/lib/Target/XCore/XCoreISelLowering.cpp @@ -29,6 +29,7 @@ #include "llvm/CodeGen/MachineFrameInfo.h" #include "llvm/CodeGen/MachineFunction.h" #include "llvm/CodeGen/MachineInstrBuilder.h" +#include "llvm/CodeGen/MachineJumpTableInfo.h" #include "llvm/CodeGen/MachineRegisterInfo.h" #include "llvm/CodeGen/SelectionDAGISel.h" #include "llvm/CodeGen/ValueTypes.h" @@ -53,6 +54,8 @@ getTargetNodeName(unsigned Opcode) const case XCoreISD::RETSP : return "XCoreISD::RETSP"; case XCoreISD::LADD : return "XCoreISD::LADD"; case XCoreISD::LSUB : return "XCoreISD::LSUB"; + case XCoreISD::BR_JT : return "XCoreISD::BR_JT"; + case XCoreISD::BR_JT32 : return "XCoreISD::BR_JT32"; default : return NULL; } } @@ -106,9 +109,8 @@ XCoreTargetLowering::XCoreTargetLowering(XCoreTargetMachine &XTM) setOperationAction(ISD::TRAP, MVT::Other, Legal); - // Expand jump tables for now - setOperationAction(ISD::BR_JT, MVT::Other, Expand); - setOperationAction(ISD::JumpTable, MVT::i32, Custom); + // Jump tables. + setOperationAction(ISD::BR_JT, MVT::Other, Custom); setOperationAction(ISD::GlobalAddress, MVT::i32, Custom); setOperationAction(ISD::BlockAddress, MVT::i32 , Custom); @@ -157,7 +159,7 @@ LowerOperation(SDValue Op, SelectionDAG &DAG) { case ISD::GlobalTLSAddress: return LowerGlobalTLSAddress(Op, DAG); case ISD::BlockAddress: return LowerBlockAddress(Op, DAG); case ISD::ConstantPool: return LowerConstantPool(Op, DAG); - case ISD::JumpTable: return LowerJumpTable(Op, DAG); + case ISD::BR_JT: return LowerBR_JT(Op, DAG); case ISD::LOAD: return LowerLOAD(Op, DAG); case ISD::STORE: return LowerSTORE(Op, DAG); case ISD::SELECT_CC: return LowerSELECT_CC(Op, DAG); @@ -315,14 +317,27 @@ LowerConstantPool(SDValue Op, SelectionDAG &DAG) } SDValue XCoreTargetLowering:: -LowerJumpTable(SDValue Op, SelectionDAG &DAG) +LowerBR_JT(SDValue Op, SelectionDAG &DAG) { - // FIXME there isn't really debug info here + SDValue Chain = Op.getOperand(0); + SDValue Table = Op.getOperand(1); + SDValue Index = Op.getOperand(2); DebugLoc dl = Op.getDebugLoc(); - EVT PtrVT = Op.getValueType(); - JumpTableSDNode *JT = cast<JumpTableSDNode>(Op); - SDValue JTI = DAG.getTargetJumpTable(JT->getIndex(), PtrVT); - return DAG.getNode(XCoreISD::DPRelativeWrapper, dl, MVT::i32, JTI); + JumpTableSDNode *JT = cast<JumpTableSDNode>(Table); + unsigned JTI = JT->getIndex(); + MachineFunction &MF = DAG.getMachineFunction(); + const MachineJumpTableInfo *MJTI = MF.getJumpTableInfo(); + SDValue TargetJT = DAG.getTargetJumpTable(JT->getIndex(), MVT::i32); + + unsigned NumEntries = MJTI->getJumpTables()[JTI].MBBs.size(); + if (NumEntries <= 32) { + return DAG.getNode(XCoreISD::BR_JT, dl, MVT::Other, Chain, TargetJT, Index); + } + assert((NumEntries >> 31) == 0); + SDValue ScaledIndex = DAG.getNode(ISD::SHL, dl, MVT::i32, Index, + DAG.getConstant(1, MVT::i32)); + return DAG.getNode(XCoreISD::BR_JT32, dl, MVT::Other, Chain, TargetJT, + ScaledIndex); } static bool @@ -458,7 +473,7 @@ LowerLOAD(SDValue Op, SelectionDAG &DAG) false, false, 0, CallingConv::C, false, /*isReturnValueUsed=*/true, DAG.getExternalSymbol("__misaligned_load", getPointerTy()), - Args, DAG, dl, DAG.GetOrdering(Chain.getNode())); + Args, DAG, dl); SDValue Ops[] = { CallResult.first, CallResult.second }; @@ -521,7 +536,7 @@ LowerSTORE(SDValue Op, SelectionDAG &DAG) false, false, 0, CallingConv::C, false, /*isReturnValueUsed=*/true, DAG.getExternalSymbol("__misaligned_store", getPointerTy()), - Args, DAG, dl, DAG.GetOrdering(Chain.getNode())); + Args, DAG, dl); return CallResult.second; } @@ -1146,10 +1161,8 @@ static inline bool isImmUs4(int64_t val) bool XCoreTargetLowering::isLegalAddressingMode(const AddrMode &AM, const Type *Ty) const { - // Be conservative with void - // FIXME: Can we be more aggressive? if (Ty->getTypeID() == Type::VoidTyID) - return false; + return AM.Scale == 0 && isImmUs(AM.BaseOffs) && isImmUs4(AM.BaseOffs); const TargetData *TD = TM.getTargetData(); unsigned Size = TD->getTypeAllocSize(Ty); diff --git a/lib/Target/XCore/XCoreISelLowering.h b/lib/Target/XCore/XCoreISelLowering.h index 5095f36..0c638af 100644 --- a/lib/Target/XCore/XCoreISelLowering.h +++ b/lib/Target/XCore/XCoreISelLowering.h @@ -52,7 +52,13 @@ namespace llvm { LADD, // Corresponds to LSUB instruction - LSUB + LSUB, + + // Jumptable branch. + BR_JT, + + // Jumptable branch using long branches for each entry. + BR_JT32 }; } @@ -122,7 +128,7 @@ namespace llvm { SDValue LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG); SDValue LowerBlockAddress(SDValue Op, SelectionDAG &DAG); SDValue LowerConstantPool(SDValue Op, SelectionDAG &DAG); - SDValue LowerJumpTable(SDValue Op, SelectionDAG &DAG); + SDValue LowerBR_JT(SDValue Op, SelectionDAG &DAG); SDValue LowerSELECT_CC(SDValue Op, SelectionDAG &DAG); SDValue LowerVAARG(SDValue Op, SelectionDAG &DAG); SDValue LowerVASTART(SDValue Op, SelectionDAG &DAG); diff --git a/lib/Target/XCore/XCoreInstrInfo.cpp b/lib/Target/XCore/XCoreInstrInfo.cpp index 5a54844..722e747 100644 --- a/lib/Target/XCore/XCoreInstrInfo.cpp +++ b/lib/Target/XCore/XCoreInstrInfo.cpp @@ -145,6 +145,11 @@ static inline bool IsCondBranch(unsigned BrOpc) { return IsBRF(BrOpc) || IsBRT(BrOpc); } +static inline bool IsBR_JT(unsigned BrOpc) { + return BrOpc == XCore::BR_JT + || BrOpc == XCore::BR_JT32; +} + /// GetCondFromBranchOpc - Return the XCore CC that matches /// the correspondent Branch instruction opcode. static XCore::CondCode GetCondFromBranchOpc(unsigned BrOpc) @@ -271,6 +276,14 @@ XCoreInstrInfo::AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB, return false; } + // Likewise if it ends with a branch table followed by an unconditional branch. + if (IsBR_JT(SecondLastInst->getOpcode()) && IsBRU(LastInst->getOpcode())) { + I = LastInst; + if (AllowModify) + I->eraseFromParent(); + return true; + } + // Otherwise, can't handle this. return true; } diff --git a/lib/Target/XCore/XCoreInstrInfo.td b/lib/Target/XCore/XCoreInstrInfo.td index 10dc18c..46805d5 100644 --- a/lib/Target/XCore/XCoreInstrInfo.td +++ b/lib/Target/XCore/XCoreInstrInfo.td @@ -34,6 +34,15 @@ def XCoreBranchLink : SDNode<"XCoreISD::BL",SDT_XCoreBranchLink, def XCoreRetsp : SDNode<"XCoreISD::RETSP", SDTNone, [SDNPHasChain, SDNPOptInFlag]>; +def SDT_XCoreBR_JT : SDTypeProfile<0, 2, + [SDTCisVT<0, i32>, SDTCisVT<1, i32>]>; + +def XCoreBR_JT : SDNode<"XCoreISD::BR_JT", SDT_XCoreBR_JT, + [SDNPHasChain]>; + +def XCoreBR_JT32 : SDNode<"XCoreISD::BR_JT32", SDT_XCoreBR_JT, + [SDNPHasChain]>; + def SDT_XCoreAddress : SDTypeProfile<1, 1, [SDTCisSameAs<0, 1>, SDTCisPtrTy<0>]>; @@ -185,6 +194,15 @@ def MEMii : Operand<i32> { let MIOperandInfo = (ops i32imm, i32imm); } +// Jump tables. +def InlineJT : Operand<i32> { + let PrintMethod = "printInlineJT"; +} + +def InlineJT32 : Operand<i32> { + let PrintMethod = "printInlineJT32"; +} + //===----------------------------------------------------------------------===// // Instruction Class Templates //===----------------------------------------------------------------------===// @@ -624,7 +642,7 @@ defm RETSP : FU6_LU6<"retsp", XCoreRetsp>; // TODO extdp, kentsp, krestsp, blat, setsr // clrsr, getsr, kalli -let isBranch = 1, isTerminator = 1 in { +let isBranch = 1, isTerminator = 1, isBarrier = 1 in { def BRBU_u6 : _FU6< (outs), (ins brtarget:$target), @@ -756,24 +774,34 @@ def CLZ_l2r : _FL2R<(outs GRRegs:$dst), (ins GRRegs:$src), // One operand short // TODO edu, eeu, waitet, waitef, freer, tstart, msync, mjoin, syncr, clrtp -// bru, setdp, setcp, setv, setev, kcall +// setdp, setcp, setv, setev, kcall // dgetreg -let isBranch=1, isIndirectBranch=1, isTerminator=1 in +let isBranch=1, isIndirectBranch=1, isTerminator=1, isBarrier = 1 in def BAU_1r : _F1R<(outs), (ins GRRegs:$addr), "bau $addr", [(brind GRRegs:$addr)]>; +let isBranch=1, isIndirectBranch=1, isTerminator=1, isBarrier = 1 in +def BR_JT : PseudoInstXCore<(outs), (ins InlineJT:$t, GRRegs:$i), + "bru $i\n$t", + [(XCoreBR_JT tjumptable:$t, GRRegs:$i)]>; + +let isBranch=1, isIndirectBranch=1, isTerminator=1, isBarrier = 1 in +def BR_JT32 : PseudoInstXCore<(outs), (ins InlineJT32:$t, GRRegs:$i), + "bru $i\n$t", + [(XCoreBR_JT32 tjumptable:$t, GRRegs:$i)]>; + let Defs=[SP], neverHasSideEffects=1 in def SETSP_1r : _F1R<(outs), (ins GRRegs:$src), "set sp, $src", []>; -let isBarrier = 1, hasCtrlDep = 1 in +let hasCtrlDep = 1 in def ECALLT_1r : _F1R<(outs), (ins GRRegs:$src), "ecallt $src", []>; -let isBarrier = 1, hasCtrlDep = 1 in +let hasCtrlDep = 1 in def ECALLF_1r : _F1R<(outs), (ins GRRegs:$src), "ecallf $src", []>; diff --git a/lib/Transforms/Hello/Hello.cpp b/lib/Transforms/Hello/Hello.cpp index eac4e17..37d7a00 100644 --- a/lib/Transforms/Hello/Hello.cpp +++ b/lib/Transforms/Hello/Hello.cpp @@ -15,7 +15,6 @@ #define DEBUG_TYPE "hello" #include "llvm/Pass.h" #include "llvm/Function.h" -#include "llvm/ADT/StringExtras.h" #include "llvm/Support/raw_ostream.h" #include "llvm/ADT/Statistic.h" using namespace llvm; diff --git a/lib/Transforms/IPO/ArgumentPromotion.cpp b/lib/Transforms/IPO/ArgumentPromotion.cpp index 325d353..7cb1367 100644 --- a/lib/Transforms/IPO/ArgumentPromotion.cpp +++ b/lib/Transforms/IPO/ArgumentPromotion.cpp @@ -124,7 +124,7 @@ CallGraphNode *ArgPromotion::PromoteArguments(CallGraphNode *CGN) { unsigned ArgNo = 0; for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end(); I != E; ++I, ++ArgNo) - if (isa<PointerType>(I->getType())) + if (I->getType()->isPointerTy()) PointerArgs.push_back(std::pair<Argument*, unsigned>(I, ArgNo)); if (PointerArgs.empty()) return 0; @@ -317,7 +317,7 @@ bool ArgPromotion::isSafeToPromoteArgument(Argument *Arg, bool isByVal) const { GEPIndicesSet ToPromote; // If the pointer is always valid, any load with first index 0 is valid. - if(isByVal || AllCalleesPassInValidPointerForArgument(Arg)) + if (isByVal || AllCalleesPassInValidPointerForArgument(Arg)) SafeToUnconditionallyLoad.insert(IndicesVector(1, 0)); // First, iterate the entry block and mark loads of (geps of) arguments as @@ -673,7 +673,7 @@ CallGraphNode *ArgPromotion::DoPromotion(Function *F, IE = SI->end(); II != IE; ++II) { // Use i32 to index structs, and i64 for others (pointers/arrays). // This satisfies GEP constraints. - const Type *IdxTy = (isa<StructType>(ElTy) ? + const Type *IdxTy = (ElTy->isStructTy() ? Type::getInt32Ty(F->getContext()) : Type::getInt64Ty(F->getContext())); Ops.push_back(ConstantInt::get(IdxTy, *II)); diff --git a/lib/Transforms/IPO/DeadArgumentElimination.cpp b/lib/Transforms/IPO/DeadArgumentElimination.cpp index 1749b1e..f386ed7 100644 --- a/lib/Transforms/IPO/DeadArgumentElimination.cpp +++ b/lib/Transforms/IPO/DeadArgumentElimination.cpp @@ -796,7 +796,7 @@ bool DAE::RemoveDeadStuffFromFunction(Function *F) { // Replace by null for now. Call->replaceAllUsesWith(Constant::getNullValue(Call->getType())); } else { - assert(isa<StructType>(RetTy) && + assert(RetTy->isStructTy() && "Return type changed, but not into a void. The old return type" " must have been a struct!"); Instruction *InsertPt = Call; @@ -870,7 +870,7 @@ bool DAE::RemoveDeadStuffFromFunction(Function *F) { if (NFTy->getReturnType() == Type::getVoidTy(F->getContext())) { RetVal = 0; } else { - assert (isa<StructType>(RetTy)); + assert (RetTy->isStructTy()); // The original return value was a struct, insert // extractvalue/insertvalue chains to extract only the values we need // to return and insert them into our new result. diff --git a/lib/Transforms/IPO/FunctionAttrs.cpp b/lib/Transforms/IPO/FunctionAttrs.cpp index 64a6d78..298d5cf 100644 --- a/lib/Transforms/IPO/FunctionAttrs.cpp +++ b/lib/Transforms/IPO/FunctionAttrs.cpp @@ -175,7 +175,7 @@ bool FunctionAttrs::AddReadAttrs(const std::vector<CallGraphNode *> &SCC) { for (CallSite::arg_iterator CI = CS.arg_begin(), CE = CS.arg_end(); CI != CE; ++CI) { Value *Arg = *CI; - if (isa<PointerType>(Arg->getType()) && !PointsToLocalMemory(Arg)) + if (Arg->getType()->isPointerTy() && !PointsToLocalMemory(Arg)) // Writes memory. Just give up. return false; } @@ -257,7 +257,7 @@ bool FunctionAttrs::AddNoCaptureAttrs(const std::vector<CallGraphNode *> &SCC) { continue; for (Function::arg_iterator A = F->arg_begin(), E = F->arg_end(); A!=E; ++A) - if (isa<PointerType>(A->getType()) && !A->hasNoCaptureAttr() && + if (A->getType()->isPointerTy() && !A->hasNoCaptureAttr() && !PointerMayBeCaptured(A, true, /*StoreCaptures=*/false)) { A->addAttr(Attribute::NoCapture); ++NumNoCapture; @@ -362,7 +362,7 @@ bool FunctionAttrs::AddNoAliasAttrs(const std::vector<CallGraphNode *> &SCC) { // We annotate noalias return values, which are only applicable to // pointer types. - if (!isa<PointerType>(F->getReturnType())) + if (!F->getReturnType()->isPointerTy()) continue; if (!IsFunctionMallocLike(F, SCCNodes)) @@ -372,7 +372,7 @@ bool FunctionAttrs::AddNoAliasAttrs(const std::vector<CallGraphNode *> &SCC) { bool MadeChange = false; for (unsigned i = 0, e = SCC.size(); i != e; ++i) { Function *F = SCC[i]->getFunction(); - if (F->doesNotAlias(0) || !isa<PointerType>(F->getReturnType())) + if (F->doesNotAlias(0) || !F->getReturnType()->isPointerTy()) continue; F->setDoesNotAlias(0); diff --git a/lib/Transforms/IPO/GlobalOpt.cpp b/lib/Transforms/IPO/GlobalOpt.cpp index df060eb..7b1e9c0 100644 --- a/lib/Transforms/IPO/GlobalOpt.cpp +++ b/lib/Transforms/IPO/GlobalOpt.cpp @@ -303,7 +303,7 @@ static bool CleanupConstantGlobalUsers(Value *V, Constant *Init) { SubInit = ConstantFoldLoadThroughGEPConstantExpr(Init, CE); Changed |= CleanupConstantGlobalUsers(CE, SubInit); } else if (CE->getOpcode() == Instruction::BitCast && - isa<PointerType>(CE->getType())) { + CE->getType()->isPointerTy()) { // Pointer cast, delete any stores and memsets to the global. Changed |= CleanupConstantGlobalUsers(CE, 0); } @@ -431,7 +431,7 @@ static bool IsUserOfGlobalSafeForSRA(User *U, GlobalValue *GV) { else if (const VectorType *SubVectorTy = dyn_cast<VectorType>(*GEPI)) NumElements = SubVectorTy->getNumElements(); else { - assert(isa<StructType>(*GEPI) && + assert((*GEPI)->isStructTy() && "Indexed GEP type is not array, vector, or struct!"); continue; } @@ -543,7 +543,7 @@ static GlobalVariable *SRAGlobal(GlobalVariable *GV, const TargetData &TD) { if (NewGlobals.empty()) return 0; - + DEBUG(dbgs() << "PERFORMING GLOBAL SRA ON: " << *GV); Constant *NullInt =Constant::getNullValue(Type::getInt32Ty(GV->getContext())); @@ -642,7 +642,7 @@ static bool AllUsesOfValueWillTrapIfNull(Value *V, return false; } else if (isa<ICmpInst>(*UI) && isa<ConstantPointerNull>(UI->getOperand(1))) { - // Ignore setcc X, null + // Ignore icmp X, null } else { //cerr << "NONTRAPPING USE: " << **UI; return false; @@ -813,57 +813,47 @@ static void ConstantPropUsersOf(Value *V) { static GlobalVariable *OptimizeGlobalAddressOfMalloc(GlobalVariable *GV, CallInst *CI, const Type *AllocTy, - Value* NElems, + ConstantInt *NElements, TargetData* TD) { - DEBUG(dbgs() << "PROMOTING GLOBAL: " << *GV << " CALL = " << *CI << '\n'); - - const Type *IntPtrTy = TD->getIntPtrType(GV->getContext()); + DEBUG(errs() << "PROMOTING GLOBAL: " << *GV << " CALL = " << *CI << '\n'); - // CI has either 0 or 1 bitcast uses (getMallocType() would otherwise have - // returned NULL and we would not be here). - BitCastInst *BCI = NULL; - for (Value::use_iterator UI = CI->use_begin(), E = CI->use_end(); UI != E; ) - if ((BCI = dyn_cast<BitCastInst>(cast<Instruction>(*UI++)))) - break; - - ConstantInt *NElements = cast<ConstantInt>(NElems); - if (NElements->getZExtValue() != 1) { - // If we have an array allocation, transform it to a single element - // allocation to make the code below simpler. - Type *NewTy = ArrayType::get(AllocTy, NElements->getZExtValue()); - unsigned TypeSize = TD->getTypeAllocSize(NewTy); - if (const StructType *ST = dyn_cast<StructType>(NewTy)) - TypeSize = TD->getStructLayout(ST)->getSizeInBytes(); - Instruction *NewCI = CallInst::CreateMalloc(CI, IntPtrTy, NewTy, - ConstantInt::get(IntPtrTy, TypeSize)); - Value* Indices[2]; - Indices[0] = Indices[1] = Constant::getNullValue(IntPtrTy); - Value *NewGEP = GetElementPtrInst::Create(NewCI, Indices, Indices + 2, - NewCI->getName()+".el0", CI); - Value *Cast = new BitCastInst(NewGEP, CI->getType(), "el0", CI); - if (BCI) BCI->replaceAllUsesWith(NewGEP); - CI->replaceAllUsesWith(Cast); - if (BCI) BCI->eraseFromParent(); - CI->eraseFromParent(); - BCI = dyn_cast<BitCastInst>(NewCI); - CI = BCI ? extractMallocCallFromBitCast(BCI) : cast<CallInst>(NewCI); - } + const Type *GlobalType; + if (NElements->getZExtValue() == 1) + GlobalType = AllocTy; + else + // If we have an array allocation, the global variable is of an array. + GlobalType = ArrayType::get(AllocTy, NElements->getZExtValue()); // Create the new global variable. The contents of the malloc'd memory is // undefined, so initialize with an undef value. - const Type *MAT = getMallocAllocatedType(CI); - Constant *Init = UndefValue::get(MAT); GlobalVariable *NewGV = new GlobalVariable(*GV->getParent(), - MAT, false, - GlobalValue::InternalLinkage, Init, + GlobalType, false, + GlobalValue::InternalLinkage, + UndefValue::get(GlobalType), GV->getName()+".body", GV, GV->isThreadLocal()); - // Anything that used the malloc or its bitcast now uses the global directly. - if (BCI) BCI->replaceAllUsesWith(NewGV); - CI->replaceAllUsesWith(new BitCastInst(NewGV, CI->getType(), "newgv", CI)); - + // If there are bitcast users of the malloc (which is typical, usually we have + // a malloc + bitcast) then replace them with uses of the new global. Update + // other users to use the global as well. + BitCastInst *TheBC = 0; + while (!CI->use_empty()) { + Instruction *User = cast<Instruction>(CI->use_back()); + if (BitCastInst *BCI = dyn_cast<BitCastInst>(User)) { + if (BCI->getType() == NewGV->getType()) { + BCI->replaceAllUsesWith(NewGV); + BCI->eraseFromParent(); + } else { + BCI->setOperand(0, NewGV); + } + } else { + if (TheBC == 0) + TheBC = new BitCastInst(NewGV, CI->getType(), "newgv", CI); + User->replaceUsesOfWith(CI, TheBC); + } + } + Constant *RepValue = NewGV; if (NewGV->getType() != GV->getType()->getElementType()) RepValue = ConstantExpr::getBitCast(RepValue, @@ -879,60 +869,60 @@ static GlobalVariable *OptimizeGlobalAddressOfMalloc(GlobalVariable *GV, bool InitBoolUsed = false; // Loop over all uses of GV, processing them in turn. - std::vector<StoreInst*> Stores; - while (!GV->use_empty()) - if (LoadInst *LI = dyn_cast<LoadInst>(GV->use_back())) { - while (!LI->use_empty()) { - Use &LoadUse = LI->use_begin().getUse(); - if (!isa<ICmpInst>(LoadUse.getUser())) - LoadUse = RepValue; - else { - ICmpInst *ICI = cast<ICmpInst>(LoadUse.getUser()); - // Replace the cmp X, 0 with a use of the bool value. - Value *LV = new LoadInst(InitBool, InitBool->getName()+".val", ICI); - InitBoolUsed = true; - switch (ICI->getPredicate()) { - default: llvm_unreachable("Unknown ICmp Predicate!"); - case ICmpInst::ICMP_ULT: - case ICmpInst::ICMP_SLT: // X < null -> always false - LV = ConstantInt::getFalse(GV->getContext()); - break; - case ICmpInst::ICMP_ULE: - case ICmpInst::ICMP_SLE: - case ICmpInst::ICMP_EQ: - LV = BinaryOperator::CreateNot(LV, "notinit", ICI); - break; - case ICmpInst::ICMP_NE: - case ICmpInst::ICMP_UGE: - case ICmpInst::ICMP_SGE: - case ICmpInst::ICMP_UGT: - case ICmpInst::ICMP_SGT: - break; // no change. - } - ICI->replaceAllUsesWith(LV); - ICI->eraseFromParent(); - } - } - LI->eraseFromParent(); - } else { - StoreInst *SI = cast<StoreInst>(GV->use_back()); + while (!GV->use_empty()) { + if (StoreInst *SI = dyn_cast<StoreInst>(GV->use_back())) { // The global is initialized when the store to it occurs. new StoreInst(ConstantInt::getTrue(GV->getContext()), InitBool, SI); SI->eraseFromParent(); + continue; } + + LoadInst *LI = cast<LoadInst>(GV->use_back()); + while (!LI->use_empty()) { + Use &LoadUse = LI->use_begin().getUse(); + if (!isa<ICmpInst>(LoadUse.getUser())) { + LoadUse = RepValue; + continue; + } + + ICmpInst *ICI = cast<ICmpInst>(LoadUse.getUser()); + // Replace the cmp X, 0 with a use of the bool value. + Value *LV = new LoadInst(InitBool, InitBool->getName()+".val", ICI); + InitBoolUsed = true; + switch (ICI->getPredicate()) { + default: llvm_unreachable("Unknown ICmp Predicate!"); + case ICmpInst::ICMP_ULT: + case ICmpInst::ICMP_SLT: // X < null -> always false + LV = ConstantInt::getFalse(GV->getContext()); + break; + case ICmpInst::ICMP_ULE: + case ICmpInst::ICMP_SLE: + case ICmpInst::ICMP_EQ: + LV = BinaryOperator::CreateNot(LV, "notinit", ICI); + break; + case ICmpInst::ICMP_NE: + case ICmpInst::ICMP_UGE: + case ICmpInst::ICMP_SGE: + case ICmpInst::ICMP_UGT: + case ICmpInst::ICMP_SGT: + break; // no change. + } + ICI->replaceAllUsesWith(LV); + ICI->eraseFromParent(); + } + LI->eraseFromParent(); + } // If the initialization boolean was used, insert it, otherwise delete it. if (!InitBoolUsed) { while (!InitBool->use_empty()) // Delete initializations - cast<Instruction>(InitBool->use_back())->eraseFromParent(); + cast<StoreInst>(InitBool->use_back())->eraseFromParent(); delete InitBool; } else GV->getParent()->getGlobalList().insert(GV, InitBool); - - // Now the GV is dead, nuke it and the malloc (both CI and BCI). + // Now the GV is dead, nuke it and the malloc.. GV->eraseFromParent(); - if (BCI) BCI->eraseFromParent(); CI->eraseFromParent(); // To further other optimizations, loop over all users of NewGV and try to @@ -1303,9 +1293,7 @@ static GlobalVariable *PerformHeapAllocSRoA(GlobalVariable *GV, CallInst *CI, ConstantInt::get(IntPtrTy, TypeSize), NElems, CI->getName() + ".f" + Twine(FieldNo)); - CallInst *NCI = dyn_cast<BitCastInst>(NMI) ? - extractMallocCallFromBitCast(NMI) : cast<CallInst>(NMI); - FieldMallocs.push_back(NCI); + FieldMallocs.push_back(NMI); new StoreInst(NMI, NGV, CI); } @@ -1497,7 +1485,7 @@ static bool TryToOptimizeStoreOfMallocToGlobal(GlobalVariable *GV, // something. if (TD && NElements->getZExtValue() * TD->getTypeAllocSize(AllocTy) < 2048) { - GVI = OptimizeGlobalAddressOfMalloc(GV, CI, AllocTy, NElems, TD); + GVI = OptimizeGlobalAddressOfMalloc(GV, CI, AllocTy, NElements, TD); return true; } @@ -1556,7 +1544,7 @@ static bool OptimizeOnceStoredGlobal(GlobalVariable *GV, Value *StoredOnceVal, // only has one (non-null) value stored into it, then we can optimize any // users of the loaded value (often calls and loads) that would trap if the // value was null. - if (isa<PointerType>(GV->getInitializer()->getType()) && + if (GV->getInitializer()->getType()->isPointerTy() && GV->getInitializer()->isNullValue()) { if (Constant *SOVC = dyn_cast<Constant>(StoredOnceVal)) { if (GV->getInitializer()->getType() != SOVC->getType()) @@ -1591,7 +1579,7 @@ static bool TryToShrinkGlobalToBoolean(GlobalVariable *GV, Constant *OtherVal) { // where v1 and v2 both require constant pool loads, a big loss. if (GVElType == Type::getInt1Ty(GV->getContext()) || GVElType->isFloatingPointTy() || - isa<PointerType>(GVElType) || isa<VectorType>(GVElType)) + GVElType->isPointerTy() || GVElType->isVectorTy()) return false; // Walk the use list of the global seeing if all the uses are load or store. @@ -2148,7 +2136,7 @@ static Constant *EvaluateStoreInto(Constant *Init, Constant *Val, Elts[CI->getZExtValue()] = EvaluateStoreInto(Elts[CI->getZExtValue()], Val, Addr, OpNo+1); - if (isa<ArrayType>(Init->getType())) + if (Init->getType()->isArrayTy()) return ConstantArray::get(cast<ArrayType>(InitTy), Elts); else return ConstantVector::get(&Elts[0], Elts.size()); diff --git a/lib/Transforms/InstCombine/InstCombineAddSub.cpp b/lib/Transforms/InstCombine/InstCombineAddSub.cpp index 2da17f1..4d2c89e 100644 --- a/lib/Transforms/InstCombine/InstCombineAddSub.cpp +++ b/lib/Transforms/InstCombine/InstCombineAddSub.cpp @@ -373,10 +373,10 @@ Instruction *InstCombiner::visitFAdd(BinaryOperator &I) { if (CFP->getValueAPF().isPosZero() && CannotBeNegativeZero(LHS)) return ReplaceInstUsesWith(I, LHS); - // Check for (add double (sitofp x), y), see if we can merge this into an + // Check for (fadd double (sitofp x), y), see if we can merge this into an // integer add followed by a promotion. if (SIToFPInst *LHSConv = dyn_cast<SIToFPInst>(LHS)) { - // (add double (sitofp x), fpcst) --> (sitofp (add int x, intcst)) + // (fadd double (sitofp x), fpcst) --> (sitofp (add int x, intcst)) // ... if the constant fits in the integer value. This is useful for things // like (double)(x & 1234) + 4.0 -> (double)((X & 1234)+4) which no longer // requires a constant pool load, and generally allows the add to be better @@ -394,7 +394,7 @@ Instruction *InstCombiner::visitFAdd(BinaryOperator &I) { } } - // (add double (sitofp x), (sitofp y)) --> (sitofp (add int x, y)) + // (fadd double (sitofp x), (sitofp y)) --> (sitofp (add int x, y)) if (SIToFPInst *RHSConv = dyn_cast<SIToFPInst>(RHS)) { // Only do this if x/y have the same type, if at last one of them has a // single use (so we don't increase the number of int->fp conversions), diff --git a/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp b/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp index 5e47953..86673f8 100644 --- a/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp +++ b/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp @@ -1464,8 +1464,7 @@ Instruction *InstCombiner::visitOr(BinaryOperator &I) { if (Value *V = SimplifyOrInst(Op0, Op1, TD)) return ReplaceInstUsesWith(I, V); - - + // See if we can simplify any instructions used by the instruction whose sole // purpose is to compute bits we don't care about. if (SimplifyDemandedInstructionBits(I)) @@ -1474,7 +1473,9 @@ Instruction *InstCombiner::visitOr(BinaryOperator &I) { if (ConstantInt *RHS = dyn_cast<ConstantInt>(Op1)) { ConstantInt *C1 = 0; Value *X = 0; // (X & C1) | C2 --> (X | C2) & (C1|C2) + // iff (C1 & C2) == 0. if (match(Op0, m_And(m_Value(X), m_ConstantInt(C1))) && + (RHS->getValue() & C1->getValue()) != 0 && Op0->hasOneUse()) { Value *Or = Builder->CreateOr(X, RHS); Or->takeName(Op0); @@ -1497,6 +1498,7 @@ Instruction *InstCombiner::visitOr(BinaryOperator &I) { if (SelectInst *SI = dyn_cast<SelectInst>(Op0)) if (Instruction *R = FoldOpIntoSelect(I, SI)) return R; + if (isa<PHINode>(Op0)) if (Instruction *NV = FoldOpIntoPhi(I)) return NV; @@ -1618,7 +1620,7 @@ Instruction *InstCombiner::visitOr(BinaryOperator &I) { // (A & (C0?-1:0)) | (B & ~(C0?-1:0)) -> C0 ? A : B, and commuted variants. // Don't do this for vector select idioms, the code generator doesn't handle // them well yet. - if (!isa<VectorType>(I.getType())) { + if (!I.getType()->isVectorTy()) { if (Instruction *Match = MatchSelectFromAndOr(A, B, C, D)) return Match; if (Instruction *Match = MatchSelectFromAndOr(B, A, D, C)) @@ -1755,7 +1757,7 @@ Instruction *InstCombiner::visitXor(BinaryOperator &I) { // purpose is to compute bits we don't care about. if (SimplifyDemandedInstructionBits(I)) return &I; - if (isa<VectorType>(I.getType())) + if (I.getType()->isVectorTy()) if (isa<ConstantAggregateZero>(Op1)) return ReplaceInstUsesWith(I, Op0); // X ^ <0,0> -> X diff --git a/lib/Transforms/InstCombine/InstCombineCalls.cpp b/lib/Transforms/InstCombine/InstCombineCalls.cpp index d7efdcf..835d149 100644 --- a/lib/Transforms/InstCombine/InstCombineCalls.cpp +++ b/lib/Transforms/InstCombine/InstCombineCalls.cpp @@ -319,7 +319,7 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) { if (GlobalVariable *GV = dyn_cast<GlobalVariable>(Op1)) { if (GV->hasDefinitiveInitializer()) { Constant *C = GV->getInitializer(); - size_t globalSize = TD->getTypeAllocSize(C->getType()); + uint64_t globalSize = TD->getTypeAllocSize(C->getType()); return ReplaceInstUsesWith(CI, ConstantInt::get(ReturnTy, globalSize)); } else { Constant *RetVal = ConstantInt::get(ReturnTy, Min ? 0 : -1ULL); @@ -341,16 +341,21 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) { // Get what we're pointing to and its size. const PointerType *BaseType = cast<PointerType>(Operand->getType()); - size_t Size = TD->getTypeAllocSize(BaseType->getElementType()); + uint64_t Size = TD->getTypeAllocSize(BaseType->getElementType()); // Get the current byte offset into the thing. Use the original // operand in case we're looking through a bitcast. SmallVector<Value*, 8> Ops(CE->op_begin()+1, CE->op_end()); const PointerType *OffsetType = cast<PointerType>(GEP->getPointerOperand()->getType()); - size_t Offset = TD->getIndexedOffset(OffsetType, &Ops[0], Ops.size()); + uint64_t Offset = TD->getIndexedOffset(OffsetType, &Ops[0], Ops.size()); - assert(Size >= Offset); + if (Size < Offset) { + // Out of bound reference? Negative index normalized to large + // index? Just return "I don't know". + Constant *RetVal = ConstantInt::get(ReturnTy, Min ? 0 : -1ULL); + return ReplaceInstUsesWith(CI, RetVal); + } Constant *RetVal = ConstantInt::get(ReturnTy, Size-Offset); return ReplaceInstUsesWith(CI, RetVal); @@ -831,7 +836,7 @@ bool InstCombiner::transformConstExprCastCall(CallSite CS) { const Type *OldRetTy = Caller->getType(); const Type *NewRetTy = FT->getReturnType(); - if (isa<StructType>(NewRetTy)) + if (NewRetTy->isStructTy()) return false; // TODO: Handle multiple return values. // Check to see if we are changing the return type... @@ -839,9 +844,9 @@ bool InstCombiner::transformConstExprCastCall(CallSite CS) { if (Callee->isDeclaration() && // Conversion is ok if changing from one pointer type to another or from // a pointer to an integer of the same size. - !((isa<PointerType>(OldRetTy) || !TD || + !((OldRetTy->isPointerTy() || !TD || OldRetTy == TD->getIntPtrType(Caller->getContext())) && - (isa<PointerType>(NewRetTy) || !TD || + (NewRetTy->isPointerTy() || !TD || NewRetTy == TD->getIntPtrType(Caller->getContext())))) return false; // Cannot transform this return value. @@ -888,9 +893,9 @@ bool InstCombiner::transformConstExprCastCall(CallSite CS) { // Converting from one pointer type to another or between a pointer and an // integer of the same size is safe even if we do not have a body. bool isConvertible = ActTy == ParamTy || - (TD && ((isa<PointerType>(ParamTy) || + (TD && ((ParamTy->isPointerTy() || ParamTy == TD->getIntPtrType(Caller->getContext())) && - (isa<PointerType>(ActTy) || + (ActTy->isPointerTy() || ActTy == TD->getIntPtrType(Caller->getContext())))); if (Callee->isDeclaration() && !isConvertible) return false; } diff --git a/lib/Transforms/InstCombine/InstCombineCasts.cpp b/lib/Transforms/InstCombine/InstCombineCasts.cpp index bb4a0e9..a68fc6d 100644 --- a/lib/Transforms/InstCombine/InstCombineCasts.cpp +++ b/lib/Transforms/InstCombine/InstCombineCasts.cpp @@ -272,7 +272,7 @@ bool InstCombiner::ShouldOptimizeCast(Instruction::CastOps opc, const Value *V, // If this is a vector sext from a compare, then we don't want to break the // idiom where each element of the extended vector is either zero or all ones. - if (opc == Instruction::SExt && isa<CmpInst>(V) && isa<VectorType>(Ty)) + if (opc == Instruction::SExt && isa<CmpInst>(V) && Ty->isVectorTy()) return false; return true; @@ -303,8 +303,8 @@ Instruction *InstCombiner::commonCastTransforms(CastInst &CI) { if (isa<PHINode>(Src)) { // We don't do this if this would create a PHI node with an illegal type if // it is currently legal. - if (!isa<IntegerType>(Src->getType()) || - !isa<IntegerType>(CI.getType()) || + if (!Src->getType()->isIntegerTy() || + !CI.getType()->isIntegerTy() || ShouldChangeType(CI.getType(), Src->getType())) if (Instruction *NV = FoldOpIntoPhi(CI)) return NV; @@ -436,7 +436,7 @@ Instruction *InstCombiner::visitTrunc(TruncInst &CI) { // type. Only do this if the dest type is a simple type, don't convert the // expression tree to something weird like i93 unless the source is also // strange. - if ((isa<VectorType>(DestTy) || ShouldChangeType(SrcTy, DestTy)) && + if ((DestTy->isVectorTy() || ShouldChangeType(SrcTy, DestTy)) && CanEvaluateTruncated(Src, DestTy)) { // If this cast is a truncate, evaluting in a different type always @@ -728,7 +728,7 @@ Instruction *InstCombiner::visitZExt(ZExtInst &CI) { // expression tree to something weird like i93 unless the source is also // strange. unsigned BitsToClear; - if ((isa<VectorType>(DestTy) || ShouldChangeType(SrcTy, DestTy)) && + if ((DestTy->isVectorTy() || ShouldChangeType(SrcTy, DestTy)) && CanEvaluateZExtd(Src, DestTy, BitsToClear)) { assert(BitsToClear < SrcTy->getScalarSizeInBits() && "Unreasonable BitsToClear"); @@ -936,7 +936,7 @@ Instruction *InstCombiner::visitSExt(SExtInst &CI) { // type. Only do this if the dest type is a simple type, don't convert the // expression tree to something weird like i93 unless the source is also // strange. - if ((isa<VectorType>(DestTy) || ShouldChangeType(SrcTy, DestTy)) && + if ((DestTy->isVectorTy() || ShouldChangeType(SrcTy, DestTy)) && CanEvaluateSExtd(Src, DestTy)) { // Okay, we can transform this! Insert the new expression now. DEBUG(dbgs() << "ICE: EvaluateInDifferentType converting expression type" @@ -1289,7 +1289,7 @@ Instruction *InstCombiner::visitBitCast(BitCastInst &CI) { Constant::getNullValue(Type::getInt32Ty(CI.getContext())); unsigned NumZeros = 0; while (SrcElTy != DstElTy && - isa<CompositeType>(SrcElTy) && !isa<PointerType>(SrcElTy) && + isa<CompositeType>(SrcElTy) && !SrcElTy->isPointerTy() && SrcElTy->getNumContainedTypes() /* not "{}" */) { SrcElTy = cast<CompositeType>(SrcElTy)->getTypeAtIndex(ZeroUInt); ++NumZeros; @@ -1304,7 +1304,7 @@ Instruction *InstCombiner::visitBitCast(BitCastInst &CI) { } if (const VectorType *DestVTy = dyn_cast<VectorType>(DestTy)) { - if (DestVTy->getNumElements() == 1 && !isa<VectorType>(SrcTy)) { + if (DestVTy->getNumElements() == 1 && !SrcTy->isVectorTy()) { Value *Elem = Builder->CreateBitCast(Src, DestVTy->getElementType()); return InsertElementInst::Create(UndefValue::get(DestTy), Elem, Constant::getNullValue(Type::getInt32Ty(CI.getContext()))); @@ -1313,7 +1313,7 @@ Instruction *InstCombiner::visitBitCast(BitCastInst &CI) { } if (const VectorType *SrcVTy = dyn_cast<VectorType>(SrcTy)) { - if (SrcVTy->getNumElements() == 1 && !isa<VectorType>(DestTy)) { + if (SrcVTy->getNumElements() == 1 && !DestTy->isVectorTy()) { Value *Elem = Builder->CreateExtractElement(Src, Constant::getNullValue(Type::getInt32Ty(CI.getContext()))); @@ -1324,7 +1324,7 @@ Instruction *InstCombiner::visitBitCast(BitCastInst &CI) { if (ShuffleVectorInst *SVI = dyn_cast<ShuffleVectorInst>(Src)) { // Okay, we have (bitcast (shuffle ..)). Check to see if this is // a bitconvert to a vector with the same # elts. - if (SVI->hasOneUse() && isa<VectorType>(DestTy) && + if (SVI->hasOneUse() && DestTy->isVectorTy() && cast<VectorType>(DestTy)->getNumElements() == SVI->getType()->getNumElements() && SVI->getType()->getNumElements() == @@ -1346,7 +1346,7 @@ Instruction *InstCombiner::visitBitCast(BitCastInst &CI) { } } - if (isa<PointerType>(SrcTy)) + if (SrcTy->isPointerTy()) return commonPointerCastTransforms(CI); return commonCastTransforms(CI); } diff --git a/lib/Transforms/InstCombine/InstCombineCompares.cpp b/lib/Transforms/InstCombine/InstCombineCompares.cpp index 72af80f..518af74 100644 --- a/lib/Transforms/InstCombine/InstCombineCompares.cpp +++ b/lib/Transforms/InstCombine/InstCombineCompares.cpp @@ -1988,7 +1988,7 @@ Instruction *InstCombiner::visitICmpInst(ICmpInst &I) { // values. If the ptr->ptr cast can be stripped off both arguments, we do so // now. if (BitCastInst *CI = dyn_cast<BitCastInst>(Op0)) { - if (isa<PointerType>(Op0->getType()) && + if (Op0->getType()->isPointerTy() && (isa<Constant>(Op1) || isa<BitCastInst>(Op1))) { // We keep moving the cast from the left operand over to the right // operand, where it can often be eliminated completely. @@ -2458,17 +2458,17 @@ Instruction *InstCombiner::visitFCmpInst(FCmpInst &I) { return SelectInst::Create(LHSI->getOperand(0), Op1, Op2); break; } - case Instruction::Load: - if (GetElementPtrInst *GEP = - dyn_cast<GetElementPtrInst>(LHSI->getOperand(0))) { - if (GlobalVariable *GV = dyn_cast<GlobalVariable>(GEP->getOperand(0))) - if (GV->isConstant() && GV->hasDefinitiveInitializer() && - !cast<LoadInst>(LHSI)->isVolatile()) - if (Instruction *Res = FoldCmpLoadFromIndexedGlobal(GEP, GV, I)) - return Res; + case Instruction::Load: + if (GetElementPtrInst *GEP = + dyn_cast<GetElementPtrInst>(LHSI->getOperand(0))) { + if (GlobalVariable *GV = dyn_cast<GlobalVariable>(GEP->getOperand(0))) + if (GV->isConstant() && GV->hasDefinitiveInitializer() && + !cast<LoadInst>(LHSI)->isVolatile()) + if (Instruction *Res = FoldCmpLoadFromIndexedGlobal(GEP, GV, I)) + return Res; + } + break; } - break; - } } return Changed ? &I : 0; diff --git a/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp b/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp index e6c59c7..0f2a24f 100644 --- a/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp +++ b/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp @@ -87,8 +87,8 @@ static Instruction *InstCombineLoadCast(InstCombiner &IC, LoadInst &LI, const Type *SrcPTy = SrcTy->getElementType(); - if (DestPTy->isIntegerTy() || isa<PointerType>(DestPTy) || - isa<VectorType>(DestPTy)) { + if (DestPTy->isIntegerTy() || DestPTy->isPointerTy() || + DestPTy->isVectorTy()) { // If the source is an array, the code below will not succeed. Check to // see if a trivial 'gep P, 0, 0' will help matters. Only do this for // constants. @@ -104,11 +104,11 @@ static Instruction *InstCombineLoadCast(InstCombiner &IC, LoadInst &LI, } if (IC.getTargetData() && - (SrcPTy->isIntegerTy() || isa<PointerType>(SrcPTy) || - isa<VectorType>(SrcPTy)) && + (SrcPTy->isIntegerTy() || SrcPTy->isPointerTy() || + SrcPTy->isVectorTy()) && // Do not allow turning this into a load of an integer, which is then // casted to a pointer, this pessimizes pointer analysis a lot. - (isa<PointerType>(SrcPTy) == isa<PointerType>(LI.getType())) && + (SrcPTy->isPointerTy() == LI.getType()->isPointerTy()) && IC.getTargetData()->getTypeSizeInBits(SrcPTy) == IC.getTargetData()->getTypeSizeInBits(DestPTy)) { @@ -243,7 +243,7 @@ static Instruction *InstCombineStoreToCast(InstCombiner &IC, StoreInst &SI) { const Type *SrcPTy = SrcTy->getElementType(); - if (!DestPTy->isIntegerTy() && !isa<PointerType>(DestPTy)) + if (!DestPTy->isIntegerTy() && !DestPTy->isPointerTy()) return 0; /// NewGEPIndices - If SrcPTy is an aggregate type, we can emit a "noop gep" @@ -255,7 +255,7 @@ static Instruction *InstCombineStoreToCast(InstCombiner &IC, StoreInst &SI) { // If the source is an array, the code below will not succeed. Check to // see if a trivial 'gep P, 0, 0' will help matters. Only do this for // constants. - if (isa<ArrayType>(SrcPTy) || isa<StructType>(SrcPTy)) { + if (SrcPTy->isArrayTy() || SrcPTy->isStructTy()) { // Index through pointer. Constant *Zero = Constant::getNullValue(Type::getInt32Ty(SI.getContext())); NewGEPIndices.push_back(Zero); @@ -277,7 +277,7 @@ static Instruction *InstCombineStoreToCast(InstCombiner &IC, StoreInst &SI) { SrcTy = PointerType::get(SrcPTy, SrcTy->getAddressSpace()); } - if (!SrcPTy->isIntegerTy() && !isa<PointerType>(SrcPTy)) + if (!SrcPTy->isIntegerTy() && !SrcPTy->isPointerTy()) return 0; // If the pointers point into different address spaces or if they point to @@ -297,11 +297,11 @@ static Instruction *InstCombineStoreToCast(InstCombiner &IC, StoreInst &SI) { Instruction::CastOps opcode = Instruction::BitCast; const Type* CastSrcTy = SIOp0->getType(); const Type* CastDstTy = SrcPTy; - if (isa<PointerType>(CastDstTy)) { + if (CastDstTy->isPointerTy()) { if (CastSrcTy->isIntegerTy()) opcode = Instruction::IntToPtr; - } else if (isa<IntegerType>(CastDstTy)) { - if (isa<PointerType>(SIOp0->getType())) + } else if (CastDstTy->isIntegerTy()) { + if (SIOp0->getType()->isPointerTy()) opcode = Instruction::PtrToInt; } @@ -413,7 +413,7 @@ Instruction *InstCombiner::visitStoreInst(StoreInst &SI) { // Don't count debug info directives, lest they affect codegen, // and we skip pointer-to-pointer bitcasts, which are NOPs. if (isa<DbgInfoIntrinsic>(BBI) || - (isa<BitCastInst>(BBI) && isa<PointerType>(BBI->getType()))) { + (isa<BitCastInst>(BBI) && BBI->getType()->isPointerTy())) { ScanInsts++; continue; } @@ -483,7 +483,7 @@ Instruction *InstCombiner::visitStoreInst(StoreInst &SI) { do { ++BBI; } while (isa<DbgInfoIntrinsic>(BBI) || - (isa<BitCastInst>(BBI) && isa<PointerType>(BBI->getType()))); + (isa<BitCastInst>(BBI) && BBI->getType()->isPointerTy())); if (BranchInst *BI = dyn_cast<BranchInst>(BBI)) if (BI->isUnconditional()) if (SimplifyStoreAtEndOfBlock(SI)) @@ -544,7 +544,7 @@ bool InstCombiner::SimplifyStoreAtEndOfBlock(StoreInst &SI) { --BBI; // Skip over debugging info. while (isa<DbgInfoIntrinsic>(BBI) || - (isa<BitCastInst>(BBI) && isa<PointerType>(BBI->getType()))) { + (isa<BitCastInst>(BBI) && BBI->getType()->isPointerTy())) { if (BBI==OtherBB->begin()) return false; --BBI; diff --git a/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp b/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp index 668c34f..b3974e8 100644 --- a/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp +++ b/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp @@ -76,7 +76,7 @@ Instruction *InstCombiner::visitMul(BinaryOperator &I) { return BinaryOperator::CreateShl(Op0, ConstantInt::get(Op0->getType(), Val.logBase2())); } - } else if (isa<VectorType>(Op1C->getType())) { + } else if (Op1C->getType()->isVectorTy()) { if (Op1C->isNullValue()) return ReplaceInstUsesWith(I, Op1C); @@ -173,7 +173,7 @@ Instruction *InstCombiner::visitMul(BinaryOperator &I) { // If one of the operands of the multiply is a cast from a boolean value, then // we know the bool is either zero or one, so this is a 'masking' multiply. // X * Y (where Y is 0 or 1) -> X & (0-Y) - if (!isa<VectorType>(I.getType())) { + if (!I.getType()->isVectorTy()) { // -2 is "-1 << 1" so it is all bits set except the low one. APInt Negative2(I.getType()->getPrimitiveSizeInBits(), (uint64_t)-2, true); @@ -203,8 +203,8 @@ Instruction *InstCombiner::visitFMul(BinaryOperator &I) { // "In IEEE floating point, x*1 is not equivalent to x for nans. However, // ANSI says we can drop signals, so we can do this anyway." (from GCC) if (Op1F->isExactlyValue(1.0)) - return ReplaceInstUsesWith(I, Op0); // Eliminate 'mul double %X, 1.0' - } else if (isa<VectorType>(Op1C->getType())) { + return ReplaceInstUsesWith(I, Op0); // Eliminate 'fmul double %X, 1.0' + } else if (Op1C->getType()->isVectorTy()) { if (ConstantVector *Op1V = dyn_cast<ConstantVector>(Op1C)) { // As above, vector X*splat(1.0) -> X in all defined cases. if (Constant *Splat = Op1V->getSplatValue()) { diff --git a/lib/Transforms/InstCombine/InstCombinePHI.cpp b/lib/Transforms/InstCombine/InstCombinePHI.cpp index bb7632f..fba8354 100644 --- a/lib/Transforms/InstCombine/InstCombinePHI.cpp +++ b/lib/Transforms/InstCombine/InstCombinePHI.cpp @@ -371,7 +371,7 @@ Instruction *InstCombiner::FoldPHIArgOpIntoPHI(PHINode &PN) { // Be careful about transforming integer PHIs. We don't want to pessimize // the code by turning an i32 into an i1293. - if (isa<IntegerType>(PN.getType()) && isa<IntegerType>(CastSrcTy)) { + if (PN.getType()->isIntegerTy() && CastSrcTy->isIntegerTy()) { if (!ShouldChangeType(PN.getType(), CastSrcTy)) return 0; } @@ -832,7 +832,7 @@ Instruction *InstCombiner::visitPHINode(PHINode &PN) { // it is only used by trunc or trunc(lshr) operations. If so, we split the // PHI into the various pieces being extracted. This sort of thing is // introduced when SROA promotes an aggregate to a single large integer type. - if (isa<IntegerType>(PN.getType()) && TD && + if (PN.getType()->isIntegerTy() && TD && !TD->isLegalInteger(PN.getType()->getPrimitiveSizeInBits())) if (Instruction *Res = SliceUpIllegalIntegerPHI(PN)) return Res; diff --git a/lib/Transforms/InstCombine/InstCombineSelect.cpp b/lib/Transforms/InstCombine/InstCombineSelect.cpp index 7807d9a..2fc9325 100644 --- a/lib/Transforms/InstCombine/InstCombineSelect.cpp +++ b/lib/Transforms/InstCombine/InstCombineSelect.cpp @@ -539,9 +539,18 @@ Instruction *InstCombiner::visitSelectInst(SelectInst &SI) { !CFPf->getValueAPF().isZero())) return ReplaceInstUsesWith(SI, FalseVal); } - // Transform (X != Y) ? X : Y -> X - if (FCI->getPredicate() == FCmpInst::FCMP_ONE) + // Transform (X une Y) ? X : Y -> X + if (FCI->getPredicate() == FCmpInst::FCMP_UNE) { + // This is not safe in general for floating point: + // consider X== -0, Y== +0. + // It becomes safe if either operand is a nonzero constant. + ConstantFP *CFPt, *CFPf; + if (((CFPt = dyn_cast<ConstantFP>(TrueVal)) && + !CFPt->getValueAPF().isZero()) || + ((CFPf = dyn_cast<ConstantFP>(FalseVal)) && + !CFPf->getValueAPF().isZero())) return ReplaceInstUsesWith(SI, TrueVal); + } // NOTE: if we wanted to, this is where to detect MIN/MAX } else if (FCI->getOperand(0) == FalseVal && FCI->getOperand(1) == TrueVal){ @@ -557,9 +566,18 @@ Instruction *InstCombiner::visitSelectInst(SelectInst &SI) { !CFPf->getValueAPF().isZero())) return ReplaceInstUsesWith(SI, FalseVal); } - // Transform (X != Y) ? Y : X -> Y - if (FCI->getPredicate() == FCmpInst::FCMP_ONE) - return ReplaceInstUsesWith(SI, TrueVal); + // Transform (X une Y) ? Y : X -> Y + if (FCI->getPredicate() == FCmpInst::FCMP_UNE) { + // This is not safe in general for floating point: + // consider X== -0, Y== +0. + // It becomes safe if either operand is a nonzero constant. + ConstantFP *CFPt, *CFPf; + if (((CFPt = dyn_cast<ConstantFP>(TrueVal)) && + !CFPt->getValueAPF().isZero()) || + ((CFPf = dyn_cast<ConstantFP>(FalseVal)) && + !CFPf->getValueAPF().isZero())) + return ReplaceInstUsesWith(SI, TrueVal); + } // NOTE: if we wanted to, this is where to detect MIN/MAX } // NOTE: if we wanted to, this is where to detect ABS diff --git a/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp b/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp index 5e9a52f..cd41844 100644 --- a/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp +++ b/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp @@ -104,7 +104,7 @@ Value *InstCombiner::SimplifyDemandedUseBits(Value *V, APInt DemandedMask, assert(Depth <= 6 && "Limit Search Depth"); uint32_t BitWidth = DemandedMask.getBitWidth(); const Type *VTy = V->getType(); - assert((TD || !isa<PointerType>(VTy)) && + assert((TD || !VTy->isPointerTy()) && "SimplifyDemandedBits needs to know bit widths!"); assert((!TD || TD->getTypeSizeInBits(VTy->getScalarType()) == BitWidth) && (!VTy->isIntOrIntVectorTy() || @@ -413,7 +413,7 @@ Value *InstCombiner::SimplifyDemandedUseBits(Value *V, APInt DemandedMask, } else // Don't touch a scalar-to-vector bitcast. return 0; - } else if (isa<VectorType>(I->getOperand(0)->getType())) + } else if (I->getOperand(0)->getType()->isVectorTy()) // Don't touch a vector-to-scalar bitcast. return 0; diff --git a/lib/Transforms/InstCombine/InstCombineVectorOps.cpp b/lib/Transforms/InstCombine/InstCombineVectorOps.cpp index 20fda1a..a58124d 100644 --- a/lib/Transforms/InstCombine/InstCombineVectorOps.cpp +++ b/lib/Transforms/InstCombine/InstCombineVectorOps.cpp @@ -78,7 +78,7 @@ static std::vector<unsigned> getShuffleMask(const ShuffleVectorInst *SVI) { /// value is already around as a register, for example if it were inserted then /// extracted from the vector. static Value *FindScalarElement(Value *V, unsigned EltNo) { - assert(isa<VectorType>(V->getType()) && "Not looking at a vector?"); + assert(V->getType()->isVectorTy() && "Not looking at a vector?"); const VectorType *PTy = cast<VectorType>(V->getType()); unsigned Width = PTy->getNumElements(); if (EltNo >= Width) // Out of range access. @@ -322,7 +322,7 @@ static bool CollectSingleShuffleElements(Value *V, Value *LHS, Value *RHS, /// that computes V and the LHS value of the shuffle. static Value *CollectShuffleElements(Value *V, std::vector<Constant*> &Mask, Value *&RHS) { - assert(isa<VectorType>(V->getType()) && + assert(V->getType()->isVectorTy() && (RHS == 0 || V->getType() == RHS->getType()) && "Invalid shuffle!"); unsigned NumElts = cast<VectorType>(V->getType())->getNumElements(); diff --git a/lib/Transforms/InstCombine/InstructionCombining.cpp b/lib/Transforms/InstCombine/InstructionCombining.cpp index 96c0342..af9ec5c 100644 --- a/lib/Transforms/InstCombine/InstructionCombining.cpp +++ b/lib/Transforms/InstCombine/InstructionCombining.cpp @@ -73,7 +73,7 @@ void InstCombiner::getAnalysisUsage(AnalysisUsage &AU) const { /// from 'From' to 'To'. We don't want to convert from a legal to an illegal /// type for example, or from a smaller to a larger illegal type. bool InstCombiner::ShouldChangeType(const Type *From, const Type *To) const { - assert(isa<IntegerType>(From) && isa<IntegerType>(To)); + assert(From->isIntegerTy() && To->isIntegerTy()); // If we don't have TD, we don't know if the source/dest are legal. if (!TD) return false; @@ -478,7 +478,7 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) { bool EndsWithSequential = false; for (gep_type_iterator I = gep_type_begin(*Src), E = gep_type_end(*Src); I != E; ++I) - EndsWithSequential = !isa<StructType>(*I); + EndsWithSequential = !(*I)->isStructTy(); // Can we combine the two pointer arithmetics offsets? if (EndsWithSequential) { @@ -578,7 +578,7 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) { // into: %t1 = getelementptr [2 x i32]* %str, i32 0, i32 %V; bitcast const Type *SrcElTy = StrippedPtrTy->getElementType(); const Type *ResElTy=cast<PointerType>(PtrOp->getType())->getElementType(); - if (TD && isa<ArrayType>(SrcElTy) && + if (TD && SrcElTy->isArrayTy() && TD->getTypeAllocSize(cast<ArrayType>(SrcElTy)->getElementType()) == TD->getTypeAllocSize(ResElTy)) { Value *Idx[2]; @@ -596,7 +596,7 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) { // (where tmp = 8*tmp2) into: // getelementptr [100 x double]* %arr, i32 0, i32 %tmp2; bitcast - if (TD && isa<ArrayType>(SrcElTy) && ResElTy->isIntegerTy(8)) { + if (TD && SrcElTy->isArrayTy() && ResElTy->isIntegerTy(8)) { uint64_t ArrayEltSize = TD->getTypeAllocSize(cast<ArrayType>(SrcElTy)->getElementType()); diff --git a/lib/Transforms/Scalar/ABCD.cpp b/lib/Transforms/Scalar/ABCD.cpp index cf5e8c0..ea8e5c3 100644 --- a/lib/Transforms/Scalar/ABCD.cpp +++ b/lib/Transforms/Scalar/ABCD.cpp @@ -505,7 +505,7 @@ void ABCD::executeABCD(Function &F) { continue; ICmpInst *ICI = dyn_cast<ICmpInst>(TI->getOperand(0)); - if (!ICI || !isa<IntegerType>(ICI->getOperand(0)->getType())) + if (!ICI || !ICI->getOperand(0)->getType()->isIntegerTy()) continue; createConstraintCmpInst(ICI, TI); @@ -713,7 +713,7 @@ void ABCD::createConstraintCmpInst(ICmpInst *ICI, TerminatorInst *TI) { Value *V_op1 = ICI->getOperand(0); Value *V_op2 = ICI->getOperand(1); - if (!isa<IntegerType>(V_op1->getType())) + if (!V_op1->getType()->isIntegerTy()) return; Instruction *I_op1 = dyn_cast<Instruction>(V_op1); diff --git a/lib/Transforms/Scalar/CodeGenPrepare.cpp b/lib/Transforms/Scalar/CodeGenPrepare.cpp index 21e6f89..7ceda1f 100644 --- a/lib/Transforms/Scalar/CodeGenPrepare.cpp +++ b/lib/Transforms/Scalar/CodeGenPrepare.cpp @@ -612,7 +612,7 @@ bool CodeGenPrepare::OptimizeMemoryInst(Instruction *MemoryInst, Value *Addr, // we'd end up sinking both muls. if (AddrMode.BaseReg) { Value *V = AddrMode.BaseReg; - if (isa<PointerType>(V->getType())) + if (V->getType()->isPointerTy()) V = new PtrToIntInst(V, IntPtrTy, "sunkaddr", InsertPt); if (V->getType() != IntPtrTy) V = CastInst::CreateIntegerCast(V, IntPtrTy, /*isSigned=*/true, @@ -625,7 +625,7 @@ bool CodeGenPrepare::OptimizeMemoryInst(Instruction *MemoryInst, Value *Addr, Value *V = AddrMode.ScaledReg; if (V->getType() == IntPtrTy) { // done. - } else if (isa<PointerType>(V->getType())) { + } else if (V->getType()->isPointerTy()) { V = new PtrToIntInst(V, IntPtrTy, "sunkaddr", InsertPt); } else if (cast<IntegerType>(IntPtrTy)->getBitWidth() < cast<IntegerType>(V->getType())->getBitWidth()) { diff --git a/lib/Transforms/Scalar/GVN.cpp b/lib/Transforms/Scalar/GVN.cpp index 3ce7482..fcb802a 100644 --- a/lib/Transforms/Scalar/GVN.cpp +++ b/lib/Transforms/Scalar/GVN.cpp @@ -662,11 +662,10 @@ namespace { bool runOnFunction(Function &F); public: static char ID; // Pass identification, replacement for typeid - explicit GVN(bool nopre = false, bool noloads = false) - : FunctionPass(&ID), NoPRE(nopre), NoLoads(noloads), MD(0) { } + explicit GVN(bool noloads = false) + : FunctionPass(&ID), NoLoads(noloads), MD(0) { } private: - bool NoPRE; bool NoLoads; MemoryDependenceAnalysis *MD; DominatorTree *DT; @@ -674,6 +673,9 @@ namespace { ValueTable VN; DenseMap<BasicBlock*, ValueNumberScope*> localAvail; + // List of critical edges to be split between iterations. + SmallVector<std::pair<TerminatorInst*, unsigned>, 4> toSplit; + // This transformation requires dominator postdominator info virtual void getAnalysisUsage(AnalysisUsage &AU) const { AU.addRequired<DominatorTree>(); @@ -701,14 +703,15 @@ namespace { Value *lookupNumber(BasicBlock *BB, uint32_t num); void cleanupGlobalSets(); void verifyRemoved(const Instruction *I) const; + bool splitCriticalEdges(); }; char GVN::ID = 0; } // createGVNPass - The public interface to this file... -FunctionPass *llvm::createGVNPass(bool NoPRE, bool NoLoads) { - return new GVN(NoPRE, NoLoads); +FunctionPass *llvm::createGVNPass(bool NoLoads) { + return new GVN(NoLoads); } static RegisterPass<GVN> X("gvn", @@ -836,9 +839,9 @@ static bool CanCoerceMustAliasedValueToLoad(Value *StoredVal, const TargetData &TD) { // If the loaded or stored value is an first class array or struct, don't try // to transform them. We need to be able to bitcast to integer. - if (isa<StructType>(LoadTy) || isa<ArrayType>(LoadTy) || - isa<StructType>(StoredVal->getType()) || - isa<ArrayType>(StoredVal->getType())) + if (LoadTy->isStructTy() || LoadTy->isArrayTy() || + StoredVal->getType()->isStructTy() || + StoredVal->getType()->isArrayTy()) return false; // The store has to be at least as big as the load. @@ -870,26 +873,26 @@ static Value *CoerceAvailableValueToLoadType(Value *StoredVal, // If the store and reload are the same size, we can always reuse it. if (StoreSize == LoadSize) { - if (isa<PointerType>(StoredValTy) && isa<PointerType>(LoadedTy)) { + if (StoredValTy->isPointerTy() && LoadedTy->isPointerTy()) { // Pointer to Pointer -> use bitcast. return new BitCastInst(StoredVal, LoadedTy, "", InsertPt); } // Convert source pointers to integers, which can be bitcast. - if (isa<PointerType>(StoredValTy)) { + if (StoredValTy->isPointerTy()) { StoredValTy = TD.getIntPtrType(StoredValTy->getContext()); StoredVal = new PtrToIntInst(StoredVal, StoredValTy, "", InsertPt); } const Type *TypeToCastTo = LoadedTy; - if (isa<PointerType>(TypeToCastTo)) + if (TypeToCastTo->isPointerTy()) TypeToCastTo = TD.getIntPtrType(StoredValTy->getContext()); if (StoredValTy != TypeToCastTo) StoredVal = new BitCastInst(StoredVal, TypeToCastTo, "", InsertPt); // Cast to pointer if the load needs a pointer type. - if (isa<PointerType>(LoadedTy)) + if (LoadedTy->isPointerTy()) StoredVal = new IntToPtrInst(StoredVal, LoadedTy, "", InsertPt); return StoredVal; @@ -901,13 +904,13 @@ static Value *CoerceAvailableValueToLoadType(Value *StoredVal, assert(StoreSize >= LoadSize && "CanCoerceMustAliasedValueToLoad fail"); // Convert source pointers to integers, which can be manipulated. - if (isa<PointerType>(StoredValTy)) { + if (StoredValTy->isPointerTy()) { StoredValTy = TD.getIntPtrType(StoredValTy->getContext()); StoredVal = new PtrToIntInst(StoredVal, StoredValTy, "", InsertPt); } // Convert vectors and fp to integer, which can be manipulated. - if (!isa<IntegerType>(StoredValTy)) { + if (!StoredValTy->isIntegerTy()) { StoredValTy = IntegerType::get(StoredValTy->getContext(), StoreSize); StoredVal = new BitCastInst(StoredVal, StoredValTy, "", InsertPt); } @@ -927,7 +930,7 @@ static Value *CoerceAvailableValueToLoadType(Value *StoredVal, return StoredVal; // If the result is a pointer, inttoptr. - if (isa<PointerType>(LoadedTy)) + if (LoadedTy->isPointerTy()) return new IntToPtrInst(StoredVal, LoadedTy, "inttoptr", InsertPt); // Otherwise, bitcast. @@ -989,7 +992,7 @@ static int AnalyzeLoadFromClobberingWrite(const Type *LoadTy, Value *LoadPtr, const TargetData &TD) { // If the loaded or stored value is an first class array or struct, don't try // to transform them. We need to be able to bitcast to integer. - if (isa<StructType>(LoadTy) || isa<ArrayType>(LoadTy)) + if (LoadTy->isStructTy() || LoadTy->isArrayTy()) return -1; int64_t StoreOffset = 0, LoadOffset = 0; @@ -1064,8 +1067,8 @@ static int AnalyzeLoadFromClobberingStore(const Type *LoadTy, Value *LoadPtr, StoreInst *DepSI, const TargetData &TD) { // Cannot handle reading from store of first-class aggregate yet. - if (isa<StructType>(DepSI->getOperand(0)->getType()) || - isa<ArrayType>(DepSI->getOperand(0)->getType())) + if (DepSI->getOperand(0)->getType()->isStructTy() || + DepSI->getOperand(0)->getType()->isArrayTy()) return -1; Value *StorePtr = DepSI->getPointerOperand(); @@ -1136,9 +1139,9 @@ static Value *GetStoreValueForLoad(Value *SrcVal, unsigned Offset, // Compute which bits of the stored value are being used by the load. Convert // to an integer type to start with. - if (isa<PointerType>(SrcVal->getType())) + if (SrcVal->getType()->isPointerTy()) SrcVal = Builder.CreatePtrToInt(SrcVal, TD.getIntPtrType(Ctx), "tmp"); - if (!isa<IntegerType>(SrcVal->getType())) + if (!SrcVal->getType()->isIntegerTy()) SrcVal = Builder.CreateBitCast(SrcVal, IntegerType::get(Ctx, StoreSize*8), "tmp"); @@ -1323,7 +1326,7 @@ static Value *ConstructSSAForLoadSet(LoadInst *LI, Value *V = SSAUpdate.GetValueInMiddleOfBlock(LI->getParent()); // If new PHI nodes were created, notify alias analysis. - if (isa<PointerType>(V->getType())) + if (V->getType()->isPointerTy()) for (unsigned i = 0, e = NewPHIs.size(); i != e; ++i) AA->copyValue(LI, NewPHIs[i]); @@ -1491,8 +1494,9 @@ bool GVN::processNonLocalLoad(LoadInst *LI, if (isa<PHINode>(V)) V->takeName(LI); - if (isa<PointerType>(V->getType())) + if (V->getType()->isPointerTy()) MD->invalidateCachedPointerInfo(V); + VN.erase(LI); toErase.push_back(LI); NumGVNLoad++; return true; @@ -1538,11 +1542,13 @@ bool GVN::processNonLocalLoad(LoadInst *LI, // at least one of the values is LI. Since this means that we won't be able // to eliminate LI even if we insert uses in the other predecessors, we will // end up increasing code size. Reject this by scanning for LI. - if (!EnableFullLoadPRE) { - for (unsigned i = 0, e = ValuesPerBlock.size(); i != e; ++i) - if (ValuesPerBlock[i].isSimpleValue() && - ValuesPerBlock[i].getSimpleValue() == LI) + for (unsigned i = 0, e = ValuesPerBlock.size(); i != e; ++i) { + if (ValuesPerBlock[i].isSimpleValue() && + ValuesPerBlock[i].getSimpleValue() == LI) { + // Skip cases where LI is the only definition, even for EnableFullLoadPRE. + if (!EnableFullLoadPRE || e == 1) return false; + } } // FIXME: It is extremely unclear what this loop is doing, other than @@ -1576,6 +1582,7 @@ bool GVN::processNonLocalLoad(LoadInst *LI, for (unsigned i = 0, e = UnavailableBlocks.size(); i != e; ++i) FullyAvailableBlocks[UnavailableBlocks[i]] = false; + bool NeedToSplitEdges = false; for (pred_iterator PI = pred_begin(LoadBB), E = pred_end(LoadBB); PI != E; ++PI) { BasicBlock *Pred = *PI; @@ -1583,13 +1590,20 @@ bool GVN::processNonLocalLoad(LoadInst *LI, continue; } PredLoads[Pred] = 0; - // We don't currently handle critical edges :( + if (Pred->getTerminator()->getNumSuccessors() != 1) { - DEBUG(dbgs() << "COULD NOT PRE LOAD BECAUSE OF CRITICAL EDGE '" - << Pred->getName() << "': " << *LI << '\n'); - return false; + if (isa<IndirectBrInst>(Pred->getTerminator())) { + DEBUG(dbgs() << "COULD NOT PRE LOAD BECAUSE OF INDBR CRITICAL EDGE '" + << Pred->getName() << "': " << *LI << '\n'); + return false; + } + unsigned SuccNum = GetSuccessorNumber(Pred, LoadBB); + toSplit.push_back(std::make_pair(Pred->getTerminator(), SuccNum)); + NeedToSplitEdges = true; } } + if (NeedToSplitEdges) + return false; // Decide whether PRE is profitable for this load. unsigned NumUnavailablePreds = PredLoads.size(); @@ -1623,13 +1637,8 @@ bool GVN::processNonLocalLoad(LoadInst *LI, LoadPtr = Address.PHITranslateWithInsertion(LoadBB, UnavailablePred, *DT, NewInsts); } else { - Address.PHITranslateValue(LoadBB, UnavailablePred); + Address.PHITranslateValue(LoadBB, UnavailablePred, DT); LoadPtr = Address.getAddr(); - - // Make sure the value is live in the predecessor. - if (Instruction *Inst = dyn_cast_or_null<Instruction>(LoadPtr)) - if (!DT->dominates(Inst->getParent(), UnavailablePred)) - LoadPtr = 0; } // If we couldn't find or insert a computation of this phi translated value, @@ -1697,6 +1706,8 @@ bool GVN::processNonLocalLoad(LoadInst *LI, // Add the newly created load. ValuesPerBlock.push_back(AvailableValueInBlock::get(UnavailablePred, NewLoad)); + MD->invalidateCachedPointerInfo(LoadPtr); + DEBUG(dbgs() << "GVN INSERTED " << *NewLoad << '\n'); } // Perform PHI construction. @@ -1705,8 +1716,9 @@ bool GVN::processNonLocalLoad(LoadInst *LI, LI->replaceAllUsesWith(V); if (isa<PHINode>(V)) V->takeName(LI); - if (isa<PointerType>(V->getType())) + if (V->getType()->isPointerTy()) MD->invalidateCachedPointerInfo(V); + VN.erase(LI); toErase.push_back(LI); NumPRELoad++; return true; @@ -1765,8 +1777,9 @@ bool GVN::processLoad(LoadInst *L, SmallVectorImpl<Instruction*> &toErase) { // Replace the load! L->replaceAllUsesWith(AvailVal); - if (isa<PointerType>(AvailVal->getType())) + if (AvailVal->getType()->isPointerTy()) MD->invalidateCachedPointerInfo(AvailVal); + VN.erase(L); toErase.push_back(L); NumGVNLoad++; return true; @@ -1810,8 +1823,9 @@ bool GVN::processLoad(LoadInst *L, SmallVectorImpl<Instruction*> &toErase) { // Remove it! L->replaceAllUsesWith(StoredVal); - if (isa<PointerType>(StoredVal->getType())) + if (StoredVal->getType()->isPointerTy()) MD->invalidateCachedPointerInfo(StoredVal); + VN.erase(L); toErase.push_back(L); NumGVNLoad++; return true; @@ -1839,8 +1853,9 @@ bool GVN::processLoad(LoadInst *L, SmallVectorImpl<Instruction*> &toErase) { // Remove it! L->replaceAllUsesWith(AvailableVal); - if (isa<PointerType>(DepLI->getType())) + if (DepLI->getType()->isPointerTy()) MD->invalidateCachedPointerInfo(DepLI); + VN.erase(L); toErase.push_back(L); NumGVNLoad++; return true; @@ -1851,6 +1866,7 @@ bool GVN::processLoad(LoadInst *L, SmallVectorImpl<Instruction*> &toErase) { // intervening stores, for example. if (isa<AllocaInst>(DepInst) || isMalloc(DepInst)) { L->replaceAllUsesWith(UndefValue::get(L->getType())); + VN.erase(L); toErase.push_back(L); NumGVNLoad++; return true; @@ -1861,6 +1877,7 @@ bool GVN::processLoad(LoadInst *L, SmallVectorImpl<Instruction*> &toErase) { if (IntrinsicInst* II = dyn_cast<IntrinsicInst>(DepInst)) { if (II->getIntrinsicID() == Intrinsic::lifetime_start) { L->replaceAllUsesWith(UndefValue::get(L->getType())); + VN.erase(L); toErase.push_back(L); NumGVNLoad++; return true; @@ -1943,7 +1960,7 @@ bool GVN::processInstruction(Instruction *I, if (constVal) { p->replaceAllUsesWith(constVal); - if (MD && isa<PointerType>(constVal->getType())) + if (MD && constVal->getType()->isPointerTy()) MD->invalidateCachedPointerInfo(constVal); VN.erase(p); @@ -1964,7 +1981,7 @@ bool GVN::processInstruction(Instruction *I, // Remove it! VN.erase(I); I->replaceAllUsesWith(repl); - if (MD && isa<PointerType>(repl->getType())) + if (MD && repl->getType()->isPointerTy()) MD->invalidateCachedPointerInfo(repl); toErase.push_back(I); return true; @@ -2004,6 +2021,8 @@ bool GVN::runOnFunction(Function& F) { while (ShouldContinue) { DEBUG(dbgs() << "GVN iteration: " << Iteration << "\n"); ShouldContinue = iterateOnFunction(F); + if (splitCriticalEdges()) + ShouldContinue = true; Changed |= ShouldContinue; ++Iteration; } @@ -2070,7 +2089,6 @@ bool GVN::processBlock(BasicBlock *BB) { /// control flow patterns and attempts to perform simple PRE at the join point. bool GVN::performPRE(Function &F) { bool Changed = false; - SmallVector<std::pair<TerminatorInst*, unsigned>, 4> toSplit; DenseMap<BasicBlock*, Value*> predMap; for (df_iterator<BasicBlock*> DI = df_begin(&F.getEntryBlock()), DE = df_end(&F.getEntryBlock()); DI != DE; ++DI) { @@ -2141,14 +2159,7 @@ bool GVN::performPRE(Function &F) { // We can't do PRE safely on a critical edge, so instead we schedule // the edge to be split and perform the PRE the next time we iterate // on the function. - unsigned SuccNum = 0; - for (unsigned i = 0, e = PREPred->getTerminator()->getNumSuccessors(); - i != e; ++i) - if (PREPred->getTerminator()->getSuccessor(i) == CurrentBlock) { - SuccNum = i; - break; - } - + unsigned SuccNum = GetSuccessorNumber(PREPred, CurrentBlock); if (isCriticalEdge(PREPred->getTerminator(), SuccNum)) { toSplit.push_back(std::make_pair(PREPred->getTerminator(), SuccNum)); continue; @@ -2204,7 +2215,7 @@ bool GVN::performPRE(Function &F) { localAvail[CurrentBlock]->table[ValNo] = Phi; CurInst->replaceAllUsesWith(Phi); - if (MD && isa<PointerType>(Phi->getType())) + if (MD && Phi->getType()->isPointerTy()) MD->invalidateCachedPointerInfo(Phi); VN.erase(CurInst); @@ -2216,11 +2227,23 @@ bool GVN::performPRE(Function &F) { } } - for (SmallVector<std::pair<TerminatorInst*, unsigned>, 4>::iterator - I = toSplit.begin(), E = toSplit.end(); I != E; ++I) - SplitCriticalEdge(I->first, I->second, this); + if (splitCriticalEdges()) + Changed = true; - return Changed || toSplit.size(); + return Changed; +} + +/// splitCriticalEdges - Split critical edges found during the previous +/// iteration that may enable further optimization. +bool GVN::splitCriticalEdges() { + if (toSplit.empty()) + return false; + do { + std::pair<TerminatorInst*, unsigned> Edge = toSplit.pop_back_val(); + SplitCriticalEdge(Edge.first, Edge.second, this); + } while (!toSplit.empty()); + if (MD) MD->invalidateCachedPredecessors(); + return true; } /// iterateOnFunction - Executes one iteration of GVN diff --git a/lib/Transforms/Scalar/IndVarSimplify.cpp b/lib/Transforms/Scalar/IndVarSimplify.cpp index 5302fdc..cb563c3 100644 --- a/lib/Transforms/Scalar/IndVarSimplify.cpp +++ b/lib/Transforms/Scalar/IndVarSimplify.cpp @@ -103,11 +103,9 @@ namespace { BasicBlock *ExitingBlock, BranchInst *BI, SCEVExpander &Rewriter); - void RewriteLoopExitValues(Loop *L, const SCEV *BackedgeTakenCount, - SCEVExpander &Rewriter); + void RewriteLoopExitValues(Loop *L, SCEVExpander &Rewriter); - void RewriteIVExpressions(Loop *L, const Type *LargestType, - SCEVExpander &Rewriter); + void RewriteIVExpressions(Loop *L, SCEVExpander &Rewriter); void SinkUnusedInvariants(Loop *L); @@ -190,7 +188,7 @@ ICmpInst *IndVarSimplify::LinearFunctionTestReplace(Loop *L, ICmpInst *Cond = new ICmpInst(BI, Opcode, CmpIndVar, ExitCnt, "exitcond"); - Instruction *OrigCond = cast<Instruction>(BI->getCondition()); + Value *OrigCond = BI->getCondition(); // It's tempting to use replaceAllUsesWith here to fully replace the old // comparison, but that's not immediately safe, since users of the old // comparison may not be dominated by the new comparison. Instead, just @@ -215,7 +213,6 @@ ICmpInst *IndVarSimplify::LinearFunctionTestReplace(Loop *L, /// able to brute-force evaluate arbitrary instructions as long as they have /// constant operands at the beginning of the loop. void IndVarSimplify::RewriteLoopExitValues(Loop *L, - const SCEV *BackedgeTakenCount, SCEVExpander &Rewriter) { // Verify the input to the pass in already in LCSSA form. assert(L->isLCSSAForm()); @@ -241,15 +238,24 @@ void IndVarSimplify::RewriteLoopExitValues(Loop *L, while ((PN = dyn_cast<PHINode>(BBI++))) { if (PN->use_empty()) continue; // dead use, don't replace it + + // SCEV only supports integer expressions for now. + if (!PN->getType()->isIntegerTy() && !PN->getType()->isPointerTy()) + continue; + + // It's necessary to tell ScalarEvolution about this explicitly so that + // it can walk the def-use list and forget all SCEVs, as it may not be + // watching the PHI itself. Once the new exit value is in place, there + // may not be a def-use connection between the loop and every instruction + // which got a SCEVAddRecExpr for that loop. + SE->forgetValue(PN); + // Iterate over all of the values in all the PHI nodes. for (unsigned i = 0; i != NumPreds; ++i) { // If the value being merged in is not integer or is not defined // in the loop, skip it. Value *InVal = PN->getIncomingValue(i); - if (!isa<Instruction>(InVal) || - // SCEV only supports integer expressions for now. - (!isa<IntegerType>(InVal->getType()) && - !isa<PointerType>(InVal->getType()))) + if (!isa<Instruction>(InVal)) continue; // If this pred is for a subloop, not L itself, skip it. @@ -349,7 +355,7 @@ bool IndVarSimplify::runOnLoop(Loop *L, LPPassManager &LPM) { // the current expressions. // if (!isa<SCEVCouldNotCompute>(BackedgeTakenCount)) - RewriteLoopExitValues(L, BackedgeTakenCount, Rewriter); + RewriteLoopExitValues(L, Rewriter); // Compute the type of the largest recurrence expression, and decide whether // a canonical induction variable should be inserted. @@ -378,17 +384,18 @@ bool IndVarSimplify::runOnLoop(Loop *L, LPPassManager &LPM) { // in this loop, insert a canonical induction variable of the largest size. Value *IndVar = 0; if (NeedCannIV) { - // Check to see if the loop already has a canonical-looking induction - // variable. If one is present and it's wider than the planned canonical - // induction variable, temporarily remove it, so that the Rewriter - // doesn't attempt to reuse it. - PHINode *OldCannIV = L->getCanonicalInductionVariable(); - if (OldCannIV) { + // Check to see if the loop already has any canonical-looking induction + // variables. If any are present and wider than the planned canonical + // induction variable, temporarily remove them, so that the Rewriter + // doesn't attempt to reuse them. + SmallVector<PHINode *, 2> OldCannIVs; + while (PHINode *OldCannIV = L->getCanonicalInductionVariable()) { if (SE->getTypeSizeInBits(OldCannIV->getType()) > SE->getTypeSizeInBits(LargestType)) OldCannIV->removeFromParent(); else - OldCannIV = 0; + break; + OldCannIVs.push_back(OldCannIV); } IndVar = Rewriter.getOrInsertCanonicalInductionVariable(L, LargestType); @@ -398,17 +405,21 @@ bool IndVarSimplify::runOnLoop(Loop *L, LPPassManager &LPM) { DEBUG(dbgs() << "INDVARS: New CanIV: " << *IndVar << '\n'); // Now that the official induction variable is established, reinsert - // the old canonical-looking variable after it so that the IR remains - // consistent. It will be deleted as part of the dead-PHI deletion at + // any old canonical-looking variables after it so that the IR remains + // consistent. They will be deleted as part of the dead-PHI deletion at // the end of the pass. - if (OldCannIV) - OldCannIV->insertAfter(cast<Instruction>(IndVar)); + while (!OldCannIVs.empty()) { + PHINode *OldCannIV = OldCannIVs.pop_back_val(); + OldCannIV->insertBefore(L->getHeader()->getFirstNonPHI()); + } } // If we have a trip count expression, rewrite the loop's exit condition // using it. We can currently only handle loops with a single exit. ICmpInst *NewICmp = 0; - if (!isa<SCEVCouldNotCompute>(BackedgeTakenCount) && ExitingBlock) { + if (!isa<SCEVCouldNotCompute>(BackedgeTakenCount) && + !BackedgeTakenCount->isZero() && + ExitingBlock) { assert(NeedCannIV && "LinearFunctionTestReplace requires a canonical induction variable"); // Can't rewrite non-branch yet. @@ -418,7 +429,7 @@ bool IndVarSimplify::runOnLoop(Loop *L, LPPassManager &LPM) { } // Rewrite IV-derived expressions. Clears the rewriter cache. - RewriteIVExpressions(L, LargestType, Rewriter); + RewriteIVExpressions(L, Rewriter); // The Rewriter may not be used from this point on. @@ -438,8 +449,7 @@ bool IndVarSimplify::runOnLoop(Loop *L, LPPassManager &LPM) { return Changed; } -void IndVarSimplify::RewriteIVExpressions(Loop *L, const Type *LargestType, - SCEVExpander &Rewriter) { +void IndVarSimplify::RewriteIVExpressions(Loop *L, SCEVExpander &Rewriter) { SmallVector<WeakVH, 16> DeadInsts; // Rewrite all induction variable expressions in terms of the canonical @@ -584,8 +594,8 @@ void IndVarSimplify::SinkUnusedInvariants(Loop *L) { } } -/// Return true if it is OK to use SIToFPInst for an inducation variable -/// with given inital and exit values. +/// Return true if it is OK to use SIToFPInst for an induction variable +/// with given initial and exit values. static bool useSIToFPInst(ConstantFP &InitV, ConstantFP &ExitV, uint64_t intIV, uint64_t intEV) { @@ -638,7 +648,7 @@ void IndVarSimplify::HandleFloatingPointIV(Loop *L, PHINode *PH) { if (!convertToInt(InitValue->getValueAPF(), &newInitValue)) return; - // Check IV increment. Reject this PH if increement operation is not + // Check IV increment. Reject this PH if increment operation is not // an add or increment value can not be represented by an integer. BinaryOperator *Incr = dyn_cast<BinaryOperator>(PH->getIncomingValue(BackEdge)); @@ -674,7 +684,7 @@ void IndVarSimplify::HandleFloatingPointIV(Loop *L, PHINode *PH) { if (BI->getCondition() != EC) return; } - // Find exit value. If exit value can not be represented as an interger then + // Find exit value. If exit value can not be represented as an integer then // do not handle this floating point PH. ConstantFP *EV = NULL; unsigned EVIndex = 1; @@ -736,11 +746,11 @@ void IndVarSimplify::HandleFloatingPointIV(Loop *L, PHINode *PH) { ICmpInst *NewEC = new ICmpInst(EC->getParent()->getTerminator(), NewPred, LHS, RHS, EC->getName()); - // In the following deltions, PH may become dead and may be deleted. + // In the following deletions, PH may become dead and may be deleted. // Use a WeakVH to observe whether this happens. WeakVH WeakPH = PH; - // Delete old, floating point, exit comparision instruction. + // Delete old, floating point, exit comparison instruction. NewEC->takeName(EC); EC->replaceAllUsesWith(NewEC); RecursivelyDeleteTriviallyDeadInstructions(EC); diff --git a/lib/Transforms/Scalar/JumpThreading.cpp b/lib/Transforms/Scalar/JumpThreading.cpp index 8f21aac..a6489ec 100644 --- a/lib/Transforms/Scalar/JumpThreading.cpp +++ b/lib/Transforms/Scalar/JumpThreading.cpp @@ -201,7 +201,7 @@ static unsigned getJumpThreadDuplicationCost(const BasicBlock *BB) { if (isa<DbgInfoIntrinsic>(I)) continue; // If this is a pointer->pointer bitcast, it is free. - if (isa<BitCastInst>(I) && isa<PointerType>(I->getType())) + if (isa<BitCastInst>(I) && I->getType()->isPointerTy()) continue; // All other instructions count for at least one unit. @@ -214,7 +214,7 @@ static unsigned getJumpThreadDuplicationCost(const BasicBlock *BB) { if (const CallInst *CI = dyn_cast<CallInst>(I)) { if (!isa<IntrinsicInst>(CI)) Size += 3; - else if (!isa<VectorType>(CI->getType())) + else if (!CI->getType()->isVectorTy()) Size += 1; } } diff --git a/lib/Transforms/Scalar/LICM.cpp b/lib/Transforms/Scalar/LICM.cpp index 81f9ae6..d7ace342 100644 --- a/lib/Transforms/Scalar/LICM.cpp +++ b/lib/Transforms/Scalar/LICM.cpp @@ -678,7 +678,7 @@ void LICM::PromoteValuesInLoop() { // If we are promoting a pointer value, update alias information for the // inserted load. Value *LoadValue = 0; - if (isa<PointerType>(cast<PointerType>(Ptr->getType())->getElementType())) { + if (cast<PointerType>(Ptr->getType())->getElementType()->isPointerTy()) { // Locate a load or store through the pointer, and assign the same value // to LI as we are loading or storing. Since we know that the value is // stored in this loop, this will always succeed. @@ -751,7 +751,7 @@ void LICM::PromoteValuesInLoop() { LoadInst *LI = new LoadInst(PromotedValues[i].first, "", InsertPos); // If this is a pointer type, update alias info appropriately. - if (isa<PointerType>(LI->getType())) + if (LI->getType()->isPointerTy()) CurAST->copyValue(PointerValueNumbers[PVN++], LI); // Store into the memory we promoted. diff --git a/lib/Transforms/Scalar/LoopStrengthReduce.cpp b/lib/Transforms/Scalar/LoopStrengthReduce.cpp index 240b298..f920dca 100644 --- a/lib/Transforms/Scalar/LoopStrengthReduce.cpp +++ b/lib/Transforms/Scalar/LoopStrengthReduce.cpp @@ -198,7 +198,7 @@ struct Formula { } -/// DoInitialMatch - Recurrsion helper for InitialMatch. +/// DoInitialMatch - Recursion helper for InitialMatch. static void DoInitialMatch(const SCEV *S, Loop *L, SmallVectorImpl<const SCEV *> &Good, SmallVectorImpl<const SCEV *> &Bad, @@ -337,14 +337,42 @@ void Formula::dump() const { print(errs()); errs() << '\n'; } -/// getSDiv - Return an expression for LHS /s RHS, if it can be determined, -/// or null otherwise. If IgnoreSignificantBits is true, expressions like -/// (X * Y) /s Y are simplified to Y, ignoring that the multiplication may -/// overflow, which is useful when the result will be used in a context where -/// the most significant bits are ignored. -static const SCEV *getSDiv(const SCEV *LHS, const SCEV *RHS, - ScalarEvolution &SE, - bool IgnoreSignificantBits = false) { +/// isAddRecSExtable - Return true if the given addrec can be sign-extended +/// without changing its value. +static bool isAddRecSExtable(const SCEVAddRecExpr *AR, ScalarEvolution &SE) { + const Type *WideTy = + IntegerType::get(SE.getContext(), + SE.getTypeSizeInBits(AR->getType()) + 1); + return isa<SCEVAddRecExpr>(SE.getSignExtendExpr(AR, WideTy)); +} + +/// isAddSExtable - Return true if the given add can be sign-extended +/// without changing its value. +static bool isAddSExtable(const SCEVAddExpr *A, ScalarEvolution &SE) { + const Type *WideTy = + IntegerType::get(SE.getContext(), + SE.getTypeSizeInBits(A->getType()) + 1); + return isa<SCEVAddExpr>(SE.getSignExtendExpr(A, WideTy)); +} + +/// isMulSExtable - Return true if the given add can be sign-extended +/// without changing its value. +static bool isMulSExtable(const SCEVMulExpr *A, ScalarEvolution &SE) { + const Type *WideTy = + IntegerType::get(SE.getContext(), + SE.getTypeSizeInBits(A->getType()) + 1); + return isa<SCEVMulExpr>(SE.getSignExtendExpr(A, WideTy)); +} + +/// getExactSDiv - Return an expression for LHS /s RHS, if it can be determined +/// and if the remainder is known to be zero, or null otherwise. If +/// IgnoreSignificantBits is true, expressions like (X * Y) /s Y are simplified +/// to Y, ignoring that the multiplication may overflow, which is useful when +/// the result will be used in a context where the most significant bits are +/// ignored. +static const SCEV *getExactSDiv(const SCEV *LHS, const SCEV *RHS, + ScalarEvolution &SE, + bool IgnoreSignificantBits = false) { // Handle the trivial case, which works for any SCEV type. if (LHS == RHS) return SE.getIntegerSCEV(1, LHS->getType()); @@ -365,39 +393,44 @@ static const SCEV *getSDiv(const SCEV *LHS, const SCEV *RHS, .sdiv(RC->getValue()->getValue())); } - // Distribute the sdiv over addrec operands. + // Distribute the sdiv over addrec operands, if the addrec doesn't overflow. if (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(LHS)) { - const SCEV *Start = getSDiv(AR->getStart(), RHS, SE, - IgnoreSignificantBits); - if (!Start) return 0; - const SCEV *Step = getSDiv(AR->getStepRecurrence(SE), RHS, SE, - IgnoreSignificantBits); - if (!Step) return 0; - return SE.getAddRecExpr(Start, Step, AR->getLoop()); + if (IgnoreSignificantBits || isAddRecSExtable(AR, SE)) { + const SCEV *Start = getExactSDiv(AR->getStart(), RHS, SE, + IgnoreSignificantBits); + if (!Start) return 0; + const SCEV *Step = getExactSDiv(AR->getStepRecurrence(SE), RHS, SE, + IgnoreSignificantBits); + if (!Step) return 0; + return SE.getAddRecExpr(Start, Step, AR->getLoop()); + } } - // Distribute the sdiv over add operands. + // Distribute the sdiv over add operands, if the add doesn't overflow. if (const SCEVAddExpr *Add = dyn_cast<SCEVAddExpr>(LHS)) { - SmallVector<const SCEV *, 8> Ops; - for (SCEVAddExpr::op_iterator I = Add->op_begin(), E = Add->op_end(); - I != E; ++I) { - const SCEV *Op = getSDiv(*I, RHS, SE, - IgnoreSignificantBits); - if (!Op) return 0; - Ops.push_back(Op); + if (IgnoreSignificantBits || isAddSExtable(Add, SE)) { + SmallVector<const SCEV *, 8> Ops; + for (SCEVAddExpr::op_iterator I = Add->op_begin(), E = Add->op_end(); + I != E; ++I) { + const SCEV *Op = getExactSDiv(*I, RHS, SE, + IgnoreSignificantBits); + if (!Op) return 0; + Ops.push_back(Op); + } + return SE.getAddExpr(Ops); } - return SE.getAddExpr(Ops); } // Check for a multiply operand that we can pull RHS out of. if (const SCEVMulExpr *Mul = dyn_cast<SCEVMulExpr>(LHS)) - if (IgnoreSignificantBits || Mul->hasNoSignedWrap()) { + if (IgnoreSignificantBits || isMulSExtable(Mul, SE)) { SmallVector<const SCEV *, 4> Ops; bool Found = false; for (SCEVMulExpr::op_iterator I = Mul->op_begin(), E = Mul->op_end(); I != E; ++I) { if (!Found) - if (const SCEV *Q = getSDiv(*I, RHS, SE, IgnoreSignificantBits)) { + if (const SCEV *Q = getExactSDiv(*I, RHS, SE, + IgnoreSignificantBits)) { Ops.push_back(Q); Found = true; continue; @@ -640,9 +673,9 @@ void Cost::RateRegister(const SCEV *Reg, /// RatePrimaryRegister - Record this register in the set. If we haven't seen it /// before, rate it. void Cost::RatePrimaryRegister(const SCEV *Reg, - SmallPtrSet<const SCEV *, 16> &Regs, - const Loop *L, - ScalarEvolution &SE, DominatorTree &DT) { + SmallPtrSet<const SCEV *, 16> &Regs, + const Loop *L, + ScalarEvolution &SE, DominatorTree &DT) { if (Regs.insert(Reg)) RateRegister(Reg, Regs, L, SE, DT); } @@ -879,7 +912,7 @@ public: MaxOffset(INT64_MIN), AllFixupsOutsideLoop(true) {} - bool InsertFormula(size_t LUIdx, const Formula &F); + bool InsertFormula(const Formula &F); void check() const; @@ -889,7 +922,7 @@ public: /// InsertFormula - If the given formula has not yet been inserted, add it to /// the list, and return true. Return false otherwise. -bool LSRUse::InsertFormula(size_t LUIdx, const Formula &F) { +bool LSRUse::InsertFormula(const Formula &F) { SmallVector<const SCEV *, 2> Key = F.BaseRegs; if (F.ScaledReg) Key.push_back(F.ScaledReg); // Unstable sort by host order ok, because this is only used for uniquifying. @@ -925,7 +958,7 @@ void LSRUse::print(raw_ostream &OS) const { case ICmpZero: OS << "ICmpZero"; break; case Address: OS << "Address of "; - if (isa<PointerType>(AccessTy)) + if (AccessTy->isPointerTy()) OS << "pointer"; // the full pointer type could be really verbose else OS << *AccessTy; @@ -1024,8 +1057,7 @@ static bool isAlwaysFoldable(int64_t BaseOffs, GlobalValue *BaseGV, bool HasBaseReg, LSRUse::KindType Kind, const Type *AccessTy, - const TargetLowering *TLI, - ScalarEvolution &SE) { + const TargetLowering *TLI) { // Fast-path: zero is always foldable. if (BaseOffs == 0 && !BaseGV) return true; @@ -1153,7 +1185,7 @@ class LSRInstance { const Type *AccessTy); public: - void InsertInitialFormula(const SCEV *S, Loop *L, LSRUse &LU, size_t LUIdx); + void InsertInitialFormula(const SCEV *S, LSRUse &LU, size_t LUIdx); void InsertSupplementalFormula(const SCEV *S, LSRUse &LU, size_t LUIdx); void CountRegisters(const Formula &F, size_t LUIdx); bool InsertFormula(LSRUse &LU, unsigned LUIdx, const Formula &F); @@ -1184,16 +1216,18 @@ public: Value *Expand(const LSRFixup &LF, const Formula &F, - BasicBlock::iterator IP, Loop *L, Instruction *IVIncInsertPos, + BasicBlock::iterator IP, SCEVExpander &Rewriter, - SmallVectorImpl<WeakVH> &DeadInsts, - ScalarEvolution &SE, DominatorTree &DT) const; + SmallVectorImpl<WeakVH> &DeadInsts) const; + void RewriteForPHI(PHINode *PN, const LSRFixup &LF, + const Formula &F, + SCEVExpander &Rewriter, + SmallVectorImpl<WeakVH> &DeadInsts, + Pass *P) const; void Rewrite(const LSRFixup &LF, const Formula &F, - Loop *L, Instruction *IVIncInsertPos, SCEVExpander &Rewriter, SmallVectorImpl<WeakVH> &DeadInsts, - ScalarEvolution &SE, DominatorTree &DT, Pass *P) const; void ImplementSolution(const SmallVectorImpl<const Formula *> &Solution, Pass *P); @@ -1212,7 +1246,7 @@ public: } /// OptimizeShadowIV - If IV is used in a int-to-float cast -/// inside the loop then try to eliminate the cast opeation. +/// inside the loop then try to eliminate the cast operation. void LSRInstance::OptimizeShadowIV() { const SCEV *BackedgeTakenCount = SE.getBackedgeTakenCount(L); if (isa<SCEVCouldNotCompute>(BackedgeTakenCount)) @@ -1522,7 +1556,7 @@ LSRInstance::OptimizeLoopTermCond() { A = SE.getSignExtendExpr(A, B->getType()); } if (const SCEVConstant *D = - dyn_cast_or_null<SCEVConstant>(getSDiv(B, A, SE))) { + dyn_cast_or_null<SCEVConstant>(getExactSDiv(B, A, SE))) { // Stride of one or negative one can have reuse with non-addresses. if (D->getValue()->isOne() || D->getValue()->isAllOnesValue()) @@ -1615,12 +1649,12 @@ LSRInstance::reconcileNewOffset(LSRUse &LU, int64_t NewOffset, // Conservatively assume HasBaseReg is true for now. if (NewOffset < LU.MinOffset) { if (!isAlwaysFoldable(LU.MaxOffset - NewOffset, 0, /*HasBaseReg=*/true, - Kind, AccessTy, TLI, SE)) + Kind, AccessTy, TLI)) return false; NewMinOffset = NewOffset; } else if (NewOffset > LU.MaxOffset) { if (!isAlwaysFoldable(NewOffset - LU.MinOffset, 0, /*HasBaseReg=*/true, - Kind, AccessTy, TLI, SE)) + Kind, AccessTy, TLI)) return false; NewMaxOffset = NewOffset; } @@ -1639,7 +1673,7 @@ LSRInstance::reconcileNewOffset(LSRUse &LU, int64_t NewOffset, /// getUse - Return an LSRUse index and an offset value for a fixup which /// needs the given expression, with the given kind and optional access type. -/// Either reuse an exisitng use or create a new one, as needed. +/// Either reuse an existing use or create a new one, as needed. std::pair<size_t, int64_t> LSRInstance::getUse(const SCEV *&Expr, LSRUse::KindType Kind, const Type *AccessTy) { @@ -1647,8 +1681,7 @@ LSRInstance::getUse(const SCEV *&Expr, int64_t Offset = ExtractImmediate(Expr, SE); // Basic uses can't accept any offset, for example. - if (!isAlwaysFoldable(Offset, 0, /*HasBaseReg=*/true, - Kind, AccessTy, TLI, SE)) { + if (!isAlwaysFoldable(Offset, 0, /*HasBaseReg=*/true, Kind, AccessTy, TLI)) { Expr = Copy; Offset = 0; } @@ -1683,21 +1716,29 @@ LSRInstance::getUse(const SCEV *&Expr, void LSRInstance::CollectInterestingTypesAndFactors() { SmallSetVector<const SCEV *, 4> Strides; - // Collect interesting types and factors. + // Collect interesting types and strides. for (IVUsers::const_iterator UI = IU.begin(), E = IU.end(); UI != E; ++UI) { const SCEV *Stride = UI->getStride(); // Collect interesting types. Types.insert(SE.getEffectiveSCEVType(Stride->getType())); - // Collect interesting factors. + // Add the stride for this loop. + Strides.insert(Stride); + + // Add strides for other mentioned loops. + for (const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(UI->getOffset()); + AR; AR = dyn_cast<SCEVAddRecExpr>(AR->getStart())) + Strides.insert(AR->getStepRecurrence(SE)); + } + + // Compute interesting factors from the set of interesting strides. + for (SmallSetVector<const SCEV *, 4>::const_iterator + I = Strides.begin(), E = Strides.end(); I != E; ++I) for (SmallSetVector<const SCEV *, 4>::const_iterator NewStrideIter = - Strides.begin(), SEnd = Strides.end(); NewStrideIter != SEnd; - ++NewStrideIter) { - const SCEV *OldStride = Stride; + next(I); NewStrideIter != E; ++NewStrideIter) { + const SCEV *OldStride = *I; const SCEV *NewStride = *NewStrideIter; - if (OldStride == NewStride) - continue; if (SE.getTypeSizeInBits(OldStride->getType()) != SE.getTypeSizeInBits(NewStride->getType())) { @@ -1708,19 +1749,18 @@ void LSRInstance::CollectInterestingTypesAndFactors() { OldStride = SE.getSignExtendExpr(OldStride, NewStride->getType()); } if (const SCEVConstant *Factor = - dyn_cast_or_null<SCEVConstant>(getSDiv(NewStride, OldStride, - SE, true))) { + dyn_cast_or_null<SCEVConstant>(getExactSDiv(NewStride, OldStride, + SE, true))) { if (Factor->getValue()->getValue().getMinSignedBits() <= 64) Factors.insert(Factor->getValue()->getValue().getSExtValue()); } else if (const SCEVConstant *Factor = - dyn_cast_or_null<SCEVConstant>(getSDiv(OldStride, NewStride, - SE, true))) { + dyn_cast_or_null<SCEVConstant>(getExactSDiv(OldStride, + NewStride, + SE, true))) { if (Factor->getValue()->getValue().getMinSignedBits() <= 64) Factors.insert(Factor->getValue()->getValue().getSExtValue()); } } - Strides.insert(Stride); - } // If all uses use the same type, don't bother looking for truncation-based // reuse. @@ -1788,7 +1828,7 @@ void LSRInstance::CollectFixupsAndInitialFormulae() { // If this is the first use of this LSRUse, give it a formula. if (LU.Formulae.empty()) { - InsertInitialFormula(S, L, LU, LF.LUIdx); + InsertInitialFormula(S, LU, LF.LUIdx); CountRegisters(LU.Formulae.back(), LF.LUIdx); } } @@ -1797,8 +1837,7 @@ void LSRInstance::CollectFixupsAndInitialFormulae() { } void -LSRInstance::InsertInitialFormula(const SCEV *S, Loop *L, - LSRUse &LU, size_t LUIdx) { +LSRInstance::InsertInitialFormula(const SCEV *S, LSRUse &LU, size_t LUIdx) { Formula F; F.InitialMatch(S, L, SE, DT); bool Inserted = InsertFormula(LU, LUIdx, F); @@ -1828,7 +1867,7 @@ void LSRInstance::CountRegisters(const Formula &F, size_t LUIdx) { /// InsertFormula - If the given formula has not yet been inserted, add it to /// the list, and return true. Return false otherwise. bool LSRInstance::InsertFormula(LSRUse &LU, unsigned LUIdx, const Formula &F) { - if (!LU.InsertFormula(LUIdx, F)) + if (!LU.InsertFormula(F)) return false; CountRegisters(F, LUIdx); @@ -1996,7 +2035,7 @@ void LSRInstance::GenerateReassociations(LSRUse &LU, unsigned LUIdx, /// loop-dominating registers added into a single register. void LSRInstance::GenerateCombinations(LSRUse &LU, unsigned LUIdx, Formula Base) { - // This method is only intersting on a plurality of registers. + // This method is only interesting on a plurality of registers. if (Base.BaseRegs.size() <= 1) return; Formula F = Base; @@ -2015,7 +2054,7 @@ void LSRInstance::GenerateCombinations(LSRUse &LU, unsigned LUIdx, const SCEV *Sum = SE.getAddExpr(Ops); // TODO: If Sum is zero, it probably means ScalarEvolution missed an // opportunity to fold something. For now, just ignore such cases - // rather than procede with zero in a register. + // rather than proceed with zero in a register. if (!Sum->isZero()) { F.BaseRegs.push_back(Sum); (void)InsertFormula(LU, LUIdx, F); @@ -2105,14 +2144,18 @@ void LSRInstance::GenerateICmpZeroScales(LSRUse &LU, unsigned LUIdx, Formula F = Base; // Check that the multiplication doesn't overflow. + if (F.AM.BaseOffs == INT64_MIN && Factor == -1) + continue; F.AM.BaseOffs = (uint64_t)Base.AM.BaseOffs * Factor; - if ((int64_t)F.AM.BaseOffs / Factor != Base.AM.BaseOffs) + if (F.AM.BaseOffs / Factor != Base.AM.BaseOffs) continue; // Check that multiplying with the use offset doesn't overflow. int64_t Offset = LU.MinOffset; + if (Offset == INT64_MIN && Factor == -1) + continue; Offset = (uint64_t)Offset * Factor; - if ((int64_t)Offset / Factor != LU.MinOffset) + if (Offset / Factor != LU.MinOffset) continue; // Check that this scale is legal. @@ -2127,14 +2170,14 @@ void LSRInstance::GenerateICmpZeroScales(LSRUse &LU, unsigned LUIdx, // Check that multiplying with each base register doesn't overflow. for (size_t i = 0, e = F.BaseRegs.size(); i != e; ++i) { F.BaseRegs[i] = SE.getMulExpr(F.BaseRegs[i], FactorS); - if (getSDiv(F.BaseRegs[i], FactorS, SE) != Base.BaseRegs[i]) + if (getExactSDiv(F.BaseRegs[i], FactorS, SE) != Base.BaseRegs[i]) goto next; } // Check that multiplying with the scaled register doesn't overflow. if (F.ScaledReg) { F.ScaledReg = SE.getMulExpr(F.ScaledReg, FactorS); - if (getSDiv(F.ScaledReg, FactorS, SE) != Base.ScaledReg) + if (getExactSDiv(F.ScaledReg, FactorS, SE) != Base.ScaledReg) continue; } @@ -2189,7 +2232,7 @@ void LSRInstance::GenerateScales(LSRUse &LU, unsigned LUIdx, continue; // Divide out the factor, ignoring high bits, since we'll be // scaling the value back up in the end. - if (const SCEV *Quotient = getSDiv(AR, FactorS, SE, true)) { + if (const SCEV *Quotient = getExactSDiv(AR, FactorS, SE, true)) { // TODO: This could be optimized to avoid all the copying. Formula F = Base; F.ScaledReg = Quotient; @@ -2358,7 +2401,7 @@ void LSRInstance::GenerateCrossUseConstantOffsets() { const SCEV *NegImmS = SE.getSCEV(ConstantInt::get(IntTy, -(uint64_t)Imm)); unsigned BitWidth = SE.getTypeSizeInBits(IntTy); - // TODO: Use a more targetted data structure. + // TODO: Use a more targeted data structure. for (size_t L = 0, LE = LU.Formulae.size(); L != LE; ++L) { Formula F = LU.Formulae[L]; // Use the immediate in the scaled register. @@ -2526,9 +2569,9 @@ void LSRInstance::FilterOutUndesirableDedicatedRegisters() { }); } -/// NarrowSearchSpaceUsingHeuristics - If there are an extrordinary number of +/// NarrowSearchSpaceUsingHeuristics - If there are an extraordinary number of /// formulae to choose from, use some rough heuristics to prune down the number -/// of formulae. This keeps the main solver from taking an extrordinary amount +/// of formulae. This keeps the main solver from taking an extraordinary amount /// of time in some worst-case scenarios. void LSRInstance::NarrowSearchSpaceUsingHeuristics() { // This is a rough guess that seems to work fairly well. @@ -2578,7 +2621,7 @@ void LSRInstance::NarrowSearchSpaceUsingHeuristics() { } DEBUG(dbgs() << "Narrowing the search space by assuming " << *Best - << " will yeild profitable reuse.\n"); + << " will yield profitable reuse.\n"); Taken.insert(Best); // In any use with formulae which references this register, delete formulae @@ -2625,7 +2668,7 @@ void LSRInstance::SolveRecurse(SmallVectorImpl<const Formula *> &Solution, // - sort the formula so that the most profitable solutions are found first // - sort the uses too // - search faster: - // - dont compute a cost, and then compare. compare while computing a cost + // - don't compute a cost, and then compare. compare while computing a cost // and bail early. // - track register sets with SmallBitVector @@ -2736,10 +2779,8 @@ static BasicBlock *getImmediateDominator(BasicBlock *BB, DominatorTree &DT) { Value *LSRInstance::Expand(const LSRFixup &LF, const Formula &F, BasicBlock::iterator IP, - Loop *L, Instruction *IVIncInsertPos, SCEVExpander &Rewriter, - SmallVectorImpl<WeakVH> &DeadInsts, - ScalarEvolution &SE, DominatorTree &DT) const { + SmallVectorImpl<WeakVH> &DeadInsts) const { const LSRUse &LU = Uses[LF.LUIdx]; // Then, collect some instructions which we will remain dominated by when @@ -2752,8 +2793,12 @@ Value *LSRInstance::Expand(const LSRFixup &LF, if (Instruction *I = dyn_cast<Instruction>(cast<ICmpInst>(LF.UserInst)->getOperand(1))) Inputs.push_back(I); - if (LF.PostIncLoop && !L->contains(LF.UserInst)) - Inputs.push_back(L->getLoopLatch()->getTerminator()); + if (LF.PostIncLoop) { + if (!L->contains(LF.UserInst)) + Inputs.push_back(L->getLoopLatch()->getTerminator()); + else + Inputs.push_back(IVIncInsertPos); + } // Then, climb up the immediate dominator tree as far as we can go while // still being dominated by the input positions. @@ -2816,8 +2861,10 @@ Value *LSRInstance::Expand(const LSRFixup &LF, if (AR->getLoop() == LF.PostIncLoop) { Reg = SE.getAddExpr(Reg, AR->getStepRecurrence(SE)); // If the user is inside the loop, insert the code after the increment - // so that it is dominated by its operand. - if (L->contains(LF.UserInst)) + // so that it is dominated by its operand. If the original insert point + // was already dominated by the increment, keep it, because there may + // be loop-variant operands that need to be respected also. + if (L->contains(LF.UserInst) && !DT.dominates(IVIncInsertPos, IP)) IP = IVIncInsertPos; break; } @@ -2827,6 +2874,13 @@ Value *LSRInstance::Expand(const LSRFixup &LF, Ops.push_back(SE.getUnknown(Rewriter.expandCodeFor(Reg, 0, IP))); } + // Flush the operand list to suppress SCEVExpander hoisting. + if (!Ops.empty()) { + Value *FullV = Rewriter.expandCodeFor(SE.getAddExpr(Ops), Ty, IP); + Ops.clear(); + Ops.push_back(SE.getUnknown(FullV)); + } + // Expand the ScaledReg portion. Value *ICmpScaledV = 0; if (F.AM.Scale != 0) { @@ -2853,12 +2907,25 @@ Value *LSRInstance::Expand(const LSRFixup &LF, SE.getIntegerSCEV(F.AM.Scale, ScaledS->getType())); Ops.push_back(ScaledS); + + // Flush the operand list to suppress SCEVExpander hoisting. + Value *FullV = Rewriter.expandCodeFor(SE.getAddExpr(Ops), Ty, IP); + Ops.clear(); + Ops.push_back(SE.getUnknown(FullV)); } } - // Expand the immediate portions. - if (F.AM.BaseGV) - Ops.push_back(SE.getSCEV(F.AM.BaseGV)); + // Expand the GV portion. + if (F.AM.BaseGV) { + Ops.push_back(SE.getUnknown(F.AM.BaseGV)); + + // Flush the operand list to suppress SCEVExpander hoisting. + Value *FullV = Rewriter.expandCodeFor(SE.getAddExpr(Ops), Ty, IP); + Ops.clear(); + Ops.push_back(SE.getUnknown(FullV)); + } + + // Expand the immediate portion. int64_t Offset = (uint64_t)F.AM.BaseOffs + LF.Offset; if (Offset != 0) { if (LU.Kind == LSRUse::ICmpZero) { @@ -2873,7 +2940,7 @@ Value *LSRInstance::Expand(const LSRFixup &LF, } else { // Just add the immediate values. These again are expected to be matched // as part of the address. - Ops.push_back(SE.getIntegerSCEV(Offset, IntTy)); + Ops.push_back(SE.getUnknown(ConstantInt::getSigned(IntTy, Offset))); } } @@ -2921,73 +2988,81 @@ Value *LSRInstance::Expand(const LSRFixup &LF, return FullV; } +/// RewriteForPHI - Helper for Rewrite. PHI nodes are special because the use +/// of their operands effectively happens in their predecessor blocks, so the +/// expression may need to be expanded in multiple places. +void LSRInstance::RewriteForPHI(PHINode *PN, + const LSRFixup &LF, + const Formula &F, + SCEVExpander &Rewriter, + SmallVectorImpl<WeakVH> &DeadInsts, + Pass *P) const { + DenseMap<BasicBlock *, Value *> Inserted; + for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) + if (PN->getIncomingValue(i) == LF.OperandValToReplace) { + BasicBlock *BB = PN->getIncomingBlock(i); + + // If this is a critical edge, split the edge so that we do not insert + // the code on all predecessor/successor paths. We do this unless this + // is the canonical backedge for this loop, which complicates post-inc + // users. + if (e != 1 && BB->getTerminator()->getNumSuccessors() > 1 && + !isa<IndirectBrInst>(BB->getTerminator()) && + (PN->getParent() != L->getHeader() || !L->contains(BB))) { + // Split the critical edge. + BasicBlock *NewBB = SplitCriticalEdge(BB, PN->getParent(), P); + + // If PN is outside of the loop and BB is in the loop, we want to + // move the block to be immediately before the PHI block, not + // immediately after BB. + if (L->contains(BB) && !L->contains(PN)) + NewBB->moveBefore(PN->getParent()); + + // Splitting the edge can reduce the number of PHI entries we have. + e = PN->getNumIncomingValues(); + BB = NewBB; + i = PN->getBasicBlockIndex(BB); + } + + std::pair<DenseMap<BasicBlock *, Value *>::iterator, bool> Pair = + Inserted.insert(std::make_pair(BB, static_cast<Value *>(0))); + if (!Pair.second) + PN->setIncomingValue(i, Pair.first->second); + else { + Value *FullV = Expand(LF, F, BB->getTerminator(), Rewriter, DeadInsts); + + // If this is reuse-by-noop-cast, insert the noop cast. + const Type *OpTy = LF.OperandValToReplace->getType(); + if (FullV->getType() != OpTy) + FullV = + CastInst::Create(CastInst::getCastOpcode(FullV, false, + OpTy, false), + FullV, LF.OperandValToReplace->getType(), + "tmp", BB->getTerminator()); + + PN->setIncomingValue(i, FullV); + Pair.first->second = FullV; + } + } +} + /// Rewrite - Emit instructions for the leading candidate expression for this /// LSRUse (this is called "expanding"), and update the UserInst to reference /// the newly expanded value. void LSRInstance::Rewrite(const LSRFixup &LF, const Formula &F, - Loop *L, Instruction *IVIncInsertPos, SCEVExpander &Rewriter, SmallVectorImpl<WeakVH> &DeadInsts, - ScalarEvolution &SE, DominatorTree &DT, Pass *P) const { - const Type *OpTy = LF.OperandValToReplace->getType(); - // First, find an insertion point that dominates UserInst. For PHI nodes, // find the nearest block which dominates all the relevant uses. if (PHINode *PN = dyn_cast<PHINode>(LF.UserInst)) { - DenseMap<BasicBlock *, Value *> Inserted; - for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) - if (PN->getIncomingValue(i) == LF.OperandValToReplace) { - BasicBlock *BB = PN->getIncomingBlock(i); - - // If this is a critical edge, split the edge so that we do not insert - // the code on all predecessor/successor paths. We do this unless this - // is the canonical backedge for this loop, which complicates post-inc - // users. - if (e != 1 && BB->getTerminator()->getNumSuccessors() > 1 && - !isa<IndirectBrInst>(BB->getTerminator()) && - (PN->getParent() != L->getHeader() || !L->contains(BB))) { - // Split the critical edge. - BasicBlock *NewBB = SplitCriticalEdge(BB, PN->getParent(), P); - - // If PN is outside of the loop and BB is in the loop, we want to - // move the block to be immediately before the PHI block, not - // immediately after BB. - if (L->contains(BB) && !L->contains(PN)) - NewBB->moveBefore(PN->getParent()); - - // Splitting the edge can reduce the number of PHI entries we have. - e = PN->getNumIncomingValues(); - BB = NewBB; - i = PN->getBasicBlockIndex(BB); - } - - std::pair<DenseMap<BasicBlock *, Value *>::iterator, bool> Pair = - Inserted.insert(std::make_pair(BB, static_cast<Value *>(0))); - if (!Pair.second) - PN->setIncomingValue(i, Pair.first->second); - else { - Value *FullV = Expand(LF, F, BB->getTerminator(), L, IVIncInsertPos, - Rewriter, DeadInsts, SE, DT); - - // If this is reuse-by-noop-cast, insert the noop cast. - if (FullV->getType() != OpTy) - FullV = - CastInst::Create(CastInst::getCastOpcode(FullV, false, - OpTy, false), - FullV, LF.OperandValToReplace->getType(), - "tmp", BB->getTerminator()); - - PN->setIncomingValue(i, FullV); - Pair.first->second = FullV; - } - } + RewriteForPHI(PN, LF, F, Rewriter, DeadInsts, P); } else { - Value *FullV = Expand(LF, F, LF.UserInst, L, IVIncInsertPos, - Rewriter, DeadInsts, SE, DT); + Value *FullV = Expand(LF, F, LF.UserInst, Rewriter, DeadInsts); // If this is reuse-by-noop-cast, insert the noop cast. + const Type *OpTy = LF.OperandValToReplace->getType(); if (FullV->getType() != OpTy) { Instruction *Cast = CastInst::Create(CastInst::getCastOpcode(FullV, false, OpTy, false), @@ -3024,8 +3099,7 @@ LSRInstance::ImplementSolution(const SmallVectorImpl<const Formula *> &Solution, for (size_t i = 0, e = Fixups.size(); i != e; ++i) { size_t LUIdx = Fixups[i].LUIdx; - Rewrite(Fixups[i], *Solution[LUIdx], L, IVIncInsertPos, Rewriter, - DeadInsts, SE, DT, P); + Rewrite(Fixups[i], *Solution[LUIdx], Rewriter, DeadInsts, P); Changed = true; } @@ -3054,7 +3128,7 @@ LSRInstance::LSRInstance(const TargetLowering *tli, Loop *l, Pass *P) dbgs() << ":\n"); /// OptimizeShadowIV - If IV is used in a int-to-float cast - /// inside the loop then try to eliminate the cast opeation. + /// inside the loop then try to eliminate the cast operation. OptimizeShadowIV(); // Change loop terminating condition to use the postinc iv when possible. diff --git a/lib/Transforms/Scalar/LoopUnswitch.cpp b/lib/Transforms/Scalar/LoopUnswitch.cpp index 990e0c4..071e9b7 100644 --- a/lib/Transforms/Scalar/LoopUnswitch.cpp +++ b/lib/Transforms/Scalar/LoopUnswitch.cpp @@ -170,7 +170,7 @@ Pass *llvm::createLoopUnswitchPass(bool Os) { /// Otherwise, return null. static Value *FindLIVLoopCondition(Value *Cond, Loop *L, bool &Changed) { // We can never unswitch on vector conditions. - if (isa<VectorType>(Cond->getType())) + if (Cond->getType()->isVectorTy()) return 0; // Constants should be folded, not unswitched on! diff --git a/lib/Transforms/Scalar/Reassociate.cpp b/lib/Transforms/Scalar/Reassociate.cpp index 187216a..12827b6 100644 --- a/lib/Transforms/Scalar/Reassociate.cpp +++ b/lib/Transforms/Scalar/Reassociate.cpp @@ -930,7 +930,7 @@ void Reassociate::ReassociateBB(BasicBlock *BB) { // Reject cases where it is pointless to do this. if (!isa<BinaryOperator>(BI) || BI->getType()->isFloatingPointTy() || - isa<VectorType>(BI->getType())) + BI->getType()->isVectorTy()) continue; // Floating point ops are not associative. // Do not reassociate boolean (i1) expressions. We want to preserve the diff --git a/lib/Transforms/Scalar/SCCP.cpp b/lib/Transforms/Scalar/SCCP.cpp index 02b45a1..7e37938 100644 --- a/lib/Transforms/Scalar/SCCP.cpp +++ b/lib/Transforms/Scalar/SCCP.cpp @@ -295,7 +295,7 @@ public: } void markOverdefined(Value *V) { - assert(!isa<StructType>(V->getType()) && "Should use other method"); + assert(!V->getType()->isStructTy() && "Should use other method"); markOverdefined(ValueState[V], V); } @@ -321,12 +321,12 @@ private: } void markConstant(Value *V, Constant *C) { - assert(!isa<StructType>(V->getType()) && "Should use other method"); + assert(!V->getType()->isStructTy() && "Should use other method"); markConstant(ValueState[V], V, C); } void markForcedConstant(Value *V, Constant *C) { - assert(!isa<StructType>(V->getType()) && "Should use other method"); + assert(!V->getType()->isStructTy() && "Should use other method"); ValueState[V].markForcedConstant(C); DEBUG(dbgs() << "markForcedConstant: " << *C << ": " << *V << '\n'); InstWorkList.push_back(V); @@ -360,7 +360,7 @@ private: } void mergeInValue(Value *V, LatticeVal MergeWithV) { - assert(!isa<StructType>(V->getType()) && "Should use other method"); + assert(!V->getType()->isStructTy() && "Should use other method"); mergeInValue(ValueState[V], V, MergeWithV); } @@ -369,7 +369,7 @@ private: /// value. This function handles the case when the value hasn't been seen yet /// by properly seeding constants etc. LatticeVal &getValueState(Value *V) { - assert(!isa<StructType>(V->getType()) && "Should use getStructValueState"); + assert(!V->getType()->isStructTy() && "Should use getStructValueState"); std::pair<DenseMap<Value*, LatticeVal>::iterator, bool> I = ValueState.insert(std::make_pair(V, LatticeVal())); @@ -392,7 +392,7 @@ private: /// value/field pair. This function handles the case when the value hasn't /// been seen yet by properly seeding constants etc. LatticeVal &getStructValueState(Value *V, unsigned i) { - assert(isa<StructType>(V->getType()) && "Should use getValueState"); + assert(V->getType()->isStructTy() && "Should use getValueState"); assert(i < cast<StructType>(V->getType())->getNumElements() && "Invalid element #"); @@ -666,7 +666,7 @@ bool SCCPSolver::isEdgeFeasible(BasicBlock *From, BasicBlock *To) { void SCCPSolver::visitPHINode(PHINode &PN) { // If this PN returns a struct, just mark the result overdefined. // TODO: We could do a lot better than this if code actually uses this. - if (isa<StructType>(PN.getType())) + if (PN.getType()->isStructTy()) return markAnythingOverdefined(&PN); if (getValueState(&PN).isOverdefined()) { @@ -742,7 +742,7 @@ void SCCPSolver::visitReturnInst(ReturnInst &I) { Value *ResultOp = I.getOperand(0); // If we are tracking the return value of this function, merge it in. - if (!TrackedRetVals.empty() && !isa<StructType>(ResultOp->getType())) { + if (!TrackedRetVals.empty() && !ResultOp->getType()->isStructTy()) { DenseMap<Function*, LatticeVal>::iterator TFRVI = TrackedRetVals.find(F); if (TFRVI != TrackedRetVals.end()) { @@ -787,7 +787,7 @@ void SCCPSolver::visitCastInst(CastInst &I) { void SCCPSolver::visitExtractValueInst(ExtractValueInst &EVI) { // If this returns a struct, mark all elements over defined, we don't track // structs in structs. - if (isa<StructType>(EVI.getType())) + if (EVI.getType()->isStructTy()) return markAnythingOverdefined(&EVI); // If this is extracting from more than one level of struct, we don't know. @@ -795,7 +795,7 @@ void SCCPSolver::visitExtractValueInst(ExtractValueInst &EVI) { return markOverdefined(&EVI); Value *AggVal = EVI.getAggregateOperand(); - if (isa<StructType>(AggVal->getType())) { + if (AggVal->getType()->isStructTy()) { unsigned i = *EVI.idx_begin(); LatticeVal EltVal = getStructValueState(AggVal, i); mergeInValue(getValueState(&EVI), &EVI, EltVal); @@ -828,7 +828,7 @@ void SCCPSolver::visitInsertValueInst(InsertValueInst &IVI) { } Value *Val = IVI.getInsertedValueOperand(); - if (isa<StructType>(Val->getType())) + if (Val->getType()->isStructTy()) // We don't track structs in structs. markOverdefined(getStructValueState(&IVI, i), &IVI); else { @@ -841,7 +841,7 @@ void SCCPSolver::visitInsertValueInst(InsertValueInst &IVI) { void SCCPSolver::visitSelectInst(SelectInst &I) { // If this select returns a struct, just mark the result overdefined. // TODO: We could do a lot better than this if code actually uses this. - if (isa<StructType>(I.getType())) + if (I.getType()->isStructTy()) return markAnythingOverdefined(&I); LatticeVal CondValue = getValueState(I.getCondition()); @@ -1166,7 +1166,7 @@ void SCCPSolver::visitGetElementPtrInst(GetElementPtrInst &I) { void SCCPSolver::visitStoreInst(StoreInst &SI) { // If this store is of a struct, ignore it. - if (isa<StructType>(SI.getOperand(0)->getType())) + if (SI.getOperand(0)->getType()->isStructTy()) return; if (TrackedGlobals.empty() || !isa<GlobalVariable>(SI.getOperand(1))) @@ -1187,7 +1187,7 @@ void SCCPSolver::visitStoreInst(StoreInst &SI) { // global, we can replace the load with the loaded constant value! void SCCPSolver::visitLoadInst(LoadInst &I) { // If this load is of a struct, just mark the result overdefined. - if (isa<StructType>(I.getType())) + if (I.getType()->isStructTy()) return markAnythingOverdefined(&I); LatticeVal PtrVal = getValueState(I.getOperand(0)); @@ -1241,7 +1241,7 @@ CallOverdefined: // Otherwise, if we have a single return value case, and if the function is // a declaration, maybe we can constant fold it. - if (F && F->isDeclaration() && !isa<StructType>(I->getType()) && + if (F && F->isDeclaration() && !I->getType()->isStructTy() && canConstantFoldCallTo(F)) { SmallVector<Constant*, 8> Operands; @@ -1352,7 +1352,7 @@ void SCCPSolver::Solve() { // since all of its users will have already been marked as overdefined. // Update all of the users of this instruction's value. // - if (isa<StructType>(I->getType()) || !getValueState(I).isOverdefined()) + if (I->getType()->isStructTy() || !getValueState(I).isOverdefined()) for (Value::use_iterator UI = I->use_begin(), E = I->use_end(); UI != E; ++UI) if (Instruction *I = dyn_cast<Instruction>(*UI)) @@ -1418,7 +1418,7 @@ bool SCCPSolver::ResolvedUndefsIn(Function &F) { if (!LV.isUndefined()) continue; // No instructions using structs need disambiguation. - if (isa<StructType>(I->getOperand(0)->getType())) + if (I->getOperand(0)->getType()->isStructTy()) continue; // Get the lattice values of the first two operands for use below. @@ -1426,7 +1426,7 @@ bool SCCPSolver::ResolvedUndefsIn(Function &F) { LatticeVal Op1LV; if (I->getNumOperands() == 2) { // No instructions using structs need disambiguation. - if (isa<StructType>(I->getOperand(1)->getType())) + if (I->getOperand(1)->getType()->isStructTy()) continue; // If this is a two-operand instruction, and if both operands are @@ -1656,7 +1656,7 @@ bool SCCP::runOnFunction(Function &F) { continue; // TODO: Reconstruct structs from their elements. - if (isa<StructType>(Inst->getType())) + if (Inst->getType()->isStructTy()) continue; LatticeVal IV = Solver.getLatticeValueFor(Inst); @@ -1792,7 +1792,7 @@ bool IPSCCP::runOnModule(Module &M) { if (Solver.isBlockExecutable(F->begin())) { for (Function::arg_iterator AI = F->arg_begin(), E = F->arg_end(); AI != E; ++AI) { - if (AI->use_empty() || isa<StructType>(AI->getType())) continue; + if (AI->use_empty() || AI->getType()->isStructTy()) continue; // TODO: Could use getStructLatticeValueFor to find out if the entire // result is a constant and replace it entirely if so. @@ -1835,7 +1835,7 @@ bool IPSCCP::runOnModule(Module &M) { for (BasicBlock::iterator BI = BB->begin(), E = BB->end(); BI != E; ) { Instruction *Inst = BI++; - if (Inst->getType()->isVoidTy() || isa<StructType>(Inst->getType())) + if (Inst->getType()->isVoidTy() || Inst->getType()->isStructTy()) continue; // TODO: Could use getStructLatticeValueFor to find out if the entire @@ -1918,6 +1918,14 @@ bool IPSCCP::runOnModule(Module &M) { // all call uses with the inferred value. This means we don't need to bother // actually returning anything from the function. Replace all return // instructions with return undef. + // + // Do this in two stages: first identify the functions we should process, then + // actually zap their returns. This is important because we can only do this + // if the address of the function isn't taken. In cases where a return is the + // last use of a function, the order of processing functions would affect + // whether other functions are optimizable. + SmallVector<ReturnInst*, 8> ReturnsToZap; + // TODO: Process multiple value ret instructions also. const DenseMap<Function*, LatticeVal> &RV = Solver.getTrackedRetVals(); for (DenseMap<Function*, LatticeVal>::const_iterator I = RV.begin(), @@ -1933,7 +1941,13 @@ bool IPSCCP::runOnModule(Module &M) { for (Function::iterator BB = F->begin(), E = F->end(); BB != E; ++BB) if (ReturnInst *RI = dyn_cast<ReturnInst>(BB->getTerminator())) if (!isa<UndefValue>(RI->getOperand(0))) - RI->setOperand(0, UndefValue::get(F->getReturnType())); + ReturnsToZap.push_back(RI); + } + + // Zap all returns which we've identified as zap to change. + for (unsigned i = 0, e = ReturnsToZap.size(); i != e; ++i) { + Function *F = ReturnsToZap[i]->getParent()->getParent(); + ReturnsToZap[i]->setOperand(0, UndefValue::get(F->getReturnType())); } // If we infered constant or undef values for globals variables, we can delete diff --git a/lib/Transforms/Scalar/ScalarReplAggregates.cpp b/lib/Transforms/Scalar/ScalarReplAggregates.cpp index 822712e..bbe6270 100644 --- a/lib/Transforms/Scalar/ScalarReplAggregates.cpp +++ b/lib/Transforms/Scalar/ScalarReplAggregates.cpp @@ -302,7 +302,7 @@ bool SROA::performScalarRepl(Function &F) { // random stuff that doesn't use vectors (e.g. <9 x double>) because then // we just get a lot of insert/extracts. If at least one vector is // involved, then we probably really do have a union of vector/array. - if (VectorTy && isa<VectorType>(VectorTy) && HadAVector) { + if (VectorTy && VectorTy->isVectorTy() && HadAVector) { DEBUG(dbgs() << "CONVERT TO VECTOR: " << *AI << "\n TYPE = " << *VectorTy << '\n'); @@ -449,7 +449,7 @@ void SROA::isSafeGEP(GetElementPtrInst *GEPI, AllocaInst *AI, // into. for (; GEPIt != E; ++GEPIt) { // Ignore struct elements, no extra checking needed for these. - if (isa<StructType>(*GEPIt)) + if ((*GEPIt)->isStructTy()) continue; ConstantInt *IdxVal = dyn_cast<ConstantInt>(GEPIt.getOperand()); @@ -480,7 +480,7 @@ void SROA::isSafeMemAccess(AllocaInst *AI, uint64_t Offset, uint64_t MemSize, // (which are essentially the same as the MemIntrinsics, especially with // regard to copying padding between elements), or references using the // aggregate type of the alloca. - if (!MemOpType || isa<IntegerType>(MemOpType) || UsesAggregateType) { + if (!MemOpType || MemOpType->isIntegerTy() || UsesAggregateType) { if (!UsesAggregateType) { if (isStore) Info.isMemCpyDst = true; @@ -565,7 +565,7 @@ void SROA::RewriteForScalarRepl(Instruction *I, AllocaInst *AI, uint64_t Offset, } LI->replaceAllUsesWith(Insert); DeadInsts.push_back(LI); - } else if (isa<IntegerType>(LIType) && + } else if (LIType->isIntegerTy() && TD->getTypeAllocSize(LIType) == TD->getTypeAllocSize(AI->getAllocatedType())) { // If this is a load of the entire alloca to an integer, rewrite it. @@ -588,7 +588,7 @@ void SROA::RewriteForScalarRepl(Instruction *I, AllocaInst *AI, uint64_t Offset, new StoreInst(Extract, NewElts[i], SI); } DeadInsts.push_back(SI); - } else if (isa<IntegerType>(SIType) && + } else if (SIType->isIntegerTy() && TD->getTypeAllocSize(SIType) == TD->getTypeAllocSize(AI->getAllocatedType())) { // If this is a store of the entire alloca from an integer, rewrite it. @@ -833,7 +833,7 @@ void SROA::RewriteMemIntrinUserOfAlloca(MemIntrinsic *MI, Instruction *Inst, // Convert the integer value to the appropriate type. StoreVal = ConstantInt::get(Context, TotalVal); - if (isa<PointerType>(ValTy)) + if (ValTy->isPointerTy()) StoreVal = ConstantExpr::getIntToPtr(StoreVal, ValTy); else if (ValTy->isFloatingPointTy()) StoreVal = ConstantExpr::getBitCast(StoreVal, ValTy); @@ -939,7 +939,7 @@ void SROA::RewriteStoreUserOfWholeAlloca(StoreInst *SI, AllocaInst *AI, Value *DestField = NewElts[i]; if (EltVal->getType() == FieldTy) { // Storing to an integer field of this size, just do it. - } else if (FieldTy->isFloatingPointTy() || isa<VectorType>(FieldTy)) { + } else if (FieldTy->isFloatingPointTy() || FieldTy->isVectorTy()) { // Bitcast to the right element type (for fp/vector values). EltVal = new BitCastInst(EltVal, FieldTy, "", SI); } else { @@ -984,7 +984,7 @@ void SROA::RewriteStoreUserOfWholeAlloca(StoreInst *SI, AllocaInst *AI, if (EltVal->getType() == ArrayEltTy) { // Storing to an integer field of this size, just do it. } else if (ArrayEltTy->isFloatingPointTy() || - isa<VectorType>(ArrayEltTy)) { + ArrayEltTy->isVectorTy()) { // Bitcast to the right element type (for fp/vector values). EltVal = new BitCastInst(EltVal, ArrayEltTy, "", SI); } else { @@ -1044,8 +1044,8 @@ void SROA::RewriteLoadUserOfWholeAlloca(LoadInst *LI, AllocaInst *AI, const IntegerType *FieldIntTy = IntegerType::get(LI->getContext(), FieldSizeBits); - if (!isa<IntegerType>(FieldTy) && !FieldTy->isFloatingPointTy() && - !isa<VectorType>(FieldTy)) + if (!FieldTy->isIntegerTy() && !FieldTy->isFloatingPointTy() && + !FieldTy->isVectorTy()) SrcField = new BitCastInst(SrcField, PointerType::getUnqual(FieldIntTy), "", LI); @@ -1183,7 +1183,7 @@ static void MergeInType(const Type *In, uint64_t Offset, const Type *&VecTy, return; } } else if (In->isFloatTy() || In->isDoubleTy() || - (isa<IntegerType>(In) && In->getPrimitiveSizeInBits() >= 8 && + (In->isIntegerTy() && In->getPrimitiveSizeInBits() >= 8 && isPowerOf2_32(In->getPrimitiveSizeInBits()))) { // If we're accessing something that could be an element of a vector, see // if the implied vector agrees with what we already have and if Offset is @@ -1227,7 +1227,7 @@ bool SROA::CanConvertToScalar(Value *V, bool &IsNotTrivial, const Type *&VecTy, return false; MergeInType(LI->getType(), Offset, VecTy, AllocaSize, *TD, V->getContext()); - SawVec |= isa<VectorType>(LI->getType()); + SawVec |= LI->getType()->isVectorTy(); continue; } @@ -1236,7 +1236,7 @@ bool SROA::CanConvertToScalar(Value *V, bool &IsNotTrivial, const Type *&VecTy, if (SI->getOperand(0) == V || SI->isVolatile()) return 0; MergeInType(SI->getOperand(0)->getType(), Offset, VecTy, AllocaSize, *TD, V->getContext()); - SawVec |= isa<VectorType>(SI->getOperand(0)->getType()); + SawVec |= SI->getOperand(0)->getType()->isVectorTy(); continue; } @@ -1438,7 +1438,7 @@ Value *SROA::ConvertScalar_ExtractValue(Value *FromVal, const Type *ToType, // If the result alloca is a vector type, this is either an element // access or a bitcast to another vector type of the same size. if (const VectorType *VTy = dyn_cast<VectorType>(FromVal->getType())) { - if (isa<VectorType>(ToType)) + if (ToType->isVectorTy()) return Builder.CreateBitCast(FromVal, ToType, "tmp"); // Otherwise it must be an element access. @@ -1521,9 +1521,9 @@ Value *SROA::ConvertScalar_ExtractValue(Value *FromVal, const Type *ToType, LIBitWidth), "tmp"); // If the result is an integer, this is a trunc or bitcast. - if (isa<IntegerType>(ToType)) { + if (ToType->isIntegerTy()) { // Should be done. - } else if (ToType->isFloatingPointTy() || isa<VectorType>(ToType)) { + } else if (ToType->isFloatingPointTy() || ToType->isVectorTy()) { // Just do a bitcast, we know the sizes match up. FromVal = Builder.CreateBitCast(FromVal, ToType, "tmp"); } else { @@ -1601,10 +1601,10 @@ Value *SROA::ConvertScalar_InsertValue(Value *SV, Value *Old, unsigned DestWidth = TD->getTypeSizeInBits(AllocaType); unsigned SrcStoreWidth = TD->getTypeStoreSizeInBits(SV->getType()); unsigned DestStoreWidth = TD->getTypeStoreSizeInBits(AllocaType); - if (SV->getType()->isFloatingPointTy() || isa<VectorType>(SV->getType())) + if (SV->getType()->isFloatingPointTy() || SV->getType()->isVectorTy()) SV = Builder.CreateBitCast(SV, IntegerType::get(SV->getContext(),SrcWidth), "tmp"); - else if (isa<PointerType>(SV->getType())) + else if (SV->getType()->isPointerTy()) SV = Builder.CreatePtrToInt(SV, TD->getIntPtrType(SV->getContext()), "tmp"); // Zero extend or truncate the value if needed. diff --git a/lib/Transforms/Scalar/SimplifyLibCalls.cpp b/lib/Transforms/Scalar/SimplifyLibCalls.cpp index 54b4380..cde214b 100644 --- a/lib/Transforms/Scalar/SimplifyLibCalls.cpp +++ b/lib/Transforms/Scalar/SimplifyLibCalls.cpp @@ -357,7 +357,7 @@ void LibCallOptimization::EmitFPutC(Value *Char, Value *File, IRBuilder<> &B) { AWI[0] = AttributeWithIndex::get(2, Attribute::NoCapture); AWI[1] = AttributeWithIndex::get(~0u, Attribute::NoUnwind); Constant *F; - if (isa<PointerType>(File->getType())) + if (File->getType()->isPointerTy()) F = M->getOrInsertFunction("fputc", AttrListPtr::get(AWI, 2), Type::getInt32Ty(*Context), Type::getInt32Ty(*Context), File->getType(), @@ -384,7 +384,7 @@ void LibCallOptimization::EmitFPutS(Value *Str, Value *File, IRBuilder<> &B) { AWI[1] = AttributeWithIndex::get(2, Attribute::NoCapture); AWI[2] = AttributeWithIndex::get(~0u, Attribute::NoUnwind); Constant *F; - if (isa<PointerType>(File->getType())) + if (File->getType()->isPointerTy()) F = M->getOrInsertFunction("fputs", AttrListPtr::get(AWI, 3), Type::getInt32Ty(*Context), Type::getInt8PtrTy(*Context), @@ -409,7 +409,7 @@ void LibCallOptimization::EmitFWrite(Value *Ptr, Value *Size, Value *File, AWI[1] = AttributeWithIndex::get(4, Attribute::NoCapture); AWI[2] = AttributeWithIndex::get(~0u, Attribute::NoUnwind); Constant *F; - if (isa<PointerType>(File->getType())) + if (File->getType()->isPointerTy()) F = M->getOrInsertFunction("fwrite", AttrListPtr::get(AWI, 3), TD->getIntPtrType(*Context), Type::getInt8PtrTy(*Context), @@ -548,7 +548,7 @@ static uint64_t GetStringLengthH(Value *V, SmallPtrSet<PHINode*, 32> &PHIs) { /// GetStringLength - If we can compute the length of the string pointed to by /// the specified pointer, return 'len+1'. If we can't, return 0. static uint64_t GetStringLength(Value *V) { - if (!isa<PointerType>(V->getType())) return 0; + if (!V->getType()->isPointerTy()) return 0; SmallPtrSet<PHINode*, 32> PHIs; uint64_t Len = GetStringLengthH(V, PHIs); @@ -638,7 +638,7 @@ struct StrNCatOpt : public StrCatOpt { FT->getReturnType() != Type::getInt8PtrTy(*Context) || FT->getParamType(0) != FT->getReturnType() || FT->getParamType(1) != FT->getReturnType() || - !isa<IntegerType>(FT->getParamType(2))) + !FT->getParamType(2)->isIntegerTy()) return 0; // Extract some information from the instruction @@ -790,7 +790,7 @@ struct StrNCmpOpt : public LibCallOptimization { !FT->getReturnType()->isIntegerTy(32) || FT->getParamType(0) != FT->getParamType(1) || FT->getParamType(0) != Type::getInt8PtrTy(*Context) || - !isa<IntegerType>(FT->getParamType(2))) + !FT->getParamType(2)->isIntegerTy()) return 0; Value *Str1P = CI->getOperand(1), *Str2P = CI->getOperand(2); @@ -866,7 +866,7 @@ struct StrNCpyOpt : public LibCallOptimization { if (FT->getNumParams() != 3 || FT->getReturnType() != FT->getParamType(0) || FT->getParamType(0) != FT->getParamType(1) || FT->getParamType(0) != Type::getInt8PtrTy(*Context) || - !isa<IntegerType>(FT->getParamType(2))) + !FT->getParamType(2)->isIntegerTy()) return 0; Value *Dst = CI->getOperand(1); @@ -915,7 +915,7 @@ struct StrLenOpt : public LibCallOptimization { const FunctionType *FT = Callee->getFunctionType(); if (FT->getNumParams() != 1 || FT->getParamType(0) != Type::getInt8PtrTy(*Context) || - !isa<IntegerType>(FT->getReturnType())) + !FT->getReturnType()->isIntegerTy()) return 0; Value *Src = CI->getOperand(1); @@ -939,8 +939,8 @@ struct StrToOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { const FunctionType *FT = Callee->getFunctionType(); if ((FT->getNumParams() != 2 && FT->getNumParams() != 3) || - !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1))) + !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy()) return 0; Value *EndPtr = CI->getOperand(2); @@ -960,9 +960,9 @@ struct StrStrOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { const FunctionType *FT = Callee->getFunctionType(); if (FT->getNumParams() != 2 || - !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1)) || - !isa<PointerType>(FT->getReturnType())) + !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy() || + !FT->getReturnType()->isPointerTy()) return 0; // fold strstr(x, x) -> x. @@ -1006,8 +1006,8 @@ struct StrStrOpt : public LibCallOptimization { struct MemCmpOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { const FunctionType *FT = Callee->getFunctionType(); - if (FT->getNumParams() != 3 || !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1)) || + if (FT->getNumParams() != 3 || !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy() || !FT->getReturnType()->isIntegerTy(32)) return 0; @@ -1055,8 +1055,8 @@ struct MemCpyOpt : public LibCallOptimization { const FunctionType *FT = Callee->getFunctionType(); if (FT->getNumParams() != 3 || FT->getReturnType() != FT->getParamType(0) || - !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1)) || + !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy() || FT->getParamType(2) != TD->getIntPtrType(*Context)) return 0; @@ -1076,8 +1076,8 @@ struct MemMoveOpt : public LibCallOptimization { const FunctionType *FT = Callee->getFunctionType(); if (FT->getNumParams() != 3 || FT->getReturnType() != FT->getParamType(0) || - !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1)) || + !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy() || FT->getParamType(2) != TD->getIntPtrType(*Context)) return 0; @@ -1097,8 +1097,8 @@ struct MemSetOpt : public LibCallOptimization { const FunctionType *FT = Callee->getFunctionType(); if (FT->getNumParams() != 3 || FT->getReturnType() != FT->getParamType(0) || - !isa<PointerType>(FT->getParamType(0)) || - !isa<IntegerType>(FT->getParamType(1)) || + !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isIntegerTy() || FT->getParamType(2) != TD->getIntPtrType(*Context)) return 0; @@ -1124,9 +1124,9 @@ struct MemCpyChkOpt : public LibCallOptimization { const FunctionType *FT = Callee->getFunctionType(); if (FT->getNumParams() != 4 || FT->getReturnType() != FT->getParamType(0) || - !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1)) || - !isa<IntegerType>(FT->getParamType(3)) || + !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy() || + !FT->getParamType(3)->isIntegerTy() || FT->getParamType(2) != TD->getIntPtrType(*Context)) return 0; @@ -1152,9 +1152,9 @@ struct MemSetChkOpt : public LibCallOptimization { const FunctionType *FT = Callee->getFunctionType(); if (FT->getNumParams() != 4 || FT->getReturnType() != FT->getParamType(0) || - !isa<PointerType>(FT->getParamType(0)) || - !isa<IntegerType>(FT->getParamType(1)) || - !isa<IntegerType>(FT->getParamType(3)) || + !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isIntegerTy() || + !FT->getParamType(3)->isIntegerTy() || FT->getParamType(2) != TD->getIntPtrType(*Context)) return 0; @@ -1182,9 +1182,9 @@ struct MemMoveChkOpt : public LibCallOptimization { const FunctionType *FT = Callee->getFunctionType(); if (FT->getNumParams() != 4 || FT->getReturnType() != FT->getParamType(0) || - !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1)) || - !isa<IntegerType>(FT->getParamType(3)) || + !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy() || + !FT->getParamType(3)->isIntegerTy() || FT->getParamType(2) != TD->getIntPtrType(*Context)) return 0; @@ -1205,8 +1205,8 @@ struct StrCpyChkOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { const FunctionType *FT = Callee->getFunctionType(); if (FT->getNumParams() != 3 || FT->getReturnType() != FT->getParamType(0) || - !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1))) + !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy()) return 0; ConstantInt *SizeCI = dyn_cast<ConstantInt>(CI->getOperand(3)); @@ -1376,7 +1376,7 @@ struct FFSOpt : public LibCallOptimization { // result type. if (FT->getNumParams() != 1 || !FT->getReturnType()->isIntegerTy(32) || - !isa<IntegerType>(FT->getParamType(0))) + !FT->getParamType(0)->isIntegerTy()) return 0; Value *Op = CI->getOperand(1); @@ -1410,7 +1410,7 @@ struct IsDigitOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { const FunctionType *FT = Callee->getFunctionType(); // We require integer(i32) - if (FT->getNumParams() != 1 || !isa<IntegerType>(FT->getReturnType()) || + if (FT->getNumParams() != 1 || !FT->getReturnType()->isIntegerTy() || !FT->getParamType(0)->isIntegerTy(32)) return 0; @@ -1431,7 +1431,7 @@ struct IsAsciiOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { const FunctionType *FT = Callee->getFunctionType(); // We require integer(i32) - if (FT->getNumParams() != 1 || !isa<IntegerType>(FT->getReturnType()) || + if (FT->getNumParams() != 1 || !FT->getReturnType()->isIntegerTy() || !FT->getParamType(0)->isIntegerTy(32)) return 0; @@ -1450,7 +1450,7 @@ struct AbsOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { const FunctionType *FT = Callee->getFunctionType(); // We require integer(integer) where the types agree. - if (FT->getNumParams() != 1 || !isa<IntegerType>(FT->getReturnType()) || + if (FT->getNumParams() != 1 || !FT->getReturnType()->isIntegerTy() || FT->getParamType(0) != FT->getReturnType()) return 0; @@ -1493,8 +1493,8 @@ struct PrintFOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { // Require one fixed pointer argument and an integer/void result. const FunctionType *FT = Callee->getFunctionType(); - if (FT->getNumParams() < 1 || !isa<PointerType>(FT->getParamType(0)) || - !(isa<IntegerType>(FT->getReturnType()) || + if (FT->getNumParams() < 1 || !FT->getParamType(0)->isPointerTy() || + !(FT->getReturnType()->isIntegerTy() || FT->getReturnType()->isVoidTy())) return 0; @@ -1534,7 +1534,7 @@ struct PrintFOpt : public LibCallOptimization { // Optimize specific format strings. // printf("%c", chr) --> putchar(*(i8*)dst) if (FormatStr == "%c" && CI->getNumOperands() > 2 && - isa<IntegerType>(CI->getOperand(2)->getType())) { + CI->getOperand(2)->getType()->isIntegerTy()) { Value *Res = EmitPutChar(CI->getOperand(2), B); if (CI->use_empty()) return CI; @@ -1543,7 +1543,7 @@ struct PrintFOpt : public LibCallOptimization { // printf("%s\n", str) --> puts(str) if (FormatStr == "%s\n" && CI->getNumOperands() > 2 && - isa<PointerType>(CI->getOperand(2)->getType()) && + CI->getOperand(2)->getType()->isPointerTy() && CI->use_empty()) { EmitPutS(CI->getOperand(2), B); return CI; @@ -1559,9 +1559,9 @@ struct SPrintFOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { // Require two fixed pointer arguments and an integer result. const FunctionType *FT = Callee->getFunctionType(); - if (FT->getNumParams() != 2 || !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1)) || - !isa<IntegerType>(FT->getReturnType())) + if (FT->getNumParams() != 2 || !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy() || + !FT->getReturnType()->isIntegerTy()) return 0; // Check for a fixed format string. @@ -1595,7 +1595,7 @@ struct SPrintFOpt : public LibCallOptimization { // Decode the second character of the format string. if (FormatStr[1] == 'c') { // sprintf(dst, "%c", chr) --> *(i8*)dst = chr; *((i8*)dst+1) = 0 - if (!isa<IntegerType>(CI->getOperand(3)->getType())) return 0; + if (!CI->getOperand(3)->getType()->isIntegerTy()) return 0; Value *V = B.CreateTrunc(CI->getOperand(3), Type::getInt8Ty(*Context), "char"); Value *Ptr = CastToCStr(CI->getOperand(1), B); @@ -1612,7 +1612,7 @@ struct SPrintFOpt : public LibCallOptimization { if (!TD) return 0; // sprintf(dest, "%s", str) -> llvm.memcpy(dest, str, strlen(str)+1, 1) - if (!isa<PointerType>(CI->getOperand(3)->getType())) return 0; + if (!CI->getOperand(3)->getType()->isPointerTy()) return 0; Value *Len = EmitStrLen(CI->getOperand(3), B); Value *IncLen = B.CreateAdd(Len, @@ -1634,11 +1634,11 @@ struct FWriteOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { // Require a pointer, an integer, an integer, a pointer, returning integer. const FunctionType *FT = Callee->getFunctionType(); - if (FT->getNumParams() != 4 || !isa<PointerType>(FT->getParamType(0)) || - !isa<IntegerType>(FT->getParamType(1)) || - !isa<IntegerType>(FT->getParamType(2)) || - !isa<PointerType>(FT->getParamType(3)) || - !isa<IntegerType>(FT->getReturnType())) + if (FT->getNumParams() != 4 || !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isIntegerTy() || + !FT->getParamType(2)->isIntegerTy() || + !FT->getParamType(3)->isPointerTy() || + !FT->getReturnType()->isIntegerTy()) return 0; // Get the element size and count. @@ -1672,8 +1672,8 @@ struct FPutsOpt : public LibCallOptimization { // Require two pointers. Also, we can't optimize if return value is used. const FunctionType *FT = Callee->getFunctionType(); - if (FT->getNumParams() != 2 || !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1)) || + if (FT->getNumParams() != 2 || !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy() || !CI->use_empty()) return 0; @@ -1694,9 +1694,9 @@ struct FPrintFOpt : public LibCallOptimization { virtual Value *CallOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) { // Require two fixed paramters as pointers and integer result. const FunctionType *FT = Callee->getFunctionType(); - if (FT->getNumParams() != 2 || !isa<PointerType>(FT->getParamType(0)) || - !isa<PointerType>(FT->getParamType(1)) || - !isa<IntegerType>(FT->getReturnType())) + if (FT->getNumParams() != 2 || !FT->getParamType(0)->isPointerTy() || + !FT->getParamType(1)->isPointerTy() || + !FT->getReturnType()->isIntegerTy()) return 0; // All the optimizations depend on the format string. @@ -1728,14 +1728,14 @@ struct FPrintFOpt : public LibCallOptimization { // Decode the second character of the format string. if (FormatStr[1] == 'c') { // fprintf(F, "%c", chr) --> *(i8*)dst = chr - if (!isa<IntegerType>(CI->getOperand(3)->getType())) return 0; + if (!CI->getOperand(3)->getType()->isIntegerTy()) return 0; EmitFPutC(CI->getOperand(3), CI->getOperand(1), B); return ConstantInt::get(CI->getType(), 1); } if (FormatStr[1] == 's') { // fprintf(F, "%s", str) -> fputs(str, F) - if (!isa<PointerType>(CI->getOperand(3)->getType()) || !CI->use_empty()) + if (!CI->getOperand(3)->getType()->isPointerTy() || !CI->use_empty()) return 0; EmitFPutS(CI->getOperand(3), CI->getOperand(1), B); return CI; @@ -2000,7 +2000,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 's': if (Name == "strlen") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setOnlyReadsMemory(F); setDoesNotThrow(F); @@ -2018,14 +2018,14 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "strncpy" || Name == "strtoull") { if (FTy->getNumParams() < 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); } else if (Name == "strxfrm") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2038,8 +2038,8 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "strcasecmp" || Name == "strncasecmp") { if (FTy->getNumParams() < 2 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setOnlyReadsMemory(F); setDoesNotThrow(F); @@ -2048,7 +2048,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { } else if (Name == "strstr" || Name == "strpbrk") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setOnlyReadsMemory(F); setDoesNotThrow(F); @@ -2056,7 +2056,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { } else if (Name == "strtok" || Name == "strtok_r") { if (FTy->getNumParams() < 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); @@ -2064,15 +2064,15 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "setbuf" || Name == "setvbuf") { if (FTy->getNumParams() < 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "strdup" || Name == "strndup") { if (FTy->getNumParams() < 1 || - !isa<PointerType>(FTy->getReturnType()) || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getReturnType()->isPointerTy() || + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); @@ -2082,31 +2082,31 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "sprintf" || Name == "statvfs") { if (FTy->getNumParams() < 2 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); setDoesNotCapture(F, 2); } else if (Name == "snprintf") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(2))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(2)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); setDoesNotCapture(F, 3); } else if (Name == "setitimer") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(1)) || - !isa<PointerType>(FTy->getParamType(2))) + !FTy->getParamType(1)->isPointerTy() || + !FTy->getParamType(2)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); setDoesNotCapture(F, 3); } else if (Name == "system") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; // May throw; "system" is a valid pthread cancellation point. setDoesNotCapture(F, 1); @@ -2115,14 +2115,14 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'm': if (Name == "malloc") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getReturnType())) + !FTy->getReturnType()->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); } else if (Name == "memcmp") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setOnlyReadsMemory(F); setDoesNotThrow(F); @@ -2141,18 +2141,18 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "memccpy" || Name == "memmove") { if (FTy->getNumParams() < 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); } else if (Name == "memalign") { - if (!isa<PointerType>(FTy->getReturnType())) + if (!FTy->getReturnType()->isPointerTy()) continue; setDoesNotAlias(F, 0); } else if (Name == "mkdir" || Name == "mktime") { if (FTy->getNumParams() == 0 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2161,15 +2161,15 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'r': if (Name == "realloc") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getReturnType())) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getReturnType()->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); setDoesNotCapture(F, 1); } else if (Name == "read") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; // May throw; "read" is a valid pthread cancellation point. setDoesNotCapture(F, 2); @@ -2178,15 +2178,15 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "remove" || Name == "realpath") { if (FTy->getNumParams() < 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "rename" || Name == "readlink") { if (FTy->getNumParams() < 2 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2196,7 +2196,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'w': if (Name == "write") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; // May throw; "write" is a valid pthread cancellation point. setDoesNotCapture(F, 2); @@ -2205,16 +2205,16 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'b': if (Name == "bcopy") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); setDoesNotCapture(F, 2); } else if (Name == "bcmp") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setOnlyReadsMemory(F); @@ -2222,7 +2222,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { setDoesNotCapture(F, 2); } else if (Name == "bzero") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2231,7 +2231,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'c': if (Name == "calloc") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getReturnType())) + !FTy->getReturnType()->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); @@ -2241,7 +2241,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "clearerr" || Name == "closedir") { if (FTy->getNumParams() == 0 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2253,14 +2253,14 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "atof" || Name == "atoll") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setOnlyReadsMemory(F); setDoesNotCapture(F, 1); } else if (Name == "access") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2269,9 +2269,9 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'f': if (Name == "fopen") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getReturnType()) || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getReturnType()->isPointerTy() || + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); @@ -2279,8 +2279,8 @@ bool SimplifyLibCalls::doInitialization(Module &M) { setDoesNotCapture(F, 2); } else if (Name == "fdopen") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getReturnType()) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getReturnType()->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); @@ -2300,13 +2300,13 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "funlockfile" || Name == "ftrylockfile") { if (FTy->getNumParams() == 0 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "ferror") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2318,22 +2318,22 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "frexpl" || Name == "fstatvfs") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); } else if (Name == "fgets") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(2))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(2)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 3); } else if (Name == "fread" || Name == "fwrite") { if (FTy->getNumParams() != 4 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(3))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(3)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2343,8 +2343,8 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "fprintf" || Name == "fgetpos") { if (FTy->getNumParams() < 2 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2356,13 +2356,13 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "getlogin_r" || Name == "getc_unlocked") { if (FTy->getNumParams() == 0 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "getenv") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setOnlyReadsMemory(F); @@ -2372,13 +2372,13 @@ bool SimplifyLibCalls::doInitialization(Module &M) { setDoesNotThrow(F); } else if (Name == "getitimer") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); } else if (Name == "getpwnam") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2387,7 +2387,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'u': if (Name == "ungetc") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); @@ -2395,15 +2395,15 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "unlink" || Name == "unsetenv") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "utime" || Name == "utimes") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2413,7 +2413,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'p': if (Name == "putc") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); @@ -2421,14 +2421,14 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "printf" || Name == "perror") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "pread" || Name == "pwrite") { if (FTy->getNumParams() != 4 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; // May throw; these are valid pthread cancellation points. setDoesNotCapture(F, 2); @@ -2436,9 +2436,9 @@ bool SimplifyLibCalls::doInitialization(Module &M) { setDoesNotThrow(F); } else if (Name == "popen") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getReturnType()) || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getReturnType()->isPointerTy() || + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); @@ -2446,7 +2446,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { setDoesNotCapture(F, 2); } else if (Name == "pclose") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2455,43 +2455,43 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'v': if (Name == "vscanf") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "vsscanf" || Name == "vfscanf") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(1)) || - !isa<PointerType>(FTy->getParamType(2))) + !FTy->getParamType(1)->isPointerTy() || + !FTy->getParamType(2)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); setDoesNotCapture(F, 2); } else if (Name == "valloc") { - if (!isa<PointerType>(FTy->getReturnType())) + if (!FTy->getReturnType()->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); } else if (Name == "vprintf") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "vfprintf" || Name == "vsprintf") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); setDoesNotCapture(F, 2); } else if (Name == "vsnprintf") { if (FTy->getNumParams() != 4 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(2))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(2)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2501,14 +2501,14 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'o': if (Name == "open") { if (FTy->getNumParams() < 2 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; // May throw; "open" is a valid pthread cancellation point. setDoesNotCapture(F, 1); } else if (Name == "opendir") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getReturnType()) || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getReturnType()->isPointerTy() || + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); @@ -2517,13 +2517,13 @@ bool SimplifyLibCalls::doInitialization(Module &M) { break; case 't': if (Name == "tmpfile") { - if (!isa<PointerType>(FTy->getReturnType())) + if (!FTy->getReturnType()->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); } else if (Name == "times") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2546,15 +2546,15 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'l': if (Name == "lstat") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); setDoesNotCapture(F, 2); } else if (Name == "lchown") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2563,7 +2563,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 'q': if (Name == "qsort") { if (FTy->getNumParams() != 4 || - !isa<PointerType>(FTy->getParamType(3))) + !FTy->getParamType(3)->isPointerTy()) continue; // May throw; places call through function pointer. setDoesNotCapture(F, 4); @@ -2573,27 +2573,27 @@ bool SimplifyLibCalls::doInitialization(Module &M) { if (Name == "__strdup" || Name == "__strndup") { if (FTy->getNumParams() < 1 || - !isa<PointerType>(FTy->getReturnType()) || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getReturnType()->isPointerTy() || + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); setDoesNotCapture(F, 1); } else if (Name == "__strtok_r") { if (FTy->getNumParams() != 3 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); } else if (Name == "_IO_getc") { if (FTy->getNumParams() != 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "_IO_putc") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); @@ -2602,7 +2602,7 @@ bool SimplifyLibCalls::doInitialization(Module &M) { case 1: if (Name == "\1__isoc99_scanf") { if (FTy->getNumParams() < 1 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); @@ -2611,17 +2611,17 @@ bool SimplifyLibCalls::doInitialization(Module &M) { Name == "\1statvfs64" || Name == "\1__isoc99_sscanf") { if (FTy->getNumParams() < 1 || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); setDoesNotCapture(F, 2); } else if (Name == "\1fopen64") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getReturnType()) || - !isa<PointerType>(FTy->getParamType(0)) || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getReturnType()->isPointerTy() || + !FTy->getParamType(0)->isPointerTy() || + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); @@ -2630,25 +2630,25 @@ bool SimplifyLibCalls::doInitialization(Module &M) { } else if (Name == "\1fseeko64" || Name == "\1ftello64") { if (FTy->getNumParams() == 0 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 1); } else if (Name == "\1tmpfile64") { - if (!isa<PointerType>(FTy->getReturnType())) + if (!FTy->getReturnType()->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotAlias(F, 0); } else if (Name == "\1fstat64" || Name == "\1fstatvfs64") { if (FTy->getNumParams() != 2 || - !isa<PointerType>(FTy->getParamType(1))) + !FTy->getParamType(1)->isPointerTy()) continue; setDoesNotThrow(F); setDoesNotCapture(F, 2); } else if (Name == "\1open64") { if (FTy->getNumParams() < 2 || - !isa<PointerType>(FTy->getParamType(0))) + !FTy->getParamType(0)->isPointerTy()) continue; // May throw; "open" is a valid pthread cancellation point. setDoesNotCapture(F, 1); diff --git a/lib/Transforms/Utils/AddrModeMatcher.cpp b/lib/Transforms/Utils/AddrModeMatcher.cpp index 8c4aa59..be6b383 100644 --- a/lib/Transforms/Utils/AddrModeMatcher.cpp +++ b/lib/Transforms/Utils/AddrModeMatcher.cpp @@ -125,7 +125,7 @@ static bool MightBeFoldableInst(Instruction *I) { // Don't touch identity bitcasts. if (I->getType() == I->getOperand(0)->getType()) return false; - return isa<PointerType>(I->getType()) || isa<IntegerType>(I->getType()); + return I->getType()->isPointerTy() || I->getType()->isIntegerTy(); case Instruction::PtrToInt: // PtrToInt is always a noop, as we know that the int type is pointer sized. return true; @@ -167,8 +167,8 @@ bool AddressingModeMatcher::MatchOperationAddr(User *AddrInst, unsigned Opcode, case Instruction::BitCast: // BitCast is always a noop, and we can handle it as long as it is // int->int or pointer->pointer (we don't want int<->fp or something). - if ((isa<PointerType>(AddrInst->getOperand(0)->getType()) || - isa<IntegerType>(AddrInst->getOperand(0)->getType())) && + if ((AddrInst->getOperand(0)->getType()->isPointerTy() || + AddrInst->getOperand(0)->getType()->isIntegerTy()) && // Don't touch identity bitcasts. These were probably put here by LSR, // and we don't want to mess around with them. Assume it knows what it // is doing. @@ -569,7 +569,7 @@ IsProfitableToFoldIntoAddressingMode(Instruction *I, ExtAddrMode &AMBefore, // Get the access type of this use. If the use isn't a pointer, we don't // know what it accesses. Value *Address = User->getOperand(OpNo); - if (!isa<PointerType>(Address->getType())) + if (!Address->getType()->isPointerTy()) return false; const Type *AddressAccessTy = cast<PointerType>(Address->getType())->getElementType(); diff --git a/lib/Transforms/Utils/BasicBlockUtils.cpp b/lib/Transforms/Utils/BasicBlockUtils.cpp index 7bc4fcd..1f62dab 100644 --- a/lib/Transforms/Utils/BasicBlockUtils.cpp +++ b/lib/Transforms/Utils/BasicBlockUtils.cpp @@ -274,24 +274,31 @@ void llvm::RemoveSuccessor(TerminatorInst *TI, unsigned SuccNum) { ReplaceInstWithInst(TI, NewTI); } -/// SplitEdge - Split the edge connecting specified block. Pass P must -/// not be NULL. -BasicBlock *llvm::SplitEdge(BasicBlock *BB, BasicBlock *Succ, Pass *P) { - TerminatorInst *LatchTerm = BB->getTerminator(); - unsigned SuccNum = 0; +/// GetSuccessorNumber - Search for the specified successor of basic block BB +/// and return its position in the terminator instruction's list of +/// successors. It is an error to call this with a block that is not a +/// successor. +unsigned llvm::GetSuccessorNumber(BasicBlock *BB, BasicBlock *Succ) { + TerminatorInst *Term = BB->getTerminator(); #ifndef NDEBUG - unsigned e = LatchTerm->getNumSuccessors(); + unsigned e = Term->getNumSuccessors(); #endif for (unsigned i = 0; ; ++i) { assert(i != e && "Didn't find edge?"); - if (LatchTerm->getSuccessor(i) == Succ) { - SuccNum = i; - break; - } + if (Term->getSuccessor(i) == Succ) + return i; } + return 0; +} + +/// SplitEdge - Split the edge connecting specified block. Pass P must +/// not be NULL. +BasicBlock *llvm::SplitEdge(BasicBlock *BB, BasicBlock *Succ, Pass *P) { + unsigned SuccNum = GetSuccessorNumber(BB, Succ); // If this is a critical edge, let SplitCriticalEdge do it. - if (SplitCriticalEdge(BB->getTerminator(), SuccNum, P)) + TerminatorInst *LatchTerm = BB->getTerminator(); + if (SplitCriticalEdge(LatchTerm, SuccNum, P)) return LatchTerm->getSuccessor(SuccNum); // If the edge isn't critical, then BB has a single successor or Succ has a diff --git a/lib/Transforms/Utils/Local.cpp b/lib/Transforms/Utils/Local.cpp index 57ad459..d03f7a6 100644 --- a/lib/Transforms/Utils/Local.cpp +++ b/lib/Transforms/Utils/Local.cpp @@ -46,7 +46,7 @@ using namespace llvm; static Value *getUnderlyingObjectWithOffset(Value *V, const TargetData *TD, uint64_t &ByteOffset, unsigned MaxLookup = 6) { - if (!isa<PointerType>(V->getType())) + if (!V->getType()->isPointerTy()) return V; for (unsigned Count = 0; MaxLookup == 0 || Count < MaxLookup; ++Count) { if (GEPOperator *GEP = dyn_cast<GEPOperator>(V)) { @@ -65,7 +65,7 @@ static Value *getUnderlyingObjectWithOffset(Value *V, const TargetData *TD, } else { return V; } - assert(isa<PointerType>(V->getType()) && "Unexpected operand type!"); + assert(V->getType()->isPointerTy() && "Unexpected operand type!"); } return V; } diff --git a/lib/Transforms/Utils/LoopSimplify.cpp b/lib/Transforms/Utils/LoopSimplify.cpp index 57bab60..924b744 100644 --- a/lib/Transforms/Utils/LoopSimplify.cpp +++ b/lib/Transforms/Utils/LoopSimplify.cpp @@ -51,6 +51,7 @@ #include "llvm/Transforms/Utils/BasicBlockUtils.h" #include "llvm/Transforms/Utils/Local.h" #include "llvm/Support/CFG.h" +#include "llvm/Support/Debug.h" #include "llvm/ADT/SetOperations.h" #include "llvm/ADT/SetVector.h" #include "llvm/ADT/Statistic.h" @@ -147,6 +148,11 @@ ReprocessLoop: // Delete each unique out-of-loop (and thus dead) predecessor. for (SmallPtrSet<BasicBlock *, 4>::iterator I = BadPreds.begin(), E = BadPreds.end(); I != E; ++I) { + + DEBUG(dbgs() << "LoopSimplify: Deleting edge from dead predecessor "; + WriteAsOperand(dbgs(), *I, false); + dbgs() << "\n"); + // Inform each successor of each dead pred. for (succ_iterator SI = succ_begin(*I), SE = succ_end(*I); SI != SE; ++SI) (*SI)->removePredecessor(*I); @@ -159,6 +165,27 @@ ReprocessLoop: } } + // If there are exiting blocks with branches on undef, resolve the undef in + // the direction which will exit the loop. This will help simplify loop + // trip count computations. + SmallVector<BasicBlock*, 8> ExitingBlocks; + L->getExitingBlocks(ExitingBlocks); + for (SmallVectorImpl<BasicBlock *>::iterator I = ExitingBlocks.begin(), + E = ExitingBlocks.end(); I != E; ++I) + if (BranchInst *BI = dyn_cast<BranchInst>((*I)->getTerminator())) + if (BI->isConditional()) { + if (UndefValue *Cond = dyn_cast<UndefValue>(BI->getCondition())) { + + DEBUG(dbgs() << "LoopSimplify: Resolving \"br i1 undef\" to exit in "; + WriteAsOperand(dbgs(), *I, false); + dbgs() << "\n"); + + BI->setCondition(ConstantInt::get(Cond->getType(), + !L->contains(BI->getSuccessor(0)))); + Changed = true; + } + } + // Does the loop already have a preheader? If so, don't insert one. BasicBlock *Preheader = L->getLoopPreheader(); if (!Preheader) { @@ -250,8 +277,6 @@ ReprocessLoop: break; } if (UniqueExit) { - SmallVector<BasicBlock*, 8> ExitingBlocks; - L->getExitingBlocks(ExitingBlocks); for (unsigned i = 0, e = ExitingBlocks.size(); i != e; ++i) { BasicBlock *ExitingBlock = ExitingBlocks[i]; if (!ExitingBlock->getSinglePredecessor()) continue; @@ -282,6 +307,11 @@ ReprocessLoop: // Success. The block is now dead, so remove it from the loop, // update the dominator tree and dominance frontier, and delete it. + + DEBUG(dbgs() << "LoopSimplify: Eliminating exiting block "; + WriteAsOperand(dbgs(), ExitingBlock, false); + dbgs() << "\n"); + assert(pred_begin(ExitingBlock) == pred_end(ExitingBlock)); Changed = true; LI->removeBlock(ExitingBlock); @@ -335,6 +365,10 @@ BasicBlock *LoopSimplify::InsertPreheaderForLoop(Loop *L) { SplitBlockPredecessors(Header, &OutsideBlocks[0], OutsideBlocks.size(), ".preheader", this); + DEBUG(dbgs() << "LoopSimplify: Creating pre-header "; + WriteAsOperand(dbgs(), NewBB, false); + dbgs() << "\n"); + // Make sure that NewBB is put someplace intelligent, which doesn't mess up // code layout too horribly. PlaceSplitBlockCarefully(NewBB, OutsideBlocks, L); @@ -360,6 +394,10 @@ BasicBlock *LoopSimplify::RewriteLoopExitBlock(Loop *L, BasicBlock *Exit) { LoopBlocks.size(), ".loopexit", this); + DEBUG(dbgs() << "LoopSimplify: Creating dedicated exit block "; + WriteAsOperand(dbgs(), NewBB, false); + dbgs() << "\n"); + return NewBB; } @@ -480,6 +518,8 @@ Loop *LoopSimplify::SeparateNestedLoop(Loop *L, LPPassManager &LPM) { OuterLoopPreds.push_back(PN->getIncomingBlock(i)); } + DEBUG(dbgs() << "LoopSimplify: Splitting out a new outer loop\n"); + BasicBlock *Header = L->getHeader(); BasicBlock *NewBB = SplitBlockPredecessors(Header, &OuterLoopPreds[0], OuterLoopPreds.size(), @@ -574,6 +614,10 @@ LoopSimplify::InsertUniqueBackedgeBlock(Loop *L, BasicBlock *Preheader) { Header->getName()+".backedge", F); BranchInst *BETerminator = BranchInst::Create(Header, BEBlock); + DEBUG(dbgs() << "LoopSimplify: Inserting unique backedge block "; + WriteAsOperand(dbgs(), BEBlock, false); + dbgs() << "\n"); + // Move the new backedge block to right after the last backedge block. Function::iterator InsertPos = BackedgeBlocks.back(); ++InsertPos; F->getBasicBlockList().splice(InsertPos, F->getBasicBlockList(), BEBlock); diff --git a/lib/Transforms/Utils/PromoteMemoryToRegister.cpp b/lib/Transforms/Utils/PromoteMemoryToRegister.cpp index 544e20b..4f5a70b 100644 --- a/lib/Transforms/Utils/PromoteMemoryToRegister.cpp +++ b/lib/Transforms/Utils/PromoteMemoryToRegister.cpp @@ -518,7 +518,7 @@ void PromoteMem2Reg::run() { // If this PHI node merges one value and/or undefs, get the value. if (Value *V = PN->hasConstantValue(&DT)) { - if (AST && isa<PointerType>(PN->getType())) + if (AST && PN->getType()->isPointerTy()) AST->deleteValue(PN); PN->replaceAllUsesWith(V); PN->eraseFromParent(); @@ -780,7 +780,7 @@ void PromoteMem2Reg::RewriteSingleStoreAlloca(AllocaInst *AI, if (ReplVal == LI) ReplVal = UndefValue::get(LI->getType()); LI->replaceAllUsesWith(ReplVal); - if (AST && isa<PointerType>(LI->getType())) + if (AST && LI->getType()->isPointerTy()) AST->deleteValue(LI); LI->eraseFromParent(); LBI.deleteValue(LI); @@ -838,7 +838,7 @@ void PromoteMem2Reg::PromoteSingleBlockAlloca(AllocaInst *AI, AllocaInfo &Info, for (Value::use_iterator UI = AI->use_begin(), E = AI->use_end(); UI != E;) if (LoadInst *LI = dyn_cast<LoadInst>(*UI++)) { LI->replaceAllUsesWith(UndefValue::get(LI->getType())); - if (AST && isa<PointerType>(LI->getType())) + if (AST && LI->getType()->isPointerTy()) AST->deleteValue(LI); LBI.deleteValue(LI); LI->eraseFromParent(); @@ -874,7 +874,7 @@ void PromoteMem2Reg::PromoteSingleBlockAlloca(AllocaInst *AI, AllocaInfo &Info, // Otherwise, there was a store before this load, the load takes its value. --I; LI->replaceAllUsesWith(I->second->getOperand(0)); - if (AST && isa<PointerType>(LI->getType())) + if (AST && LI->getType()->isPointerTy()) AST->deleteValue(LI); LI->eraseFromParent(); LBI.deleteValue(LI); @@ -922,7 +922,7 @@ bool PromoteMem2Reg::QueuePhiNode(BasicBlock *BB, unsigned AllocaNo, InsertedPHINodes.insert(PN); - if (AST && isa<PointerType>(PN->getType())) + if (AST && PN->getType()->isPointerTy()) AST->copyValue(PointerAllocaValues[AllocaNo], PN); return true; @@ -996,7 +996,7 @@ NextIteration: // Anything using the load now uses the current value. LI->replaceAllUsesWith(V); - if (AST && isa<PointerType>(LI->getType())) + if (AST && LI->getType()->isPointerTy()) AST->deleteValue(LI); BB->getInstList().erase(LI); } else if (StoreInst *SI = dyn_cast<StoreInst>(I)) { diff --git a/lib/Transforms/Utils/SimplifyCFG.cpp b/lib/Transforms/Utils/SimplifyCFG.cpp index 2215059..f343c38 100644 --- a/lib/Transforms/Utils/SimplifyCFG.cpp +++ b/lib/Transforms/Utils/SimplifyCFG.cpp @@ -271,7 +271,7 @@ static bool DominatesMergePoint(Value *V, BasicBlock *BB, ConstantInt *SimplifyCFGOpt::GetConstantInt(Value *V) { // Normal constant int. ConstantInt *CI = dyn_cast<ConstantInt>(V); - if (CI || !TD || !isa<Constant>(V) || !isa<PointerType>(V->getType())) + if (CI || !TD || !isa<Constant>(V) || !V->getType()->isPointerTy()) return CI; // This is some kind of pointer constant. Turn it into a pointer-sized @@ -701,7 +701,7 @@ bool SimplifyCFGOpt::FoldValueComparisonIntoPredecessors(TerminatorInst *TI) { AddPredecessorToBlock(NewSuccessors[i], Pred, BB); // Convert pointer to int before we switch. - if (isa<PointerType>(CV->getType())) { + if (CV->getType()->isPointerTy()) { assert(TD && "Cannot switch on pointer without TargetData"); CV = new PtrToIntInst(CV, TD->getIntPtrType(CV->getContext()), "magicptr", PTI); @@ -915,7 +915,7 @@ static bool SpeculativelyExecuteBB(BranchInst *BI, BasicBlock *BB1) { case Instruction::Add: case Instruction::Sub: // Not worth doing for vector ops. - if (isa<VectorType>(HInst->getType())) + if (HInst->getType()->isVectorTy()) return false; break; case Instruction::And: @@ -925,7 +925,7 @@ static bool SpeculativelyExecuteBB(BranchInst *BI, BasicBlock *BB1) { case Instruction::LShr: case Instruction::AShr: // Don't mess with vector operations. - if (isa<VectorType>(HInst->getType())) + if (HInst->getType()->isVectorTy()) return false; break; // These are all cheap and non-trapping instructions. } @@ -2068,7 +2068,7 @@ bool SimplifyCFGOpt::run(BasicBlock *BB) { if (!TrueWhenEqual) std::swap(DefaultBB, EdgeBB); // Convert pointer to int before we switch. - if (isa<PointerType>(CompVal->getType())) { + if (CompVal->getType()->isPointerTy()) { assert(TD && "Cannot switch on pointer without TargetData"); CompVal = new PtrToIntInst(CompVal, TD->getIntPtrType(CompVal->getContext()), diff --git a/lib/VMCore/AsmWriter.cpp b/lib/VMCore/AsmWriter.cpp index f5ba7e7..fd74241 100644 --- a/lib/VMCore/AsmWriter.cpp +++ b/lib/VMCore/AsmWriter.cpp @@ -17,6 +17,7 @@ #include "llvm/Assembly/Writer.h" #include "llvm/Assembly/PrintModulePass.h" #include "llvm/Assembly/AsmAnnotationWriter.h" +#include "llvm/LLVMContext.h" #include "llvm/CallingConv.h" #include "llvm/Constants.h" #include "llvm/DerivedTypes.h" @@ -376,8 +377,8 @@ namespace { return; // If this is a structure or opaque type, add a name for the type. - if (((isa<StructType>(Ty) && cast<StructType>(Ty)->getNumElements()) - || isa<OpaqueType>(Ty)) && !TP.hasTypeName(Ty)) { + if (((Ty->isStructTy() && cast<StructType>(Ty)->getNumElements()) + || Ty->isOpaqueTy()) && !TP.hasTypeName(Ty)) { TP.addTypeName(Ty, "%"+utostr(unsigned(NumberedTypes.size()))); NumberedTypes.push_back(Ty); } @@ -432,7 +433,7 @@ static void AddModuleTypesToPrinter(TypePrinting &TP, if (const PointerType *PTy = dyn_cast<PointerType>(Ty)) { const Type *PETy = PTy->getElementType(); if ((PETy->isPrimitiveType() || PETy->isIntegerTy()) && - !isa<OpaqueType>(PETy)) + !PETy->isOpaqueTy()) continue; } @@ -1236,7 +1237,6 @@ class AssemblyWriter { TypePrinting TypePrinter; AssemblyAnnotationWriter *AnnotationWriter; std::vector<const Type*> NumberedTypes; - SmallVector<StringRef, 8> MDNames; public: inline AssemblyWriter(formatted_raw_ostream &o, SlotTracker &Mac, @@ -1244,8 +1244,6 @@ public: AssemblyAnnotationWriter *AAW) : Out(o), Machine(Mac), TheModule(M), AnnotationWriter(AAW) { AddModuleTypesToPrinter(TypePrinter, NumberedTypes, M); - if (M) - M->getMDKindNames(MDNames); } void printMDNodeBody(const MDNode *MD); @@ -1850,8 +1848,8 @@ void AssemblyWriter::printInstruction(const Instruction &I) { // Out << ' '; if (!FTy->isVarArg() && - (!isa<PointerType>(RetTy) || - !isa<FunctionType>(cast<PointerType>(RetTy)->getElementType()))) { + (!RetTy->isPointerTy() || + !cast<PointerType>(RetTy)->getElementType()->isFunctionTy())) { TypePrinter.print(RetTy, Out); Out << ' '; writeOperand(Operand, false); @@ -1896,8 +1894,8 @@ void AssemblyWriter::printInstruction(const Instruction &I) { // Out << ' '; if (!FTy->isVarArg() && - (!isa<PointerType>(RetTy) || - !isa<FunctionType>(cast<PointerType>(RetTy)->getElementType()))) { + (!RetTy->isPointerTy() || + !cast<PointerType>(RetTy)->getElementType()->isFunctionTy())) { TypePrinter.print(RetTy, Out); Out << ' '; writeOperand(Operand, false); @@ -1988,12 +1986,20 @@ void AssemblyWriter::printInstruction(const Instruction &I) { } // Print Metadata info. - if (!MDNames.empty()) { - SmallVector<std::pair<unsigned, MDNode*>, 4> InstMD; - I.getAllMetadata(InstMD); - for (unsigned i = 0, e = InstMD.size(); i != e; ++i) - Out << ", !" << MDNames[InstMD[i].first] - << " !" << Machine.getMetadataSlot(InstMD[i].second); + SmallVector<std::pair<unsigned, MDNode*>, 4> InstMD; + I.getAllMetadata(InstMD); + if (!InstMD.empty()) { + SmallVector<StringRef, 8> MDNames; + I.getType()->getContext().getMDKindNames(MDNames); + for (unsigned i = 0, e = InstMD.size(); i != e; ++i) { + unsigned Kind = InstMD[i].first; + if (Kind < MDNames.size()) { + Out << ", !" << MDNames[Kind]; + } else { + Out << ", !<unknown kind #" << Kind << ">"; + } + Out << " !" << Machine.getMetadataSlot(InstMD[i].second); + } } printInfoComment(I); } diff --git a/lib/VMCore/Attributes.cpp b/lib/VMCore/Attributes.cpp index ff0cc9b..a000aee 100644 --- a/lib/VMCore/Attributes.cpp +++ b/lib/VMCore/Attributes.cpp @@ -93,7 +93,7 @@ Attributes Attribute::typeIncompatible(const Type *Ty) { // Attributes that only apply to integers. Incompatible |= SExt | ZExt; - if (!isa<PointerType>(Ty)) + if (!Ty->isPointerTy()) // Attributes that only apply to pointers. Incompatible |= ByVal | Nest | NoAlias | StructRet | NoCapture; diff --git a/lib/VMCore/ConstantFold.cpp b/lib/VMCore/ConstantFold.cpp index 78a45e8..194a6d4 100644 --- a/lib/VMCore/ConstantFold.cpp +++ b/lib/VMCore/ConstantFold.cpp @@ -112,7 +112,7 @@ static Constant *FoldBitCast(Constant *V, const Type *DestTy) { IdxList.push_back(Zero); } else if (const SequentialType *STy = dyn_cast<SequentialType>(ElTy)) { - if (isa<PointerType>(ElTy)) break; // Can't index into pointers! + if (ElTy->isPointerTy()) break; // Can't index into pointers! ElTy = STy->getElementType(); IdxList.push_back(Zero); } else { @@ -189,7 +189,7 @@ static Constant *FoldBitCast(Constant *V, const Type *DestTy) { /// static Constant *ExtractConstantBytes(Constant *C, unsigned ByteStart, unsigned ByteSize) { - assert(isa<IntegerType>(C->getType()) && + assert(C->getType()->isIntegerTy() && (cast<IntegerType>(C->getType())->getBitWidth() & 7) == 0 && "Non-byte sized integer input"); unsigned CSize = cast<IntegerType>(C->getType())->getBitWidth()/8; @@ -334,11 +334,7 @@ static Constant *getFoldedSizeOf(const Type *Ty, const Type *DestTy, Constant *E = getFoldedSizeOf(ATy->getElementType(), DestTy, true); return ConstantExpr::getNUWMul(E, N); } - if (const VectorType *VTy = dyn_cast<VectorType>(Ty)) { - Constant *N = ConstantInt::get(DestTy, VTy->getNumElements()); - Constant *E = getFoldedSizeOf(VTy->getElementType(), DestTy, true); - return ConstantExpr::getNUWMul(E, N); - } + if (const StructType *STy = dyn_cast<StructType>(Ty)) if (!STy->isPacked()) { unsigned NumElems = STy->getNumElements(); @@ -361,6 +357,22 @@ static Constant *getFoldedSizeOf(const Type *Ty, const Type *DestTy, } } + if (const UnionType *UTy = dyn_cast<UnionType>(Ty)) { + unsigned NumElems = UTy->getNumElements(); + // Check for a union with all members having the same size. + Constant *MemberSize = + getFoldedSizeOf(UTy->getElementType(0), DestTy, true); + bool AllSame = true; + for (unsigned i = 1; i != NumElems; ++i) + if (MemberSize != + getFoldedSizeOf(UTy->getElementType(i), DestTy, true)) { + AllSame = false; + break; + } + if (AllSame) + return MemberSize; + } + // Pointer size doesn't depend on the pointee type, so canonicalize them // to an arbitrary pointee. if (const PointerType *PTy = dyn_cast<PointerType>(Ty)) @@ -426,6 +438,24 @@ static Constant *getFoldedAlignOf(const Type *Ty, const Type *DestTy, return MemberAlign; } + if (const UnionType *UTy = dyn_cast<UnionType>(Ty)) { + // Union alignment is the maximum alignment of any member. + // Without target data, we can't compare much, but we can check to see + // if all the members have the same alignment. + unsigned NumElems = UTy->getNumElements(); + // Check for a union with all members having the same alignment. + Constant *MemberAlign = + getFoldedAlignOf(UTy->getElementType(0), DestTy, true); + bool AllSame = true; + for (unsigned i = 1; i != NumElems; ++i) + if (MemberAlign != getFoldedAlignOf(UTy->getElementType(i), DestTy, true)) { + AllSame = false; + break; + } + if (AllSame) + return MemberAlign; + } + // Pointer alignment doesn't depend on the pointee type, so canonicalize them // to an arbitrary pointee. if (const PointerType *PTy = dyn_cast<PointerType>(Ty)) @@ -464,13 +494,7 @@ static Constant *getFoldedOffsetOf(const Type *Ty, Constant *FieldNo, Constant *E = getFoldedSizeOf(ATy->getElementType(), DestTy, true); return ConstantExpr::getNUWMul(E, N); } - if (const VectorType *VTy = dyn_cast<VectorType>(Ty)) { - Constant *N = ConstantExpr::getCast(CastInst::getCastOpcode(FieldNo, false, - DestTy, false), - FieldNo, DestTy); - Constant *E = getFoldedSizeOf(VTy->getElementType(), DestTy, true); - return ConstantExpr::getNUWMul(E, N); - } + if (const StructType *STy = dyn_cast<StructType>(Ty)) if (!STy->isPacked()) { unsigned NumElems = STy->getNumElements(); @@ -551,7 +575,7 @@ Constant *llvm::ConstantFoldCastInstruction(unsigned opc, Constant *V, // operating on each element. In the cast of bitcasts, the element // count may be mismatched; don't attempt to handle that here. if (ConstantVector *CV = dyn_cast<ConstantVector>(V)) - if (isa<VectorType>(DestTy) && + if (DestTy->isVectorTy() && cast<VectorType>(DestTy)->getNumElements() == CV->getType()->getNumElements()) { std::vector<Constant*> res; @@ -634,7 +658,7 @@ Constant *llvm::ConstantFoldCastInstruction(unsigned opc, Constant *V, } } // Handle an offsetof-like expression. - if (isa<StructType>(Ty) || isa<ArrayType>(Ty) || isa<VectorType>(Ty)){ + if (Ty->isStructTy() || Ty->isArrayTy() || Ty->isVectorTy()){ if (Constant *C = getFoldedOffsetOf(Ty, CE->getOperand(2), DestTy, false)) return C; @@ -885,7 +909,7 @@ Constant *llvm::ConstantFoldInsertValueInstruction(Constant *Agg, unsigned numOps; if (const ArrayType *AR = dyn_cast<ArrayType>(AggTy)) numOps = AR->getNumElements(); - else if (isa<UnionType>(AggTy)) + else if (AggTy->isUnionTy()) numOps = 1; else numOps = cast<StructType>(AggTy)->getNumElements(); @@ -1105,6 +1129,10 @@ Constant *llvm::ConstantFoldBinaryInstruction(unsigned Opcode, return ConstantExpr::getLShr(C1, C2); break; } + } else if (isa<ConstantInt>(C1)) { + // If C1 is a ConstantInt and C2 is not, swap the operands. + if (Instruction::isCommutative(Opcode)) + return ConstantExpr::get(Opcode, C2, C1); } // At this point we know neither constant is an UndefValue. @@ -1364,31 +1392,8 @@ Constant *llvm::ConstantFoldBinaryInstruction(unsigned Opcode, } else if (isa<ConstantExpr>(C2)) { // If C2 is a constant expr and C1 isn't, flop them around and fold the // other way if possible. - switch (Opcode) { - case Instruction::Add: - case Instruction::FAdd: - case Instruction::Mul: - case Instruction::FMul: - case Instruction::And: - case Instruction::Or: - case Instruction::Xor: - // No change of opcode required. + if (Instruction::isCommutative(Opcode)) return ConstantFoldBinaryInstruction(Opcode, C2, C1); - - case Instruction::Shl: - case Instruction::LShr: - case Instruction::AShr: - case Instruction::Sub: - case Instruction::FSub: - case Instruction::SDiv: - case Instruction::UDiv: - case Instruction::FDiv: - case Instruction::URem: - case Instruction::SRem: - case Instruction::FRem: - default: // These instructions cannot be flopped around. - break; - } } // i1 can be simplified in many cases. @@ -1427,7 +1432,7 @@ Constant *llvm::ConstantFoldBinaryInstruction(unsigned Opcode, /// isZeroSizedType - This type is zero sized if its an array or structure of /// zero sized types. The only leaf zero sized type is an empty structure. static bool isMaybeZeroSizedType(const Type *Ty) { - if (isa<OpaqueType>(Ty)) return true; // Can't say. + if (Ty->isOpaqueTy()) return true; // Can't say. if (const StructType *STy = dyn_cast<StructType>(Ty)) { // If all of elements have zero size, this does too. @@ -1667,7 +1672,7 @@ static ICmpInst::Predicate evaluateICmpRelation(Constant *V1, Constant *V2, // If the cast is not actually changing bits, and the second operand is a // null pointer, do the comparison with the pre-casted value. if (V2->isNullValue() && - (isa<PointerType>(CE1->getType()) || CE1->getType()->isIntegerTy())) { + (CE1->getType()->isPointerTy() || CE1->getType()->isIntegerTy())) { if (CE1->getOpcode() == Instruction::ZExt) isSigned = false; if (CE1->getOpcode() == Instruction::SExt) isSigned = true; return evaluateICmpRelation(CE1Op0, @@ -1914,7 +1919,7 @@ Constant *llvm::ConstantFoldCompareInstruction(unsigned short pred, return ConstantInt::get(ResultTy, R==APFloat::cmpGreaterThan || R==APFloat::cmpEqual); } - } else if (isa<VectorType>(C1->getType())) { + } else if (C1->getType()->isVectorTy()) { SmallVector<Constant*, 16> C1Elts, C2Elts; C1->getVectorElements(C1Elts); C2->getVectorElements(C2Elts); @@ -2065,7 +2070,7 @@ Constant *llvm::ConstantFoldCompareInstruction(unsigned short pred, if (ConstantExpr *CE2 = dyn_cast<ConstantExpr>(C2)) { Constant *CE2Op0 = CE2->getOperand(0); if (CE2->getOpcode() == Instruction::BitCast && - isa<VectorType>(CE2->getType())==isa<VectorType>(CE2Op0->getType())) { + CE2->getType()->isVectorTy()==CE2Op0->getType()->isVectorTy()) { Constant *Inverse = ConstantExpr::getBitCast(C1, CE2Op0->getType()); return ConstantExpr::getICmp(pred, Inverse, CE2Op0); } @@ -2184,7 +2189,7 @@ Constant *llvm::ConstantFoldGetElementPtr(Constant *C, I != E; ++I) LastTy = *I; - if ((LastTy && isa<ArrayType>(LastTy)) || Idx0->isNullValue()) { + if ((LastTy && LastTy->isArrayTy()) || Idx0->isNullValue()) { SmallVector<Value*, 16> NewIndices; NewIndices.reserve(NumIdx + CE->getNumOperands()); for (unsigned i = 1, e = CE->getNumOperands()-1; i != e; ++i) diff --git a/lib/VMCore/Constants.cpp b/lib/VMCore/Constants.cpp index 98040ea..10f8879 100644 --- a/lib/VMCore/Constants.cpp +++ b/lib/VMCore/Constants.cpp @@ -229,7 +229,7 @@ Constant::PossibleRelocationsTy Constant::getRelocationInfo() const { /// This handles breaking down a vector undef into undef elements, etc. For /// constant exprs and other cases we can't handle, we return an empty vector. void Constant::getVectorElements(SmallVectorImpl<Constant*> &Elts) const { - assert(isa<VectorType>(getType()) && "Not a vector constant!"); + assert(getType()->isVectorTy() && "Not a vector constant!"); if (const ConstantVector *CV = dyn_cast<ConstantVector>(this)) { for (unsigned i = 0, e = CV->getNumOperands(); i != e; ++i) @@ -944,7 +944,7 @@ bool ConstantFP::isValueValidForType(const Type *Ty, const APFloat& Val) { // Factory Function Implementation ConstantAggregateZero* ConstantAggregateZero::get(const Type* Ty) { - assert((isa<StructType>(Ty) || isa<ArrayType>(Ty) || isa<VectorType>(Ty)) && + assert((Ty->isStructTy() || Ty->isArrayTy() || Ty->isVectorTy()) && "Cannot create an aggregate zero of non-aggregate type!"); LLVMContextImpl *pImpl = Ty->getContext().pImpl; @@ -1239,8 +1239,8 @@ Constant *ConstantExpr::getTruncOrBitCast(Constant *C, const Type *Ty) { } Constant *ConstantExpr::getPointerCast(Constant *S, const Type *Ty) { - assert(isa<PointerType>(S->getType()) && "Invalid cast"); - assert((Ty->isIntegerTy() || isa<PointerType>(Ty)) && "Invalid cast"); + assert(S->getType()->isPointerTy() && "Invalid cast"); + assert((Ty->isIntegerTy() || Ty->isPointerTy()) && "Invalid cast"); if (Ty->isIntegerTy()) return getCast(Instruction::PtrToInt, S, Ty); @@ -1383,14 +1383,14 @@ Constant *ConstantExpr::getFPToSI(Constant *C, const Type *Ty) { } Constant *ConstantExpr::getPtrToInt(Constant *C, const Type *DstTy) { - assert(isa<PointerType>(C->getType()) && "PtrToInt source must be pointer"); + assert(C->getType()->isPointerTy() && "PtrToInt source must be pointer"); assert(DstTy->isIntegerTy() && "PtrToInt destination must be integral"); return getFoldedCast(Instruction::PtrToInt, C, DstTy); } Constant *ConstantExpr::getIntToPtr(Constant *C, const Type *DstTy) { assert(C->getType()->isIntegerTy() && "IntToPtr source must be integral"); - assert(isa<PointerType>(DstTy) && "IntToPtr destination must be a pointer"); + assert(DstTy->isPointerTy() && "IntToPtr destination must be a pointer"); return getFoldedCast(Instruction::IntToPtr, C, DstTy); } @@ -1592,7 +1592,7 @@ Constant *ConstantExpr::getGetElementPtrTy(const Type *ReqTy, Constant *C, (Constant**)Idxs, NumIdx)) return FC; // Fold a few common cases... - assert(isa<PointerType>(C->getType()) && + assert(C->getType()->isPointerTy() && "Non-pointer type for constant GetElementPtr expression"); // Look up the constant in the table first to ensure uniqueness std::vector<Constant*> ArgVec; @@ -1619,7 +1619,7 @@ Constant *ConstantExpr::getInBoundsGetElementPtrTy(const Type *ReqTy, (Constant**)Idxs, NumIdx)) return FC; // Fold a few common cases... - assert(isa<PointerType>(C->getType()) && + assert(C->getType()->isPointerTy() && "Non-pointer type for constant GetElementPtr expression"); // Look up the constant in the table first to ensure uniqueness std::vector<Constant*> ArgVec; @@ -1727,7 +1727,7 @@ Constant *ConstantExpr::getExtractElementTy(const Type *ReqTy, Constant *Val, } Constant *ConstantExpr::getExtractElement(Constant *Val, Constant *Idx) { - assert(isa<VectorType>(Val->getType()) && + assert(Val->getType()->isVectorTy() && "Tried to create extractelement operation on non-vector type!"); assert(Idx->getType()->isIntegerTy(32) && "Extractelement index must be i32 type!"); @@ -1751,7 +1751,7 @@ Constant *ConstantExpr::getInsertElementTy(const Type *ReqTy, Constant *Val, Constant *ConstantExpr::getInsertElement(Constant *Val, Constant *Elt, Constant *Idx) { - assert(isa<VectorType>(Val->getType()) && + assert(Val->getType()->isVectorTy() && "Tried to create insertelement operation on non-vector type!"); assert(Elt->getType() == cast<VectorType>(Val->getType())->getElementType() && "Insertelement types must match!"); @@ -2113,7 +2113,52 @@ void ConstantStruct::replaceUsesOfWithOnConstant(Value *From, Value *To, void ConstantUnion::replaceUsesOfWithOnConstant(Value *From, Value *To, Use *U) { - assert(false && "Implement replaceUsesOfWithOnConstant for unions"); + assert(isa<Constant>(To) && "Cannot make Constant refer to non-constant!"); + Constant *ToC = cast<Constant>(To); + + assert(U == OperandList && "Union constants can only have one use!"); + assert(getNumOperands() == 1 && "Union constants can only have one use!"); + assert(getOperand(0) == From && "ReplaceAllUsesWith broken!"); + + std::pair<LLVMContextImpl::UnionConstantsTy::MapKey, ConstantUnion*> Lookup; + Lookup.first.first = getType(); + Lookup.second = this; + Lookup.first.second = ToC; + + LLVMContext &Context = getType()->getContext(); + LLVMContextImpl *pImpl = Context.pImpl; + + Constant *Replacement = 0; + if (ToC->isNullValue()) { + Replacement = ConstantAggregateZero::get(getType()); + } else { + // Check to see if we have this union type already. + bool Exists; + LLVMContextImpl::UnionConstantsTy::MapTy::iterator I = + pImpl->UnionConstants.InsertOrGetItem(Lookup, Exists); + + if (Exists) { + Replacement = I->second; + } else { + // Okay, the new shape doesn't exist in the system yet. Instead of + // creating a new constant union, inserting it, replaceallusesof'ing the + // old with the new, then deleting the old... just update the current one + // in place! + pImpl->UnionConstants.MoveConstantToNewSlot(this, I); + + // Update to the new value. + setOperand(0, ToC); + return; + } + } + + assert(Replacement != this && "I didn't contain From!"); + + // Everyone using this now uses the replacement. + uncheckedReplaceAllUsesWith(Replacement); + + // Delete the old constant! + destroyConstant(); } void ConstantVector::replaceUsesOfWithOnConstant(Value *From, Value *To, diff --git a/lib/VMCore/Core.cpp b/lib/VMCore/Core.cpp index a044fc5..f4f65c5 100644 --- a/lib/VMCore/Core.cpp +++ b/lib/VMCore/Core.cpp @@ -55,6 +55,15 @@ void LLVMContextDispose(LLVMContextRef C) { delete unwrap(C); } +unsigned LLVMGetMDKindIDInContext(LLVMContextRef C, const char* Name, + unsigned SLen) { + return unwrap(C)->getMDKindID(StringRef(Name, SLen)); +} + +unsigned LLVMGetMDKindID(const char* Name, unsigned SLen) { + return LLVMGetMDKindIDInContext(LLVMGetGlobalContext(), Name, SLen); +} + /*===-- Operations on modules ---------------------------------------------===*/ @@ -425,6 +434,18 @@ void LLVMReplaceAllUsesWith(LLVMValueRef OldVal, LLVMValueRef NewVal) { unwrap(OldVal)->replaceAllUsesWith(unwrap(NewVal)); } +int LLVMHasMetadata(LLVMValueRef Inst) { + return unwrap<Instruction>(Inst)->hasMetadata(); +} + +LLVMValueRef LLVMGetMetadata(LLVMValueRef Inst, unsigned KindID) { + return wrap(unwrap<Instruction>(Inst)->getMetadata(KindID)); +} + +void LLVMSetMetadata(LLVMValueRef Inst, unsigned KindID, LLVMValueRef MD) { + unwrap<Instruction>(Inst)->setMetadata(KindID, MD? unwrap<MDNode>(MD) : NULL); +} + /*--.. Conversion functions ................................................--*/ #define LLVM_DEFINE_VALUE_CAST(name) \ @@ -435,7 +456,7 @@ void LLVMReplaceAllUsesWith(LLVMValueRef OldVal, LLVMValueRef NewVal) { LLVM_FOR_EACH_VALUE_SUBCLASS(LLVM_DEFINE_VALUE_CAST) /*--.. Operations on Uses ..................................................--*/ -LLVMUseIteratorRef LLVMGetFirstUse(LLVMValueRef Val) { +LLVMUseRef LLVMGetFirstUse(LLVMValueRef Val) { Value *V = unwrap(Val); Value::use_iterator I = V->use_begin(); if (I == V->use_end()) @@ -443,16 +464,19 @@ LLVMUseIteratorRef LLVMGetFirstUse(LLVMValueRef Val) { return wrap(&(I.getUse())); } -LLVMUseIteratorRef LLVMGetNextUse(LLVMUseIteratorRef UR) { - return wrap(unwrap(UR)->getNext()); +LLVMUseRef LLVMGetNextUse(LLVMUseRef U) { + Use *Next = unwrap(U)->getNext(); + if (Next) + return wrap(Next); + return 0; } -LLVMValueRef LLVMGetUser(LLVMUseIteratorRef UR) { - return wrap(unwrap(UR)->getUser()); +LLVMValueRef LLVMGetUser(LLVMUseRef U) { + return wrap(unwrap(U)->getUser()); } -LLVMValueRef LLVMGetUsedValue(LLVMUseIteratorRef UR) { - return wrap(unwrap(UR)->get()); +LLVMValueRef LLVMGetUsedValue(LLVMUseRef U) { + return wrap(unwrap(U)->get()); } /*--.. Operations on Users .................................................--*/ @@ -493,6 +517,26 @@ LLVMValueRef LLVMConstPointerNull(LLVMTypeRef Ty) { wrap(ConstantPointerNull::get(unwrap<PointerType>(Ty))); } +/*--.. Operations on metadata nodes ........................................--*/ + +LLVMValueRef LLVMMDStringInContext(LLVMContextRef C, const char *Str, + unsigned SLen) { + return wrap(MDString::get(*unwrap(C), StringRef(Str, SLen))); +} + +LLVMValueRef LLVMMDString(const char *Str, unsigned SLen) { + return LLVMMDStringInContext(LLVMGetGlobalContext(), Str, SLen); +} + +LLVMValueRef LLVMMDNodeInContext(LLVMContextRef C, LLVMValueRef *Vals, + unsigned Count) { + return wrap(MDNode::get(*unwrap(C), unwrap<Value>(Vals, Count), Count)); +} + +LLVMValueRef LLVMMDNode(LLVMValueRef *Vals, unsigned Count) { + return LLVMMDNodeInContext(LLVMGetGlobalContext(), Vals, Count); +} + /*--.. Operations on scalar constants ......................................--*/ LLVMValueRef LLVMConstInt(LLVMTypeRef IntTy, unsigned long long N, @@ -567,11 +611,13 @@ LLVMValueRef LLVMConstStruct(LLVMValueRef *ConstantVals, unsigned Count, return LLVMConstStructInContext(LLVMGetGlobalContext(), ConstantVals, Count, Packed); } - LLVMValueRef LLVMConstVector(LLVMValueRef *ScalarConstantVals, unsigned Size) { return wrap(ConstantVector::get( unwrap<Constant>(ScalarConstantVals, Size), Size)); } +LLVMValueRef LLVMConstUnion(LLVMTypeRef Ty, LLVMValueRef Val) { + return wrap(ConstantUnion::get(unwrap<UnionType>(Ty), unwrap<Constant>(Val))); +} /*--.. Constant expressions ................................................--*/ @@ -592,6 +638,17 @@ LLVMValueRef LLVMConstNeg(LLVMValueRef ConstantVal) { unwrap<Constant>(ConstantVal))); } +LLVMValueRef LLVMConstNSWNeg(LLVMValueRef ConstantVal) { + return wrap(ConstantExpr::getNSWNeg( + unwrap<Constant>(ConstantVal))); +} + +LLVMValueRef LLVMConstNUWNeg(LLVMValueRef ConstantVal) { + return wrap(ConstantExpr::getNUWNeg( + unwrap<Constant>(ConstantVal))); +} + + LLVMValueRef LLVMConstFNeg(LLVMValueRef ConstantVal) { return wrap(ConstantExpr::getFNeg( unwrap<Constant>(ConstantVal))); @@ -615,6 +672,13 @@ LLVMValueRef LLVMConstNSWAdd(LLVMValueRef LHSConstant, unwrap<Constant>(RHSConstant))); } +LLVMValueRef LLVMConstNUWAdd(LLVMValueRef LHSConstant, + LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getNUWAdd( + unwrap<Constant>(LHSConstant), + unwrap<Constant>(RHSConstant))); +} + LLVMValueRef LLVMConstFAdd(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { return wrap(ConstantExpr::getFAdd( unwrap<Constant>(LHSConstant), @@ -627,6 +691,20 @@ LLVMValueRef LLVMConstSub(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { unwrap<Constant>(RHSConstant))); } +LLVMValueRef LLVMConstNSWSub(LLVMValueRef LHSConstant, + LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getNSWSub( + unwrap<Constant>(LHSConstant), + unwrap<Constant>(RHSConstant))); +} + +LLVMValueRef LLVMConstNUWSub(LLVMValueRef LHSConstant, + LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getNUWSub( + unwrap<Constant>(LHSConstant), + unwrap<Constant>(RHSConstant))); +} + LLVMValueRef LLVMConstFSub(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { return wrap(ConstantExpr::getFSub(unwrap<Constant>(LHSConstant), unwrap<Constant>(RHSConstant))); @@ -638,6 +716,20 @@ LLVMValueRef LLVMConstMul(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { unwrap<Constant>(RHSConstant))); } +LLVMValueRef LLVMConstNSWMul(LLVMValueRef LHSConstant, + LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getNSWMul( + unwrap<Constant>(LHSConstant), + unwrap<Constant>(RHSConstant))); +} + +LLVMValueRef LLVMConstNUWMul(LLVMValueRef LHSConstant, + LLVMValueRef RHSConstant) { + return wrap(ConstantExpr::getNUWMul( + unwrap<Constant>(LHSConstant), + unwrap<Constant>(RHSConstant))); +} + LLVMValueRef LLVMConstFMul(LLVMValueRef LHSConstant, LLVMValueRef RHSConstant) { return wrap(ConstantExpr::getFMul( unwrap<Constant>(LHSConstant), @@ -924,6 +1016,10 @@ LLVMValueRef LLVMConstInlineAsm(LLVMTypeRef Ty, const char *AsmString, Constraints, HasSideEffects, IsAlignStack)); } +LLVMValueRef LLVMBlockAddress(LLVMValueRef F, LLVMBasicBlockRef BB) { + return wrap(BlockAddress::get(unwrap<Function>(F), unwrap(BB))); +} + /*--.. Operations on global variables, functions, and aliases (globals) ....--*/ LLVMModuleRef LLVMGetGlobalParent(LLVMValueRef Global) { @@ -1060,6 +1156,14 @@ LLVMValueRef LLVMAddGlobal(LLVMModuleRef M, LLVMTypeRef Ty, const char *Name) { GlobalValue::ExternalLinkage, 0, Name)); } +LLVMValueRef LLVMAddGlobalInAddressSpace(LLVMModuleRef M, LLVMTypeRef Ty, + const char *Name, + unsigned AddressSpace) { + return wrap(new GlobalVariable(*unwrap(M), unwrap(Ty), false, + GlobalValue::ExternalLinkage, 0, Name, 0, + false, AddressSpace)); +} + LLVMValueRef LLVMGetNamedGlobal(LLVMModuleRef M, const char *Name) { return wrap(unwrap(M)->getNamedGlobal(Name)); } @@ -1215,14 +1319,14 @@ void LLVMSetGC(LLVMValueRef Fn, const char *GC) { void LLVMAddFunctionAttr(LLVMValueRef Fn, LLVMAttribute PA) { Function *Func = unwrap<Function>(Fn); const AttrListPtr PAL = Func->getAttributes(); - const AttrListPtr PALnew = PAL.addAttr(0, PA); + const AttrListPtr PALnew = PAL.addAttr(~0U, PA); Func->setAttributes(PALnew); } void LLVMRemoveFunctionAttr(LLVMValueRef Fn, LLVMAttribute PA) { Function *Func = unwrap<Function>(Fn); const AttrListPtr PAL = Func->getAttributes(); - const AttrListPtr PALnew = PAL.removeAttr(0, PA); + const AttrListPtr PALnew = PAL.removeAttr(~0U, PA); Func->setAttributes(PALnew); } @@ -1563,6 +1667,21 @@ void LLVMDisposeBuilder(LLVMBuilderRef Builder) { delete unwrap(Builder); } +/*--.. Metadata builders ...................................................--*/ + +void LLVMSetCurrentDebugLocation(LLVMBuilderRef Builder, LLVMValueRef L) { + unwrap(Builder)->SetCurrentDebugLocation(L? unwrap<MDNode>(L) : NULL); +} + +LLVMValueRef LLVMGetCurrentDebugLocation(LLVMBuilderRef Builder) { + return wrap(unwrap(Builder)->getCurrentDebugLocation()); +} + +void LLVMSetInstDebugLocation(LLVMBuilderRef Builder, LLVMValueRef Inst) { + unwrap(Builder)->SetInstDebugLocation(unwrap<Instruction>(Inst)); +} + + /*--.. Instruction builders ................................................--*/ LLVMValueRef LLVMBuildRetVoid(LLVMBuilderRef B) { @@ -1592,6 +1711,11 @@ LLVMValueRef LLVMBuildSwitch(LLVMBuilderRef B, LLVMValueRef V, return wrap(unwrap(B)->CreateSwitch(unwrap(V), unwrap(Else), NumCases)); } +LLVMValueRef LLVMBuildIndirectBr(LLVMBuilderRef B, LLVMValueRef Addr, + unsigned NumDests) { + return wrap(unwrap(B)->CreateIndirectBr(unwrap(Addr), NumDests)); +} + LLVMValueRef LLVMBuildInvoke(LLVMBuilderRef B, LLVMValueRef Fn, LLVMValueRef *Args, unsigned NumArgs, LLVMBasicBlockRef Then, LLVMBasicBlockRef Catch, @@ -1614,6 +1738,10 @@ void LLVMAddCase(LLVMValueRef Switch, LLVMValueRef OnVal, unwrap<SwitchInst>(Switch)->addCase(unwrap<ConstantInt>(OnVal), unwrap(Dest)); } +void LLVMAddDestination(LLVMValueRef IndirectBr, LLVMBasicBlockRef Dest) { + unwrap<IndirectBrInst>(IndirectBr)->addDestination(unwrap(Dest)); +} + /*--.. Arithmetic ..........................................................--*/ LLVMValueRef LLVMBuildAdd(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, @@ -1626,6 +1754,11 @@ LLVMValueRef LLVMBuildNSWAdd(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RH return wrap(unwrap(B)->CreateNSWAdd(unwrap(LHS), unwrap(RHS), Name)); } +LLVMValueRef LLVMBuildNUWAdd(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name) { + return wrap(unwrap(B)->CreateNUWAdd(unwrap(LHS), unwrap(RHS), Name)); +} + LLVMValueRef LLVMBuildFAdd(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name) { return wrap(unwrap(B)->CreateFAdd(unwrap(LHS), unwrap(RHS), Name)); @@ -1636,6 +1769,16 @@ LLVMValueRef LLVMBuildSub(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, return wrap(unwrap(B)->CreateSub(unwrap(LHS), unwrap(RHS), Name)); } +LLVMValueRef LLVMBuildNSWSub(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name) { + return wrap(unwrap(B)->CreateNSWSub(unwrap(LHS), unwrap(RHS), Name)); +} + +LLVMValueRef LLVMBuildNUWSub(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name) { + return wrap(unwrap(B)->CreateNUWSub(unwrap(LHS), unwrap(RHS), Name)); +} + LLVMValueRef LLVMBuildFSub(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name) { return wrap(unwrap(B)->CreateFSub(unwrap(LHS), unwrap(RHS), Name)); @@ -1646,6 +1789,16 @@ LLVMValueRef LLVMBuildMul(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, return wrap(unwrap(B)->CreateMul(unwrap(LHS), unwrap(RHS), Name)); } +LLVMValueRef LLVMBuildNSWMul(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name) { + return wrap(unwrap(B)->CreateNSWMul(unwrap(LHS), unwrap(RHS), Name)); +} + +LLVMValueRef LLVMBuildNUWMul(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name) { + return wrap(unwrap(B)->CreateNUWMul(unwrap(LHS), unwrap(RHS), Name)); +} + LLVMValueRef LLVMBuildFMul(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, const char *Name) { return wrap(unwrap(B)->CreateFMul(unwrap(LHS), unwrap(RHS), Name)); @@ -1716,10 +1869,27 @@ LLVMValueRef LLVMBuildXor(LLVMBuilderRef B, LLVMValueRef LHS, LLVMValueRef RHS, return wrap(unwrap(B)->CreateXor(unwrap(LHS), unwrap(RHS), Name)); } +LLVMValueRef LLVMBuildBinOp(LLVMBuilderRef B, LLVMOpcode Op, + LLVMValueRef LHS, LLVMValueRef RHS, + const char *Name) { + return wrap(unwrap(B)->CreateBinOp(Instruction::BinaryOps(Op), unwrap(LHS), + unwrap(RHS), Name)); +} + LLVMValueRef LLVMBuildNeg(LLVMBuilderRef B, LLVMValueRef V, const char *Name) { return wrap(unwrap(B)->CreateNeg(unwrap(V), Name)); } +LLVMValueRef LLVMBuildNSWNeg(LLVMBuilderRef B, LLVMValueRef V, + const char *Name) { + return wrap(unwrap(B)->CreateNSWNeg(unwrap(V), Name)); +} + +LLVMValueRef LLVMBuildNUWNeg(LLVMBuilderRef B, LLVMValueRef V, + const char *Name) { + return wrap(unwrap(B)->CreateNUWNeg(unwrap(V), Name)); +} + LLVMValueRef LLVMBuildFNeg(LLVMBuilderRef B, LLVMValueRef V, const char *Name) { return wrap(unwrap(B)->CreateFNeg(unwrap(V), Name)); } @@ -1887,6 +2057,12 @@ LLVMValueRef LLVMBuildTruncOrBitCast(LLVMBuilderRef B, LLVMValueRef Val, Name)); } +LLVMValueRef LLVMBuildCast(LLVMBuilderRef B, LLVMOpcode Op, LLVMValueRef Val, + LLVMTypeRef DestTy, const char *Name) { + return wrap(unwrap(B)->CreateCast(Instruction::CastOps(Op), unwrap(Val), + unwrap(DestTy), Name)); +} + LLVMValueRef LLVMBuildPointerCast(LLVMBuilderRef B, LLVMValueRef Val, LLVMTypeRef DestTy, const char *Name) { return wrap(unwrap(B)->CreatePointerCast(unwrap(Val), unwrap(DestTy), Name)); diff --git a/lib/VMCore/Function.cpp b/lib/VMCore/Function.cpp index f00f6ee..dbc283e 100644 --- a/lib/VMCore/Function.cpp +++ b/lib/VMCore/Function.cpp @@ -73,35 +73,35 @@ unsigned Argument::getArgNo() const { /// hasByValAttr - Return true if this argument has the byval attribute on it /// in its containing function. bool Argument::hasByValAttr() const { - if (!isa<PointerType>(getType())) return false; + if (!getType()->isPointerTy()) return false; return getParent()->paramHasAttr(getArgNo()+1, Attribute::ByVal); } /// hasNestAttr - Return true if this argument has the nest attribute on /// it in its containing function. bool Argument::hasNestAttr() const { - if (!isa<PointerType>(getType())) return false; + if (!getType()->isPointerTy()) return false; return getParent()->paramHasAttr(getArgNo()+1, Attribute::Nest); } /// hasNoAliasAttr - Return true if this argument has the noalias attribute on /// it in its containing function. bool Argument::hasNoAliasAttr() const { - if (!isa<PointerType>(getType())) return false; + if (!getType()->isPointerTy()) return false; return getParent()->paramHasAttr(getArgNo()+1, Attribute::NoAlias); } /// hasNoCaptureAttr - Return true if this argument has the nocapture attribute /// on it in its containing function. bool Argument::hasNoCaptureAttr() const { - if (!isa<PointerType>(getType())) return false; + if (!getType()->isPointerTy()) return false; return getParent()->paramHasAttr(getArgNo()+1, Attribute::NoCapture); } /// hasSRetAttr - Return true if this argument has the sret attribute on /// it in its containing function. bool Argument::hasStructRetAttr() const { - if (!isa<PointerType>(getType())) return false; + if (!getType()->isPointerTy()) return false; if (this != getParent()->arg_begin()) return false; // StructRet param must be first param return getParent()->paramHasAttr(1, Attribute::StructRet); @@ -155,7 +155,7 @@ Function::Function(const FunctionType *Ty, LinkageTypes Linkage, : GlobalValue(PointerType::getUnqual(Ty), Value::FunctionVal, 0, 0, Linkage, name) { assert(FunctionType::isValidReturnType(getReturnType()) && - !isa<OpaqueType>(getReturnType()) && "invalid return type"); + !getReturnType()->isOpaqueTy() && "invalid return type"); SymTab = new ValueSymbolTable(); // If the function has arguments, mark them as lazily built. diff --git a/lib/VMCore/InlineAsm.cpp b/lib/VMCore/InlineAsm.cpp index ec21773..6355834 100644 --- a/lib/VMCore/InlineAsm.cpp +++ b/lib/VMCore/InlineAsm.cpp @@ -220,7 +220,7 @@ bool InlineAsm::Verify(const FunctionType *Ty, StringRef ConstStr) { if (!Ty->getReturnType()->isVoidTy()) return false; break; case 1: - if (isa<StructType>(Ty->getReturnType())) return false; + if (Ty->getReturnType()->isStructTy()) return false; break; default: const StructType *STy = dyn_cast<StructType>(Ty->getReturnType()); diff --git a/lib/VMCore/Instructions.cpp b/lib/VMCore/Instructions.cpp index 9d5f7a5..8f4763f 100644 --- a/lib/VMCore/Instructions.cpp +++ b/lib/VMCore/Instructions.cpp @@ -562,7 +562,7 @@ static Instruction* createFree(Value* Source, Instruction *InsertBefore, BasicBlock *InsertAtEnd) { assert(((!InsertBefore && InsertAtEnd) || (InsertBefore && !InsertAtEnd)) && "createFree needs either InsertBefore or InsertAtEnd"); - assert(isa<PointerType>(Source->getType()) && + assert(Source->getType()->isPointerTy() && "Can not free something of nonpointer type!"); BasicBlock* BB = InsertBefore ? InsertBefore->getParent() : InsertAtEnd; @@ -989,7 +989,7 @@ bool AllocaInst::isStaticAlloca() const { //===----------------------------------------------------------------------===// void LoadInst::AssertOK() { - assert(isa<PointerType>(getOperand(0)->getType()) && + assert(getOperand(0)->getType()->isPointerTy() && "Ptr must have pointer type."); } @@ -1103,7 +1103,7 @@ void LoadInst::setAlignment(unsigned Align) { void StoreInst::AssertOK() { assert(getOperand(0) && getOperand(1) && "Both operands must be non-null!"); - assert(isa<PointerType>(getOperand(1)->getType()) && + assert(getOperand(1)->getType()->isPointerTy() && "Ptr must have pointer type!"); assert(getOperand(0)->getType() == cast<PointerType>(getOperand(1)->getType())->getElementType() @@ -1285,7 +1285,7 @@ static const Type* getIndexedTypeInternal(const Type *Ptr, IndexTy const *Idxs, unsigned CurIdx = 1; for (; CurIdx != NumIdx; ++CurIdx) { const CompositeType *CT = dyn_cast<CompositeType>(Agg); - if (!CT || isa<PointerType>(CT)) return 0; + if (!CT || CT->isPointerTy()) return 0; IndexTy Index = Idxs[CurIdx]; if (!CT->indexValid(Index)) return 0; Agg = CT->getTypeAtIndex(Index); @@ -1391,7 +1391,7 @@ ExtractElementInst::ExtractElementInst(Value *Val, Value *Index, bool ExtractElementInst::isValidOperands(const Value *Val, const Value *Index) { - if (!isa<VectorType>(Val->getType()) || !Index->getType()->isIntegerTy(32)) + if (!Val->getType()->isVectorTy() || !Index->getType()->isIntegerTy(32)) return false; return true; } @@ -1432,7 +1432,7 @@ InsertElementInst::InsertElementInst(Value *Vec, Value *Elt, Value *Index, bool InsertElementInst::isValidOperands(const Value *Vec, const Value *Elt, const Value *Index) { - if (!isa<VectorType>(Vec->getType())) + if (!Vec->getType()->isVectorTy()) return false; // First operand of insertelement must be vector type. if (Elt->getType() != cast<VectorType>(Vec->getType())->getElementType()) @@ -1485,7 +1485,7 @@ ShuffleVectorInst::ShuffleVectorInst(Value *V1, Value *V2, Value *Mask, bool ShuffleVectorInst::isValidOperands(const Value *V1, const Value *V2, const Value *Mask) { - if (!isa<VectorType>(V1->getType()) || V1->getType() != V2->getType()) + if (!V1->getType()->isVectorTy() || V1->getType() != V2->getType()) return false; const VectorType *MaskTy = dyn_cast<VectorType>(Mask->getType()); @@ -1602,7 +1602,7 @@ const Type* ExtractValueInst::getIndexedType(const Type *Agg, unsigned CurIdx = 0; for (; CurIdx != NumIdx; ++CurIdx) { const CompositeType *CT = dyn_cast<CompositeType>(Agg); - if (!CT || isa<PointerType>(CT) || isa<VectorType>(CT)) return 0; + if (!CT || CT->isPointerTy() || CT->isVectorTy()) return 0; unsigned Index = Idxs[CurIdx]; if (!CT->indexValid(Index)) return 0; Agg = CT->getTypeAtIndex(Index); @@ -1693,7 +1693,7 @@ void BinaryOperator::init(BinaryOps iType) { case SDiv: assert(getType() == LHS->getType() && "Arithmetic operation should return same type as operands!"); - assert((getType()->isIntegerTy() || (isa<VectorType>(getType()) && + assert((getType()->isIntegerTy() || (getType()->isVectorTy() && cast<VectorType>(getType())->getElementType()->isIntegerTy())) && "Incorrect operand type (not integer) for S/UDIV"); break; @@ -1707,7 +1707,7 @@ void BinaryOperator::init(BinaryOps iType) { case SRem: assert(getType() == LHS->getType() && "Arithmetic operation should return same type as operands!"); - assert((getType()->isIntegerTy() || (isa<VectorType>(getType()) && + assert((getType()->isIntegerTy() || (getType()->isVectorTy() && cast<VectorType>(getType())->getElementType()->isIntegerTy())) && "Incorrect operand type (not integer) for S/UREM"); break; @@ -1723,7 +1723,7 @@ void BinaryOperator::init(BinaryOps iType) { assert(getType() == LHS->getType() && "Shift operation should return same type as operands!"); assert((getType()->isIntegerTy() || - (isa<VectorType>(getType()) && + (getType()->isVectorTy() && cast<VectorType>(getType())->getElementType()->isIntegerTy())) && "Tried to create a shift operation on a non-integral type!"); break; @@ -1732,7 +1732,7 @@ void BinaryOperator::init(BinaryOps iType) { assert(getType() == LHS->getType() && "Logical operation should return same type as operands!"); assert((getType()->isIntegerTy() || - (isa<VectorType>(getType()) && + (getType()->isVectorTy() && cast<VectorType>(getType())->getElementType()->isIntegerTy())) && "Tried to create a logical operation on a non-integral type!"); break; @@ -1977,8 +1977,8 @@ bool CastInst::isLosslessCast() const { return true; // Pointer to pointer is always lossless. - if (isa<PointerType>(SrcTy)) - return isa<PointerType>(DstTy); + if (SrcTy->isPointerTy()) + return DstTy->isPointerTy(); return false; // Other types have no identity values } @@ -2094,7 +2094,7 @@ unsigned CastInst::isEliminableCastPair( // no-op cast in second op implies firstOp as long as the DestTy // is integer and we are not converting between a vector and a // non vector type. - if (!isa<VectorType>(SrcTy) && DstTy->isIntegerTy()) + if (!SrcTy->isVectorTy() && DstTy->isIntegerTy()) return firstOp; return 0; case 4: @@ -2148,12 +2148,12 @@ unsigned CastInst::isEliminableCastPair( case 11: // bitcast followed by ptrtoint is allowed as long as the bitcast // is a pointer to pointer cast. - if (isa<PointerType>(SrcTy) && isa<PointerType>(MidTy)) + if (SrcTy->isPointerTy() && MidTy->isPointerTy()) return secondOp; return 0; case 12: // inttoptr, bitcast -> intptr if bitcast is a ptr to ptr cast - if (isa<PointerType>(MidTy) && isa<PointerType>(DstTy)) + if (MidTy->isPointerTy() && DstTy->isPointerTy()) return firstOp; return 0; case 13: { @@ -2274,8 +2274,8 @@ CastInst *CastInst::CreateTruncOrBitCast(Value *S, const Type *Ty, CastInst *CastInst::CreatePointerCast(Value *S, const Type *Ty, const Twine &Name, BasicBlock *InsertAtEnd) { - assert(isa<PointerType>(S->getType()) && "Invalid cast"); - assert((Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert(S->getType()->isPointerTy() && "Invalid cast"); + assert((Ty->isIntegerTy() || Ty->isPointerTy()) && "Invalid cast"); if (Ty->isIntegerTy()) @@ -2287,8 +2287,8 @@ CastInst *CastInst::CreatePointerCast(Value *S, const Type *Ty, CastInst *CastInst::CreatePointerCast(Value *S, const Type *Ty, const Twine &Name, Instruction *InsertBefore) { - assert(isa<PointerType>(S->getType()) && "Invalid cast"); - assert((Ty->isIntegerTy() || isa<PointerType>(Ty)) && + assert(S->getType()->isPointerTy() && "Invalid cast"); + assert((Ty->isIntegerTy() || Ty->isPointerTy()) && "Invalid cast"); if (Ty->isIntegerTy()) @@ -2373,7 +2373,7 @@ bool CastInst::isCastable(const Type *SrcTy, const Type *DestTy) { // Casting from vector return DestBits == PTy->getBitWidth(); } else { // Casting from something else - return isa<PointerType>(SrcTy); + return SrcTy->isPointerTy(); } } else if (DestTy->isFloatingPointTy()) { // Casting to floating pt if (SrcTy->isIntegerTy()) { // Casting from integral @@ -2394,8 +2394,8 @@ bool CastInst::isCastable(const Type *SrcTy, const Type *DestTy) { } else { // Casting from something else return DestPTy->getBitWidth() == SrcBits; } - } else if (isa<PointerType>(DestTy)) { // Casting to pointer - if (isa<PointerType>(SrcTy)) { // Casting from pointer + } else if (DestTy->isPointerTy()) { // Casting to pointer + if (SrcTy->isPointerTy()) { // Casting from pointer return true; } else if (SrcTy->isIntegerTy()) { // Casting from integral return true; @@ -2449,7 +2449,7 @@ CastInst::getCastOpcode( PTy = NULL; return BitCast; // Same size, no-op cast } else { - assert(isa<PointerType>(SrcTy) && + assert(SrcTy->isPointerTy() && "Casting from a value that is not first-class type"); return PtrToInt; // ptr -> int } @@ -2486,8 +2486,8 @@ CastInst::getCastOpcode( } else { assert(!"Illegal cast to vector (wrong type or size)"); } - } else if (isa<PointerType>(DestTy)) { - if (isa<PointerType>(SrcTy)) { + } else if (DestTy->isPointerTy()) { + if (SrcTy->isPointerTy()) { return BitCast; // ptr -> ptr } else if (SrcTy->isIntegerTy()) { return IntToPtr; // int -> ptr @@ -2566,13 +2566,13 @@ CastInst::castIsValid(Instruction::CastOps op, Value *S, const Type *DstTy) { } return SrcTy->isFPOrFPVectorTy() && DstTy->isIntOrIntVectorTy(); case Instruction::PtrToInt: - return isa<PointerType>(SrcTy) && DstTy->isIntegerTy(); + return SrcTy->isPointerTy() && DstTy->isIntegerTy(); case Instruction::IntToPtr: - return SrcTy->isIntegerTy() && isa<PointerType>(DstTy); + return SrcTy->isIntegerTy() && DstTy->isPointerTy(); case Instruction::BitCast: // BitCast implies a no-op cast of type only. No bits change. // However, you can't cast pointers to anything but pointers. - if (isa<PointerType>(SrcTy) != isa<PointerType>(DstTy)) + if (SrcTy->isPointerTy() != DstTy->isPointerTy()) return false; // Now we know we're not dealing with a pointer/non-pointer mismatch. In all @@ -3150,7 +3150,7 @@ void SwitchInst::setSuccessorV(unsigned idx, BasicBlock *B) { //===----------------------------------------------------------------------===// void IndirectBrInst::init(Value *Address, unsigned NumDests) { - assert(Address && isa<PointerType>(Address->getType()) && + assert(Address && Address->getType()->isPointerTy() && "Address of indirectbr must be a pointer"); ReservedSpace = 1+NumDests; NumOperands = 1; diff --git a/lib/VMCore/LLVMContextImpl.h b/lib/VMCore/LLVMContextImpl.h index 62491d8..85bbe4a 100644 --- a/lib/VMCore/LLVMContextImpl.h +++ b/lib/VMCore/LLVMContextImpl.h @@ -229,13 +229,23 @@ public: if (I->second->use_empty()) delete I->second; } - MDNodeSet.clear(); AlwaysOpaqueTy->dropRef(); for (OpaqueTypesTy::iterator I = OpaqueTypes.begin(), E = OpaqueTypes.end(); I != E; ++I) { (*I)->AbstractTypeUsers.clear(); delete *I; } + // Destroy MDNode operands first. + for (FoldingSetIterator<MDNode> I = MDNodeSet.begin(), E = MDNodeSet.end(); + I != E;) { + MDNode *N = &(*I); + ++I; + N->replaceAllOperandsWithNull(); + } + while (!MDNodeSet.empty()) { + MDNode *N = &(*MDNodeSet.begin()); + N->destroy(); + } } }; diff --git a/lib/VMCore/Metadata.cpp b/lib/VMCore/Metadata.cpp index 07a5f3c..a08c454 100644 --- a/lib/VMCore/Metadata.cpp +++ b/lib/VMCore/Metadata.cpp @@ -257,6 +257,13 @@ void MDNode::Profile(FoldingSetNodeID &ID) const { ID.AddPointer(getOperand(i)); } +// replaceAllOperandsWithNull - This is used while destroying llvm context to +// gracefully delete all nodes. This method replaces all operands with null. +void MDNode::replaceAllOperandsWithNull() { + for (MDNodeOperand *Op = getOperandPtr(this, 0), *E = Op+NumOperands; + Op != E; ++Op) + replaceOperand(Op, 0); +} // Replace value from this node's operand list. void MDNode::replaceOperand(MDNodeOperand *Op, Value *To) { diff --git a/lib/VMCore/PassManager.cpp b/lib/VMCore/PassManager.cpp index 8a3527e..c4dfe14 100644 --- a/lib/VMCore/PassManager.cpp +++ b/lib/VMCore/PassManager.cpp @@ -1118,6 +1118,7 @@ bool BBPassManager::runOnFunction(Function &F) { for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) for (unsigned Index = 0; Index < getNumContainedPasses(); ++Index) { BasicBlockPass *BP = getContainedPass(Index); + bool LocalChanged = false; dumpPassInfo(BP, EXECUTION_MSG, ON_BASICBLOCK_MSG, I->getName()); dumpRequiredSet(BP); @@ -1129,11 +1130,12 @@ bool BBPassManager::runOnFunction(Function &F) { PassManagerPrettyStackEntry X(BP, *I); Timer *T = StartPassTimer(BP); - Changed |= BP->runOnBasicBlock(*I); + LocalChanged |= BP->runOnBasicBlock(*I); StopPassTimer(BP, T); } - if (Changed) + Changed |= LocalChanged; + if (LocalChanged) dumpPassInfo(BP, MODIFICATION_MSG, ON_BASICBLOCK_MSG, I->getName()); dumpPreservedSet(BP); @@ -1334,6 +1336,7 @@ bool FPPassManager::runOnFunction(Function &F) { for (unsigned Index = 0; Index < getNumContainedPasses(); ++Index) { FunctionPass *FP = getContainedPass(Index); + bool LocalChanged = false; dumpPassInfo(FP, EXECUTION_MSG, ON_FUNCTION_MSG, F.getName()); dumpRequiredSet(FP); @@ -1344,11 +1347,12 @@ bool FPPassManager::runOnFunction(Function &F) { PassManagerPrettyStackEntry X(FP, F); Timer *T = StartPassTimer(FP); - Changed |= FP->runOnFunction(F); + LocalChanged |= FP->runOnFunction(F); StopPassTimer(FP, T); } - if (Changed) + Changed |= LocalChanged; + if (LocalChanged) dumpPassInfo(FP, MODIFICATION_MSG, ON_FUNCTION_MSG, F.getName()); dumpPreservedSet(FP); @@ -1407,6 +1411,7 @@ MPPassManager::runOnModule(Module &M) { for (unsigned Index = 0; Index < getNumContainedPasses(); ++Index) { ModulePass *MP = getContainedPass(Index); + bool LocalChanged = false; dumpPassInfo(MP, EXECUTION_MSG, ON_MODULE_MSG, M.getModuleIdentifier()); dumpRequiredSet(MP); @@ -1416,11 +1421,12 @@ MPPassManager::runOnModule(Module &M) { { PassManagerPrettyStackEntry X(MP, M); Timer *T = StartPassTimer(MP); - Changed |= MP->runOnModule(M); + LocalChanged |= MP->runOnModule(M); StopPassTimer(MP, T); } - if (Changed) + Changed |= LocalChanged; + if (LocalChanged) dumpPassInfo(MP, MODIFICATION_MSG, ON_MODULE_MSG, M.getModuleIdentifier()); dumpPreservedSet(MP); @@ -1706,8 +1712,13 @@ LLVMPassManagerRef LLVMCreatePassManager() { return wrap(new PassManager()); } +LLVMPassManagerRef LLVMCreateFunctionPassManagerForModule(LLVMModuleRef M) { + return wrap(new FunctionPassManager(unwrap(M))); +} + LLVMPassManagerRef LLVMCreateFunctionPassManager(LLVMModuleProviderRef P) { - return wrap(new FunctionPassManager(unwrap(P))); + return LLVMCreateFunctionPassManagerForModule( + reinterpret_cast<LLVMModuleRef>(P)); } LLVMBool LLVMRunPassManager(LLVMPassManagerRef PM, LLVMModuleRef M) { diff --git a/lib/VMCore/Type.cpp b/lib/VMCore/Type.cpp index f4cd366..9b2c2ca 100644 --- a/lib/VMCore/Type.cpp +++ b/lib/VMCore/Type.cpp @@ -61,8 +61,8 @@ void Type::destroy() const { // Structures and Functions allocate their contained types past the end of // the type object itself. These need to be destroyed differently than the // other types. - if (isa<FunctionType>(this) || isa<StructType>(this) || - isa<UnionType>(this)) { + if (this->isFunctionTy() || this->isStructTy() || + this->isUnionTy()) { // First, make sure we destruct any PATypeHandles allocated by these // subclasses. They must be manually destructed. for (unsigned i = 0; i < NumContainedTys; ++i) @@ -70,9 +70,9 @@ void Type::destroy() const { // Now call the destructor for the subclass directly because we're going // to delete this as an array of char. - if (isa<FunctionType>(this)) + if (this->isFunctionTy()) static_cast<const FunctionType*>(this)->FunctionType::~FunctionType(); - else if (isa<StructType>(this)) + else if (this->isStructTy()) static_cast<const StructType*>(this)->StructType::~StructType(); else static_cast<const UnionType*>(this)->UnionType::~UnionType(); @@ -176,8 +176,8 @@ bool Type::canLosslesslyBitCastTo(const Type *Ty) const { // At this point we have only various mismatches of the first class types // remaining and ptr->ptr. Just select the lossless conversions. Everything // else is not lossless. - if (isa<PointerType>(this)) - return isa<PointerType>(Ty); + if (this->isPointerTy()) + return Ty->isPointerTy(); return false; // Other types have no identity values } @@ -220,7 +220,7 @@ int Type::getFPMantissaWidth() const { /// iff all of the members of the type are sized as well. Since asking for /// their size is relatively uncommon, move this operation out of line. bool Type::isSizedDerivedType() const { - if (isa<IntegerType>(this)) + if (this->isIntegerTy()) return true; if (const ArrayType *ATy = dyn_cast<ArrayType>(this)) @@ -229,7 +229,7 @@ bool Type::isSizedDerivedType() const { if (const VectorType *PTy = dyn_cast<VectorType>(this)) return PTy->getElementType()->isSized(); - if (!isa<StructType>(this) && !isa<UnionType>(this)) + if (!this->isStructTy() && !this->isUnionTy()) return false; // Okay, our struct is sized if all of the elements are... @@ -447,7 +447,7 @@ bool FunctionType::isValidReturnType(const Type *RetTy) { /// isValidArgumentType - Return true if the specified type is valid as an /// argument type. bool FunctionType::isValidArgumentType(const Type *ArgTy) { - return ArgTy->isFirstClassType() || isa<OpaqueType>(ArgTy); + return ArgTy->isFirstClassType() || ArgTy->isOpaqueTy(); } FunctionType::FunctionType(const Type *Result, @@ -613,7 +613,7 @@ void Type::PromoteAbstractToConcrete() { // Concrete types are leaves in the tree. Since an SCC will either be all // abstract or all concrete, we only need to check one type. if (SCC[0]->isAbstract()) { - if (isa<OpaqueType>(SCC[0])) + if (SCC[0]->isOpaqueTy()) return; // Not going to be concrete, sorry. // If all of the children of all of the types in this SCC are concrete, @@ -660,7 +660,7 @@ static bool TypesEqual(const Type *Ty, const Type *Ty2, std::map<const Type *, const Type *> &EqTypes) { if (Ty == Ty2) return true; if (Ty->getTypeID() != Ty2->getTypeID()) return false; - if (isa<OpaqueType>(Ty)) + if (Ty->isOpaqueTy()) return false; // Two unequal opaque types are never equal std::map<const Type*, const Type*>::iterator It = EqTypes.find(Ty); @@ -888,7 +888,7 @@ ArrayType *ArrayType::get(const Type *ElementType, uint64_t NumElements) { bool ArrayType::isValidElementType(const Type *ElemTy) { return ElemTy->getTypeID() != VoidTyID && ElemTy->getTypeID() != LabelTyID && - ElemTy->getTypeID() != MetadataTyID && !isa<FunctionType>(ElemTy); + ElemTy->getTypeID() != MetadataTyID && !ElemTy->isFunctionTy(); } VectorType *VectorType::get(const Type *ElementType, unsigned NumElements) { @@ -912,7 +912,7 @@ VectorType *VectorType::get(const Type *ElementType, unsigned NumElements) { bool VectorType::isValidElementType(const Type *ElemTy) { return ElemTy->isIntegerTy() || ElemTy->isFloatingPointTy() || - isa<OpaqueType>(ElemTy); + ElemTy->isOpaqueTy(); } //===----------------------------------------------------------------------===// @@ -955,7 +955,7 @@ StructType *StructType::get(LLVMContext &Context, const Type *type, ...) { bool StructType::isValidElementType(const Type *ElemTy) { return !ElemTy->isVoidTy() && !ElemTy->isLabelTy() && - !ElemTy->isMetadataTy() && !isa<FunctionType>(ElemTy); + !ElemTy->isMetadataTy() && !ElemTy->isFunctionTy(); } @@ -1303,7 +1303,7 @@ void PointerType::typeBecameConcrete(const DerivedType *AbsTy) { } bool SequentialType::indexValid(const Value *V) const { - if (isa<IntegerType>(V->getType())) + if (V->getType()->isIntegerTy()) return true; return false; } diff --git a/lib/VMCore/Value.cpp b/lib/VMCore/Value.cpp index 3759b8a..a36d262 100644 --- a/lib/VMCore/Value.cpp +++ b/lib/VMCore/Value.cpp @@ -45,11 +45,11 @@ Value::Value(const Type *ty, unsigned scid) UseList(0), Name(0) { if (isa<CallInst>(this) || isa<InvokeInst>(this)) assert((VTy->isFirstClassType() || VTy->isVoidTy() || - isa<OpaqueType>(ty) || VTy->getTypeID() == Type::StructTyID) && + ty->isOpaqueTy() || VTy->isStructTy()) && "invalid CallInst type!"); else if (!isa<Constant>(this) && !isa<BasicBlock>(this)) assert((VTy->isFirstClassType() || VTy->isVoidTy() || - isa<OpaqueType>(ty)) && + ty->isOpaqueTy()) && "Cannot create non-first-class values except for constants!"); } @@ -320,7 +320,7 @@ void Value::replaceAllUsesWith(Value *New) { } Value *Value::stripPointerCasts() { - if (!isa<PointerType>(getType())) + if (!getType()->isPointerTy()) return this; Value *V = this; do { @@ -337,12 +337,12 @@ Value *Value::stripPointerCasts() { } else { return V; } - assert(isa<PointerType>(V->getType()) && "Unexpected operand type!"); + assert(V->getType()->isPointerTy() && "Unexpected operand type!"); } while (1); } Value *Value::getUnderlyingObject(unsigned MaxLookup) { - if (!isa<PointerType>(getType())) + if (!getType()->isPointerTy()) return this; Value *V = this; for (unsigned Count = 0; MaxLookup == 0 || Count < MaxLookup; ++Count) { @@ -357,7 +357,7 @@ Value *Value::getUnderlyingObject(unsigned MaxLookup) { } else { return V; } - assert(isa<PointerType>(V->getType()) && "Unexpected operand type!"); + assert(V->getType()->isPointerTy() && "Unexpected operand type!"); } return V; } diff --git a/lib/VMCore/ValueTypes.cpp b/lib/VMCore/ValueTypes.cpp index 62b9034..a092cd1 100644 --- a/lib/VMCore/ValueTypes.cpp +++ b/lib/VMCore/ValueTypes.cpp @@ -46,7 +46,7 @@ bool EVT::isExtendedInteger() const { bool EVT::isExtendedVector() const { assert(isExtended() && "Type is not extended!"); - return isa<VectorType>(LLVMTy); + return LLVMTy->isVectorTy(); } bool EVT::isExtended64BitVector() const { @@ -126,6 +126,7 @@ std::string EVT::getEVTString() const { case MVT::v8f32: return "v8f32"; case MVT::v2f64: return "v2f64"; case MVT::v4f64: return "v4f64"; + case MVT::Metadata:return "Metadata"; } } diff --git a/lib/VMCore/Verifier.cpp b/lib/VMCore/Verifier.cpp index 2b4892b..35625a5 100644 --- a/lib/VMCore/Verifier.cpp +++ b/lib/VMCore/Verifier.cpp @@ -433,7 +433,7 @@ void Verifier::visitGlobalValue(GlobalValue &GV) { if (GV.hasAppendingLinkage()) { GlobalVariable *GVar = dyn_cast<GlobalVariable>(&GV); - Assert1(GVar && isa<ArrayType>(GVar->getType()->getElementType()), + Assert1(GVar && GVar->getType()->getElementType()->isArrayTy(), "Only global arrays can have appending linkage!", GVar); } } @@ -609,7 +609,7 @@ void Verifier::visitFunction(Function &F) { &F, FT); Assert1(F.getReturnType()->isFirstClassType() || F.getReturnType()->isVoidTy() || - isa<StructType>(F.getReturnType()), + F.getReturnType()->isStructTy(), "Functions cannot return aggregate values!", &F); Assert1(!F.hasStructRetAttr() || F.getReturnType()->isVoidTy(), @@ -838,7 +838,7 @@ void Verifier::visitTruncInst(TruncInst &I) { Assert1(SrcTy->isIntOrIntVectorTy(), "Trunc only operates on integer", &I); Assert1(DestTy->isIntOrIntVectorTy(), "Trunc only produces integer", &I); - Assert1(isa<VectorType>(SrcTy) == isa<VectorType>(DestTy), + Assert1(SrcTy->isVectorTy() == DestTy->isVectorTy(), "trunc source and destination must both be a vector or neither", &I); Assert1(SrcBitSize > DestBitSize,"DestTy too big for Trunc", &I); @@ -853,7 +853,7 @@ void Verifier::visitZExtInst(ZExtInst &I) { // Get the size of the types in bits, we'll need this later Assert1(SrcTy->isIntOrIntVectorTy(), "ZExt only operates on integer", &I); Assert1(DestTy->isIntOrIntVectorTy(), "ZExt only produces an integer", &I); - Assert1(isa<VectorType>(SrcTy) == isa<VectorType>(DestTy), + Assert1(SrcTy->isVectorTy() == DestTy->isVectorTy(), "zext source and destination must both be a vector or neither", &I); unsigned SrcBitSize = SrcTy->getScalarSizeInBits(); unsigned DestBitSize = DestTy->getScalarSizeInBits(); @@ -874,7 +874,7 @@ void Verifier::visitSExtInst(SExtInst &I) { Assert1(SrcTy->isIntOrIntVectorTy(), "SExt only operates on integer", &I); Assert1(DestTy->isIntOrIntVectorTy(), "SExt only produces an integer", &I); - Assert1(isa<VectorType>(SrcTy) == isa<VectorType>(DestTy), + Assert1(SrcTy->isVectorTy() == DestTy->isVectorTy(), "sext source and destination must both be a vector or neither", &I); Assert1(SrcBitSize < DestBitSize,"Type too small for SExt", &I); @@ -891,7 +891,7 @@ void Verifier::visitFPTruncInst(FPTruncInst &I) { Assert1(SrcTy->isFPOrFPVectorTy(),"FPTrunc only operates on FP", &I); Assert1(DestTy->isFPOrFPVectorTy(),"FPTrunc only produces an FP", &I); - Assert1(isa<VectorType>(SrcTy) == isa<VectorType>(DestTy), + Assert1(SrcTy->isVectorTy() == DestTy->isVectorTy(), "fptrunc source and destination must both be a vector or neither",&I); Assert1(SrcBitSize > DestBitSize,"DestTy too big for FPTrunc", &I); @@ -909,7 +909,7 @@ void Verifier::visitFPExtInst(FPExtInst &I) { Assert1(SrcTy->isFPOrFPVectorTy(),"FPExt only operates on FP", &I); Assert1(DestTy->isFPOrFPVectorTy(),"FPExt only produces an FP", &I); - Assert1(isa<VectorType>(SrcTy) == isa<VectorType>(DestTy), + Assert1(SrcTy->isVectorTy() == DestTy->isVectorTy(), "fpext source and destination must both be a vector or neither", &I); Assert1(SrcBitSize < DestBitSize,"DestTy too small for FPExt", &I); @@ -921,8 +921,8 @@ void Verifier::visitUIToFPInst(UIToFPInst &I) { const Type *SrcTy = I.getOperand(0)->getType(); const Type *DestTy = I.getType(); - bool SrcVec = isa<VectorType>(SrcTy); - bool DstVec = isa<VectorType>(DestTy); + bool SrcVec = SrcTy->isVectorTy(); + bool DstVec = DestTy->isVectorTy(); Assert1(SrcVec == DstVec, "UIToFP source and dest must both be vector or scalar", &I); @@ -944,8 +944,8 @@ void Verifier::visitSIToFPInst(SIToFPInst &I) { const Type *SrcTy = I.getOperand(0)->getType(); const Type *DestTy = I.getType(); - bool SrcVec = isa<VectorType>(SrcTy); - bool DstVec = isa<VectorType>(DestTy); + bool SrcVec = SrcTy->isVectorTy(); + bool DstVec = DestTy->isVectorTy(); Assert1(SrcVec == DstVec, "SIToFP source and dest must both be vector or scalar", &I); @@ -967,8 +967,8 @@ void Verifier::visitFPToUIInst(FPToUIInst &I) { const Type *SrcTy = I.getOperand(0)->getType(); const Type *DestTy = I.getType(); - bool SrcVec = isa<VectorType>(SrcTy); - bool DstVec = isa<VectorType>(DestTy); + bool SrcVec = SrcTy->isVectorTy(); + bool DstVec = DestTy->isVectorTy(); Assert1(SrcVec == DstVec, "FPToUI source and dest must both be vector or scalar", &I); @@ -990,8 +990,8 @@ void Verifier::visitFPToSIInst(FPToSIInst &I) { const Type *SrcTy = I.getOperand(0)->getType(); const Type *DestTy = I.getType(); - bool SrcVec = isa<VectorType>(SrcTy); - bool DstVec = isa<VectorType>(DestTy); + bool SrcVec = SrcTy->isVectorTy(); + bool DstVec = DestTy->isVectorTy(); Assert1(SrcVec == DstVec, "FPToSI source and dest must both be vector or scalar", &I); @@ -1013,7 +1013,7 @@ void Verifier::visitPtrToIntInst(PtrToIntInst &I) { const Type *SrcTy = I.getOperand(0)->getType(); const Type *DestTy = I.getType(); - Assert1(isa<PointerType>(SrcTy), "PtrToInt source must be pointer", &I); + Assert1(SrcTy->isPointerTy(), "PtrToInt source must be pointer", &I); Assert1(DestTy->isIntegerTy(), "PtrToInt result must be integral", &I); visitInstruction(I); @@ -1025,7 +1025,7 @@ void Verifier::visitIntToPtrInst(IntToPtrInst &I) { const Type *DestTy = I.getType(); Assert1(SrcTy->isIntegerTy(), "IntToPtr source must be an integral", &I); - Assert1(isa<PointerType>(DestTy), "IntToPtr result must be a pointer",&I); + Assert1(DestTy->isPointerTy(), "IntToPtr result must be a pointer",&I); visitInstruction(I); } @@ -1041,7 +1041,7 @@ void Verifier::visitBitCastInst(BitCastInst &I) { // BitCast implies a no-op cast of type only. No bits change. // However, you can't cast pointers to anything but pointers. - Assert1(isa<PointerType>(DestTy) == isa<PointerType>(DestTy), + Assert1(DestTy->isPointerTy() == DestTy->isPointerTy(), "Bitcast requires both operands to be pointer or neither", &I); Assert1(SrcBitSize == DestBitSize, "Bitcast requires types of same width",&I); @@ -1084,11 +1084,11 @@ void Verifier::visitPHINode(PHINode &PN) { void Verifier::VerifyCallSite(CallSite CS) { Instruction *I = CS.getInstruction(); - Assert1(isa<PointerType>(CS.getCalledValue()->getType()), + Assert1(CS.getCalledValue()->getType()->isPointerTy(), "Called function must be a pointer!", I); const PointerType *FPTy = cast<PointerType>(CS.getCalledValue()->getType()); - Assert1(isa<FunctionType>(FPTy->getElementType()), + Assert1(FPTy->getElementType()->isFunctionTy(), "Called function is not pointer to function type!", I); const FunctionType *FTy = cast<FunctionType>(FPTy->getElementType()); @@ -1219,7 +1219,7 @@ void Verifier::visitICmpInst(ICmpInst& IC) { Assert1(Op0Ty == Op1Ty, "Both operands to ICmp instruction are not of the same type!", &IC); // Check that the operands are the right type - Assert1(Op0Ty->isIntOrIntVectorTy() || isa<PointerType>(Op0Ty), + Assert1(Op0Ty->isIntOrIntVectorTy() || Op0Ty->isPointerTy(), "Invalid operand types for ICmp instruction", &IC); visitInstruction(IC); @@ -1286,7 +1286,7 @@ void Verifier::visitGetElementPtrInst(GetElementPtrInst &GEP) { GetElementPtrInst::getIndexedType(GEP.getOperand(0)->getType(), Idxs.begin(), Idxs.end()); Assert1(ElTy, "Invalid indices for GEP pointer type!", &GEP); - Assert2(isa<PointerType>(GEP.getType()) && + Assert2(GEP.getType()->isPointerTy() && cast<PointerType>(GEP.getType())->getElementType() == ElTy, "GEP is not of right type for indices!", &GEP, ElTy); visitInstruction(GEP); @@ -1632,7 +1632,7 @@ void Verifier::visitIntrinsicFunctionCall(Intrinsic::ID ID, CallInst &CI) { if (ID == Intrinsic::gcroot) { AllocaInst *AI = dyn_cast<AllocaInst>(CI.getOperand(1)->stripPointerCasts()); - Assert1(AI && isa<PointerType>(AI->getType()->getElementType()), + Assert1(AI && AI->getType()->getElementType()->isPointerTy(), "llvm.gcroot parameter #1 must be a pointer alloca.", &CI); Assert1(isa<Constant>(CI.getOperand(2)), "llvm.gcroot parameter #2 must be a constant.", &CI); @@ -1794,7 +1794,7 @@ bool Verifier::PerformTypeCheck(Intrinsic::ID ID, Function *F, const Type *Ty, } Suffix += ".v" + utostr(NumElts) + EVT::getEVT(EltTy).getEVTString(); } else if (VT == MVT::iPTR) { - if (!isa<PointerType>(Ty)) { + if (!Ty->isPointerTy()) { CheckFailed(IntrinsicParam(ArgNo, NumRets) + " is not a " "pointer and a pointer is required.", F); return false; diff --git a/projects/sample/configure b/projects/sample/configure index f6aa847..5a13388 100755 --- a/projects/sample/configure +++ b/projects/sample/configure @@ -2362,3 +2362,4 @@ if test "$no_create" != yes; then $ac_cs_success || { (exit 1); exit 1; } fi + diff --git a/runtime/libprofile/Makefile b/runtime/libprofile/Makefile index 92a8558..15e6779 100644 --- a/runtime/libprofile/Makefile +++ b/runtime/libprofile/Makefile @@ -10,7 +10,7 @@ LEVEL = ../.. include $(LEVEL)/Makefile.config -ifneq ($(wildcard $(LLVMGCC)),) +ifneq ($(strip $(LLVMCC)),) BYTECODE_LIBRARY = 1 endif SHARED_LIBRARY = 1 diff --git a/test/Analysis/Andersens/2007-11-19-InlineAsm.ll b/test/Analysis/Andersens/2007-11-19-InlineAsm.ll deleted file mode 100644 index 5ba3499..0000000 --- a/test/Analysis/Andersens/2007-11-19-InlineAsm.ll +++ /dev/null @@ -1,8 +0,0 @@ -; RUN: opt < %s -anders-aa -disable-output - -define void @x(i16 %Y) { -entry: - %tmp = call i16 asm "bswap $0", "=r,r"(i16 %Y) - ret void -} - diff --git a/test/Analysis/Andersens/2008-03-19-External.ll b/test/Analysis/Andersens/2008-03-19-External.ll deleted file mode 100644 index a973103..0000000 --- a/test/Analysis/Andersens/2008-03-19-External.ll +++ /dev/null @@ -1,12 +0,0 @@ -; RUN: opt < %s -anders-aa -gvn -S | not grep undef -; PR2160 - -declare void @f(i32*) - -define i32 @g() { -entry: - %tmp = alloca i32 ; <i32*> [#uses=2] - call void @f( i32* %tmp ) - %tmp2 = load i32* %tmp ; <i32> [#uses=1] - ret i32 %tmp2 -} diff --git a/test/Analysis/Andersens/2008-04-07-Memcpy.ll b/test/Analysis/Andersens/2008-04-07-Memcpy.ll deleted file mode 100644 index 5a50dd5..0000000 --- a/test/Analysis/Andersens/2008-04-07-Memcpy.ll +++ /dev/null @@ -1,14 +0,0 @@ -; RUN: opt < %s -anders-aa -gvn -S | not grep undef -; PR2169 - -declare void @llvm.memcpy.i32(i8*, i8*, i32, i32) nounwind -declare void @use(i8) - -define void @f(i8* %x) { -entry: - %copy = alloca i8 ; <i8*> [#uses=6] - call void @llvm.memcpy.i32( i8* %copy, i8* %x, i32 1, i32 4 ) - %tmp = load i8* %copy ; <i8> [#uses=1] - call void @use(i8 %tmp) - ret void -} diff --git a/test/Analysis/Andersens/2008-12-27-BuiltinWrongType.ll b/test/Analysis/Andersens/2008-12-27-BuiltinWrongType.ll deleted file mode 100644 index da67511..0000000 --- a/test/Analysis/Andersens/2008-12-27-BuiltinWrongType.ll +++ /dev/null @@ -1,19 +0,0 @@ -; RUN: opt < %s -anders-aa -; PR3262 - -@.str15 = external global [3 x i8] ; <[3 x i8]*> [#uses=1] - -declare i8* @strtok(...) -declare i8* @memmove(...) - -define void @test1(i8* %want1) nounwind { -entry: - %0 = call i8* (...)* @strtok(i32 0, i8* getelementptr ([3 x i8]* @.str15, i32 0, i32 0)) nounwind ; <i8*> [#uses=0] - unreachable -} - -define void @test2() nounwind { -entry: - %0 = call i8* (...)* @memmove() - unreachable -} diff --git a/test/Analysis/Andersens/basictest.ll b/test/Analysis/Andersens/basictest.ll deleted file mode 100644 index 47226dd..0000000 --- a/test/Analysis/Andersens/basictest.ll +++ /dev/null @@ -1,28 +0,0 @@ -; RUN: opt < %s -anders-aa -aa-eval 2>/dev/null - -define void @test1() { - %X = malloc i32* - %Y = malloc i32 - %Z = ptrtoint i32* %Y to i32 - %W = inttoptr i32 %Z to i32* - store i32* %W, i32** %X - ret void -} - -define void @test2(i32* %P) { - %X = malloc i32* - %Y = malloc i32 - store i32* %P, i32** %X - ret void -} - -define internal i32 *@test3(i32* %P) { - ret i32* %P -} - -define void @test4() { - %X = malloc i32 - %Y = call i32* @test3(i32* %X) - %ZZ = getelementptr i32* null, i32 17 - ret void -} diff --git a/test/Analysis/Andersens/dg.exp b/test/Analysis/Andersens/dg.exp deleted file mode 100644 index 1eb4755..0000000 --- a/test/Analysis/Andersens/dg.exp +++ /dev/null @@ -1,4 +0,0 @@ -load_lib llvm.exp - -RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{ll,c,cpp}]] - diff --git a/test/Analysis/Andersens/external.ll b/test/Analysis/Andersens/external.ll deleted file mode 100644 index 13c12dc..0000000 --- a/test/Analysis/Andersens/external.ll +++ /dev/null @@ -1,20 +0,0 @@ -; RUN: opt < %s -anders-aa -gvn -deadargelim -S | grep store | not grep null - -; Because the 'internal' function is passed to an external function, we don't -; know what the incoming values will alias. As such, we cannot do the -; optimization checked by the 'arg-must-alias.ll' test. - -declare void @external(i32(i32*)*) -@G = internal constant i32* null - -define internal i32 @internal(i32* %ARG) { - ;;; We *DON'T* know that ARG always points to null! - store i32* %ARG, i32** @G - ret i32 0 -} - -define i32 @foo() { - call void @external(i32(i32*)* @internal) - %V = call i32 @internal(i32* null) - ret i32 %V -} diff --git a/test/Analysis/Andersens/modreftest.ll b/test/Analysis/Andersens/modreftest.ll deleted file mode 100644 index e0c2edc..0000000 --- a/test/Analysis/Andersens/modreftest.ll +++ /dev/null @@ -1,15 +0,0 @@ -; RUN: opt < %s -anders-aa -gvn -instcombine -S \ -; RUN: | grep {ret i1 true} - -@G = internal global i32* null -declare i32 *@ext() - -define i1 @bar() { - %V1 = load i32** @G - %X2 = call i32 *@ext() - %V2 = load i32** @G - store i32* %X2, i32** @G - - %C = icmp eq i32* %V1, %V2 - ret i1 %C -} diff --git a/test/Analysis/Andersens/modreftest2.ll b/test/Analysis/Andersens/modreftest2.ll deleted file mode 100644 index 562c961..0000000 --- a/test/Analysis/Andersens/modreftest2.ll +++ /dev/null @@ -1,14 +0,0 @@ -; RUN: opt < %s -anders-aa -gvn -S \ -; RUN: | not grep {ret i32 undef} - -;; From PR 2160 -declare void @f(i32*) - -define i32 @g() { -entry: - %tmp = alloca i32 ; <i32*> [#uses=2] - call void @f( i32* %tmp ) - %tmp2 = load i32* %tmp ; <i32> [#uses=1] - ret i32 %tmp2 -} - diff --git a/test/Analysis/Andersens/trivialtest.ll b/test/Analysis/Andersens/trivialtest.ll deleted file mode 100644 index f9f938f..0000000 --- a/test/Analysis/Andersens/trivialtest.ll +++ /dev/null @@ -1,3 +0,0 @@ -; RUN: opt < %s -anders-aa -disable-output - -define void @foo() { ret void } diff --git a/test/Analysis/ScalarEvolution/trip-count10.ll b/test/Analysis/ScalarEvolution/trip-count10.ll new file mode 100644 index 0000000..0a992f9 --- /dev/null +++ b/test/Analysis/ScalarEvolution/trip-count10.ll @@ -0,0 +1,76 @@ +; RUN: opt < %s -analyze -scalar-evolution | FileCheck %s + +; Trip counts with trivial exit conditions. + +; CHECK: Determining loop execution counts for: @a +; CHECK: Loop %loop: Unpredictable backedge-taken count. +; CHECK: Loop %loop: Unpredictable max backedge-taken count. + +; CHECK: Determining loop execution counts for: @b +; CHECK: Loop %loop: backedge-taken count is false +; CHECK: Loop %loop: max backedge-taken count is false + +; CHECK: Determining loop execution counts for: @c +; CHECK: Loop %loop: backedge-taken count is false +; CHECK: Loop %loop: max backedge-taken count is false + +; CHECK: Determining loop execution counts for: @d +; CHECK: Loop %loop: Unpredictable backedge-taken count. +; CHECK: Loop %loop: Unpredictable max backedge-taken count. + +define void @a(i64 %n) nounwind { +entry: + %t0 = icmp sgt i64 %n, 0 + br i1 %t0, label %loop, label %return + +loop: + %i = phi i64 [ %i.next, %loop ], [ 0, %entry ] + %i.next = add nsw i64 %i, 1 + %exitcond = icmp eq i64 %i.next, %n + br i1 false, label %return, label %loop + +return: + ret void +} +define void @b(i64 %n) nounwind { +entry: + %t0 = icmp sgt i64 %n, 0 + br i1 %t0, label %loop, label %return + +loop: + %i = phi i64 [ %i.next, %loop ], [ 0, %entry ] + %i.next = add nsw i64 %i, 1 + %exitcond = icmp eq i64 %i.next, %n + br i1 true, label %return, label %loop + +return: + ret void +} +define void @c(i64 %n) nounwind { +entry: + %t0 = icmp sgt i64 %n, 0 + br i1 %t0, label %loop, label %return + +loop: + %i = phi i64 [ %i.next, %loop ], [ 0, %entry ] + %i.next = add nsw i64 %i, 1 + %exitcond = icmp eq i64 %i.next, %n + br i1 false, label %loop, label %return + +return: + ret void +} +define void @d(i64 %n) nounwind { +entry: + %t0 = icmp sgt i64 %n, 0 + br i1 %t0, label %loop, label %return + +loop: + %i = phi i64 [ %i.next, %loop ], [ 0, %entry ] + %i.next = add nsw i64 %i, 1 + %exitcond = icmp eq i64 %i.next, %n + br i1 true, label %loop, label %return + +return: + ret void +} diff --git a/test/Bindings/Ocaml/bitreader.ml b/test/Bindings/Ocaml/bitreader.ml index 5c23041..112ca61 100644 --- a/test/Bindings/Ocaml/bitreader.ml +++ b/test/Bindings/Ocaml/bitreader.ml @@ -41,16 +41,16 @@ let _ = true end; - (* get_module_provider *) + (* get_module *) begin let mb = Llvm.MemoryBuffer.of_file fn in - let mp = begin try - Llvm_bitreader.get_module_provider context mb + let m = begin try + Llvm_bitreader.get_module context mb with x -> Llvm.MemoryBuffer.dispose mb; raise x end in - Llvm.ModuleProvider.dispose mp + Llvm.dispose_module m end; (* corrupt the bitcode *) @@ -60,17 +60,17 @@ let _ = close_out oc end; - (* test get_module_provider exceptions *) + (* test get_module exceptions *) test begin try let mb = Llvm.MemoryBuffer.of_file fn in - let mp = begin try - Llvm_bitreader.get_module_provider context mb + let m = begin try + Llvm_bitreader.get_module context mb with x -> Llvm.MemoryBuffer.dispose mb; raise x end in - Llvm.ModuleProvider.dispose mp; + Llvm.dispose_module m; false with Llvm_bitreader.Error _ -> true diff --git a/test/Bindings/Ocaml/executionengine.ml b/test/Bindings/Ocaml/executionengine.ml index ce56c50..2caeb82 100644 --- a/test/Bindings/Ocaml/executionengine.ml +++ b/test/Bindings/Ocaml/executionengine.ml @@ -64,9 +64,8 @@ let test_executionengine () = let m2 = create_module (global_context ()) "test_module2" in define_plus m2; - let ee = ExecutionEngine.create (ModuleProvider.create m) in - let mp2 = ModuleProvider.create m2 in - ExecutionEngine.add_module_provider mp2 ee; + let ee = ExecutionEngine.create m in + ExecutionEngine.add_module m2 ee; (* run_static_ctors *) ExecutionEngine.run_static_ctors ee; @@ -94,8 +93,8 @@ let test_executionengine () = ee in if 4 != GenericValue.as_int res then bomb "plus did not work"; - (* remove_module_provider *) - Llvm.dispose_module (ExecutionEngine.remove_module_provider mp2 ee); + (* remove_module *) + Llvm.dispose_module (ExecutionEngine.remove_module m2 ee); (* run_static_dtors *) ExecutionEngine.run_static_dtors ee; diff --git a/test/Bindings/Ocaml/scalar_opts.ml b/test/Bindings/Ocaml/scalar_opts.ml index 0a65810..1b488c5 100644 --- a/test/Bindings/Ocaml/scalar_opts.ml +++ b/test/Bindings/Ocaml/scalar_opts.ml @@ -22,7 +22,6 @@ let suite name f = let filename = Sys.argv.(1) let m = create_module context filename -let mp = ModuleProvider.create m (*===-- Transforms --------------------------------------------------------===*) @@ -36,7 +35,7 @@ let test_transforms () = let td = TargetData.create (target_triple m) in - ignore (PassManager.create_function mp + ignore (PassManager.create_function m ++ TargetData.add td ++ add_instruction_combining ++ add_reassociation @@ -55,4 +54,4 @@ let test_transforms () = let _ = suite "transforms" test_transforms; - ModuleProvider.dispose mp + dispose_module m diff --git a/test/Bindings/Ocaml/vmcore.ml b/test/Bindings/Ocaml/vmcore.ml index dd0404a..506bf50 100644 --- a/test/Bindings/Ocaml/vmcore.ml +++ b/test/Bindings/Ocaml/vmcore.ml @@ -58,7 +58,6 @@ let suite name f = let filename = Sys.argv.(1) let m = create_module context filename -let mp = ModuleProvider.create m (*===-- Target ------------------------------------------------------------===*) @@ -83,181 +82,189 @@ let test_target () = (*===-- Types -------------------------------------------------------------===*) let test_types () = - (* RUN: grep {Ty01.*void} < %t.ll + (* RUN: grep {void_type.*void} < %t.ll *) group "void"; - insist (define_type_name "Ty01" void_type m); + insist (define_type_name "void_type" void_type m); insist (TypeKind.Void == classify_type void_type); - (* RUN: grep {Ty02.*i1} < %t.ll + (* RUN: grep {i1_type.*i1} < %t.ll *) group "i1"; - insist (define_type_name "Ty02" i1_type m); + insist (define_type_name "i1_type" i1_type m); insist (TypeKind.Integer == classify_type i1_type); - (* RUN: grep {Ty03.*i32} < %t.ll + (* RUN: grep {i32_type.*i32} < %t.ll *) group "i32"; - insist (define_type_name "Ty03" i32_type m); + insist (define_type_name "i32_type" i32_type m); - (* RUN: grep {Ty04.*i42} < %t.ll + (* RUN: grep {i42_type.*i42} < %t.ll *) group "i42"; let ty = integer_type context 42 in - insist (define_type_name "Ty04" ty m); + insist (define_type_name "i42_type" ty m); - (* RUN: grep {Ty05.*float} < %t.ll + (* RUN: grep {float_type.*float} < %t.ll *) group "float"; - insist (define_type_name "Ty05" float_type m); + insist (define_type_name "float_type" float_type m); insist (TypeKind.Float == classify_type float_type); - (* RUN: grep {Ty06.*double} < %t.ll + (* RUN: grep {double_type.*double} < %t.ll *) group "double"; - insist (define_type_name "Ty06" double_type m); + insist (define_type_name "double_type" double_type m); insist (TypeKind.Double == classify_type double_type); - (* RUN: grep {Ty07.*i32.*i1, double} < %t.ll + (* RUN: grep {function_type.*i32.*i1, double} < %t.ll *) group "function"; let ty = function_type i32_type [| i1_type; double_type |] in - insist (define_type_name "Ty07" ty m); + insist (define_type_name "function_type" ty m); insist (TypeKind.Function = classify_type ty); insist (not (is_var_arg ty)); insist (i32_type == return_type ty); insist (double_type == (param_types ty).(1)); - (* RUN: grep {Ty08.*\.\.\.} < %t.ll + (* RUN: grep {var_arg_type.*\.\.\.} < %t.ll *) group "var arg function"; let ty = var_arg_function_type void_type [| i32_type |] in - insist (define_type_name "Ty08" ty m); + insist (define_type_name "var_arg_type" ty m); insist (is_var_arg ty); - (* RUN: grep {Ty09.*\\\[7 x i8\\\]} < %t.ll + (* RUN: grep {array_type.*\\\[7 x i8\\\]} < %t.ll *) group "array"; let ty = array_type i8_type 7 in - insist (define_type_name "Ty09" ty m); + insist (define_type_name "array_type" ty m); insist (7 = array_length ty); insist (i8_type == element_type ty); insist (TypeKind.Array == classify_type ty); begin group "pointer"; - (* RUN: grep {UnqualPtrTy.*float\*} < %t.ll + (* RUN: grep {pointer_type.*float\*} < %t.ll *) let ty = pointer_type float_type in - insist (define_type_name "UnqualPtrTy" ty m); + insist (define_type_name "pointer_type" ty m); insist (float_type == element_type ty); insist (0 == address_space ty); insist (TypeKind.Pointer == classify_type ty) end; begin group "qualified_pointer"; - (* RUN: grep {QualPtrTy.*i8.*3.*\*} < %t.ll + (* RUN: grep {qualified_pointer_type.*i8.*3.*\*} < %t.ll *) let ty = qualified_pointer_type i8_type 3 in - insist (define_type_name "QualPtrTy" ty m); + insist (define_type_name "qualified_pointer_type" ty m); insist (i8_type == element_type ty); insist (3 == address_space ty) end; - (* RUN: grep {Ty11.*\<4 x i16\>} < %t.ll + (* RUN: grep {vector_type.*\<4 x i16\>} < %t.ll *) group "vector"; let ty = vector_type i16_type 4 in - insist (define_type_name "Ty11" ty m); + insist (define_type_name "vector_type" ty m); insist (i16_type == element_type ty); insist (4 = vector_size ty); - (* RUN: grep {Ty12.*opaque} < %t.ll + (* RUN: grep {opaque_type.*opaque} < %t.ll *) group "opaque"; let ty = opaque_type context in - insist (define_type_name "Ty12" ty m); + insist (define_type_name "opaque_type" ty m); insist (ty == ty); insist (ty <> opaque_type context); - (* RUN: grep -v {Ty13} < %t.ll + (* RUN: grep -v {delete_type} < %t.ll *) group "delete"; let ty = opaque_type context in - insist (define_type_name "Ty13" ty m); - delete_type_name "Ty13" m; + insist (define_type_name "delete_type" ty m); + delete_type_name "delete_type" m; + + (* RUN: grep {type_name.*opaque} < %t.ll + *) + group "type_name"; begin + let ty = opaque_type context in + insist (define_type_name "type_name" ty m); + insist ((type_by_name m "type_name") = Some ty) + end; - (* RUN: grep -v {RecursiveTy.*RecursiveTy} < %t.ll + (* RUN: grep -v {recursive_type.*recursive_type} < %t.ll *) group "recursive"; let ty = opaque_type context in let th = handle_to_type ty in refine_type ty (pointer_type ty); let ty = type_of_handle th in - insist (define_type_name "RecursiveTy" ty m); + insist (define_type_name "recursive_type" ty m); insist (ty == element_type ty) (*===-- Constants ---------------------------------------------------------===*) let test_constants () = - (* RUN: grep {Const01.*i32.*-1} < %t.ll + (* RUN: grep {const_int.*i32.*-1} < %t.ll *) group "int"; let c = const_int i32_type (-1) in - ignore (define_global "Const01" c m); + ignore (define_global "const_int" c m); insist (i32_type = type_of c); insist (is_constant c); - (* RUN: grep {Const02.*i64.*-1} < %t.ll + (* RUN: grep {const_sext_int.*i64.*-1} < %t.ll *) group "sext int"; let c = const_int i64_type (-1) in - ignore (define_global "Const02" c m); + ignore (define_global "const_sext_int" c m); insist (i64_type = type_of c); - (* RUN: grep {Const03.*i64.*4294967295} < %t.ll + (* RUN: grep {const_zext_int64.*i64.*4294967295} < %t.ll *) group "zext int64"; let c = const_of_int64 i64_type (Int64.of_string "4294967295") false in - ignore (define_global "Const03" c m); + ignore (define_global "const_zext_int64" c m); insist (i64_type = type_of c); - (* RUN: grep {ConstIntString.*i32.*-1} < %t.ll + (* RUN: grep {const_int_string.*i32.*-1} < %t.ll *) group "int string"; let c = const_int_of_string i32_type "-1" 10 in - ignore (define_global "ConstIntString" c m); + ignore (define_global "const_int_string" c m); insist (i32_type = type_of c); - (* RUN: grep {Const04.*"cruel\\\\00world"} < %t.ll + (* RUN: grep {const_string.*"cruel\\\\00world"} < %t.ll *) group "string"; let c = const_string context "cruel\000world" in - ignore (define_global "Const04" c m); + ignore (define_global "const_string" c m); insist ((array_type i8_type 11) = type_of c); - (* RUN: grep {Const05.*"hi\\\\00again\\\\00"} < %t.ll + (* RUN: grep {const_stringz.*"hi\\\\00again\\\\00"} < %t.ll *) group "stringz"; let c = const_stringz context "hi\000again" in - ignore (define_global "Const05" c m); + ignore (define_global "const_stringz" c m); insist ((array_type i8_type 9) = type_of c); - (* RUN: grep {ConstSingle.*2.75} < %t.ll - * RUN: grep {ConstDouble.*3.1459} < %t.ll - * RUN: grep {ConstDoubleString.*1.25} < %t.ll + (* RUN: grep {const_single.*2.75} < %t.ll + * RUN: grep {const_double.*3.1459} < %t.ll + * RUN: grep {const_double_string.*1.25} < %t.ll *) begin group "real"; let cs = const_float float_type 2.75 in - ignore (define_global "ConstSingle" cs m); + ignore (define_global "const_single" cs m); insist (float_type = type_of cs); let cd = const_float double_type 3.1459 in - ignore (define_global "ConstDouble" cd m); + ignore (define_global "const_double" cd m); insist (double_type = type_of cd); let cd = const_float_of_string double_type "1.25" in - ignore (define_global "ConstDoubleString" cd m); + ignore (define_global "const_double_string" cd m); insist (double_type = type_of cd) end; @@ -266,67 +273,93 @@ let test_constants () = let three = const_int i32_type 3 in let four = const_int i32_type 4 in - (* RUN: grep {Const07.*\\\[i32 3, i32 4\\\]} < %t.ll + (* RUN: grep {const_array.*\\\[i32 3, i32 4\\\]} < %t.ll *) group "array"; let c = const_array i32_type [| three; four |] in - ignore (define_global "Const07" c m); + ignore (define_global "const_array" c m); insist ((array_type i32_type 2) = (type_of c)); - (* RUN: grep {Const08.*<i16 1, i16 2.*>} < %t.ll + (* RUN: grep {const_vector.*<i16 1, i16 2.*>} < %t.ll *) group "vector"; let c = const_vector [| one; two; one; two; one; two; one; two |] in - ignore (define_global "Const08" c m); + ignore (define_global "const_vector" c m); insist ((vector_type i16_type 8) = (type_of c)); - (* RUN: grep {Const09.*.i16 1, i16 2, i32 3, i32 4} < %t.ll + (* RUN: grep {const_structure.*.i16 1, i16 2, i32 3, i32 4} < %t.ll *) group "structure"; let c = const_struct context [| one; two; three; four |] in - ignore (define_global "Const09" c m); + ignore (define_global "const_structure" c m); insist ((struct_type context [| i16_type; i16_type; i32_type; i32_type |]) = (type_of c)); + + group "union"; + let t = union_type context [| i1_type; i16_type; i64_type; double_type |] in + let c = const_union t one in + ignore (define_global "const_union" c m); + insist (t = (type_of c)); - (* RUN: grep {Const10.*zeroinit} < %t.ll + (* RUN: grep {const_null.*zeroinit} < %t.ll *) group "null"; let c = const_null (packed_struct_type context [| i1_type; i8_type; i64_type; double_type |]) in - ignore (define_global "Const10" c m); + ignore (define_global "const_null" c m); - (* RUN: grep {Const11.*-1} < %t.ll + (* RUN: grep {const_all_ones.*-1} < %t.ll *) group "all ones"; let c = const_all_ones i64_type in - ignore (define_global "Const11" c m); + ignore (define_global "const_all_ones" c m); + + group "pointer null"; begin + (* RUN: grep {const_pointer_null = global i64\\* null} < %t.ll + *) + let c = const_pointer_null (pointer_type i64_type) in + ignore (define_global "const_pointer_null" c m); + end; - (* RUN: grep {Const12.*undef} < %t.ll + (* RUN: grep {const_undef.*undef} < %t.ll *) group "undef"; let c = undef i1_type in - ignore (define_global "Const12" c m); + ignore (define_global "const_undef" c m); insist (i1_type = type_of c); insist (is_undef c); group "constant arithmetic"; - (* RUN: grep {ConstNeg.*sub} < %t.ll - * RUN: grep {ConstNot.*xor} < %t.ll - * RUN: grep {ConstAdd.*add} < %t.ll - * RUN: grep {ConstSub.*sub} < %t.ll - * RUN: grep {ConstMul.*mul} < %t.ll - * RUN: grep {ConstUDiv.*udiv} < %t.ll - * RUN: grep {ConstSDiv.*sdiv} < %t.ll - * RUN: grep {ConstFDiv.*fdiv} < %t.ll - * RUN: grep {ConstURem.*urem} < %t.ll - * RUN: grep {ConstSRem.*srem} < %t.ll - * RUN: grep {ConstFRem.*frem} < %t.ll - * RUN: grep {ConstAnd.*and} < %t.ll - * RUN: grep {ConstOr.*or} < %t.ll - * RUN: grep {ConstXor.*xor} < %t.ll - * RUN: grep {ConstICmp.*icmp} < %t.ll - * RUN: grep {ConstFCmp.*fcmp} < %t.ll + (* RUN: grep {@const_neg = global i64 sub} < %t.ll + * RUN: grep {@const_nsw_neg = global i64 sub nsw } < %t.ll + * RUN: grep {@const_nuw_neg = global i64 sub nuw } < %t.ll + * RUN: grep {@const_fneg = global double fsub } < %t.ll + * RUN: grep {@const_not = global i64 xor } < %t.ll + * RUN: grep {@const_add = global i64 add } < %t.ll + * RUN: grep {@const_nsw_add = global i64 add nsw } < %t.ll + * RUN: grep {@const_nuw_add = global i64 add nuw } < %t.ll + * RUN: grep {@const_fadd = global double fadd } < %t.ll + * RUN: grep {@const_sub = global i64 sub } < %t.ll + * RUN: grep {@const_nsw_sub = global i64 sub nsw } < %t.ll + * RUN: grep {@const_nuw_sub = global i64 sub nuw } < %t.ll + * RUN: grep {@const_fsub = global double fsub } < %t.ll + * RUN: grep {@const_mul = global i64 mul } < %t.ll + * RUN: grep {@const_nsw_mul = global i64 mul nsw } < %t.ll + * RUN: grep {@const_nuw_mul = global i64 mul nuw } < %t.ll + * RUN: grep {@const_fmul = global double fmul } < %t.ll + * RUN: grep {@const_udiv = global i64 udiv } < %t.ll + * RUN: grep {@const_sdiv = global i64 sdiv } < %t.ll + * RUN: grep {@const_exact_sdiv = global i64 sdiv exact } < %t.ll + * RUN: grep {@const_fdiv = global double fdiv } < %t.ll + * RUN: grep {@const_urem = global i64 urem } < %t.ll + * RUN: grep {@const_srem = global i64 srem } < %t.ll + * RUN: grep {@const_frem = global double frem } < %t.ll + * RUN: grep {@const_and = global i64 and } < %t.ll + * RUN: grep {@const_or = global i64 or } < %t.ll + * RUN: grep {@const_xor = global i64 xor } < %t.ll + * RUN: grep {@const_icmp = global i1 icmp sle } < %t.ll + * RUN: grep {@const_fcmp = global i1 fcmp ole } < %t.ll *) let void_ptr = pointer_type i8_type in let five = const_int i64_type 5 in @@ -334,82 +367,105 @@ let test_constants () = let foldbomb_gv = define_global "FoldBomb" (const_null i8_type) m in let foldbomb = const_ptrtoint foldbomb_gv i64_type in let ffoldbomb = const_uitofp foldbomb double_type in - ignore (define_global "ConstNeg" (const_neg foldbomb) m); - ignore (define_global "ConstNot" (const_not foldbomb) m); - ignore (define_global "ConstAdd" (const_add foldbomb five) m); - ignore (define_global "ConstSub" (const_sub foldbomb five) m); - ignore (define_global "ConstMul" (const_mul foldbomb five) m); - ignore (define_global "ConstUDiv" (const_udiv foldbomb five) m); - ignore (define_global "ConstSDiv" (const_sdiv foldbomb five) m); - ignore (define_global "ConstFDiv" (const_fdiv ffoldbomb ffive) m); - ignore (define_global "ConstURem" (const_urem foldbomb five) m); - ignore (define_global "ConstSRem" (const_srem foldbomb five) m); - ignore (define_global "ConstFRem" (const_frem ffoldbomb ffive) m); - ignore (define_global "ConstAnd" (const_and foldbomb five) m); - ignore (define_global "ConstOr" (const_or foldbomb five) m); - ignore (define_global "ConstXor" (const_xor foldbomb five) m); - ignore (define_global "ConstICmp" (const_icmp Icmp.Sle foldbomb five) m); - ignore (define_global "ConstFCmp" (const_fcmp Fcmp.Ole ffoldbomb ffive) m); + ignore (define_global "const_neg" (const_neg foldbomb) m); + ignore (define_global "const_nsw_neg" (const_nsw_neg foldbomb) m); + ignore (define_global "const_nuw_neg" (const_nuw_neg foldbomb) m); + ignore (define_global "const_fneg" (const_fneg ffoldbomb) m); + ignore (define_global "const_not" (const_not foldbomb) m); + ignore (define_global "const_add" (const_add foldbomb five) m); + ignore (define_global "const_nsw_add" (const_nsw_add foldbomb five) m); + ignore (define_global "const_nuw_add" (const_nuw_add foldbomb five) m); + ignore (define_global "const_fadd" (const_fadd ffoldbomb ffive) m); + ignore (define_global "const_sub" (const_sub foldbomb five) m); + ignore (define_global "const_nsw_sub" (const_nsw_sub foldbomb five) m); + ignore (define_global "const_nuw_sub" (const_nuw_sub foldbomb five) m); + ignore (define_global "const_fsub" (const_fsub ffoldbomb ffive) m); + ignore (define_global "const_mul" (const_mul foldbomb five) m); + ignore (define_global "const_nsw_mul" (const_nsw_mul foldbomb five) m); + ignore (define_global "const_nuw_mul" (const_nuw_mul foldbomb five) m); + ignore (define_global "const_fmul" (const_fmul ffoldbomb ffive) m); + ignore (define_global "const_udiv" (const_udiv foldbomb five) m); + ignore (define_global "const_sdiv" (const_sdiv foldbomb five) m); + ignore (define_global "const_exact_sdiv" (const_exact_sdiv foldbomb five) m); + ignore (define_global "const_fdiv" (const_fdiv ffoldbomb ffive) m); + ignore (define_global "const_urem" (const_urem foldbomb five) m); + ignore (define_global "const_srem" (const_srem foldbomb five) m); + ignore (define_global "const_frem" (const_frem ffoldbomb ffive) m); + ignore (define_global "const_and" (const_and foldbomb five) m); + ignore (define_global "const_or" (const_or foldbomb five) m); + ignore (define_global "const_xor" (const_xor foldbomb five) m); + ignore (define_global "const_icmp" (const_icmp Icmp.Sle foldbomb five) m); + ignore (define_global "const_fcmp" (const_fcmp Fcmp.Ole ffoldbomb ffive) m); group "constant casts"; - (* RUN: grep {ConstTrunc.*trunc} < %t.ll - * RUN: grep {ConstSExt.*sext} < %t.ll - * RUN: grep {ConstZExt.*zext} < %t.ll - * RUN: grep {ConstFPTrunc.*fptrunc} < %t.ll - * RUN: grep {ConstFPExt.*fpext} < %t.ll - * RUN: grep {ConstUIToFP.*uitofp} < %t.ll - * RUN: grep {ConstSIToFP.*sitofp} < %t.ll - * RUN: grep {ConstFPToUI.*fptoui} < %t.ll - * RUN: grep {ConstFPToSI.*fptosi} < %t.ll - * RUN: grep {ConstPtrToInt.*ptrtoint} < %t.ll - * RUN: grep {ConstIntToPtr.*inttoptr} < %t.ll - * RUN: grep {ConstBitCast.*bitcast} < %t.ll + (* RUN: grep {const_trunc.*trunc} < %t.ll + * RUN: grep {const_sext.*sext} < %t.ll + * RUN: grep {const_zext.*zext} < %t.ll + * RUN: grep {const_fptrunc.*fptrunc} < %t.ll + * RUN: grep {const_fpext.*fpext} < %t.ll + * RUN: grep {const_uitofp.*uitofp} < %t.ll + * RUN: grep {const_sitofp.*sitofp} < %t.ll + * RUN: grep {const_fptoui.*fptoui} < %t.ll + * RUN: grep {const_fptosi.*fptosi} < %t.ll + * RUN: grep {const_ptrtoint.*ptrtoint} < %t.ll + * RUN: grep {const_inttoptr.*inttoptr} < %t.ll + * RUN: grep {const_bitcast.*bitcast} < %t.ll *) let i128_type = integer_type context 128 in - ignore (define_global "ConstTrunc" (const_trunc (const_add foldbomb five) + ignore (define_global "const_trunc" (const_trunc (const_add foldbomb five) i8_type) m); - ignore (define_global "ConstSExt" (const_sext foldbomb i128_type) m); - ignore (define_global "ConstZExt" (const_zext foldbomb i128_type) m); - ignore (define_global "ConstFPTrunc" (const_fptrunc ffoldbomb float_type) m); - ignore (define_global "ConstFPExt" (const_fpext ffoldbomb fp128_type) m); - ignore (define_global "ConstUIToFP" (const_uitofp foldbomb double_type) m); - ignore (define_global "ConstSIToFP" (const_sitofp foldbomb double_type) m); - ignore (define_global "ConstFPToUI" (const_fptoui ffoldbomb i32_type) m); - ignore (define_global "ConstFPToSI" (const_fptosi ffoldbomb i32_type) m); - ignore (define_global "ConstPtrToInt" (const_ptrtoint + ignore (define_global "const_sext" (const_sext foldbomb i128_type) m); + ignore (define_global "const_zext" (const_zext foldbomb i128_type) m); + ignore (define_global "const_fptrunc" (const_fptrunc ffoldbomb float_type) m); + ignore (define_global "const_fpext" (const_fpext ffoldbomb fp128_type) m); + ignore (define_global "const_uitofp" (const_uitofp foldbomb double_type) m); + ignore (define_global "const_sitofp" (const_sitofp foldbomb double_type) m); + ignore (define_global "const_fptoui" (const_fptoui ffoldbomb i32_type) m); + ignore (define_global "const_fptosi" (const_fptosi ffoldbomb i32_type) m); + ignore (define_global "const_ptrtoint" (const_ptrtoint (const_gep (const_null (pointer_type i8_type)) [| const_int i32_type 1 |]) i32_type) m); - ignore (define_global "ConstIntToPtr" (const_inttoptr (const_add foldbomb five) + ignore (define_global "const_inttoptr" (const_inttoptr (const_add foldbomb five) void_ptr) m); - ignore (define_global "ConstBitCast" (const_bitcast ffoldbomb i64_type) m); + ignore (define_global "const_bitcast" (const_bitcast ffoldbomb i64_type) m); group "misc constants"; - (* RUN: grep {ConstSizeOf.*getelementptr.*null} < %t.ll - * RUN: grep {ConstGEP.*getelementptr} < %t.ll - * RUN: grep {ConstSelect.*select} < %t.ll - * RUN: grep {ConstExtractElement.*extractelement} < %t.ll - * RUN: grep {ConstInsertElement.*insertelement} < %t.ll - * RUN: grep {ConstShuffleVector.*shufflevector} < %t.ll + (* RUN: grep {const_size_of.*getelementptr.*null} < %t.ll + * RUN: grep {const_gep.*getelementptr} < %t.ll + * RUN: grep {const_select.*select} < %t.ll + * RUN: grep {const_extractelement.*extractelement} < %t.ll + * RUN: grep {const_insertelement.*insertelement} < %t.ll + * RUN: grep {const_shufflevector.*shufflevector} < %t.ll *) - ignore (define_global "ConstSizeOf" (size_of (pointer_type i8_type)) m); - ignore (define_global "ConstGEP" (const_gep foldbomb_gv [| five |]) m); - ignore (define_global "ConstSelect" (const_select + ignore (define_global "const_size_of" (size_of (pointer_type i8_type)) m); + ignore (define_global "const_gep" (const_gep foldbomb_gv [| five |]) m); + ignore (define_global "const_select" (const_select (const_icmp Icmp.Sle foldbomb five) (const_int i8_type (-1)) (const_int i8_type 0)) m); let zero = const_int i32_type 0 in let one = const_int i32_type 1 in - ignore (define_global "ConstExtractElement" (const_extractelement + ignore (define_global "const_extractelement" (const_extractelement (const_vector [| zero; one; zero; one |]) (const_trunc foldbomb i32_type)) m); - ignore (define_global "ConstInsertElement" (const_insertelement + ignore (define_global "const_insertelement" (const_insertelement (const_vector [| zero; one; zero; one |]) zero (const_trunc foldbomb i32_type)) m); - ignore (define_global "ConstShuffleVector" (const_shufflevector + ignore (define_global "const_shufflevector" (const_shufflevector (const_vector [| zero; one |]) (const_vector [| one; zero |]) - (const_bitcast foldbomb (vector_type i32_type 2))) m) + (const_bitcast foldbomb (vector_type i32_type 2))) m); + + group "asm"; begin + let ft = function_type void_type [| i32_type; i32_type; i32_type |] in + ignore (const_inline_asm + ft + "" + "{cx},{ax},{di},~{dirflag},~{fpsr},~{flags},~{edi},~{ecx}" + true + false) + end (*===-- Global Values -----------------------------------------------------===*) @@ -461,28 +517,46 @@ let test_global_variables () = let (++) x f = f x; x in let fourty_two32 = const_int i32_type 42 in - (* RUN: grep {GVar01.*external} < %t.ll - *) - group "declarations"; - insist (None == lookup_global "GVar01" m); - let g = declare_global i32_type "GVar01" m in - insist (is_declaration g); - insist (pointer_type float_type == - type_of (declare_global float_type "GVar01" m)); - insist (g == declare_global i32_type "GVar01" m); - insist (match lookup_global "GVar01" m with Some x -> x = g - | None -> false); + group "declarations"; begin + (* RUN: grep {GVar01.*external} < %t.ll + *) + insist (None == lookup_global "GVar01" m); + let g = declare_global i32_type "GVar01" m in + insist (is_declaration g); + insist (pointer_type float_type == + type_of (declare_global float_type "GVar01" m)); + insist (g == declare_global i32_type "GVar01" m); + insist (match lookup_global "GVar01" m with Some x -> x = g + | None -> false); + + insist (None == lookup_global "QGVar01" m); + let g = declare_qualified_global i32_type "QGVar01" 3 m in + insist (is_declaration g); + insist (qualified_pointer_type float_type 3 == + type_of (declare_qualified_global float_type "QGVar01" 3 m)); + insist (g == declare_qualified_global i32_type "QGVar01" 3 m); + insist (match lookup_global "QGVar01" m with Some x -> x = g + | None -> false); + end; - (* RUN: grep {GVar02.*42} < %t.ll - * RUN: grep {GVar03.*42} < %t.ll - *) - group "definitions"; - let g = define_global "GVar02" fourty_two32 m in - let g2 = declare_global i32_type "GVar03" m ++ + group "definitions"; begin + (* RUN: grep {GVar02.*42} < %t.ll + * RUN: grep {GVar03.*42} < %t.ll + *) + let g = define_global "GVar02" fourty_two32 m in + let g2 = declare_global i32_type "GVar03" m ++ + set_initializer fourty_two32 in + insist (not (is_declaration g)); + insist (not (is_declaration g2)); + insist ((global_initializer g) == (global_initializer g2)); + + let g = define_qualified_global "QGVar02" fourty_two32 3 m in + let g2 = declare_qualified_global i32_type "QGVar03" 3 m ++ set_initializer fourty_two32 in - insist (not (is_declaration g)); - insist (not (is_declaration g2)); - insist ((global_initializer g) == (global_initializer g2)); + insist (not (is_declaration g)); + insist (not (is_declaration g2)); + insist ((global_initializer g) == (global_initializer g2)); + end; (* RUN: grep {GVar04.*thread_local} < %t.ll *) @@ -532,6 +606,59 @@ let test_global_variables () = end +(*===-- Uses --------------------------------------------------------------===*) + +let test_uses () = + let ty = function_type i32_type [| i32_type; i32_type |] in + let fn = define_function "use_function" ty m in + let b = builder_at_end context (entry_block fn) in + + let p1 = param fn 0 in + let p2 = param fn 1 in + let v1 = build_add p1 p2 "v1" b in + let v2 = build_add p1 v1 "v2" b in + let _ = build_add v1 v2 "v3" b in + + let lf s u = value_name (user u) ^ "->" ^ s in + insist ("v2->v3->" = fold_left_uses lf "" v1); + let rf u s = value_name (user u) ^ "<-" ^ s in + insist ("v3<-v2<-" = fold_right_uses rf v1 ""); + + let lf s u = value_name (used_value u) ^ "->" ^ s in + insist ("v1->v1->" = fold_left_uses lf "" v1); + + let rf u s = value_name (used_value u) ^ "<-" ^ s in + insist ("v1<-v1<-" = fold_right_uses rf v1 ""); + + ignore (build_unreachable b) + + +(*===-- Users -------------------------------------------------------------===*) + +let test_users () = + let ty = function_type i32_type [| i32_type; i32_type |] in + let fn = define_function "user_function" ty m in + let b = builder_at_end context (entry_block fn) in + + let p1 = param fn 0 in + let p2 = param fn 1 in + let i = build_add p1 p2 "sum" b in + + insist ((operand i 0) = p1); + insist ((operand i 1) = p2); + + ignore (build_unreachable b) + + +(*===-- Aliases -----------------------------------------------------------===*) + +let test_aliases () = + (* RUN: grep {@alias = alias i32\\* @aliasee} < %t.ll + *) + let v = declare_global i32_type "aliasee" m in + ignore (add_alias m (pointer_type i32_type) v "alias") + + (*===-- Functions ---------------------------------------------------------===*) let test_functions () = @@ -845,11 +972,11 @@ let test_builder () = end; group "cond_br"; begin - (* RUN: grep {br.*Inst01.*Bb03.*Bb00} < %t.ll + (* RUN: grep {br.*build_br.*Bb03.*Bb00} < %t.ll *) let bb03 = append_block context "Bb03" fn in let b = builder_at_end context bb03 in - let cond = build_trunc p1 i1_type "Inst01" b in + let cond = build_trunc p1 i1_type "build_br" b in ignore (build_cond_br cond bb03 bb00 b) end; @@ -865,14 +992,31 @@ let test_builder () = let si = build_switch p1 bb3 1 (builder_at_end context bb1) in ignore (add_case si (const_int i32_type 2) bb2) end; + + group "indirectbr"; begin + (* RUN: grep {indirectbr i8\\* blockaddress(@X7, %IBRBlock2), \\\[label %IBRBlock2, label %IBRBlock3\\\]} < %t.ll + *) + let bb1 = append_block context "IBRBlock1" fn in + + let bb2 = append_block context "IBRBlock2" fn in + ignore (build_unreachable (builder_at_end context bb2)); + + let bb3 = append_block context "IBRBlock3" fn in + ignore (build_unreachable (builder_at_end context bb3)); + + let addr = block_address fn bb2 in + let ibr = build_indirect_br addr 2 (builder_at_end context bb1) in + ignore (add_destination ibr bb2); + ignore (add_destination ibr bb3) + end; group "invoke"; begin - (* RUN: grep {Inst02.*invoke.*P1.*P2} < %t.ll + (* RUN: grep {build_invoke.*invoke.*P1.*P2} < %t.ll * RUN: grep {to.*Bb04.*unwind.*Bb00} < %t.ll *) let bb04 = append_block context "Bb04" fn in let b = builder_at_end context bb04 in - ignore (build_invoke fn [| p1; p2 |] bb04 bb00 "Inst02" b) + ignore (build_invoke fn [| p1; p2 |] bb04 bb00 "build_invoke" b) end; group "unwind"; begin @@ -895,121 +1039,139 @@ let test_builder () = let bb07 = append_block context "Bb07" fn in let b = builder_at_end context bb07 in - (* RUN: grep {Inst03.*add.*P1.*P2} < %t.ll - * RUN: grep {Inst04.*sub.*P1.*Inst03} < %t.ll - * RUN: grep {Inst05.*mul.*P1.*Inst04} < %t.ll - * RUN: grep {Inst06.*udiv.*P1.*Inst05} < %t.ll - * RUN: grep {Inst07.*sdiv.*P1.*Inst06} < %t.ll - * RUN: grep {Inst08.*fdiv.*F1.*F2} < %t.ll - * RUN: grep {Inst09.*urem.*P1.*Inst07} < %t.ll - * RUN: grep {Inst10.*srem.*P1.*Inst09} < %t.ll - * RUN: grep {Inst11.*frem.*F1.*Inst08} < %t.ll - * RUN: grep {Inst12.*shl.*P1.*Inst10} < %t.ll - * RUN: grep {Inst13.*lshr.*P1.*Inst12} < %t.ll - * RUN: grep {Inst14.*ashr.*P1.*Inst13} < %t.ll - * RUN: grep {Inst15.*and.*P1.*Inst14} < %t.ll - * RUN: grep {Inst16.*or.*P1.*Inst15} < %t.ll - * RUN: grep {Inst17.*xor.*P1.*Inst16} < %t.ll - * RUN: grep {Inst18.*sub.*0.*Inst17} < %t.ll - * RUN: grep {Inst19.*xor.*Inst18.*-1} < %t.ll + (* RUN: grep {%build_add = add i32 %P1, %P2} < %t.ll + * RUN: grep {%build_nsw_add = add nsw i32 %P1, %P2} < %t.ll + * RUN: grep {%build_nuw_add = add nuw i32 %P1, %P2} < %t.ll + * RUN: grep {%build_fadd = fadd float %F1, %F2} < %t.ll + * RUN: grep {%build_sub = sub i32 %P1, %P2} < %t.ll + * RUN: grep {%build_nsw_sub = sub nsw i32 %P1, %P2} < %t.ll + * RUN: grep {%build_nuw_sub = sub nuw i32 %P1, %P2} < %t.ll + * RUN: grep {%build_fsub = fsub float %F1, %F2} < %t.ll + * RUN: grep {%build_mul = mul i32 %P1, %P2} < %t.ll + * RUN: grep {%build_nsw_mul = mul nsw i32 %P1, %P2} < %t.ll + * RUN: grep {%build_nuw_mul = mul nuw i32 %P1, %P2} < %t.ll + * RUN: grep {%build_fmul = fmul float %F1, %F2} < %t.ll + * RUN: grep {%build_udiv = udiv i32 %P1, %P2} < %t.ll + * RUN: grep {%build_sdiv = sdiv i32 %P1, %P2} < %t.ll + * RUN: grep {%build_exact_sdiv = sdiv exact i32 %P1, %P2} < %t.ll + * RUN: grep {%build_fdiv = fdiv float %F1, %F2} < %t.ll + * RUN: grep {%build_urem = urem i32 %P1, %P2} < %t.ll + * RUN: grep {%build_srem = srem i32 %P1, %P2} < %t.ll + * RUN: grep {%build_frem = frem float %F1, %F2} < %t.ll + * RUN: grep {%build_shl = shl i32 %P1, %P2} < %t.ll + * RUN: grep {%build_lshl = lshr i32 %P1, %P2} < %t.ll + * RUN: grep {%build_ashl = ashr i32 %P1, %P2} < %t.ll + * RUN: grep {%build_and = and i32 %P1, %P2} < %t.ll + * RUN: grep {%build_or = or i32 %P1, %P2} < %t.ll + * RUN: grep {%build_xor = xor i32 %P1, %P2} < %t.ll + * RUN: grep {%build_neg = sub i32 0, %P1} < %t.ll + * RUN: grep {%build_nsw_neg = sub nsw i32 0, %P1} < %t.ll + * RUN: grep {%build_nuw_neg = sub nuw i32 0, %P1} < %t.ll + * RUN: grep {%build_fneg = fsub float .*0.*, %F1} < %t.ll + * RUN: grep {%build_not = xor i32 %P1, -1} < %t.ll *) - let inst03 = build_add p1 p2 "Inst03" b in - let inst04 = build_sub p1 inst03 "Inst04" b in - let inst05 = build_mul p1 inst04 "Inst05" b in - let inst06 = build_udiv p1 inst05 "Inst06" b in - let inst07 = build_sdiv p1 inst06 "Inst07" b in - let inst08 = build_fdiv f1 f2 "Inst08" b in - let inst09 = build_urem p1 inst07 "Inst09" b in - let inst10 = build_srem p1 inst09 "Inst10" b in - ignore(build_frem f1 inst08 "Inst11" b); - let inst12 = build_shl p1 inst10 "Inst12" b in - let inst13 = build_lshr p1 inst12 "Inst13" b in - let inst14 = build_ashr p1 inst13 "Inst14" b in - let inst15 = build_and p1 inst14 "Inst15" b in - let inst16 = build_or p1 inst15 "Inst16" b in - let inst17 = build_xor p1 inst16 "Inst17" b in - let inst18 = build_neg inst17 "Inst18" b in - ignore (build_not inst18 "Inst19" b); - ignore (build_unreachable b) + ignore (build_add p1 p2 "build_add" b); + ignore (build_nsw_add p1 p2 "build_nsw_add" b); + ignore (build_nuw_add p1 p2 "build_nuw_add" b); + ignore (build_fadd f1 f2 "build_fadd" b); + ignore (build_sub p1 p2 "build_sub" b); + ignore (build_nsw_sub p1 p2 "build_nsw_sub" b); + ignore (build_nuw_sub p1 p2 "build_nuw_sub" b); + ignore (build_fsub f1 f2 "build_fsub" b); + ignore (build_mul p1 p2 "build_mul" b); + ignore (build_nsw_mul p1 p2 "build_nsw_mul" b); + ignore (build_nuw_mul p1 p2 "build_nuw_mul" b); + ignore (build_fmul f1 f2 "build_fmul" b); + ignore (build_udiv p1 p2 "build_udiv" b); + ignore (build_sdiv p1 p2 "build_sdiv" b); + ignore (build_exact_sdiv p1 p2 "build_exact_sdiv" b); + ignore (build_fdiv f1 f2 "build_fdiv" b); + ignore (build_urem p1 p2 "build_urem" b); + ignore (build_srem p1 p2 "build_srem" b); + ignore (build_frem f1 f2 "build_frem" b); + ignore (build_shl p1 p2 "build_shl" b); + ignore (build_lshr p1 p2 "build_lshl" b); + ignore (build_ashr p1 p2 "build_ashl" b); + ignore (build_and p1 p2 "build_and" b); + ignore (build_or p1 p2 "build_or" b); + ignore (build_xor p1 p2 "build_xor" b); + ignore (build_neg p1 "build_neg" b); + ignore (build_nsw_neg p1 "build_nsw_neg" b); + ignore (build_nuw_neg p1 "build_nuw_neg" b); + ignore (build_fneg f1 "build_fneg" b); + ignore (build_not p1 "build_not" b); + ignore (build_unreachable b) end; group "memory"; begin let bb08 = append_block context "Bb08" fn in let b = builder_at_end context bb08 in - (* RUN: grep {Inst20.*malloc} < %t.ll - * RUN: grep {Inst21.*malloc} < %t.ll - * RUN: grep {Inst22.*alloca.*i32 } < %t.ll - * RUN: grep {Inst23.*alloca.*i32.*P2} < %t.ll - * RUN: grep {free.*Inst20} < %t.ll - * RUN: grep {Inst25.*load.*Inst21} < %t.ll - * RUN: grep {store.*P2.*Inst22} < %t.ll - * RUN: grep {Inst27.*getelementptr.*Inst23.*P2} < %t.ll + (* RUN: grep {%build_alloca = alloca i32} < %t.ll + * RUN: grep {%build_array_alloca = alloca i32, i32 %P2} < %t.ll + * RUN: grep {%build_load = load i32\\* %build_array_alloca} < %t.ll + * RUN: grep {store i32 %P2, i32\\* %build_alloca} < %t.ll + * RUN: grep {%build_gep = getelementptr i32\\* %build_array_alloca, i32 %P2} < %t.ll *) - let inst20 = build_malloc i8_type "Inst20" b in - let inst21 = build_array_malloc i8_type p1 "Inst21" b in - let inst22 = build_alloca i32_type "Inst22" b in - let inst23 = build_array_alloca i32_type p2 "Inst23" b in - ignore(build_free inst20 b); - ignore(build_load inst21 "Inst25" b); - ignore(build_store p2 inst22 b); - ignore(build_gep inst23 [| p2 |] "Inst27" b); - ignore(build_unreachable b) + let alloca = build_alloca i32_type "build_alloca" b in + let array_alloca = build_array_alloca i32_type p2 "build_array_alloca" b in + ignore(build_load array_alloca "build_load" b); + ignore(build_store p2 alloca b); + ignore(build_gep array_alloca [| p2 |] "build_gep" b); + ignore(build_unreachable b) end; group "casts"; begin let void_ptr = pointer_type i8_type in - (* RUN: grep {Inst28.*trunc.*P1.*i8} < %t.ll - * RUN: grep {Inst29.*zext.*Inst28.*i32} < %t.ll - * RUN: grep {Inst30.*sext.*Inst29.*i64} < %t.ll - * RUN: grep {Inst31.*uitofp.*Inst30.*float} < %t.ll - * RUN: grep {Inst32.*sitofp.*Inst29.*double} < %t.ll - * RUN: grep {Inst33.*fptoui.*Inst31.*i32} < %t.ll - * RUN: grep {Inst34.*fptosi.*Inst32.*i64} < %t.ll - * RUN: grep {Inst35.*fptrunc.*Inst32.*float} < %t.ll - * RUN: grep {Inst36.*fpext.*Inst35.*double} < %t.ll - * RUN: grep {Inst37.*inttoptr.*P1.*i8\*} < %t.ll - * RUN: grep {Inst38.*ptrtoint.*Inst37.*i64} < %t.ll - * RUN: grep {Inst39.*bitcast.*Inst38.*double} < %t.ll + (* RUN: grep {%build_trunc = trunc i32 %P1 to i8} < %t.ll + * RUN: grep {%build_zext = zext i8 %build_trunc to i32} < %t.ll + * RUN: grep {%build_sext = sext i32 %build_zext to i64} < %t.ll + * RUN: grep {%build_uitofp = uitofp i64 %build_sext to float} < %t.ll + * RUN: grep {%build_sitofp = sitofp i32 %build_zext to double} < %t.ll + * RUN: grep {%build_fptoui = fptoui float %build_uitofp to i32} < %t.ll + * RUN: grep {%build_fptosi = fptosi double %build_sitofp to i64} < %t.ll + * RUN: grep {%build_fptrunc = fptrunc double %build_sitofp to float} < %t.ll + * RUN: grep {%build_fpext = fpext float %build_fptrunc to double} < %t.ll + * RUN: grep {%build_inttoptr = inttoptr i32 %P1 to i8\\*} < %t.ll + * RUN: grep {%build_ptrtoint = ptrtoint i8\\* %build_inttoptr to i64} < %t.ll + * RUN: grep {%build_bitcast = bitcast i64 %build_ptrtoint to double} < %t.ll *) - let inst28 = build_trunc p1 i8_type "Inst28" atentry in - let inst29 = build_zext inst28 i32_type "Inst29" atentry in - let inst30 = build_sext inst29 i64_type "Inst30" atentry in - let inst31 = build_uitofp inst30 float_type "Inst31" atentry in - let inst32 = build_sitofp inst29 double_type "Inst32" atentry in - ignore(build_fptoui inst31 i32_type "Inst33" atentry); - ignore(build_fptosi inst32 i64_type "Inst34" atentry); - let inst35 = build_fptrunc inst32 float_type "Inst35" atentry in - ignore(build_fpext inst35 double_type "Inst36" atentry); - let inst37 = build_inttoptr p1 void_ptr "Inst37" atentry in - let inst38 = build_ptrtoint inst37 i64_type "Inst38" atentry in - ignore(build_bitcast inst38 double_type "Inst39" atentry) + let inst28 = build_trunc p1 i8_type "build_trunc" atentry in + let inst29 = build_zext inst28 i32_type "build_zext" atentry in + let inst30 = build_sext inst29 i64_type "build_sext" atentry in + let inst31 = build_uitofp inst30 float_type "build_uitofp" atentry in + let inst32 = build_sitofp inst29 double_type "build_sitofp" atentry in + ignore(build_fptoui inst31 i32_type "build_fptoui" atentry); + ignore(build_fptosi inst32 i64_type "build_fptosi" atentry); + let inst35 = build_fptrunc inst32 float_type "build_fptrunc" atentry in + ignore(build_fpext inst35 double_type "build_fpext" atentry); + let inst37 = build_inttoptr p1 void_ptr "build_inttoptr" atentry in + let inst38 = build_ptrtoint inst37 i64_type "build_ptrtoint" atentry in + ignore(build_bitcast inst38 double_type "build_bitcast" atentry) end; group "comparisons"; begin - (* RUN: grep {Inst40.*icmp.*ne.*P1.*P2} < %t.ll - * RUN: grep {Inst41.*icmp.*sle.*P2.*P1} < %t.ll - * RUN: grep {Inst42.*fcmp.*false.*F1.*F2} < %t.ll - * RUN: grep {Inst43.*fcmp.*true.*F2.*F1} < %t.ll + (* RUN: grep {%build_icmp_ne = icmp ne i32 %P1, %P2} < %t.ll + * RUN: grep {%build_icmp_sle = icmp sle i32 %P2, %P1} < %t.ll + * RUN: grep {%build_icmp_false = fcmp false float %F1, %F2} < %t.ll + * RUN: grep {%build_icmp_true = fcmp true float %F2, %F1} < %t.ll *) - ignore (build_icmp Icmp.Ne p1 p2 "Inst40" atentry); - ignore (build_icmp Icmp.Sle p2 p1 "Inst41" atentry); - ignore (build_fcmp Fcmp.False f1 f2 "Inst42" atentry); - ignore (build_fcmp Fcmp.True f2 f1 "Inst43" atentry) + ignore (build_icmp Icmp.Ne p1 p2 "build_icmp_ne" atentry); + ignore (build_icmp Icmp.Sle p2 p1 "build_icmp_sle" atentry); + ignore (build_fcmp Fcmp.False f1 f2 "build_icmp_false" atentry); + ignore (build_fcmp Fcmp.True f2 f1 "build_icmp_true" atentry) end; group "miscellaneous"; begin - (* RUN: grep {CallInst.*call.*P2.*P1} < %t.ll - * RUN: grep {CallInst.*cc63} < %t.ll - * RUN: grep {Inst47.*select.*Inst46.*P1.*P2} < %t.ll - * RUN: grep {Inst48.*va_arg.*null.*i32} < %t.ll - * RUN: grep {Inst49.*extractelement.*Vec1.*P2} < %t.ll - * RUN: grep {Inst50.*insertelement.*Vec1.*P1.*P2} < %t.ll - * RUN: grep {Inst51.*shufflevector.*Vec1.*Vec2.*1.*1.*0.*0} < %t.ll - * RUN: grep {CallInst.*tail call} < %t.ll + (* RUN: grep {%build_call = tail call cc63 i32 @.*(i32 signext %P2, i32 %P1)} < %t.ll + * RUN: grep {%build_select = select i1 %build_icmp, i32 %P1, i32 %P2} < %t.ll + * RUN: grep {%build_va_arg = va_arg i8\\*\\* null, i32} < %t.ll + * RUN: grep {%build_extractelement = extractelement <4 x i32> %Vec1, i32 %P2} < %t.ll + * RUN: grep {%build_insertelement = insertelement <4 x i32> %Vec1, i32 %P1, i32 %P2} < %t.ll + * RUN: grep {%build_shufflevector = shufflevector <4 x i32> %Vec1, <4 x i32> %Vec2, <4 x i32> <i32 1, i32 1, i32 0, i32 0>} < %t.ll *) - let ci = build_call fn [| p2; p1 |] "CallInst" atentry in + let ci = build_call fn [| p2; p1 |] "build_call" atentry in insist (CallConv.c = instruction_call_conv ci); set_instruction_call_conv 63 ci; insist (63 = instruction_call_conv ci); @@ -1020,11 +1182,11 @@ let test_builder () = add_instruction_param_attr ci 2 Attribute.Noalias; remove_instruction_param_attr ci 2 Attribute.Noalias; - let inst46 = build_icmp Icmp.Eq p1 p2 "Inst46" atentry in - ignore (build_select inst46 p1 p2 "Inst47" atentry); - ignore (build_va_arg - (const_null (pointer_type (pointer_type i8_type))) - i32_type "Inst48" atentry); + let inst46 = build_icmp Icmp.Eq p1 p2 "build_icmp" atentry in + ignore (build_select inst46 p1 p2 "build_select" atentry); + ignore (build_va_arg + (const_null (pointer_type (pointer_type i8_type))) + i32_type "build_va_arg" atentry); (* Set up some vector vregs. *) let one = const_int i32_type 1 in @@ -1035,9 +1197,49 @@ let test_builder () = let vec1 = build_insertelement t1 p1 p2 "Vec1" atentry in let vec2 = build_insertelement t2 p1 p2 "Vec2" atentry in - ignore (build_extractelement vec1 p2 "Inst49" atentry); - ignore (build_insertelement vec1 p1 p2 "Inst50" atentry); - ignore (build_shufflevector vec1 vec2 t3 "Inst51" atentry); + ignore (build_extractelement vec1 p2 "build_extractelement" atentry); + ignore (build_insertelement vec1 p1 p2 "build_insertelement" atentry); + ignore (build_shufflevector vec1 vec2 t3 "build_shufflevector" atentry); + end; + + group "metadata"; begin + (* RUN: grep {%metadata = add i32 %P1, %P2, !test !0} < %t.ll + * RUN: grep {!0 = metadata !\{i32 1, metadata !"metadata test"\}} < %t.ll + *) + let i = build_add p1 p2 "metadata" atentry in + insist ((has_metadata i) = false); + + let m1 = const_int i32_type 1 in + let m2 = mdstring context "metadata test" in + let md = mdnode context [| m1; m2 |] in + + let kind = mdkind_id context "test" in + set_metadata i kind md; + + insist ((has_metadata i) = true); + insist ((metadata i kind) = Some md); + + clear_metadata i kind; + + insist ((has_metadata i) = false); + insist ((metadata i kind) = None); + + set_metadata i kind md + end; + + group "dbg"; begin + (* RUN: grep {%dbg = add i32 %P1, %P2, !dbg !1} < %t.ll + * RUN: grep {!1 = metadata !\{i32 2, metadata !"dbg test"\}} < %t.ll + *) + let m1 = const_int i32_type 2 in + let m2 = mdstring context "dbg test" in + let md = mdnode context [| m1; m2 |] in + set_current_debug_location atentry md; + + let i = build_add p1 p2 "dbg" atentry in + insist ((has_metadata i) = true); + + clear_current_debug_location atentry end; group "phi"; begin @@ -1061,14 +1263,6 @@ let test_builder () = end -(*===-- Module Provider ---------------------------------------------------===*) - -let test_module_provider () = - let m = create_module context "test" in - let mp = ModuleProvider.create m in - ModuleProvider.dispose mp - - (*===-- Pass Managers -----------------------------------------------------===*) let test_pass_manager () = @@ -1085,7 +1279,7 @@ let test_pass_manager () = let fn = define_function "FunctionPassManager" fty m in ignore (build_ret_void (builder_at_end context (entry_block fn))); - ignore (PassManager.create_function mp + ignore (PassManager.create_function m ++ PassManager.initialize ++ PassManager.run_function fn ++ PassManager.finalize @@ -1104,7 +1298,7 @@ let test_writer () = group "writer"; insist (write_bitcode_file m filename); - ModuleProvider.dispose mp + dispose_module m (*===-- Driver ------------------------------------------------------------===*) @@ -1115,12 +1309,14 @@ let _ = suite "constants" test_constants; suite "global values" test_global_values; suite "global variables" test_global_variables; + suite "uses" test_uses; + suite "users" test_users; + suite "aliases" test_aliases; suite "functions" test_functions; suite "params" test_params; suite "basic blocks" test_basic_blocks; suite "instructions" test_instructions; suite "builder" test_builder; - suite "module provider" test_module_provider; suite "pass manager" test_pass_manager; suite "writer" test_writer; (* Keep this last; it disposes m. *) exit !exit_status diff --git a/test/CodeGen/ARM/call.ll b/test/CodeGen/ARM/call.ll index 3dd66ae..c60b75b 100644 --- a/test/CodeGen/ARM/call.ll +++ b/test/CodeGen/ARM/call.ll @@ -20,3 +20,17 @@ define void @g.upgrd.1() { %tmp.upgrd.2 = tail call i32 %tmp( ) ; <i32> [#uses=0] ret void } + +define i32* @m_231b(i32, i32, i32*, i32*, i32*) nounwind { +; CHECKV4: m_231b +; CHECKV4: bx r{{.*}} +BB0: + %5 = inttoptr i32 %0 to i32* ; <i32*> [#uses=1] + %t35 = volatile load i32* %5 ; <i32> [#uses=1] + %6 = inttoptr i32 %t35 to i32** ; <i32**> [#uses=1] + %7 = getelementptr i32** %6, i32 86 ; <i32**> [#uses=1] + %8 = load i32** %7 ; <i32*> [#uses=1] + %9 = bitcast i32* %8 to i32* (i32, i32*, i32, i32*, i32*, i32*)* ; <i32* (i32, i32*, i32, i32*, i32*, i32*)*> [#uses=1] + %10 = call i32* %9(i32 %0, i32* null, i32 %1, i32* %2, i32* %3, i32* %4) ; <i32*> [#uses=1] + ret i32* %10 +} diff --git a/test/CodeGen/ARM/neon_minmax.ll b/test/CodeGen/ARM/neon_minmax.ll new file mode 100644 index 0000000..d301c6a --- /dev/null +++ b/test/CodeGen/ARM/neon_minmax.ll @@ -0,0 +1,81 @@ +; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s + +define float @fmin_ole(float %x) nounwind { +;CHECK: fmin_ole: +;CHECK: vmin.f32 + %cond = fcmp ole float 1.0, %x + %min1 = select i1 %cond, float 1.0, float %x + ret float %min1 +} + +define float @fmin_ole_zero(float %x) nounwind { +;CHECK: fmin_ole_zero: +;CHECK-NOT: vmin.f32 + %cond = fcmp ole float 0.0, %x + %min1 = select i1 %cond, float 0.0, float %x + ret float %min1 +} + +define float @fmin_ult(float %x) nounwind { +;CHECK: fmin_ult: +;CHECK: vmin.f32 + %cond = fcmp ult float %x, 1.0 + %min1 = select i1 %cond, float %x, float 1.0 + ret float %min1 +} + +define float @fmax_ogt(float %x) nounwind { +;CHECK: fmax_ogt: +;CHECK: vmax.f32 + %cond = fcmp ogt float 1.0, %x + %max1 = select i1 %cond, float 1.0, float %x + ret float %max1 +} + +define float @fmax_uge(float %x) nounwind { +;CHECK: fmax_uge: +;CHECK: vmax.f32 + %cond = fcmp uge float %x, 1.0 + %max1 = select i1 %cond, float %x, float 1.0 + ret float %max1 +} + +define float @fmax_uge_zero(float %x) nounwind { +;CHECK: fmax_uge_zero: +;CHECK-NOT: vmax.f32 + %cond = fcmp uge float %x, 0.0 + %max1 = select i1 %cond, float %x, float 0.0 + ret float %max1 +} + +define float @fmax_olt_reverse(float %x) nounwind { +;CHECK: fmax_olt_reverse: +;CHECK: vmax.f32 + %cond = fcmp olt float %x, 1.0 + %max1 = select i1 %cond, float 1.0, float %x + ret float %max1 +} + +define float @fmax_ule_reverse(float %x) nounwind { +;CHECK: fmax_ule_reverse: +;CHECK: vmax.f32 + %cond = fcmp ult float 1.0, %x + %max1 = select i1 %cond, float %x, float 1.0 + ret float %max1 +} + +define float @fmin_oge_reverse(float %x) nounwind { +;CHECK: fmin_oge_reverse: +;CHECK: vmin.f32 + %cond = fcmp oge float %x, 1.0 + %min1 = select i1 %cond, float 1.0, float %x + ret float %min1 +} + +define float @fmin_ugt_reverse(float %x) nounwind { +;CHECK: fmin_ugt_reverse: +;CHECK: vmin.f32 + %cond = fcmp ugt float 1.0, %x + %min1 = select i1 %cond, float %x, float 1.0 + ret float %min1 +} diff --git a/test/CodeGen/Alpha/add.ll b/test/CodeGen/Alpha/add.ll index 24a7418..cd883f6 100644 --- a/test/CodeGen/Alpha/add.ll +++ b/test/CodeGen/Alpha/add.ll @@ -4,9 +4,8 @@ ; RUN: grep { addl} %t.s | count 2 ; RUN: grep { addq} %t.s | count 2 ; RUN: grep { subl} %t.s | count 2 -; RUN: grep { subq} %t.s | count 1 +; RUN: grep { subq} %t.s | count 2 ; -; RUN: grep {lda \$0,-100(\$16)} %t.s | count 1 ; RUN: grep {s4addl} %t.s | count 2 ; RUN: grep {s8addl} %t.s | count 2 ; RUN: grep {s4addq} %t.s | count 2 diff --git a/test/CodeGen/Blackfin/promote-logic.ll b/test/CodeGen/Blackfin/promote-logic.ll index 46da566..1ac1408 100644 --- a/test/CodeGen/Blackfin/promote-logic.ll +++ b/test/CodeGen/Blackfin/promote-logic.ll @@ -1,5 +1,4 @@ -; RUN: llc < %s -march=bfin > %t -; XFAIL: * +; RUN: llc < %s -march=bfin ; DAGCombiner::SimplifyBinOpWithSameOpcodeHands can produce an illegal i16 OR ; operation after LegalizeOps. diff --git a/test/CodeGen/Generic/GC/argpromotion.ll b/test/CodeGen/Generic/GC/argpromotion.ll index dda376d..c63ce22 100644 --- a/test/CodeGen/Generic/GC/argpromotion.ll +++ b/test/CodeGen/Generic/GC/argpromotion.ll @@ -1,4 +1,4 @@ -; RUN: opt < %s -anders-aa -argpromotion +; RUN: opt < %s -argpromotion declare void @llvm.gcroot(i8**, i8*) diff --git a/test/CodeGen/Generic/debug-info.ll b/test/CodeGen/Generic/debug-info.ll deleted file mode 100644 index 20d9f91..0000000 --- a/test/CodeGen/Generic/debug-info.ll +++ /dev/null @@ -1,19 +0,0 @@ -; RUN: llc < %s - - %lldb.compile_unit = type { i32, i16, i16, i8*, i8*, i8*, { }* } -@d.compile_unit7 = external global %lldb.compile_unit ; <%lldb.compile_unit*> [#uses=1] - -declare void @llvm.dbg.stoppoint(i32, i32, %lldb.compile_unit*) - -define void @rb_raise(i32, ...) { -entry: - br i1 false, label %strlen.exit, label %no_exit.i - -no_exit.i: ; preds = %entry - ret void - -strlen.exit: ; preds = %entry - call void @llvm.dbg.stoppoint( i32 4358, i32 0, %lldb.compile_unit* @d.compile_unit7 ) - unreachable -} - diff --git a/test/CodeGen/MBlaze/brind.ll b/test/CodeGen/MBlaze/brind.ll new file mode 100644 index 0000000..7798e0f --- /dev/null +++ b/test/CodeGen/MBlaze/brind.ll @@ -0,0 +1,73 @@ +; Ensure that the select instruction is supported and is lowered to +; some sort of branch instruction. +; +; RUN: llc < %s -march=mblaze -mattr=+mul,+fpu,+barrel | FileCheck %s + +declare i32 @printf(i8*, ...) +@MSG = internal constant [13 x i8] c"Message: %d\0A\00" + +@BLKS = private constant [5 x i8*] + [ i8* blockaddress(@brind, %L1), + i8* blockaddress(@brind, %L2), + i8* blockaddress(@brind, %L3), + i8* blockaddress(@brind, %L4), + i8* blockaddress(@brind, %L5) ] + +define i32 @brind(i32 %a, i32 %b) +{ + ; CHECK: brind: +entry: + br label %loop + +loop: + %tmp.0 = phi i32 [ 0, %entry ], [ %tmp.8, %finish ] + %dst.0 = getelementptr [5 x i8*]* @BLKS, i32 0, i32 %tmp.0 + %dst.1 = load i8** %dst.0 + indirectbr i8* %dst.1, [ label %L1, + label %L2, + label %L3, + label %L4, + label %L5 ] + ; CHECK: br {{r[0-9]*}} + +L1: + %tmp.1 = add i32 %a, %b + br label %finish + ; CHECK: br + +L2: + %tmp.2 = sub i32 %a, %b + br label %finish + ; CHECK: br + +L3: + %tmp.3 = mul i32 %a, %b + br label %finish + ; CHECK: br + +L4: + %tmp.4 = sdiv i32 %a, %b + br label %finish + ; CHECK: br + +L5: + %tmp.5 = srem i32 %a, %b + br label %finish + ; CHECK: br + +finish: + %tmp.6 = phi i32 [ %tmp.1, %L1 ], + [ %tmp.2, %L2 ], + [ %tmp.3, %L3 ], + [ %tmp.4, %L4 ], + [ %tmp.5, %L5 ] + + call i32 (i8*,...)* @printf( i8* getelementptr([13 x i8]* @MSG,i32 0,i32 0), + i32 %tmp.6) + + %tmp.7 = add i32 %tmp.0, 1 + %tmp.8 = urem i32 %tmp.7, 5 + + br label %loop + ; CHECK: br +} diff --git a/test/CodeGen/MBlaze/callind.ll b/test/CodeGen/MBlaze/callind.ll new file mode 100644 index 0000000..bfc8d00 --- /dev/null +++ b/test/CodeGen/MBlaze/callind.ll @@ -0,0 +1,80 @@ +; Ensure that indirect calls work and that they are lowered to some +; sort of branch and link instruction. +; +; RUN: llc < %s -march=mblaze -mattr=+mul,+fpu,+barrel | FileCheck %s + +declare i32 @printf(i8*, ...) +@MSG = internal constant [13 x i8] c"Message: %d\0A\00" + +@FUNS = private constant [5 x i32 (i32,i32)*] + [ i32 (i32,i32)* @doadd, + i32 (i32,i32)* @dosub, + i32 (i32,i32)* @domul, + i32 (i32,i32)* @dodiv, + i32 (i32,i32)* @dorem ] + +define i32 @doadd(i32 %a, i32 %b) +{ + ; CHECK: doadd: + %tmp.0 = add i32 %a, %b + ret i32 %tmp.0 + ; CHECK: rtsd +} + +define i32 @dosub(i32 %a, i32 %b) +{ + ; CHECK: dosub: + %tmp.0 = sub i32 %a, %b + ret i32 %tmp.0 + ; CHECK: rtsd +} + +define i32 @domul(i32 %a, i32 %b) +{ + ; CHECK: domul: + %tmp.0 = mul i32 %a, %b + ret i32 %tmp.0 + ; CHECK: rtsd +} + +define i32 @dodiv(i32 %a, i32 %b) +{ + ; CHECK: dodiv: + %tmp.0 = sdiv i32 %a, %b + ret i32 %tmp.0 + ; CHECK: rtsd +} + +define i32 @dorem(i32 %a, i32 %b) +{ + ; CHECK: dorem: + %tmp.0 = srem i32 %a, %b + ret i32 %tmp.0 + ; CHECK: rtsd +} + +define i32 @callind(i32 %a, i32 %b) +{ + ; CHECK: callind: +entry: + br label %loop + +loop: + %tmp.0 = phi i32 [ 0, %entry ], [ %tmp.3, %loop ] + %dst.0 = getelementptr [5 x i32 (i32,i32)*]* @FUNS, i32 0, i32 %tmp.0 + %dst.1 = load i32 (i32,i32)** %dst.0 + %tmp.1 = call i32 %dst.1(i32 %a, i32 %b) + ; CHECK-NOT: brli + ; CHECK-NOT: brlai + ; CHECK: brl + + call i32 (i8*,...)* @printf( i8* getelementptr([13 x i8]* @MSG,i32 0,i32 0), + i32 %tmp.1) + ; CHECK: brl + + %tmp.2 = add i32 %tmp.0, 1 + %tmp.3 = urem i32 %tmp.2, 5 + + br label %loop + ; CHECK: br +} diff --git a/test/CodeGen/MBlaze/cc.ll b/test/CodeGen/MBlaze/cc.ll new file mode 100644 index 0000000..de55728 --- /dev/null +++ b/test/CodeGen/MBlaze/cc.ll @@ -0,0 +1,315 @@ +; Test some of the calling convention lowering done by the MBlaze backend. +; We test that integer values are passed in the correct registers and +; returned in the correct registers. Additionally, we test that the stack +; is used as appropriate for passing arguments that cannot be placed into +; registers. +; +; RUN: llc < %s -march=mblaze | FileCheck %s + +declare i32 @printf(i8*, ...) +@MSG = internal constant [13 x i8] c"Message: %d\0A\00" + +define void @params0_noret() { + ; CHECK: params0_noret: + ret void + ; CHECK-NOT: {{.* r3, r0, 1}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i8 @params0_8bitret() { + ; CHECK: params0_8bitret: + ret i8 1 + ; CHECK: {{.* r3, r0, 1}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i16 @params0_16bitret() { + ; CHECK: params0_16bitret: + ret i16 1 + ; CHECK: {{.* r3, r0, 1}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params0_32bitret() { + ; CHECK: params0_32bitret: + ret i32 1 + ; CHECK: {{.* r3, r0, 1}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i64 @params0_64bitret() { + ; CHECK: params0_64bitret: + ret i64 1 + ; CHECK: {{.* r3, r0, .*}} + ; CHECK: {{.* r4, r0, 1}} + ; CHECK: rtsd +} + +define i32 @params1_32bitret(i32 %a) { + ; CHECK: params1_32bitret: + ret i32 %a + ; CHECK: {{.* r3, r5, r0}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params2_32bitret(i32 %a, i32 %b) { + ; CHECK: params2_32bitret: + ret i32 %b + ; CHECK: {{.* r3, r6, r0}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params3_32bitret(i32 %a, i32 %b, i32 %c) { + ; CHECK: params3_32bitret: + ret i32 %c + ; CHECK: {{.* r3, r7, r0}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params4_32bitret(i32 %a, i32 %b, i32 %c, i32 %d) { + ; CHECK: params4_32bitret: + ret i32 %d + ; CHECK: {{.* r3, r8, r0}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params5_32bitret(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) { + ; CHECK: params5_32bitret: + ret i32 %e + ; CHECK: {{.* r3, r9, r0}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params6_32bitret(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e, i32 %f) { + ; CHECK: params6_32bitret: + ret i32 %f + ; CHECK: {{.* r3, r10, r0}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params7_32bitret(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e, i32 %f, + i32 %g) { + ; CHECK: params7_32bitret: + ret i32 %g + ; CHECK: {{lwi? r3, r1, 8}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params8_32bitret(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e, i32 %f, + i32 %g, i32 %h) { + ; CHECK: params8_32bitret: + ret i32 %h + ; CHECK: {{lwi? r3, r1, 12}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params9_32bitret(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e, i32 %f, + i32 %g, i32 %h, i32 %i) { + ; CHECK: params9_32bitret: + ret i32 %i + ; CHECK: {{lwi? r3, r1, 16}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define i32 @params10_32bitret(i32 %a, i32 %b, i32 %c, i32 %d, i32 %e, i32 %f, + i32 %g, i32 %h, i32 %i, i32 %j) { + ; CHECK: params10_32bitret: + ret i32 %j + ; CHECK: {{lwi? r3, r1, 20}} + ; CHECK-NOT: {{.* r4, .*, .*}} + ; CHECK: rtsd +} + +define void @testing() { + %MSG.1 = getelementptr [13 x i8]* @MSG, i32 0, i32 0 + + call void @params0_noret() + ; CHECK: brlid + + %tmp.1 = call i8 @params0_8bitret() + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i8 %tmp.1) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.2 = call i16 @params0_16bitret() + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i16 %tmp.2) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.3 = call i32 @params0_32bitret() + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.3) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.4 = call i64 @params0_64bitret() + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i64 %tmp.4) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK: {{.* r7, r4, r0}} + ; CHECK: brlid + + %tmp.5 = call i32 @params1_32bitret(i32 1) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.5) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.6 = call i32 @params2_32bitret(i32 1, i32 2) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.6) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.7 = call i32 @params3_32bitret(i32 1, i32 2, i32 3) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, .*, .*}} + ; CHECK: {{.* r7, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.7) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.8 = call i32 @params4_32bitret(i32 1, i32 2, i32 3, i32 4) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, .*, .*}} + ; CHECK: {{.* r7, .*, .*}} + ; CHECK: {{.* r8, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.8) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.9 = call i32 @params5_32bitret(i32 1, i32 2, i32 3, i32 4, i32 5) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, .*, .*}} + ; CHECK: {{.* r7, .*, .*}} + ; CHECK: {{.* r8, .*, .*}} + ; CHECK: {{.* r9, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.9) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.10 = call i32 @params6_32bitret(i32 1, i32 2, i32 3, i32 4, i32 5, + i32 6) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, .*, .*}} + ; CHECK: {{.* r7, .*, .*}} + ; CHECK: {{.* r8, .*, .*}} + ; CHECK: {{.* r9, .*, .*}} + ; CHECK: {{.* r10, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.10) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.11 = call i32 @params7_32bitret(i32 1, i32 2, i32 3, i32 4, i32 5, + i32 6, i32 7) + ; CHECK: {{swi? .*, r1, 4}} + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, .*, .*}} + ; CHECK: {{.* r7, .*, .*}} + ; CHECK: {{.* r8, .*, .*}} + ; CHECK: {{.* r9, .*, .*}} + ; CHECK: {{.* r10, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.11) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.12 = call i32 @params8_32bitret(i32 1, i32 2, i32 3, i32 4, i32 5, + i32 6, i32 7, i32 8) + ; CHECK: {{swi? .*, r1, 4}} + ; CHECK: {{swi? .*, r1, 8}} + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, .*, .*}} + ; CHECK: {{.* r7, .*, .*}} + ; CHECK: {{.* r8, .*, .*}} + ; CHECK: {{.* r9, .*, .*}} + ; CHECK: {{.* r10, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.12) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.13 = call i32 @params9_32bitret(i32 1, i32 2, i32 3, i32 4, i32 5, + i32 6, i32 7, i32 8, i32 9) + ; CHECK: {{swi? .*, r1, 4}} + ; CHECK: {{swi? .*, r1, 8}} + ; CHECK: {{swi? .*, r1, 12}} + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, .*, .*}} + ; CHECK: {{.* r7, .*, .*}} + ; CHECK: {{.* r8, .*, .*}} + ; CHECK: {{.* r9, .*, .*}} + ; CHECK: {{.* r10, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.13) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + %tmp.14 = call i32 @params10_32bitret(i32 1, i32 2, i32 3, i32 4, i32 5, + i32 6, i32 7, i32 8, i32 9, i32 10) + ; CHECK: {{swi? .*, r1, 4}} + ; CHECK: {{swi? .*, r1, 8}} + ; CHECK: {{swi? .*, r1, 12}} + ; CHECK: {{swi? .*, r1, 16}} + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, .*, .*}} + ; CHECK: {{.* r7, .*, .*}} + ; CHECK: {{.* r8, .*, .*}} + ; CHECK: {{.* r9, .*, .*}} + ; CHECK: {{.* r10, .*, .*}} + ; CHECK: brlid + call i32 (i8*,...)* @printf(i8* %MSG.1, i32 %tmp.14) + ; CHECK: {{.* r5, .*, .*}} + ; CHECK: {{.* r6, r3, r0}} + ; CHECK-NOT: {{.* r7, .*, .*}} + ; CHECK: brlid + + ret void +} diff --git a/test/CodeGen/MBlaze/dg.exp b/test/CodeGen/MBlaze/dg.exp new file mode 100644 index 0000000..bfd5e47 --- /dev/null +++ b/test/CodeGen/MBlaze/dg.exp @@ -0,0 +1,5 @@ +load_lib llvm.exp + +if { [llvm_supports_target MBlaze] } { + RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{ll,c,cpp}]] +} diff --git a/test/CodeGen/MBlaze/div.ll b/test/CodeGen/MBlaze/div.ll new file mode 100644 index 0000000..fae9830 --- /dev/null +++ b/test/CodeGen/MBlaze/div.ll @@ -0,0 +1,75 @@ +; Ensure that multiplication is lowered to function calls when the multiplier +; unit is not available in the hardware and that function calls are not used +; when the multiplier unit is available in the hardware. +; +; RUN: llc < %s -march=mblaze | FileCheck -check-prefix=FUN %s +; RUN: llc < %s -march=mblaze -mattr=+div | FileCheck -check-prefix=DIV %s + +define i8 @test_i8(i8 %a, i8 %b) { + ; FUN: test_i8: + ; DIV: test_i8: + + %tmp.1 = udiv i8 %a, %b + ; FUN-NOT: idiv + ; FUN: brlid + ; DIV-NOT: brlid + ; DIV: idivu + + %tmp.2 = sdiv i8 %a, %b + ; FUN-NOT: idiv + ; FUN: brlid + ; DIV-NOT: brlid + ; DIV-NOT: idivu + ; DIV: idiv + + %tmp.3 = add i8 %tmp.1, %tmp.2 + ret i8 %tmp.3 + ; FUN: rtsd + ; DIV: rtsd +} + +define i16 @test_i16(i16 %a, i16 %b) { + ; FUN: test_i16: + ; DIV: test_i16: + + %tmp.1 = udiv i16 %a, %b + ; FUN-NOT: idiv + ; FUN: brlid + ; DIV-NOT: brlid + ; DIV: idivu + + %tmp.2 = sdiv i16 %a, %b + ; FUN-NOT: idiv + ; FUN: brlid + ; DIV-NOT: brlid + ; DIV-NOT: idivu + ; DIV: idiv + + %tmp.3 = add i16 %tmp.1, %tmp.2 + ret i16 %tmp.3 + ; FUN: rtsd + ; DIV: rtsd +} + +define i32 @test_i32(i32 %a, i32 %b) { + ; FUN: test_i32: + ; DIV: test_i32: + + %tmp.1 = udiv i32 %a, %b + ; FUN-NOT: idiv + ; FUN: brlid + ; DIV-NOT: brlid + ; DIV: idivu + + %tmp.2 = sdiv i32 %a, %b + ; FUN-NOT: idiv + ; FUN: brlid + ; DIV-NOT: brlid + ; DIV-NOT: idivu + ; DIV: idiv + + %tmp.3 = add i32 %tmp.1, %tmp.2 + ret i32 %tmp.3 + ; FUN: rtsd + ; DIV: rtsd +} diff --git a/test/CodeGen/MBlaze/fpu.ll b/test/CodeGen/MBlaze/fpu.ll new file mode 100644 index 0000000..83f4d83 --- /dev/null +++ b/test/CodeGen/MBlaze/fpu.ll @@ -0,0 +1,66 @@ +; Ensure that floating point operations are lowered to function calls when the +; FPU is not available in the hardware and that function calls are not used +; when the FPU is available in the hardware. +; +; RUN: llc < %s -march=mblaze | FileCheck -check-prefix=FUN %s +; RUN: llc < %s -march=mblaze -mattr=+fpu | FileCheck -check-prefix=FPU %s + +define float @test_add(float %a, float %b) { + ; FUN: test_add: + ; FPU: test_add: + + %tmp.1 = fadd float %a, %b + ; FUN-NOT: fadd + ; FUN: brlid + ; FPU-NOT: brlid + ; FPU: fadd + + ret float %tmp.1 + ; FUN: rtsd + ; FPU: rtsd +} + +define float @test_sub(float %a, float %b) { + ; FUN: test_sub: + ; FPU: test_sub: + + %tmp.1 = fsub float %a, %b + ; FUN-NOT: frsub + ; FUN: brlid + ; FPU-NOT: brlid + ; FPU: frsub + + ret float %tmp.1 + ; FUN: rtsd + ; FPU: rtsd +} + +define float @test_mul(float %a, float %b) { + ; FUN: test_mul: + ; FPU: test_mul: + + %tmp.1 = fmul float %a, %b + ; FUN-NOT: fmul + ; FUN: brlid + ; FPU-NOT: brlid + ; FPU: fmul + + ret float %tmp.1 + ; FUN: rtsd + ; FPU: rtsd +} + +define float @test_div(float %a, float %b) { + ; FUN: test_div: + ; FPU: test_div: + + %tmp.1 = fdiv float %a, %b + ; FUN-NOT: fdiv + ; FUN: brlid + ; FPU-NOT: brlid + ; FPU: fdiv + + ret float %tmp.1 + ; FUN: rtsd + ; FPU: rtsd +} diff --git a/test/CodeGen/MBlaze/fsl.ll b/test/CodeGen/MBlaze/fsl.ll new file mode 100644 index 0000000..f9c6205 --- /dev/null +++ b/test/CodeGen/MBlaze/fsl.ll @@ -0,0 +1,323 @@ +; Ensure that the FSL instrinsic instruction generate single FSL instructions +; at the machine level. Additionally, ensure that dynamic values use the +; dynamic version of the instructions and that constant values use the +; constant version of the instructions. +; +; RUN: llc < %s -march=mblaze | FileCheck %s + +declare i32 @llvm.mblaze.fsl.get(i32 %port) +declare i32 @llvm.mblaze.fsl.aget(i32 %port) +declare i32 @llvm.mblaze.fsl.cget(i32 %port) +declare i32 @llvm.mblaze.fsl.caget(i32 %port) +declare i32 @llvm.mblaze.fsl.eget(i32 %port) +declare i32 @llvm.mblaze.fsl.eaget(i32 %port) +declare i32 @llvm.mblaze.fsl.ecget(i32 %port) +declare i32 @llvm.mblaze.fsl.ecaget(i32 %port) +declare i32 @llvm.mblaze.fsl.nget(i32 %port) +declare i32 @llvm.mblaze.fsl.naget(i32 %port) +declare i32 @llvm.mblaze.fsl.ncget(i32 %port) +declare i32 @llvm.mblaze.fsl.ncaget(i32 %port) +declare i32 @llvm.mblaze.fsl.neget(i32 %port) +declare i32 @llvm.mblaze.fsl.neaget(i32 %port) +declare i32 @llvm.mblaze.fsl.necget(i32 %port) +declare i32 @llvm.mblaze.fsl.necaget(i32 %port) +declare i32 @llvm.mblaze.fsl.tget(i32 %port) +declare i32 @llvm.mblaze.fsl.taget(i32 %port) +declare i32 @llvm.mblaze.fsl.tcget(i32 %port) +declare i32 @llvm.mblaze.fsl.tcaget(i32 %port) +declare i32 @llvm.mblaze.fsl.teget(i32 %port) +declare i32 @llvm.mblaze.fsl.teaget(i32 %port) +declare i32 @llvm.mblaze.fsl.tecget(i32 %port) +declare i32 @llvm.mblaze.fsl.tecaget(i32 %port) +declare i32 @llvm.mblaze.fsl.tnget(i32 %port) +declare i32 @llvm.mblaze.fsl.tnaget(i32 %port) +declare i32 @llvm.mblaze.fsl.tncget(i32 %port) +declare i32 @llvm.mblaze.fsl.tncaget(i32 %port) +declare i32 @llvm.mblaze.fsl.tneget(i32 %port) +declare i32 @llvm.mblaze.fsl.tneaget(i32 %port) +declare i32 @llvm.mblaze.fsl.tnecget(i32 %port) +declare i32 @llvm.mblaze.fsl.tnecaget(i32 %port) + +declare void @llvm.mblaze.fsl.put(i32 %value, i32 %port) +declare void @llvm.mblaze.fsl.aput(i32 %value, i32 %port) +declare void @llvm.mblaze.fsl.cput(i32 %value, i32 %port) +declare void @llvm.mblaze.fsl.caput(i32 %value, i32 %port) +declare void @llvm.mblaze.fsl.nput(i32 %value, i32 %port) +declare void @llvm.mblaze.fsl.naput(i32 %value, i32 %port) +declare void @llvm.mblaze.fsl.ncput(i32 %value, i32 %port) +declare void @llvm.mblaze.fsl.ncaput(i32 %value, i32 %port) +declare void @llvm.mblaze.fsl.tput(i32 %port) +declare void @llvm.mblaze.fsl.taput(i32 %port) +declare void @llvm.mblaze.fsl.tcput(i32 %port) +declare void @llvm.mblaze.fsl.tcaput(i32 %port) +declare void @llvm.mblaze.fsl.tnput(i32 %port) +declare void @llvm.mblaze.fsl.tnaput(i32 %port) +declare void @llvm.mblaze.fsl.tncput(i32 %port) +declare void @llvm.mblaze.fsl.tncaput(i32 %port) + +define i32 @fsl_get(i32 %port) +{ + ; CHECK: fsl_get: + %v0 = call i32 @llvm.mblaze.fsl.get(i32 %port) + ; CHECK: getd + %v1 = call i32 @llvm.mblaze.fsl.aget(i32 %port) + ; CHECK-NEXT: agetd + %v2 = call i32 @llvm.mblaze.fsl.cget(i32 %port) + ; CHECK-NEXT: cgetd + %v3 = call i32 @llvm.mblaze.fsl.caget(i32 %port) + ; CHECK-NEXT: cagetd + %v4 = call i32 @llvm.mblaze.fsl.eget(i32 %port) + ; CHECK-NEXT: egetd + %v5 = call i32 @llvm.mblaze.fsl.eaget(i32 %port) + ; CHECK-NEXT: eagetd + %v6 = call i32 @llvm.mblaze.fsl.ecget(i32 %port) + ; CHECK-NEXT: ecgetd + %v7 = call i32 @llvm.mblaze.fsl.ecaget(i32 %port) + ; CHECK-NEXT: ecagetd + %v8 = call i32 @llvm.mblaze.fsl.nget(i32 %port) + ; CHECK-NEXT: ngetd + %v9 = call i32 @llvm.mblaze.fsl.naget(i32 %port) + ; CHECK-NEXT: nagetd + %v10 = call i32 @llvm.mblaze.fsl.ncget(i32 %port) + ; CHECK-NEXT: ncgetd + %v11 = call i32 @llvm.mblaze.fsl.ncaget(i32 %port) + ; CHECK-NEXT: ncagetd + %v12 = call i32 @llvm.mblaze.fsl.neget(i32 %port) + ; CHECK-NEXT: negetd + %v13 = call i32 @llvm.mblaze.fsl.neaget(i32 %port) + ; CHECK-NEXT: neagetd + %v14 = call i32 @llvm.mblaze.fsl.necget(i32 %port) + ; CHECK-NEXT: necgetd + %v15 = call i32 @llvm.mblaze.fsl.necaget(i32 %port) + ; CHECK-NEXT: necagetd + %v16 = call i32 @llvm.mblaze.fsl.tget(i32 %port) + ; CHECK-NEXT: tgetd + %v17 = call i32 @llvm.mblaze.fsl.taget(i32 %port) + ; CHECK-NEXT: tagetd + %v18 = call i32 @llvm.mblaze.fsl.tcget(i32 %port) + ; CHECK-NEXT: tcgetd + %v19 = call i32 @llvm.mblaze.fsl.tcaget(i32 %port) + ; CHECK-NEXT: tcagetd + %v20 = call i32 @llvm.mblaze.fsl.teget(i32 %port) + ; CHECK-NEXT: tegetd + %v21 = call i32 @llvm.mblaze.fsl.teaget(i32 %port) + ; CHECK-NEXT: teagetd + %v22 = call i32 @llvm.mblaze.fsl.tecget(i32 %port) + ; CHECK-NEXT: tecgetd + %v23 = call i32 @llvm.mblaze.fsl.tecaget(i32 %port) + ; CHECK-NEXT: tecagetd + %v24 = call i32 @llvm.mblaze.fsl.tnget(i32 %port) + ; CHECK-NEXT: tngetd + %v25 = call i32 @llvm.mblaze.fsl.tnaget(i32 %port) + ; CHECK-NEXT: tnagetd + %v26 = call i32 @llvm.mblaze.fsl.tncget(i32 %port) + ; CHECK-NEXT: tncgetd + %v27 = call i32 @llvm.mblaze.fsl.tncaget(i32 %port) + ; CHECK-NEXT: tncagetd + %v28 = call i32 @llvm.mblaze.fsl.tneget(i32 %port) + ; CHECK-NEXT: tnegetd + %v29 = call i32 @llvm.mblaze.fsl.tneaget(i32 %port) + ; CHECK-NEXT: tneagetd + %v30 = call i32 @llvm.mblaze.fsl.tnecget(i32 %port) + ; CHECK-NEXT: tnecgetd + %v31 = call i32 @llvm.mblaze.fsl.tnecaget(i32 %port) + ; CHECK-NEXT: tnecagetd + ret i32 1 + ; CHECK: rtsd +} + +define i32 @fslc_get() +{ + ; CHECK: fslc_get: + %v0 = call i32 @llvm.mblaze.fsl.get(i32 1) + ; CHECK: get + %v1 = call i32 @llvm.mblaze.fsl.aget(i32 1) + ; CHECK-NOT: agetd + ; CHECK: aget + %v2 = call i32 @llvm.mblaze.fsl.cget(i32 1) + ; CHECK-NOT: cgetd + ; CHECK: cget + %v3 = call i32 @llvm.mblaze.fsl.caget(i32 1) + ; CHECK-NOT: cagetd + ; CHECK: caget + %v4 = call i32 @llvm.mblaze.fsl.eget(i32 1) + ; CHECK-NOT: egetd + ; CHECK: eget + %v5 = call i32 @llvm.mblaze.fsl.eaget(i32 1) + ; CHECK-NOT: eagetd + ; CHECK: eaget + %v6 = call i32 @llvm.mblaze.fsl.ecget(i32 1) + ; CHECK-NOT: ecgetd + ; CHECK: ecget + %v7 = call i32 @llvm.mblaze.fsl.ecaget(i32 1) + ; CHECK-NOT: ecagetd + ; CHECK: ecaget + %v8 = call i32 @llvm.mblaze.fsl.nget(i32 1) + ; CHECK-NOT: ngetd + ; CHECK: nget + %v9 = call i32 @llvm.mblaze.fsl.naget(i32 1) + ; CHECK-NOT: nagetd + ; CHECK: naget + %v10 = call i32 @llvm.mblaze.fsl.ncget(i32 1) + ; CHECK-NOT: ncgetd + ; CHECK: ncget + %v11 = call i32 @llvm.mblaze.fsl.ncaget(i32 1) + ; CHECK-NOT: ncagetd + ; CHECK: ncaget + %v12 = call i32 @llvm.mblaze.fsl.neget(i32 1) + ; CHECK-NOT: negetd + ; CHECK: neget + %v13 = call i32 @llvm.mblaze.fsl.neaget(i32 1) + ; CHECK-NOT: neagetd + ; CHECK: neaget + %v14 = call i32 @llvm.mblaze.fsl.necget(i32 1) + ; CHECK-NOT: necgetd + ; CHECK: necget + %v15 = call i32 @llvm.mblaze.fsl.necaget(i32 1) + ; CHECK-NOT: necagetd + ; CHECK: necaget + %v16 = call i32 @llvm.mblaze.fsl.tget(i32 1) + ; CHECK-NOT: tgetd + ; CHECK: tget + %v17 = call i32 @llvm.mblaze.fsl.taget(i32 1) + ; CHECK-NOT: tagetd + ; CHECK: taget + %v18 = call i32 @llvm.mblaze.fsl.tcget(i32 1) + ; CHECK-NOT: tcgetd + ; CHECK: tcget + %v19 = call i32 @llvm.mblaze.fsl.tcaget(i32 1) + ; CHECK-NOT: tcagetd + ; CHECK: tcaget + %v20 = call i32 @llvm.mblaze.fsl.teget(i32 1) + ; CHECK-NOT: tegetd + ; CHECK: teget + %v21 = call i32 @llvm.mblaze.fsl.teaget(i32 1) + ; CHECK-NOT: teagetd + ; CHECK: teaget + %v22 = call i32 @llvm.mblaze.fsl.tecget(i32 1) + ; CHECK-NOT: tecgetd + ; CHECK: tecget + %v23 = call i32 @llvm.mblaze.fsl.tecaget(i32 1) + ; CHECK-NOT: tecagetd + ; CHECK: tecaget + %v24 = call i32 @llvm.mblaze.fsl.tnget(i32 1) + ; CHECK-NOT: tngetd + ; CHECK: tnget + %v25 = call i32 @llvm.mblaze.fsl.tnaget(i32 1) + ; CHECK-NOT: tnagetd + ; CHECK: tnaget + %v26 = call i32 @llvm.mblaze.fsl.tncget(i32 1) + ; CHECK-NOT: tncgetd + ; CHECK: tncget + %v27 = call i32 @llvm.mblaze.fsl.tncaget(i32 1) + ; CHECK-NOT: tncagetd + ; CHECK: tncaget + %v28 = call i32 @llvm.mblaze.fsl.tneget(i32 1) + ; CHECK-NOT: tnegetd + ; CHECK: tneget + %v29 = call i32 @llvm.mblaze.fsl.tneaget(i32 1) + ; CHECK-NOT: tneagetd + ; CHECK: tneaget + %v30 = call i32 @llvm.mblaze.fsl.tnecget(i32 1) + ; CHECK-NOT: tnecgetd + ; CHECK: tnecget + %v31 = call i32 @llvm.mblaze.fsl.tnecaget(i32 1) + ; CHECK-NOT: tnecagetd + ; CHECK: tnecaget + ret i32 1 + ; CHECK: rtsd +} + +define void @putfsl(i32 %value, i32 %port) +{ + ; CHECK: putfsl: + call void @llvm.mblaze.fsl.put(i32 %value, i32 %port) + ; CHECK: putd + call void @llvm.mblaze.fsl.aput(i32 %value, i32 %port) + ; CHECK-NEXT: aputd + call void @llvm.mblaze.fsl.cput(i32 %value, i32 %port) + ; CHECK-NEXT: cputd + call void @llvm.mblaze.fsl.caput(i32 %value, i32 %port) + ; CHECK-NEXT: caputd + call void @llvm.mblaze.fsl.nput(i32 %value, i32 %port) + ; CHECK-NEXT: nputd + call void @llvm.mblaze.fsl.naput(i32 %value, i32 %port) + ; CHECK-NEXT: naputd + call void @llvm.mblaze.fsl.ncput(i32 %value, i32 %port) + ; CHECK-NEXT: ncputd + call void @llvm.mblaze.fsl.ncaput(i32 %value, i32 %port) + ; CHECK-NEXT: ncaputd + call void @llvm.mblaze.fsl.tput(i32 %port) + ; CHECK-NEXT: tputd + call void @llvm.mblaze.fsl.taput(i32 %port) + ; CHECK-NEXT: taputd + call void @llvm.mblaze.fsl.tcput(i32 %port) + ; CHECK-NEXT: tcputd + call void @llvm.mblaze.fsl.tcaput(i32 %port) + ; CHECK-NEXT: tcaputd + call void @llvm.mblaze.fsl.tnput(i32 %port) + ; CHECK-NEXT: tnputd + call void @llvm.mblaze.fsl.tnaput(i32 %port) + ; CHECK-NEXT: tnaputd + call void @llvm.mblaze.fsl.tncput(i32 %port) + ; CHECK-NEXT: tncputd + call void @llvm.mblaze.fsl.tncaput(i32 %port) + ; CHECK-NEXT: tncaputd + ret void + ; CHECK: rtsd +} + +define void @putfsl_const(i32 %value) +{ + ; CHECK: putfsl_const: + call void @llvm.mblaze.fsl.put(i32 %value, i32 1) + ; CHECK-NOT: putd + ; CHECK: put + call void @llvm.mblaze.fsl.aput(i32 %value, i32 1) + ; CHECK-NOT: aputd + ; CHECK: aput + call void @llvm.mblaze.fsl.cput(i32 %value, i32 1) + ; CHECK-NOT: cputd + ; CHECK: cput + call void @llvm.mblaze.fsl.caput(i32 %value, i32 1) + ; CHECK-NOT: caputd + ; CHECK: caput + call void @llvm.mblaze.fsl.nput(i32 %value, i32 1) + ; CHECK-NOT: nputd + ; CHECK: nput + call void @llvm.mblaze.fsl.naput(i32 %value, i32 1) + ; CHECK-NOT: naputd + ; CHECK: naput + call void @llvm.mblaze.fsl.ncput(i32 %value, i32 1) + ; CHECK-NOT: ncputd + ; CHECK: ncput + call void @llvm.mblaze.fsl.ncaput(i32 %value, i32 1) + ; CHECK-NOT: ncaputd + ; CHECK: ncaput + call void @llvm.mblaze.fsl.tput(i32 1) + ; CHECK-NOT: tputd + ; CHECK: tput + call void @llvm.mblaze.fsl.taput(i32 1) + ; CHECK-NOT: taputd + ; CHECK: taput + call void @llvm.mblaze.fsl.tcput(i32 1) + ; CHECK-NOT: tcputd + ; CHECK: tcput + call void @llvm.mblaze.fsl.tcaput(i32 1) + ; CHECK-NOT: tcaputd + ; CHECK: tcaput + call void @llvm.mblaze.fsl.tnput(i32 1) + ; CHECK-NOT: tnputd + ; CHECK: tnput + call void @llvm.mblaze.fsl.tnaput(i32 1) + ; CHECK-NOT: tnaputd + ; CHECK: tnaput + call void @llvm.mblaze.fsl.tncput(i32 1) + ; CHECK-NOT: tncputd + ; CHECK: tncput + call void @llvm.mblaze.fsl.tncaput(i32 1) + ; CHECK-NOT: tncaputd + ; CHECK: tncaput + ret void + ; CHECK: rtsd +} diff --git a/test/CodeGen/MBlaze/imm.ll b/test/CodeGen/MBlaze/imm.ll new file mode 100644 index 0000000..85fad17 --- /dev/null +++ b/test/CodeGen/MBlaze/imm.ll @@ -0,0 +1,70 @@ +; Ensure that all immediate values that are 32-bits or less can be loaded +; using a single instruction and that immediate values 64-bits or less can +; be loaded using two instructions. +; +; RUN: llc < %s -march=mblaze | FileCheck %s +; RUN: llc < %s -march=mblaze -mattr=+fpu | FileCheck -check-prefix=FPU %s + +define i8 @retimm_i8() { + ; CHECK: retimm_i8: + ; CHECK: add + ; CHECK-NEXT: rtsd + ; FPU: retimm_i8: + ; FPU: add + ; FPU-NEXT: rtsd + ret i8 123 +} + +define i16 @retimm_i16() { + ; CHECK: retimm_i16: + ; CHECK: add + ; CHECK-NEXT: rtsd + ; FPU: retimm_i16: + ; FPU: add + ; FPU-NEXT: rtsd + ret i16 38212 +} + +define i32 @retimm_i32() { + ; CHECK: retimm_i32: + ; CHECK: add + ; CHECK-NEXT: rtsd + ; FPU: retimm_i32: + ; FPU: add + ; FPU-NEXT: rtsd + ret i32 2938128 +} + +define i64 @retimm_i64() { + ; CHECK: retimm_i64: + ; CHECK: add + ; CHECK-NEXT: add + ; CHECK-NEXT: rtsd + ; FPU: retimm_i64: + ; FPU: add + ; FPU-NEXT: add + ; FPU-NEXT: rtsd + ret i64 94581823 +} + +define float @retimm_float() { + ; CHECK: retimm_float: + ; CHECK: add + ; CHECK-NEXT: rtsd + ; FPU: retimm_float: + ; FPU: or + ; FPU: rtsd + ret float 12.0 +} + +define double @retimm_double() { + ; CHECK: retimm_double: + ; CHECK: add + ; CHECK-NEXT: add + ; CHECK-NEXT: rtsd + ; FPU: retimm_double: + ; FPU: add + ; FPU-NEXT: add + ; FPU-NEXT: rtsd + ret double 598382.39283873 +} diff --git a/test/CodeGen/MBlaze/jumptable.ll b/test/CodeGen/MBlaze/jumptable.ll new file mode 100644 index 0000000..3f27c12 --- /dev/null +++ b/test/CodeGen/MBlaze/jumptable.ll @@ -0,0 +1,79 @@ +; Ensure that jump tables can be handled by the mblaze backend. The +; jump table should be lowered to a "br" instruction using one of the +; available registers. +; +; RUN: llc < %s -march=mblaze | FileCheck %s + +define i32 @jmptable(i32 %arg) +{ + ; CHECK: jmptable: + switch i32 %arg, label %DEFAULT [ i32 0, label %L0 + i32 1, label %L1 + i32 2, label %L2 + i32 3, label %L3 + i32 4, label %L4 + i32 5, label %L5 + i32 6, label %L6 + i32 7, label %L7 + i32 8, label %L8 + i32 9, label %L9 ] + + ; CHECK: lw [[REG:r[0-9]*]] + ; CHECK: br [[REG]] +L0: + %var0 = add i32 %arg, 0 + br label %DONE + +L1: + %var1 = add i32 %arg, 1 + br label %DONE + +L2: + %var2 = add i32 %arg, 2 + br label %DONE + +L3: + %var3 = add i32 %arg, 3 + br label %DONE + +L4: + %var4 = add i32 %arg, 4 + br label %DONE + +L5: + %var5 = add i32 %arg, 5 + br label %DONE + +L6: + %var6 = add i32 %arg, 6 + br label %DONE + +L7: + %var7 = add i32 %arg, 7 + br label %DONE + +L8: + %var8 = add i32 %arg, 8 + br label %DONE + +L9: + %var9 = add i32 %arg, 9 + br label %DONE + +DEFAULT: + unreachable + +DONE: + %rval = phi i32 [ %var0, %L0 ], + [ %var1, %L1 ], + [ %var2, %L2 ], + [ %var3, %L3 ], + [ %var4, %L4 ], + [ %var5, %L5 ], + [ %var6, %L6 ], + [ %var7, %L7 ], + [ %var8, %L8 ], + [ %var9, %L9 ] + ret i32 %rval + ; CHECK: rtsd +} diff --git a/test/CodeGen/MBlaze/loop.ll b/test/CodeGen/MBlaze/loop.ll new file mode 100644 index 0000000..b473020 --- /dev/null +++ b/test/CodeGen/MBlaze/loop.ll @@ -0,0 +1,47 @@ +; Test some complicated looping constructs to ensure that they +; compile successfully and that some sort of branching is used +; in the resulting code. +; +; RUN: llc < %s -march=mblaze -mattr=+mul,+fpu,+barrel | FileCheck %s + +declare i32 @printf(i8*, ...) +@MSG = internal constant [19 x i8] c"Message: %d %d %d\0A\00" + +define i32 @loop(i32 %a, i32 %b) +{ + ; CHECK: loop: +entry: + br label %loop_outer + +loop_outer: + %outer.0 = phi i32 [ 0, %entry ], [ %outer.2, %loop_outer_finish ] + br label %loop_inner + +loop_inner: + %inner.0 = phi i32 [ %a, %loop_outer ], [ %inner.3, %loop_inner_finish ] + %inner.1 = phi i32 [ %b, %loop_outer ], [ %inner.4, %loop_inner_finish ] + %inner.2 = phi i32 [ 0, %loop_outer ], [ %inner.5, %loop_inner_finish ] + %inner.3 = add i32 %inner.0, %inner.1 + %inner.4 = mul i32 %inner.2, 11 + br label %loop_inner_finish + +loop_inner_finish: + %inner.5 = add i32 %inner.2, 1 + ; CHECK: addi {{.*, 1}} + + call i32 (i8*,...)* @printf( i8* getelementptr([19 x i8]* @MSG,i32 0,i32 0), + i32 %inner.0, i32 %inner.1, i32 %inner.2 ) + ; CHECK: brlid + + %inner.6 = icmp eq i32 %inner.5, 100 + ; CHECK: cmp + + br i1 %inner.6, label %loop_inner, label %loop_outer_finish + ; CHECK: {{beq|bne}} + +loop_outer_finish: + %outer.1 = add i32 %outer.0, 1 + %outer.2 = urem i32 %outer.1, 1500 + br label %loop_outer + ; CHECK: br +} diff --git a/test/CodeGen/MBlaze/mul.ll b/test/CodeGen/MBlaze/mul.ll new file mode 100644 index 0000000..65d3e22 --- /dev/null +++ b/test/CodeGen/MBlaze/mul.ll @@ -0,0 +1,51 @@ +; Ensure that multiplication is lowered to function calls when the multiplier +; unit is not available in the hardware and that function calls are not used +; when the multiplier unit is available in the hardware. +; +; RUN: llc < %s -march=mblaze | FileCheck -check-prefix=FUN %s +; RUN: llc < %s -march=mblaze -mattr=+mul | FileCheck -check-prefix=MUL %s + +define i8 @test_i8(i8 %a, i8 %b) { + ; FUN: test_i8: + ; MUL: test_i8: + + %tmp.1 = mul i8 %a, %b + ; FUN-NOT: mul + ; FUN: brlid + ; MUL-NOT: brlid + ; MUL: mul + + ret i8 %tmp.1 + ; FUN: rtsd + ; MUL: rtsd +} + +define i16 @test_i16(i16 %a, i16 %b) { + ; FUN: test_i16: + ; MUL: test_i16: + + %tmp.1 = mul i16 %a, %b + ; FUN-NOT: mul + ; FUN: brlid + ; MUL-NOT: brlid + ; MUL: mul + + ret i16 %tmp.1 + ; FUN: rtsd + ; MUL: rtsd +} + +define i32 @test_i32(i32 %a, i32 %b) { + ; FUN: test_i32: + ; MUL: test_i32: + + %tmp.1 = mul i32 %a, %b + ; FUN-NOT: mul + ; FUN: brlid + ; MUL-NOT: brlid + ; MUL: mul + + ret i32 %tmp.1 + ; FUN: rtsd + ; MUL: rtsd +} diff --git a/test/CodeGen/MBlaze/mul64.ll b/test/CodeGen/MBlaze/mul64.ll new file mode 100644 index 0000000..e0ef413 --- /dev/null +++ b/test/CodeGen/MBlaze/mul64.ll @@ -0,0 +1,23 @@ +; Ensure that multiplication is lowered to function calls when the 64-bit +; multiplier unit is not available in the hardware and that function calls +; are not used when the 64-bit multiplier unit is available in the hardware. +; +; RUN: llc < %s -march=mblaze | FileCheck -check-prefix=FUN %s +; RUN: llc < %s -march=mblaze -mattr=+mul,+mul64 | \ +; RUN: FileCheck -check-prefix=MUL %s + +define i64 @test_i64(i64 %a, i64 %b) { + ; FUN: test_i64: + ; MUL: test_i64: + + %tmp.1 = mul i64 %a, %b + ; FUN-NOT: mul + ; FUN: brlid + ; MUL-NOT: brlid + ; MUL: mulh + ; MUL: mul + + ret i64 %tmp.1 + ; FUN: rtsd + ; MUL: rtsd +} diff --git a/test/CodeGen/MBlaze/select.ll b/test/CodeGen/MBlaze/select.ll new file mode 100644 index 0000000..47a88a1 --- /dev/null +++ b/test/CodeGen/MBlaze/select.ll @@ -0,0 +1,15 @@ +; Ensure that the select instruction is supported and is lowered to +; some sort of branch instruction. +; +; RUN: llc < %s -march=mblaze | FileCheck %s + +define i32 @testsel(i32 %a, i32 %b) +{ + ; CHECK: testsel: + %tmp.1 = icmp eq i32 %a, %b + ; CHECK: cmp + %tmp.2 = select i1 %tmp.1, i32 %a, i32 %b + ; CHECK: {{bne|beq}} + ret i32 %tmp.2 + ; CHECK: rtsd +} diff --git a/test/CodeGen/MBlaze/shift.ll b/test/CodeGen/MBlaze/shift.ll new file mode 100644 index 0000000..186115e --- /dev/null +++ b/test/CodeGen/MBlaze/shift.ll @@ -0,0 +1,117 @@ +; Ensure that shifts are lowered to loops when the barrel shifter unit is +; not available in the hardware and that loops are not used when the +; barrel shifter unit is available in the hardware. +; +; RUN: llc < %s -march=mblaze | FileCheck -check-prefix=FUN %s +; RUN: llc < %s -march=mblaze -mattr=+barrel | FileCheck -check-prefix=SHT %s + +define i8 @test_i8(i8 %a, i8 %b) { + ; FUN: test_i8: + ; SHT: test_i8: + + %tmp.1 = shl i8 %a, %b + ; FUN-NOT: bsll + ; FUN: andi + ; FUN: add + ; FUN: bnei + ; SHT-NOT: andi + ; SHT-NOT: bnei + ; SHT: bsll + + ret i8 %tmp.1 + ; FUN: rtsd + ; SHT: rtsd +} + +define i8 @testc_i8(i8 %a, i8 %b) { + ; FUN: testc_i8: + ; SHT: testc_i8: + + %tmp.1 = shl i8 %a, 5 + ; FUN-NOT: bsll + ; FUN: andi + ; FUN: add + ; FUN: bnei + ; SHT-NOT: andi + ; SHT-NOT: add + ; SHT-NOT: bnei + ; SHT: bslli + + ret i8 %tmp.1 + ; FUN: rtsd + ; SHT: rtsd +} + +define i16 @test_i16(i16 %a, i16 %b) { + ; FUN: test_i16: + ; SHT: test_i16: + + %tmp.1 = shl i16 %a, %b + ; FUN-NOT: bsll + ; FUN: andi + ; FUN: add + ; FUN: bnei + ; SHT-NOT: andi + ; SHT-NOT: bnei + ; SHT: bsll + + ret i16 %tmp.1 + ; FUN: rtsd + ; SHT: rtsd +} + +define i16 @testc_i16(i16 %a, i16 %b) { + ; FUN: testc_i16: + ; SHT: testc_i16: + + %tmp.1 = shl i16 %a, 5 + ; FUN-NOT: bsll + ; FUN: andi + ; FUN: add + ; FUN: bnei + ; SHT-NOT: andi + ; SHT-NOT: add + ; SHT-NOT: bnei + ; SHT: bslli + + ret i16 %tmp.1 + ; FUN: rtsd + ; SHT: rtsd +} + +define i32 @test_i32(i32 %a, i32 %b) { + ; FUN: test_i32: + ; SHT: test_i32: + + %tmp.1 = shl i32 %a, %b + ; FUN-NOT: bsll + ; FUN: andi + ; FUN: add + ; FUN: bnei + ; SHT-NOT: andi + ; SHT-NOT: bnei + ; SHT: bsll + + ret i32 %tmp.1 + ; FUN: rtsd + ; SHT: rtsd +} + +define i32 @testc_i32(i32 %a, i32 %b) { + ; FUN: testc_i32: + ; SHT: testc_i32: + + %tmp.1 = shl i32 %a, 5 + ; FUN-NOT: bsll + ; FUN: andi + ; FUN: add + ; FUN: bnei + ; SHT-NOT: andi + ; SHT-NOT: add + ; SHT-NOT: bnei + ; SHT: bslli + + ret i32 %tmp.1 + ; FUN: rtsd + ; SHT: rtsd +} diff --git a/test/CodeGen/MSP430/AddrMode-bis-rx.ll b/test/CodeGen/MSP430/AddrMode-bis-rx.ll index 3340494..115464d 100644 --- a/test/CodeGen/MSP430/AddrMode-bis-rx.ll +++ b/test/CodeGen/MSP430/AddrMode-bis-rx.ll @@ -1,4 +1,4 @@ -; RUN: llvm-as < %s | llc -march=msp430 | FileCheck %s +; RUN: llc < %s -march=msp430 | FileCheck %s target datalayout = "e-p:16:16:16-i1:8:8-i8:8:8-i16:16:16-i32:16:16" target triple = "msp430-generic-generic" diff --git a/test/CodeGen/MSP430/AddrMode-bis-xr.ll b/test/CodeGen/MSP430/AddrMode-bis-xr.ll index ca79fb6..3baf332 100644 --- a/test/CodeGen/MSP430/AddrMode-bis-xr.ll +++ b/test/CodeGen/MSP430/AddrMode-bis-xr.ll @@ -1,4 +1,4 @@ -; RUN: llvm-as < %s | llc -march=msp430 | FileCheck %s +; RUN: llc < %s -march=msp430 | FileCheck %s target datalayout = "e-p:16:16:16-i8:8:8-i16:16:16-i32:16:16" target triple = "msp430-generic-generic" diff --git a/test/CodeGen/MSP430/AddrMode-mov-rx.ll b/test/CodeGen/MSP430/AddrMode-mov-rx.ll index 67cbb02..9144f9a 100644 --- a/test/CodeGen/MSP430/AddrMode-mov-rx.ll +++ b/test/CodeGen/MSP430/AddrMode-mov-rx.ll @@ -1,4 +1,4 @@ -; RUN: llvm-as < %s | llc -march=msp430 | FileCheck %s +; RUN: llc < %s -march=msp430 | FileCheck %s target datalayout = "e-p:16:16:16-i1:8:8-i8:8:8-i16:16:16-i32:16:16" target triple = "msp430-generic-generic" diff --git a/test/CodeGen/MSP430/AddrMode-mov-xr.ll b/test/CodeGen/MSP430/AddrMode-mov-xr.ll index b8155d3..333c800 100644 --- a/test/CodeGen/MSP430/AddrMode-mov-xr.ll +++ b/test/CodeGen/MSP430/AddrMode-mov-xr.ll @@ -1,4 +1,4 @@ -; RUN: llvm-as < %s | llc -march=msp430 | FileCheck %s +; RUN: llc < %s -march=msp430 | FileCheck %s target datalayout = "e-p:16:16:16-i1:8:8-i8:8:8-i16:16:16-i32:16:16" target triple = "msp430-generic-generic" diff --git a/test/CodeGen/MSP430/Inst16mm.ll b/test/CodeGen/MSP430/Inst16mm.ll index 510afe3..2337c2c 100644 --- a/test/CodeGen/MSP430/Inst16mm.ll +++ b/test/CodeGen/MSP430/Inst16mm.ll @@ -1,4 +1,4 @@ -; RUN: llc -march=msp430 < %s | FileCheck %s +; RUN: llc -march=msp430 -combiner-alias-analysis < %s | FileCheck %s target datalayout = "e-p:16:8:8-i8:8:8-i16:8:8-i32:8:8" target triple = "msp430-generic-generic" @foo = common global i16 0, align 2 @@ -52,3 +52,18 @@ define void @xor() nounwind { ret void } +define i16 @mov2() nounwind { +entry: + %retval = alloca i16 ; <i16*> [#uses=3] + %x = alloca i32, align 2 ; <i32*> [#uses=1] + %y = alloca i32, align 2 ; <i32*> [#uses=1] + store i16 0, i16* %retval + %tmp = load i32* %y ; <i32> [#uses=1] + store i32 %tmp, i32* %x + store i16 0, i16* %retval + %0 = load i16* %retval ; <i16> [#uses=1] + ret i16 %0 +; CHECK: mov2: +; CHECK: mov.w 0(r1), 4(r1) +; CHECK: mov.w 2(r1), 6(r1) +} diff --git a/test/CodeGen/MSP430/Inst8rr.ll b/test/CodeGen/MSP430/Inst8rr.ll index 74feaae..0f5fc12 100644 --- a/test/CodeGen/MSP430/Inst8rr.ll +++ b/test/CodeGen/MSP430/Inst8rr.ll @@ -10,7 +10,7 @@ define i8 @mov(i8 %a, i8 %b) nounwind { define i8 @add(i8 %a, i8 %b) nounwind { ; CHECK: add: -; CHECK: add.b r14, r15 +; CHECK: add.b r12, r15 %1 = add i8 %a, %b ret i8 %1 } diff --git a/test/CodeGen/MSP430/bit.ll b/test/CodeGen/MSP430/bit.ll index cd664a1..03d672b 100644 --- a/test/CodeGen/MSP430/bit.ll +++ b/test/CodeGen/MSP430/bit.ll @@ -1,4 +1,4 @@ -; RUN: llvm-as < %s | llc -march=msp430 | FileCheck %s +; RUN: llc < %s -march=msp430 | FileCheck %s target datalayout = "e-p:16:16:16-i1:8:8-i8:8:8-i16:16:16-i32:16:32" target triple = "msp430-generic-generic" diff --git a/test/CodeGen/MSP430/setcc.ll b/test/CodeGen/MSP430/setcc.ll index 9db51cc..c99b17e 100644 --- a/test/CodeGen/MSP430/setcc.ll +++ b/test/CodeGen/MSP430/setcc.ll @@ -10,9 +10,9 @@ define i16 @sccweqand(i16 %a, i16 %b) nounwind { } ; CHECK: sccweqand: ; CHECK: bit.w r14, r15 -; CHECK-NEXT: mov.w r2, r15 -; CHECK-NEXT: and.w #1, r15 -; CHECK-NEXT: xor.w #1, r15 +; CHECK: mov.w r2, r15 +; CHECK: rra.w r15 +; CHECK: and.w #1, r15 define i16 @sccwneand(i16 %a, i16 %b) nounwind { %t1 = and i16 %a, %b @@ -22,8 +22,8 @@ define i16 @sccwneand(i16 %a, i16 %b) nounwind { } ; CHECK: sccwneand: ; CHECK: bit.w r14, r15 -; CHECK-NEXT: mov.w r2, r15 -; CHECK-NEXT: and.w #1, r15 +; CHECK: mov.w r2, r15 +; CHECK: and.w #1, r15 define i16 @sccwne(i16 %a, i16 %b) nounwind { %t1 = icmp ne i16 %a, %b @@ -32,9 +32,10 @@ define i16 @sccwne(i16 %a, i16 %b) nounwind { } ; CHECK:sccwne: ; CHECK: cmp.w r14, r15 -; CHECK-NEXT: mov.w r2, r15 -; CHECK-NEXT: rra.w r15 -; CHECK-NEXT: and.w #1, r15 +; CHECK: mov.w r2, r15 +; CHECK: rra.w r15 +; CHECK: and.w #1, r15 +; CHECK: xor.w #1, r15 define i16 @sccweq(i16 %a, i16 %b) nounwind { %t1 = icmp eq i16 %a, %b @@ -43,10 +44,9 @@ define i16 @sccweq(i16 %a, i16 %b) nounwind { } ; CHECK:sccweq: ; CHECK: cmp.w r14, r15 -; CHECK-NEXT: mov.w r2, r15 -; CHECK-NEXT: rra.w r15 -; CHECK-NEXT: and.w #1, r15 -; CHECK-NEXT: xor.w #1, r15 +; CHECK: mov.w r2, r15 +; CHECK: rra.w r15 +; CHECK: and.w #1, r15 define i16 @sccwugt(i16 %a, i16 %b) nounwind { %t1 = icmp ugt i16 %a, %b @@ -55,9 +55,9 @@ define i16 @sccwugt(i16 %a, i16 %b) nounwind { } ; CHECK:sccwugt: ; CHECK: cmp.w r15, r14 -; CHECK-NEXT: mov.w r2, r15 -; CHECK-NEXT: and.w #1, r15 -; CHECK-NEXT: xor.w #1, r15 +; CHECK: mov.w r2, r15 +; CHECK: and.w #1, r15 +; CHECK: xor.w #1, r15 define i16 @sccwuge(i16 %a, i16 %b) nounwind { %t1 = icmp uge i16 %a, %b @@ -66,8 +66,8 @@ define i16 @sccwuge(i16 %a, i16 %b) nounwind { } ; CHECK:sccwuge: ; CHECK: cmp.w r14, r15 -; CHECK-NEXT: mov.w r2, r15 -; CHECK-NEXT: and.w #1, r15 +; CHECK: mov.w r2, r15 +; CHECK: and.w #1, r15 define i16 @sccwult(i16 %a, i16 %b) nounwind { %t1 = icmp ult i16 %a, %b @@ -76,9 +76,9 @@ define i16 @sccwult(i16 %a, i16 %b) nounwind { } ; CHECK:sccwult: ; CHECK: cmp.w r14, r15 -; CHECK-NEXT: mov.w r2, r15 -; CHECK-NEXT: and.w #1, r15 -; CHECK-NEXT: xor.w #1, r15 +; CHECK: mov.w r2, r15 +; CHECK: and.w #1, r15 +; CHECK: xor.w #1, r15 define i16 @sccwule(i16 %a, i16 %b) nounwind { %t1 = icmp ule i16 %a, %b @@ -87,8 +87,8 @@ define i16 @sccwule(i16 %a, i16 %b) nounwind { } ; CHECK:sccwule: ; CHECK: cmp.w r15, r14 -; CHECK-NEXT: mov.w r2, r15 -; CHECK-NEXT: and.w #1, r15 +; CHECK: mov.w r2, r15 +; CHECK: and.w #1, r15 define i16 @sccwsgt(i16 %a, i16 %b) nounwind { %t1 = icmp sgt i16 %a, %b diff --git a/test/CodeGen/PIC16/C16-11.ll b/test/CodeGen/PIC16/C16-11.ll index e70092b..8a5a0ac 100644 --- a/test/CodeGen/PIC16/C16-11.ll +++ b/test/CodeGen/PIC16/C16-11.ll @@ -1,4 +1,7 @@ -;RUN: llc < %s -march=pic16 +; RUN: llc < %s -march=pic16 +; XFAIL: * +; This fails because PIC16 doesn't define a (xor reg, reg) pattern. +; @c612.auto.a.b = internal global i1 false ; <i1*> [#uses=2] @c612.auto.A.b = internal global i1 false ; <i1*> [#uses=2] diff --git a/test/CodeGen/PIC16/C16-15.ll b/test/CodeGen/PIC16/C16-15.ll index 2e1dc0c..5ca2d4a 100644 --- a/test/CodeGen/PIC16/C16-15.ll +++ b/test/CodeGen/PIC16/C16-15.ll @@ -1,4 +1,4 @@ -; RUN: llc < %s -march=pic16 | grep "extern @.lib.unordered.f32" | count 3 +; RUN: llc < %s -march=pic16 | grep "extern" | grep "@.lib.unordered.f32" | count 3 @pc = global i8* inttoptr (i64 160 to i8*), align 1 ; <i8**> [#uses=2] @aa = common global i16 0, align 1 ; <i16*> [#uses=0] diff --git a/test/CodeGen/PowerPC/2007-04-30-InlineAsmEarlyClobber.ll b/test/CodeGen/PowerPC/2007-04-30-InlineAsmEarlyClobber.ll index d1d28ae..be28a9a 100644 --- a/test/CodeGen/PowerPC/2007-04-30-InlineAsmEarlyClobber.ll +++ b/test/CodeGen/PowerPC/2007-04-30-InlineAsmEarlyClobber.ll @@ -1,6 +1,6 @@ ; RUN: llc < %s | grep {subfc r3,r5,r4} -; RUN: llc < %s | grep {subfze r4,r2} -; RUN: llc < %s -regalloc=local | grep {subfc r2,r5,r4} +; RUN: llc < %s | grep {subfze r4,r6} +; RUN: llc < %s -regalloc=local | grep {subfc r6,r5,r4} ; RUN: llc < %s -regalloc=local | grep {subfze r3,r3} ; The first argument of subfc must not be the same as any other register. diff --git a/test/CodeGen/PowerPC/2009-08-17-inline-asm-addr-mode-breakage.ll b/test/CodeGen/PowerPC/2009-08-17-inline-asm-addr-mode-breakage.ll index 5d09696..50a0278 100644 --- a/test/CodeGen/PowerPC/2009-08-17-inline-asm-addr-mode-breakage.ll +++ b/test/CodeGen/PowerPC/2009-08-17-inline-asm-addr-mode-breakage.ll @@ -1,4 +1,4 @@ -; RUN: llc < %s -march=ppc32 | FileCheck %s +; RUN: llc < %s -march=ppc32 -mtriple=powerpc-apple-darwin10 -mcpu=g5 | FileCheck %s ; ModuleID = '<stdin>' target datalayout = "E-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f128:64:128" target triple = "powerpc-apple-darwin10.0" @@ -10,8 +10,8 @@ target triple = "powerpc-apple-darwin10.0" define void @foo(i32 %y) nounwind ssp { entry: ; CHECK: foo -; CHECK: add r2 -; CHECK: 0(r2) +; CHECK: add r4 +; CHECK: 0(r4) %y_addr = alloca i32 ; <i32*> [#uses=2] %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] store i32 %y, i32* %y_addr diff --git a/test/CodeGen/PowerPC/2010-02-26-FoldFloats.ll b/test/CodeGen/PowerPC/2010-02-26-FoldFloats.ll new file mode 100644 index 0000000..f43f5ca --- /dev/null +++ b/test/CodeGen/PowerPC/2010-02-26-FoldFloats.ll @@ -0,0 +1,433 @@ +; RUN: llc < %s -O3 | FileCheck %s +target datalayout = "E-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f128:64:128-n32" +target triple = "powerpc-apple-darwin9.6" + +; There should be no stfs spills +; CHECK: main: +; CHECK-NOT: stfs +; CHECK: .section + +@.str66 = external constant [3 x i8], align 4 ; <[3 x i8]*> [#uses=1] +@.str31 = external constant [6 x i8], align 4 ; <[6 x i8]*> [#uses=1] +@.str61 = external constant [21 x i8], align 4 ; <[21 x i8]*> [#uses=1] +@.str101 = external constant [61 x i8], align 4 ; <[61 x i8]*> [#uses=1] +@.str104 = external constant [31 x i8], align 4 ; <[31 x i8]*> [#uses=1] +@.str105 = external constant [45 x i8], align 4 ; <[45 x i8]*> [#uses=1] +@.str112 = external constant [38 x i8], align 4 ; <[38 x i8]*> [#uses=1] +@.str121 = external constant [36 x i8], align 4 ; <[36 x i8]*> [#uses=1] +@.str12293 = external constant [67 x i8], align 4 ; <[67 x i8]*> [#uses=1] +@.str123 = external constant [68 x i8], align 4 ; <[68 x i8]*> [#uses=1] +@.str124 = external constant [52 x i8], align 4 ; <[52 x i8]*> [#uses=1] +@.str125 = external constant [51 x i8], align 4 ; <[51 x i8]*> [#uses=1] + +define i32 @main(i32 %argc, i8** %argv) noreturn nounwind { +entry: + br i1 undef, label %bb4.i1, label %my_fopen.exit + +bb4.i1: ; preds = %entry + unreachable + +my_fopen.exit: ; preds = %entry + br i1 undef, label %bb.i, label %bb1.i + +bb.i: ; preds = %my_fopen.exit + unreachable + +bb1.i: ; preds = %my_fopen.exit + br label %bb134.i + +bb2.i: ; preds = %bb134.i + %0 = icmp eq i32 undef, 0 ; <i1> [#uses=1] + br i1 %0, label %bb20.i, label %bb21.i + +bb20.i: ; preds = %bb2.i + br label %bb134.i + +bb21.i: ; preds = %bb2.i + %1 = call i32 @strcmp(i8* undef, i8* getelementptr inbounds ([6 x i8]* @.str31, i32 0, i32 0)) nounwind readonly ; <i32> [#uses=0] + br i1 undef, label %bb30.i, label %bb31.i + +bb30.i: ; preds = %bb21.i + br label %bb134.i + +bb31.i: ; preds = %bb21.i + br i1 undef, label %bb41.i, label %bb44.i + +bb41.i: ; preds = %bb31.i + %2 = icmp slt i32 undef, %argc ; <i1> [#uses=1] + br i1 %2, label %bb1.i77.i, label %bb2.i78.i + +bb1.i77.i: ; preds = %bb41.i + %3 = load float* undef, align 4 ; <float> [#uses=2] + %4 = fcmp ugt float %3, 0.000000e+00 ; <i1> [#uses=1] + br i1 %4, label %bb43.i, label %bb42.i + +bb2.i78.i: ; preds = %bb41.i + unreachable + +bb42.i: ; preds = %bb1.i77.i + unreachable + +bb43.i: ; preds = %bb1.i77.i + br label %bb134.i + +bb44.i: ; preds = %bb31.i + br i1 undef, label %bb45.i, label %bb49.i + +bb45.i: ; preds = %bb44.i + %5 = icmp slt i32 undef, %argc ; <i1> [#uses=1] + br i1 %5, label %bb1.i72.i, label %bb2.i73.i + +bb1.i72.i: ; preds = %bb45.i + %6 = load float* undef, align 4 ; <float> [#uses=3] + %7 = fcmp ult float %6, 1.000000e+00 ; <i1> [#uses=1] + %or.cond.i = and i1 undef, %7 ; <i1> [#uses=1] + br i1 %or.cond.i, label %bb48.i, label %bb47.i + +bb2.i73.i: ; preds = %bb45.i + unreachable + +bb47.i: ; preds = %bb1.i72.i + unreachable + +bb48.i: ; preds = %bb1.i72.i + br label %bb134.i + +bb49.i: ; preds = %bb44.i + br i1 undef, label %bb50.i, label %bb53.i + +bb50.i: ; preds = %bb49.i + br i1 false, label %bb1.i67.i, label %bb2.i68.i + +bb1.i67.i: ; preds = %bb50.i + br i1 false, label %read_float_option.exit69.i, label %bb1.i67.bb2.i68_crit_edge.i + +bb1.i67.bb2.i68_crit_edge.i: ; preds = %bb1.i67.i + br label %bb2.i68.i + +bb2.i68.i: ; preds = %bb1.i67.bb2.i68_crit_edge.i, %bb50.i + unreachable + +read_float_option.exit69.i: ; preds = %bb1.i67.i + br i1 undef, label %bb52.i, label %bb51.i + +bb51.i: ; preds = %read_float_option.exit69.i + unreachable + +bb52.i: ; preds = %read_float_option.exit69.i + br label %bb134.i + +bb53.i: ; preds = %bb49.i + %8 = call i32 @strcmp(i8* undef, i8* getelementptr inbounds ([21 x i8]* @.str61, i32 0, i32 0)) nounwind readonly ; <i32> [#uses=0] + br i1 false, label %bb89.i, label %bb92.i + +bb89.i: ; preds = %bb53.i + br i1 undef, label %bb1.i27.i, label %bb2.i28.i + +bb1.i27.i: ; preds = %bb89.i + unreachable + +bb2.i28.i: ; preds = %bb89.i + unreachable + +bb92.i: ; preds = %bb53.i + br i1 undef, label %bb93.i, label %bb96.i + +bb93.i: ; preds = %bb92.i + br i1 undef, label %bb1.i22.i, label %bb2.i23.i + +bb1.i22.i: ; preds = %bb93.i + br i1 undef, label %bb95.i, label %bb94.i + +bb2.i23.i: ; preds = %bb93.i + unreachable + +bb94.i: ; preds = %bb1.i22.i + unreachable + +bb95.i: ; preds = %bb1.i22.i + br label %bb134.i + +bb96.i: ; preds = %bb92.i + br i1 undef, label %bb97.i, label %bb100.i + +bb97.i: ; preds = %bb96.i + %9 = icmp slt i32 undef, %argc ; <i1> [#uses=1] + br i1 %9, label %bb1.i17.i, label %bb2.i18.i + +bb1.i17.i: ; preds = %bb97.i + %10 = call i32 (i8*, i8*, ...)* @"\01_sscanf$LDBL128"(i8* undef, i8* getelementptr inbounds ([3 x i8]* @.str66, i32 0, i32 0), float* undef) nounwind ; <i32> [#uses=1] + %phitmp.i16.i = icmp eq i32 %10, 1 ; <i1> [#uses=1] + br i1 %phitmp.i16.i, label %read_float_option.exit19.i, label %bb1.i17.bb2.i18_crit_edge.i + +bb1.i17.bb2.i18_crit_edge.i: ; preds = %bb1.i17.i + br label %bb2.i18.i + +bb2.i18.i: ; preds = %bb1.i17.bb2.i18_crit_edge.i, %bb97.i + unreachable + +read_float_option.exit19.i: ; preds = %bb1.i17.i + br i1 false, label %bb99.i, label %bb98.i + +bb98.i: ; preds = %read_float_option.exit19.i + unreachable + +bb99.i: ; preds = %read_float_option.exit19.i + br label %bb134.i + +bb100.i: ; preds = %bb96.i + br i1 false, label %bb101.i, label %bb104.i + +bb101.i: ; preds = %bb100.i + br i1 false, label %bb1.i12.i, label %bb2.i13.i + +bb1.i12.i: ; preds = %bb101.i + br i1 undef, label %bb102.i, label %bb103.i + +bb2.i13.i: ; preds = %bb101.i + unreachable + +bb102.i: ; preds = %bb1.i12.i + unreachable + +bb103.i: ; preds = %bb1.i12.i + br label %bb134.i + +bb104.i: ; preds = %bb100.i + unreachable + +bb134.i: ; preds = %bb103.i, %bb99.i, %bb95.i, %bb52.i, %bb48.i, %bb43.i, %bb30.i, %bb20.i, %bb1.i + %annealing_sched.1.0 = phi float [ 1.000000e+01, %bb1.i ], [ %annealing_sched.1.0, %bb20.i ], [ 1.000000e+00, %bb30.i ], [ %annealing_sched.1.0, %bb43.i ], [ %annealing_sched.1.0, %bb48.i ], [ %annealing_sched.1.0, %bb52.i ], [ %annealing_sched.1.0, %bb95.i ], [ %annealing_sched.1.0, %bb99.i ], [ %annealing_sched.1.0, %bb103.i ] ; <float> [#uses=8] + %annealing_sched.2.0 = phi float [ 1.000000e+02, %bb1.i ], [ %annealing_sched.2.0, %bb20.i ], [ %annealing_sched.2.0, %bb30.i ], [ %3, %bb43.i ], [ %annealing_sched.2.0, %bb48.i ], [ %annealing_sched.2.0, %bb52.i ], [ %annealing_sched.2.0, %bb95.i ], [ %annealing_sched.2.0, %bb99.i ], [ %annealing_sched.2.0, %bb103.i ] ; <float> [#uses=8] + %annealing_sched.3.0 = phi float [ 0x3FE99999A0000000, %bb1.i ], [ %annealing_sched.3.0, %bb20.i ], [ %annealing_sched.3.0, %bb30.i ], [ %annealing_sched.3.0, %bb43.i ], [ %6, %bb48.i ], [ %annealing_sched.3.0, %bb52.i ], [ %annealing_sched.3.0, %bb95.i ], [ %annealing_sched.3.0, %bb99.i ], [ %annealing_sched.3.0, %bb103.i ] ; <float> [#uses=8] + %annealing_sched.4.0 = phi float [ 0x3F847AE140000000, %bb1.i ], [ %annealing_sched.4.0, %bb20.i ], [ %annealing_sched.4.0, %bb30.i ], [ %annealing_sched.4.0, %bb43.i ], [ %annealing_sched.4.0, %bb48.i ], [ 0.000000e+00, %bb52.i ], [ %annealing_sched.4.0, %bb95.i ], [ %annealing_sched.4.0, %bb99.i ], [ %annealing_sched.4.0, %bb103.i ] ; <float> [#uses=8] + %router_opts.0.0 = phi float [ 0.000000e+00, %bb1.i ], [ %router_opts.0.0, %bb20.i ], [ 1.000000e+04, %bb30.i ], [ %router_opts.0.0, %bb43.i ], [ %router_opts.0.0, %bb48.i ], [ %router_opts.0.0, %bb52.i ], [ %router_opts.0.0, %bb95.i ], [ %router_opts.0.0, %bb99.i ], [ %router_opts.0.0, %bb103.i ] ; <float> [#uses=8] + %router_opts.1.0 = phi float [ 5.000000e-01, %bb1.i ], [ %router_opts.1.0, %bb20.i ], [ 1.000000e+04, %bb30.i ], [ %router_opts.1.0, %bb43.i ], [ %router_opts.1.0, %bb48.i ], [ %router_opts.1.0, %bb52.i ], [ undef, %bb95.i ], [ %router_opts.1.0, %bb99.i ], [ %router_opts.1.0, %bb103.i ] ; <float> [#uses=7] + %router_opts.2.0 = phi float [ 1.500000e+00, %bb1.i ], [ %router_opts.2.0, %bb20.i ], [ %router_opts.2.0, %bb30.i ], [ %router_opts.2.0, %bb43.i ], [ %router_opts.2.0, %bb48.i ], [ %router_opts.2.0, %bb52.i ], [ %router_opts.2.0, %bb95.i ], [ undef, %bb99.i ], [ %router_opts.2.0, %bb103.i ] ; <float> [#uses=8] + %router_opts.3.0 = phi float [ 0x3FC99999A0000000, %bb1.i ], [ %router_opts.3.0, %bb20.i ], [ %router_opts.3.0, %bb30.i ], [ %router_opts.3.0, %bb43.i ], [ %router_opts.3.0, %bb48.i ], [ %router_opts.3.0, %bb52.i ], [ %router_opts.3.0, %bb95.i ], [ %router_opts.3.0, %bb99.i ], [ 0.000000e+00, %bb103.i ] ; <float> [#uses=8] + %11 = phi float [ 0x3FC99999A0000000, %bb1.i ], [ %11, %bb20.i ], [ %11, %bb30.i ], [ %11, %bb43.i ], [ %11, %bb48.i ], [ %11, %bb52.i ], [ %11, %bb95.i ], [ %11, %bb99.i ], [ 0.000000e+00, %bb103.i ] ; <float> [#uses=8] + %12 = phi float [ 1.500000e+00, %bb1.i ], [ %12, %bb20.i ], [ %12, %bb30.i ], [ %12, %bb43.i ], [ %12, %bb48.i ], [ %12, %bb52.i ], [ %12, %bb95.i ], [ undef, %bb99.i ], [ %12, %bb103.i ] ; <float> [#uses=8] + %13 = phi float [ 5.000000e-01, %bb1.i ], [ %13, %bb20.i ], [ 1.000000e+04, %bb30.i ], [ %13, %bb43.i ], [ %13, %bb48.i ], [ %13, %bb52.i ], [ undef, %bb95.i ], [ %13, %bb99.i ], [ %13, %bb103.i ] ; <float> [#uses=7] + %14 = phi float [ 0.000000e+00, %bb1.i ], [ %14, %bb20.i ], [ 1.000000e+04, %bb30.i ], [ %14, %bb43.i ], [ %14, %bb48.i ], [ %14, %bb52.i ], [ %14, %bb95.i ], [ %14, %bb99.i ], [ %14, %bb103.i ] ; <float> [#uses=8] + %15 = phi float [ 0x3FE99999A0000000, %bb1.i ], [ %15, %bb20.i ], [ %15, %bb30.i ], [ %15, %bb43.i ], [ %6, %bb48.i ], [ %15, %bb52.i ], [ %15, %bb95.i ], [ %15, %bb99.i ], [ %15, %bb103.i ] ; <float> [#uses=8] + %16 = phi float [ 0x3F847AE140000000, %bb1.i ], [ %16, %bb20.i ], [ %16, %bb30.i ], [ %16, %bb43.i ], [ %16, %bb48.i ], [ 0.000000e+00, %bb52.i ], [ %16, %bb95.i ], [ %16, %bb99.i ], [ %16, %bb103.i ] ; <float> [#uses=8] + %17 = phi float [ 1.000000e+01, %bb1.i ], [ %17, %bb20.i ], [ 1.000000e+00, %bb30.i ], [ %17, %bb43.i ], [ %17, %bb48.i ], [ %17, %bb52.i ], [ %17, %bb95.i ], [ %17, %bb99.i ], [ %17, %bb103.i ] ; <float> [#uses=8] + %18 = icmp slt i32 undef, %argc ; <i1> [#uses=1] + br i1 %18, label %bb2.i, label %bb135.i + +bb135.i: ; preds = %bb134.i + br i1 undef, label %bb141.i, label %bb142.i + +bb141.i: ; preds = %bb135.i + unreachable + +bb142.i: ; preds = %bb135.i + br i1 undef, label %bb145.i, label %bb144.i + +bb144.i: ; preds = %bb142.i + unreachable + +bb145.i: ; preds = %bb142.i + br i1 undef, label %bb146.i, label %bb147.i + +bb146.i: ; preds = %bb145.i + unreachable + +bb147.i: ; preds = %bb145.i + br i1 undef, label %bb148.i, label %bb155.i + +bb148.i: ; preds = %bb147.i + br label %bb155.i + +bb155.i: ; preds = %bb148.i, %bb147.i + br i1 undef, label %bb156.i, label %bb161.i + +bb156.i: ; preds = %bb155.i + unreachable + +bb161.i: ; preds = %bb155.i + br i1 undef, label %bb162.i, label %bb163.i + +bb162.i: ; preds = %bb161.i + %19 = fpext float %17 to double ; <double> [#uses=1] + %20 = call i32 (i8*, ...)* @"\01_printf$LDBL128"(i8* getelementptr inbounds ([61 x i8]* @.str101, i32 0, i32 0), double %19) nounwind ; <i32> [#uses=0] + unreachable + +bb163.i: ; preds = %bb161.i + %21 = fpext float %16 to double ; <double> [#uses=1] + %22 = call i32 (i8*, ...)* @"\01_printf$LDBL128"(i8* getelementptr inbounds ([31 x i8]* @.str104, i32 0, i32 0), double %21) nounwind ; <i32> [#uses=0] + %23 = fpext float %15 to double ; <double> [#uses=1] + %24 = call i32 (i8*, ...)* @"\01_printf$LDBL128"(i8* getelementptr inbounds ([45 x i8]* @.str105, i32 0, i32 0), double %23) nounwind ; <i32> [#uses=0] + %25 = call i32 (i8*, ...)* @"\01_printf$LDBL128"(i8* getelementptr inbounds ([38 x i8]* @.str112, i32 0, i32 0), double undef) nounwind ; <i32> [#uses=0] + br i1 undef, label %parse_command.exit, label %bb176.i + +bb176.i: ; preds = %bb163.i + br i1 undef, label %bb177.i, label %bb178.i + +bb177.i: ; preds = %bb176.i + unreachable + +bb178.i: ; preds = %bb176.i + %26 = call i32 (i8*, ...)* @"\01_printf$LDBL128"(i8* getelementptr inbounds ([36 x i8]* @.str121, i32 0, i32 0), double undef) nounwind ; <i32> [#uses=0] + %27 = fpext float %14 to double ; <double> [#uses=1] + %28 = call i32 (i8*, ...)* @"\01_printf$LDBL128"(i8* getelementptr inbounds ([67 x i8]* @.str12293, i32 0, i32 0), double %27) nounwind ; <i32> [#uses=0] + %29 = fpext float %13 to double ; <double> [#uses=1] + %30 = call i32 (i8*, ...)* @"\01_printf$LDBL128"(i8* getelementptr inbounds ([68 x i8]* @.str123, i32 0, i32 0), double %29) nounwind ; <i32> [#uses=0] + %31 = fpext float %12 to double ; <double> [#uses=1] + %32 = call i32 (i8*, ...)* @"\01_printf$LDBL128"(i8* getelementptr inbounds ([52 x i8]* @.str124, i32 0, i32 0), double %31) nounwind ; <i32> [#uses=0] + %33 = fpext float %11 to double ; <double> [#uses=1] + %34 = call i32 (i8*, ...)* @"\01_printf$LDBL128"(i8* getelementptr inbounds ([51 x i8]* @.str125, i32 0, i32 0), double %33) nounwind ; <i32> [#uses=0] + unreachable + +parse_command.exit: ; preds = %bb163.i + br i1 undef, label %bb4.i152.i, label %my_fopen.exit.i + +bb4.i152.i: ; preds = %parse_command.exit + unreachable + +my_fopen.exit.i: ; preds = %parse_command.exit + br i1 undef, label %bb.i6.i99, label %bb49.preheader.i.i + +bb.i6.i99: ; preds = %my_fopen.exit.i + br i1 undef, label %bb3.i.i100, label %bb1.i8.i + +bb1.i8.i: ; preds = %bb.i6.i99 + unreachable + +bb3.i.i100: ; preds = %bb.i6.i99 + unreachable + +bb49.preheader.i.i: ; preds = %my_fopen.exit.i + br i1 undef, label %bb7.i11.i, label %bb50.i.i + +bb7.i11.i: ; preds = %bb49.preheader.i.i + unreachable + +bb50.i.i: ; preds = %bb49.preheader.i.i + br i1 undef, label %bb.i.i.i20.i, label %my_calloc.exit.i.i.i + +bb.i.i.i20.i: ; preds = %bb50.i.i + unreachable + +my_calloc.exit.i.i.i: ; preds = %bb50.i.i + br i1 undef, label %bb.i.i37.i.i, label %alloc_hash_table.exit.i21.i + +bb.i.i37.i.i: ; preds = %my_calloc.exit.i.i.i + unreachable + +alloc_hash_table.exit.i21.i: ; preds = %my_calloc.exit.i.i.i + br i1 undef, label %bb51.i.i, label %bb3.i23.i.i + +bb51.i.i: ; preds = %alloc_hash_table.exit.i21.i + unreachable + +bb3.i23.i.i: ; preds = %alloc_hash_table.exit.i21.i + br i1 undef, label %bb.i8.i.i, label %bb.nph.i.i + +bb.nph.i.i: ; preds = %bb3.i23.i.i + unreachable + +bb.i8.i.i: ; preds = %bb3.i.i34.i, %bb3.i23.i.i + br i1 undef, label %bb3.i.i34.i, label %bb1.i.i32.i + +bb1.i.i32.i: ; preds = %bb.i8.i.i + unreachable + +bb3.i.i34.i: ; preds = %bb.i8.i.i + br i1 undef, label %free_hash_table.exit.i.i, label %bb.i8.i.i + +free_hash_table.exit.i.i: ; preds = %bb3.i.i34.i + br i1 undef, label %check_netlist.exit.i, label %bb59.i.i + +bb59.i.i: ; preds = %free_hash_table.exit.i.i + unreachable + +check_netlist.exit.i: ; preds = %free_hash_table.exit.i.i + br label %bb.i.i3.i + +bb.i.i3.i: ; preds = %bb3.i.i4.i, %check_netlist.exit.i + br i1 false, label %bb3.i.i4.i, label %bb1.i.i.i122 + +bb1.i.i.i122: ; preds = %bb1.i.i.i122, %bb.i.i3.i + br i1 false, label %bb3.i.i4.i, label %bb1.i.i.i122 + +bb3.i.i4.i: ; preds = %bb1.i.i.i122, %bb.i.i3.i + br i1 undef, label %read_net.exit, label %bb.i.i3.i + +read_net.exit: ; preds = %bb3.i.i4.i + br i1 undef, label %bb.i44, label %bb3.i47 + +bb.i44: ; preds = %read_net.exit + unreachable + +bb3.i47: ; preds = %read_net.exit + br i1 false, label %bb9.i50, label %bb8.i49 + +bb8.i49: ; preds = %bb3.i47 + unreachable + +bb9.i50: ; preds = %bb3.i47 + br i1 undef, label %bb11.i51, label %bb12.i52 + +bb11.i51: ; preds = %bb9.i50 + unreachable + +bb12.i52: ; preds = %bb9.i50 + br i1 undef, label %bb.i.i53, label %my_malloc.exit.i54 + +bb.i.i53: ; preds = %bb12.i52 + unreachable + +my_malloc.exit.i54: ; preds = %bb12.i52 + br i1 undef, label %bb.i2.i55, label %my_malloc.exit3.i56 + +bb.i2.i55: ; preds = %my_malloc.exit.i54 + unreachable + +my_malloc.exit3.i56: ; preds = %my_malloc.exit.i54 + br i1 undef, label %bb.i.i.i57, label %my_malloc.exit.i.i + +bb.i.i.i57: ; preds = %my_malloc.exit3.i56 + unreachable + +my_malloc.exit.i.i: ; preds = %my_malloc.exit3.i56 + br i1 undef, label %bb, label %bb10 + +bb: ; preds = %my_malloc.exit.i.i + unreachable + +bb10: ; preds = %my_malloc.exit.i.i + br i1 false, label %bb12, label %bb11 + +bb11: ; preds = %bb10 + unreachable + +bb12: ; preds = %bb10 + store float %annealing_sched.1.0, float* null, align 4 + store float %annealing_sched.2.0, float* undef, align 8 + store float %annealing_sched.3.0, float* undef, align 4 + store float %annealing_sched.4.0, float* undef, align 8 + store float %router_opts.0.0, float* undef, align 8 + store float %router_opts.1.0, float* undef, align 4 + store float %router_opts.2.0, float* null, align 8 + store float %router_opts.3.0, float* undef, align 4 + br i1 undef, label %place_and_route.exit, label %bb7.i22 + +bb7.i22: ; preds = %bb12 + br i1 false, label %bb8.i23, label %bb9.i26 + +bb8.i23: ; preds = %bb7.i22 + unreachable + +bb9.i26: ; preds = %bb7.i22 + unreachable + +place_and_route.exit: ; preds = %bb12 + unreachable +} + +declare i32 @"\01_printf$LDBL128"(i8*, ...) nounwind + +declare i32 @strcmp(i8* nocapture, i8* nocapture) nounwind readonly + +declare i32 @"\01_sscanf$LDBL128"(i8*, i8*, ...) nounwind diff --git a/test/CodeGen/PowerPC/Frames-alloca.ll b/test/CodeGen/PowerPC/Frames-alloca.ll index aed4fdb..466ae80 100644 --- a/test/CodeGen/PowerPC/Frames-alloca.ll +++ b/test/CodeGen/PowerPC/Frames-alloca.ll @@ -24,7 +24,7 @@ ; CHECK-PPC64-NOFP: ld r1, 0(r1) ; CHECK-PPC64-NOFP: ld r31, -8(r1) -define i32* @f1(i32 %n) { +define i32* @f1(i32 %n) nounwind { %tmp = alloca i32, i32 %n ; <i32*> [#uses=1] ret i32* %tmp } diff --git a/test/CodeGen/PowerPC/LargeAbsoluteAddr.ll b/test/CodeGen/PowerPC/LargeAbsoluteAddr.ll index 0f7acac..b10a996 100644 --- a/test/CodeGen/PowerPC/LargeAbsoluteAddr.ll +++ b/test/CodeGen/PowerPC/LargeAbsoluteAddr.ll @@ -3,14 +3,14 @@ ; RUN: llc < %s -march=ppc64 -mtriple=powerpc-apple-darwin | \ ; RUN: grep {stw r3, 32751} ; RUN: llc < %s -march=ppc64 -mtriple=powerpc-apple-darwin | \ -; RUN: grep {std r2, 9024} +; RUN: grep {std r3, 9024} -define void @test() { +define void @test() nounwind { store i32 0, i32* inttoptr (i64 48725999 to i32*) ret void } -define void @test2() { +define void @test2() nounwind { store i64 0, i64* inttoptr (i64 74560 to i64*) ret void } diff --git a/test/CodeGen/PowerPC/addc.ll b/test/CodeGen/PowerPC/addc.ll index 09a7fbd..8c928ce 100644 --- a/test/CodeGen/PowerPC/addc.ll +++ b/test/CodeGen/PowerPC/addc.ll @@ -1,26 +1,33 @@ ; All of these should be codegen'd without loading immediates -; RUN: llc < %s -march=ppc32 -o %t -; RUN: grep addc %t | count 1 -; RUN: grep adde %t | count 1 -; RUN: grep addze %t | count 1 -; RUN: grep addme %t | count 1 -; RUN: grep addic %t | count 2 +; RUN: llc < %s -mtriple=powerpc-apple-darwin | FileCheck %s -define i64 @add_ll(i64 %a, i64 %b) { +define i64 @add_ll(i64 %a, i64 %b) nounwind { entry: %tmp.2 = add i64 %b, %a ; <i64> [#uses=1] ret i64 %tmp.2 +; CHECK: add_ll: +; CHECK: addc r4, r6, r4 +; CHECK: adde r3, r5, r3 +; CHECK: blr } -define i64 @add_l_5(i64 %a) { +define i64 @add_l_5(i64 %a) nounwind { entry: %tmp.1 = add i64 %a, 5 ; <i64> [#uses=1] ret i64 %tmp.1 +; CHECK: add_l_5: +; CHECK: addic r4, r4, 5 +; CHECK: addze r3, r3 +; CHECK: blr } -define i64 @add_l_m5(i64 %a) { +define i64 @add_l_m5(i64 %a) nounwind { entry: %tmp.1 = add i64 %a, -5 ; <i64> [#uses=1] ret i64 %tmp.1 +; CHECK: add_l_m5: +; CHECK: addic r4, r4, -5 +; CHECK: addme r3, r3 +; CHECK: blr } diff --git a/test/CodeGen/PowerPC/indirectbr.ll b/test/CodeGen/PowerPC/indirectbr.ll index fbc7bd2..2094e10 100644 --- a/test/CodeGen/PowerPC/indirectbr.ll +++ b/test/CodeGen/PowerPC/indirectbr.ll @@ -43,13 +43,13 @@ L2: ; preds = %L3, %bb2 L1: ; preds = %L2, %bb2 %res.3 = phi i32 [ %phitmp, %L2 ], [ 2, %bb2 ] ; <i32> [#uses=1] -; PIC: addis r4, r2, ha16(L_BA4__foo_L5-"L1$pb") -; PIC: li r5, lo16(L_BA4__foo_L5-"L1$pb") -; PIC: add r4, r4, r5 -; PIC: stw r4 -; STATIC: li r2, lo16(L_BA4__foo_L5) -; STATIC: addis r2, r2, ha16(L_BA4__foo_L5) -; STATIC: stw r2 +; PIC: addis r5, r4, ha16(L_BA4__foo_L5-"L1$pb") +; PIC: li r6, lo16(L_BA4__foo_L5-"L1$pb") +; PIC: add r5, r5, r6 +; PIC: stw r5 +; STATIC: li r4, lo16(L_BA4__foo_L5) +; STATIC: addis r4, r4, ha16(L_BA4__foo_L5) +; STATIC: stw r4 store i8* blockaddress(@foo, %L5), i8** @nextaddr, align 4 ret i32 %res.3 } diff --git a/test/CodeGen/PowerPC/lsr-postinc-pos.ll b/test/CodeGen/PowerPC/lsr-postinc-pos.ll new file mode 100644 index 0000000..f441e42 --- /dev/null +++ b/test/CodeGen/PowerPC/lsr-postinc-pos.ll @@ -0,0 +1,32 @@ +; RUN: llc < %s -print-lsr-output |& FileCheck %s + +; The icmp is a post-inc use, and the increment is in %bb11, but the +; scevgep needs to be inserted in %bb so that it is dominated by %t. + +; CHECK: %t = load i8** undef +; CHECK: %scevgep = getelementptr i8* %t, i32 %lsr.iv.next +; CHECK: %c1 = icmp ult i8* %scevgep, undef + +target datalayout = "E-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f128:64:128-n32" +target triple = "powerpc-apple-darwin9" + +define void @foo() nounwind { +entry: + br label %bb11 + +bb11: + %i = phi i32 [ 0, %entry ], [ %i.next, %bb ] ; <i32> [#uses=3] + %ii = shl i32 %i, 2 ; <i32> [#uses=1] + %c0 = icmp eq i32 %i, undef ; <i1> [#uses=1] + br i1 %c0, label %bb13, label %bb + +bb: + %t = load i8** undef, align 16 ; <i8*> [#uses=1] + %p = getelementptr i8* %t, i32 %ii ; <i8*> [#uses=1] + %c1 = icmp ult i8* %p, undef ; <i1> [#uses=1] + %i.next = add i32 %i, 1 ; <i32> [#uses=1] + br i1 %c1, label %bb11, label %bb13 + +bb13: + unreachable +} diff --git a/test/CodeGen/PowerPC/mem_update.ll b/test/CodeGen/PowerPC/mem_update.ll index b267719..17e7e28 100644 --- a/test/CodeGen/PowerPC/mem_update.ll +++ b/test/CodeGen/PowerPC/mem_update.ll @@ -3,66 +3,66 @@ ; RUN: llc < %s -march=ppc64 -enable-ppc-preinc | \ ; RUN: not grep addi -@Glob = global i64 4 ; <i64*> [#uses=2] +@Glob = global i64 4 -define i32* @test0(i32* %X, i32* %dest) { - %Y = getelementptr i32* %X, i32 4 ; <i32*> [#uses=2] - %A = load i32* %Y ; <i32> [#uses=1] +define i32* @test0(i32* %X, i32* %dest) nounwind { + %Y = getelementptr i32* %X, i32 4 + %A = load i32* %Y store i32 %A, i32* %dest ret i32* %Y } -define i32* @test1(i32* %X, i32* %dest) { - %Y = getelementptr i32* %X, i32 4 ; <i32*> [#uses=2] - %A = load i32* %Y ; <i32> [#uses=1] +define i32* @test1(i32* %X, i32* %dest) nounwind { + %Y = getelementptr i32* %X, i32 4 + %A = load i32* %Y store i32 %A, i32* %dest ret i32* %Y } -define i16* @test2(i16* %X, i32* %dest) { - %Y = getelementptr i16* %X, i32 4 ; <i16*> [#uses=2] - %A = load i16* %Y ; <i16> [#uses=1] - %B = sext i16 %A to i32 ; <i32> [#uses=1] +define i16* @test2(i16* %X, i32* %dest) nounwind { + %Y = getelementptr i16* %X, i32 4 + %A = load i16* %Y + %B = sext i16 %A to i32 store i32 %B, i32* %dest ret i16* %Y } -define i16* @test3(i16* %X, i32* %dest) { - %Y = getelementptr i16* %X, i32 4 ; <i16*> [#uses=2] - %A = load i16* %Y ; <i16> [#uses=1] - %B = zext i16 %A to i32 ; <i32> [#uses=1] +define i16* @test3(i16* %X, i32* %dest) nounwind { + %Y = getelementptr i16* %X, i32 4 + %A = load i16* %Y + %B = zext i16 %A to i32 store i32 %B, i32* %dest ret i16* %Y } -define i16* @test3a(i16* %X, i64* %dest) { - %Y = getelementptr i16* %X, i32 4 ; <i16*> [#uses=2] - %A = load i16* %Y ; <i16> [#uses=1] - %B = sext i16 %A to i64 ; <i64> [#uses=1] +define i16* @test3a(i16* %X, i64* %dest) nounwind { + %Y = getelementptr i16* %X, i32 4 + %A = load i16* %Y + %B = sext i16 %A to i64 store i64 %B, i64* %dest ret i16* %Y } -define i64* @test4(i64* %X, i64* %dest) { - %Y = getelementptr i64* %X, i32 4 ; <i64*> [#uses=2] - %A = load i64* %Y ; <i64> [#uses=1] +define i64* @test4(i64* %X, i64* %dest) nounwind { + %Y = getelementptr i64* %X, i32 4 + %A = load i64* %Y store i64 %A, i64* %dest ret i64* %Y } -define i16* @test5(i16* %X) { - %Y = getelementptr i16* %X, i32 4 ; <i16*> [#uses=2] +define i16* @test5(i16* %X) nounwind { + %Y = getelementptr i16* %X, i32 4 store i16 7, i16* %Y ret i16* %Y } -define i64* @test6(i64* %X, i64 %A) { - %Y = getelementptr i64* %X, i32 4 ; <i64*> [#uses=2] +define i64* @test6(i64* %X, i64 %A) nounwind { + %Y = getelementptr i64* %X, i32 4 store i64 %A, i64* %Y ret i64* %Y } -define i64* @test7(i64* %X, i64 %A) { +define i64* @test7(i64* %X, i64 %A) nounwind { store i64 %A, i64* @Glob ret i64* @Glob } diff --git a/test/CodeGen/PowerPC/retaddr.ll b/test/CodeGen/PowerPC/retaddr.ll index 9f8647d..cf16b4c 100644 --- a/test/CodeGen/PowerPC/retaddr.ll +++ b/test/CodeGen/PowerPC/retaddr.ll @@ -4,7 +4,7 @@ target triple = "powerpc-apple-darwin8" -define void @foo(i8** %X) { +define void @foo(i8** %X) nounwind { entry: %tmp = tail call i8* @llvm.returnaddress( i32 0 ) ; <i8*> [#uses=1] store i8* %tmp, i8** %X, align 4 diff --git a/test/CodeGen/Thumb2/2010-02-24-BigStack.ll b/test/CodeGen/Thumb2/2010-02-24-BigStack.ll new file mode 100644 index 0000000..533546b --- /dev/null +++ b/test/CodeGen/Thumb2/2010-02-24-BigStack.ll @@ -0,0 +1,15 @@ +; RUN: llc < %s -O0 -relocation-model=pic -disable-fp-elim -mcpu=cortex-a8 -mattr=+vfp2 +; This test creates a big stack frame without spilling any callee-saved registers. +; Make sure the whole stack frame is addrerssable wiothout scavenger crashes. +target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:64:64-v128:128:128-a0:0:32-n32" +target triple = "thumbv7-apple-darwin3.0.0-iphoneos" + +define arm_apcscc void @FindMin(double* %panelTDEL, i8* %dclOfRow, i32 %numRows, i32 %numCols, double* %retMin_RES_TDEL) { +entry: + %panelTDEL.addr = alloca double*, align 4 ; <double**> [#uses=1] + %panelResTDEL = alloca [2560 x double], align 4 ; <[2560 x double]*> [#uses=0] + store double* %panelTDEL, double** %panelTDEL.addr + store double* %retMin_RES_TDEL, double** undef + store i32 0, i32* undef + unreachable +} diff --git a/test/CodeGen/Thumb2/ldr-str-imm12.ll b/test/CodeGen/Thumb2/ldr-str-imm12.ll index 47d85b1..f007b5c 100644 --- a/test/CodeGen/Thumb2/ldr-str-imm12.ll +++ b/test/CodeGen/Thumb2/ldr-str-imm12.ll @@ -52,7 +52,7 @@ bb420: ; preds = %bb20, %bb20 ; CHECK: str r{{[0-7]}}, [sp] ; CHECK: str r{{[0-7]}}, [sp, #+4] ; CHECK: str r{{[0-7]}}, [sp, #+8] -; CHECK: str r{{[0-7]}}, [sp, #+24] +; CHECK: str{{(.w)?}} r{{[0-9]+}}, [sp, #+24] store %union.rec* null, %union.rec** @zz_hold, align 4 store %union.rec* null, %union.rec** @zz_res, align 4 store %union.rec* %x, %union.rec** @zz_hold, align 4 diff --git a/test/CodeGen/X86/2005-01-17-CycleInDAG.ll b/test/CodeGen/X86/2005-01-17-CycleInDAG.ll index 32fafc6..fe6674d 100644 --- a/test/CodeGen/X86/2005-01-17-CycleInDAG.ll +++ b/test/CodeGen/X86/2005-01-17-CycleInDAG.ll @@ -7,7 +7,7 @@ @GLOBAL = external global i32 ; <i32*> [#uses=1] -define i32 @test(i32* %P1, i32* %P2, i32* %P3) { +define i32 @test(i32* %P1, i32* %P2, i32* %P3) nounwind { %L = load i32* @GLOBAL ; <i32> [#uses=1] store i32 12, i32* %P2 %Y = load i32* %P3 ; <i32> [#uses=1] diff --git a/test/CodeGen/X86/2006-10-07-ScalarSSEMiscompile.ll b/test/CodeGen/X86/2006-10-07-ScalarSSEMiscompile.ll index bf9fa57..d09d061 100644 --- a/test/CodeGen/X86/2006-10-07-ScalarSSEMiscompile.ll +++ b/test/CodeGen/X86/2006-10-07-ScalarSSEMiscompile.ll @@ -5,7 +5,7 @@ target datalayout = "e-p:32:32" target triple = "i686-apple-darwin8.7.2" -define <4 x float> @test(<4 x float> %A, <4 x float>* %B) { +define <4 x float> @test(<4 x float> %A, <4 x float>* %B) nounwind { %BV = load <4 x float>* %B ; <<4 x float>> [#uses=1] %tmp28 = tail call <4 x float> @llvm.x86.sse.sub.ss( <4 x float> %A, <4 x float> %BV ) ; <<4 x float>> [#uses=1] ret <4 x float> %tmp28 diff --git a/test/CodeGen/X86/2008-02-22-ReMatBug.ll b/test/CodeGen/X86/2008-02-22-ReMatBug.ll index 8d6bb0d..a91ac27 100644 --- a/test/CodeGen/X86/2008-02-22-ReMatBug.ll +++ b/test/CodeGen/X86/2008-02-22-ReMatBug.ll @@ -1,4 +1,4 @@ -; RUN: llc < %s -march=x86 -stats |& grep {Number of re-materialization} | grep 3 +; RUN: llc < %s -march=x86 -stats |& grep {Number of re-materialization} | grep 2 ; rdar://5761454 %struct.quad_struct = type { i32, i32, %struct.quad_struct*, %struct.quad_struct*, %struct.quad_struct*, %struct.quad_struct*, %struct.quad_struct* } diff --git a/test/CodeGen/X86/2009-02-12-DebugInfoVLA.ll b/test/CodeGen/X86/2009-02-12-DebugInfoVLA.ll index 72c7ee9..0dca14d 100644 --- a/test/CodeGen/X86/2009-02-12-DebugInfoVLA.ll +++ b/test/CodeGen/X86/2009-02-12-DebugInfoVLA.ll @@ -3,74 +3,83 @@ ; PR3538 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" target triple = "i386-apple-darwin9" - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.block.type = type { i32, { }* } - %llvm.dbg.compile_unit.type = type { i32, { }*, i32, i8*, i8*, i8*, i1, i1, i8* } - %llvm.dbg.composite.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, { }*, { }* } - %llvm.dbg.subprogram.type = type { i32, { }*, { }*, i8*, i8*, i8*, { }*, i32, { }*, i1, i1 } - %llvm.dbg.subrange.type = type { i32, i64, i64 } - %llvm.dbg.variable.type = type { i32, { }*, i8*, { }*, i32, { }* } -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str = internal constant [4 x i8] c"t.c\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@.str1 = internal constant [2 x i8] c".\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@.str2 = internal constant [6 x i8] c"clang\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { i32 458769, { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to { }*), i32 1, i8* getelementptr ([4 x i8]* @.str, i32 0, i32 0), i8* getelementptr ([2 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str3 = internal constant [4 x i8] c"int\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type { i32 458788, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* getelementptr ([4 x i8]* @.str3, i32 0, i32 0), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 5 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str4 = internal constant [5 x i8] c"test\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.subprogram = internal constant %llvm.dbg.subprogram.type { i32 458798, { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to { }*), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* getelementptr ([5 x i8]* @.str4, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str4, i32 0, i32 0), i8* null, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 3, { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str5 = internal constant [2 x i8] c"X\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.variable = internal constant %llvm.dbg.variable.type { i32 459009, { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*), i8* getelementptr ([2 x i8]* @.str5, i32 0, i32 0), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 3, { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.block = internal constant %llvm.dbg.block.type { i32 458763, { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*) }, section "llvm.metadata" ; <%llvm.dbg.block.type*> [#uses=1] -@llvm.dbg.subrange = internal constant %llvm.dbg.subrange.type { i32 458785, i64 0, i64 0 }, section "llvm.metadata" ; <%llvm.dbg.subrange.type*> [#uses=1] -@llvm.dbg.array = internal constant [1 x { }*] [{ }* bitcast (%llvm.dbg.subrange.type* @llvm.dbg.subrange to { }*)], section "llvm.metadata" ; <[1 x { }*]*> [#uses=1] -@llvm.dbg.composite = internal constant %llvm.dbg.composite.type { i32 458753, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* null, { }* null, i32 0, i64 0, i64 32, i64 0, i32 0, { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*), { }* bitcast ([1 x { }*]* @llvm.dbg.array to { }*) }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str6 = internal constant [2 x i8] c"Y\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.variable7 = internal constant %llvm.dbg.variable.type { i32 459008, { }* bitcast (%llvm.dbg.block.type* @llvm.dbg.block to { }*), i8* getelementptr ([2 x i8]* @.str6, i32 0, i32 0), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 4, { }* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite to { }*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] - -define i32 @test(i32 %X) nounwind { +define signext i8 @foo(i8* %s1) nounwind ssp { entry: - %retval = alloca i32 ; <i32*> [#uses=1] - %X.addr = alloca i32 ; <i32*> [#uses=3] - %saved_stack = alloca i8* ; <i8**> [#uses=2] - call void @llvm.dbg.func.start({ }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*)) - store i32 %X, i32* %X.addr - %0 = bitcast i32* %X.addr to { }* ; <{ }*> [#uses=1] - call void @llvm.dbg.declare({ }* %0, { }* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable to { }*)) - call void @llvm.dbg.region.start({ }* bitcast (%llvm.dbg.block.type* @llvm.dbg.block to { }*)) - call void @llvm.dbg.stoppoint(i32 4, i32 3, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - %1 = call i8* @llvm.stacksave() ; <i8*> [#uses=1] - store i8* %1, i8** %saved_stack - %tmp = load i32* %X.addr ; <i32> [#uses=1] - %2 = mul i32 4, %tmp ; <i32> [#uses=1] - %vla = alloca i8, i32 %2 ; <i8*> [#uses=1] - %tmp1 = bitcast i8* %vla to i32* ; <i32*> [#uses=1] - %3 = bitcast i32* %tmp1 to { }* ; <{ }*> [#uses=1] - call void @llvm.dbg.declare({ }* %3, { }* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable7 to { }*)) - call void @llvm.dbg.stoppoint(i32 5, i32 1, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - call void @llvm.dbg.region.end({ }* bitcast (%llvm.dbg.block.type* @llvm.dbg.block to { }*)) - br label %cleanup + %s1_addr = alloca i8* ; <i8**> [#uses=2] + %retval = alloca i32 ; <i32*> [#uses=2] + %saved_stack.1 = alloca i8* ; <i8**> [#uses=2] + %0 = alloca i32 ; <i32*> [#uses=2] + %str.0 = alloca [0 x i8]* ; <[0 x i8]**> [#uses=3] + %1 = alloca i64 ; <i64*> [#uses=2] + %2 = alloca i64 ; <i64*> [#uses=1] + %3 = alloca i64 ; <i64*> [#uses=6] + %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] + call void @llvm.dbg.declare(metadata !{i8** %s1_addr}, metadata !0), !dbg !7 + store i8* %s1, i8** %s1_addr + call void @llvm.dbg.declare(metadata !{[0 x i8]** %str.0}, metadata !8), !dbg !7 + %4 = call i8* @llvm.stacksave(), !dbg !7 ; <i8*> [#uses=1] + store i8* %4, i8** %saved_stack.1, align 8, !dbg !7 + %5 = load i8** %s1_addr, align 8, !dbg !13 ; <i8*> [#uses=1] + %6 = call i64 @strlen(i8* %5) nounwind readonly, !dbg !13 ; <i64> [#uses=1] + %7 = add i64 %6, 1, !dbg !13 ; <i64> [#uses=1] + store i64 %7, i64* %3, align 8, !dbg !13 + %8 = load i64* %3, align 8, !dbg !13 ; <i64> [#uses=1] + %9 = sub nsw i64 %8, 1, !dbg !13 ; <i64> [#uses=0] + %10 = load i64* %3, align 8, !dbg !13 ; <i64> [#uses=1] + %11 = mul i64 %10, 8, !dbg !13 ; <i64> [#uses=0] + %12 = load i64* %3, align 8, !dbg !13 ; <i64> [#uses=1] + store i64 %12, i64* %2, align 8, !dbg !13 + %13 = load i64* %3, align 8, !dbg !13 ; <i64> [#uses=1] + %14 = mul i64 %13, 8, !dbg !13 ; <i64> [#uses=0] + %15 = load i64* %3, align 8, !dbg !13 ; <i64> [#uses=1] + store i64 %15, i64* %1, align 8, !dbg !13 + %16 = load i64* %1, align 8, !dbg !13 ; <i64> [#uses=1] + %17 = trunc i64 %16 to i32, !dbg !13 ; <i32> [#uses=1] + %18 = alloca i8, i32 %17, !dbg !13 ; <i8*> [#uses=1] + %19 = bitcast i8* %18 to [0 x i8]*, !dbg !13 ; <[0 x i8]*> [#uses=1] + store [0 x i8]* %19, [0 x i8]** %str.0, align 8, !dbg !13 + %20 = load [0 x i8]** %str.0, align 8, !dbg !15 ; <[0 x i8]*> [#uses=1] + %21 = getelementptr inbounds [0 x i8]* %20, i64 0, i64 0, !dbg !15 ; <i8*> [#uses=1] + store i8 0, i8* %21, align 1, !dbg !15 + %22 = load [0 x i8]** %str.0, align 8, !dbg !16 ; <[0 x i8]*> [#uses=1] + %23 = getelementptr inbounds [0 x i8]* %22, i64 0, i64 0, !dbg !16 ; <i8*> [#uses=1] + %24 = load i8* %23, align 1, !dbg !16 ; <i8> [#uses=1] + %25 = sext i8 %24 to i32, !dbg !16 ; <i32> [#uses=1] + store i32 %25, i32* %0, align 4, !dbg !16 + %26 = load i8** %saved_stack.1, align 8, !dbg !16 ; <i8*> [#uses=1] + call void @llvm.stackrestore(i8* %26), !dbg !16 + %27 = load i32* %0, align 4, !dbg !16 ; <i32> [#uses=1] + store i32 %27, i32* %retval, align 4, !dbg !16 + br label %return, !dbg !16 -cleanup: ; preds = %entry - %tmp2 = load i8** %saved_stack ; <i8*> [#uses=1] - call void @llvm.stackrestore(i8* %tmp2) - call void @llvm.dbg.region.end({ }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*)) - %4 = load i32* %retval ; <i32> [#uses=1] - ret i32 %4 +return: ; preds = %entry + %retval1 = load i32* %retval, !dbg !16 ; <i32> [#uses=1] + %retval12 = trunc i32 %retval1 to i8, !dbg !16 ; <i8> [#uses=1] + ret i8 %retval12, !dbg !16 } -declare void @llvm.dbg.func.start({ }*) nounwind - -declare void @llvm.dbg.declare({ }*, { }*) nounwind - -declare void @llvm.dbg.region.start({ }*) nounwind - -declare void @llvm.dbg.stoppoint(i32, i32, { }*) nounwind +declare void @llvm.dbg.declare(metadata, metadata) nounwind readnone declare i8* @llvm.stacksave() nounwind +declare i64 @strlen(i8*) nounwind readonly + declare void @llvm.stackrestore(i8*) nounwind -declare void @llvm.dbg.region.end({ }*) nounwind +!0 = metadata !{i32 459009, metadata !1, metadata !"s1", metadata !2, i32 2, metadata !6} ; [ DW_TAG_arg_variable ] +!1 = metadata !{i32 458798, i32 0, metadata !2, metadata !"foo", metadata !"foo", metadata !"foo", metadata !2, i32 2, metadata !3, i1 false, i1 true, i32 0, i32 0, null, i1 false} ; [ DW_TAG_subprogram ] +!2 = metadata !{i32 458769, i32 0, i32 1, metadata !"vla.c", metadata !"/tmp/", metadata !"4.2.1 (Based on Apple Inc. build 5658) (LLVM build)", i1 true, i1 false, metadata !"", i32 0} ; [ DW_TAG_compile_unit ] +!3 = metadata !{i32 458773, metadata !2, metadata !"", metadata !2, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !4, i32 0, null} ; [ DW_TAG_subroutine_type ] +!4 = metadata !{metadata !5, metadata !6} +!5 = metadata !{i32 458788, metadata !2, metadata !"char", metadata !2, i32 0, i64 8, i64 8, i64 0, i32 0, i32 6} ; [ DW_TAG_base_type ] +!6 = metadata !{i32 458767, metadata !2, metadata !"", metadata !2, i32 0, i64 64, i64 64, i64 0, i32 0, metadata !5} ; [ DW_TAG_pointer_type ] +!7 = metadata !{i32 2, i32 0, metadata !1, null} +!8 = metadata !{i32 459008, metadata !1, metadata !"str.0", metadata !2, i32 3, metadata !9} ; [ DW_TAG_auto_variable ] +!9 = metadata !{i32 458767, metadata !2, metadata !"", metadata !2, i32 0, i64 64, i64 64, i64 0, i32 64, metadata !10} ; [ DW_TAG_pointer_type ] +!10 = metadata !{i32 458753, metadata !2, metadata !"", metadata !2, i32 0, i64 8, i64 8, i64 0, i32 0, metadata !5, metadata !11, i32 0, null} ; [ DW_TAG_array_type ] +!11 = metadata !{metadata !12} +!12 = metadata !{i32 458785, i64 0, i64 0} ; [ DW_TAG_subrange_type ] +!13 = metadata !{i32 3, i32 0, metadata !14, null} +!14 = metadata !{i32 458763, metadata !1, i32 0, i32 0} ; [ DW_TAG_lexical_block ] +!15 = metadata !{i32 4, i32 0, metadata !14, null} +!16 = metadata !{i32 5, i32 0, metadata !14, null} diff --git a/test/CodeGen/X86/2009-07-16-LoadFoldingBug.ll b/test/CodeGen/X86/2009-07-16-LoadFoldingBug.ll deleted file mode 100644 index e21c892..0000000 --- a/test/CodeGen/X86/2009-07-16-LoadFoldingBug.ll +++ /dev/null @@ -1,102 +0,0 @@ -; RUN: llc < %s -mtriple=x86_64-apple-darwin10 | FileCheck %s - -; CHECK: _foo: -; CHECK: pavgw LCPI1_4(%rip) - -; rdar://7057804 - -define void @foo(i16* %out8x8, i16* %in8x8, i32 %lastrow) optsize ssp { -entry: - %0 = call <8 x i16> @llvm.x86.sse2.pmulh.w(<8 x i16> <i16 6518, i16 6518, i16 6518, i16 6518, i16 6518, i16 6518, i16 6518, i16 6518>, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=2] - %1 = call <8 x i16> @llvm.x86.sse2.pcmpeq.w(<8 x i16> %0, <8 x i16> <i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384>) nounwind readnone ; <<8 x i16>> [#uses=1] - %2 = call <8 x i16> @llvm.x86.sse2.psrli.w(<8 x i16> zeroinitializer, i32 14) nounwind readnone ; <<8 x i16>> [#uses=1] - %3 = call <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16> %2, <8 x i16> zeroinitializer) nounwind readnone ; <<8 x i16>> [#uses=1] - %tmp.i.i10 = add <8 x i16> %0, %3 ; <<8 x i16>> [#uses=1] - %4 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> zeroinitializer, <8 x i16> %1) nounwind readnone ; <<8 x i16>> [#uses=1] - %5 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %tmp.i.i10, <8 x i16> %4) nounwind readnone ; <<8 x i16>> [#uses=3] - %6 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %5, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=1] - %7 = call <8 x i16> @llvm.x86.sse2.pmulh.w(<8 x i16> <i16 6518, i16 6518, i16 6518, i16 6518, i16 6518, i16 6518, i16 6518, i16 6518>, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=2] - %8 = call <8 x i16> @llvm.x86.sse2.pcmpeq.w(<8 x i16> %7, <8 x i16> <i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384>) nounwind readnone ; <<8 x i16>> [#uses=1] - %9 = call <8 x i16> @llvm.x86.sse2.psrli.w(<8 x i16> zeroinitializer, i32 14) nounwind readnone ; <<8 x i16>> [#uses=1] - %10 = call <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16> %9, <8 x i16> zeroinitializer) nounwind readnone ; <<8 x i16>> [#uses=1] - %tmp.i.i8 = add <8 x i16> %7, %10 ; <<8 x i16>> [#uses=1] - %11 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> undef, <8 x i16> %8) nounwind readnone ; <<8 x i16>> [#uses=1] - %12 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %tmp.i.i8, <8 x i16> %11) nounwind readnone ; <<8 x i16>> [#uses=1] - %13 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> undef, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=1] - %14 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %5, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=1] - %15 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %5, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=1] - %16 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %6, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=1] - %17 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %12, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=1] - %18 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %13, <8 x i16> %15) nounwind readnone ; <<8 x i16>> [#uses=1] - %19 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> undef, <8 x i16> %14) nounwind readnone ; <<8 x i16>> [#uses=2] - %20 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> undef, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=4] - %21 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> undef, <8 x i16> %17) nounwind readnone ; <<8 x i16>> [#uses=1] - %22 = bitcast <8 x i16> %21 to <2 x i64> ; <<2 x i64>> [#uses=1] - %23 = call <8 x i16> @llvm.x86.sse2.pmulh.w(<8 x i16> <i16 23170, i16 23170, i16 23170, i16 23170, i16 23170, i16 23170, i16 23170, i16 23170>, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=2] - %24 = call <8 x i16> @llvm.x86.sse2.pcmpeq.w(<8 x i16> %23, <8 x i16> <i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384>) nounwind readnone ; <<8 x i16>> [#uses=1] - %25 = call <8 x i16> @llvm.x86.sse2.psrli.w(<8 x i16> zeroinitializer, i32 14) nounwind readnone ; <<8 x i16>> [#uses=1] - %26 = call <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16> %25, <8 x i16> zeroinitializer) nounwind readnone ; <<8 x i16>> [#uses=1] - %tmp.i.i6 = add <8 x i16> %23, %26 ; <<8 x i16>> [#uses=1] - %27 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> undef, <8 x i16> %24) nounwind readnone ; <<8 x i16>> [#uses=1] - %28 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %tmp.i.i6, <8 x i16> %27) nounwind readnone ; <<8 x i16>> [#uses=1] - %29 = call <8 x i16> @llvm.x86.sse2.pmulh.w(<8 x i16> <i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170>, <8 x i16> undef) nounwind readnone ; <<8 x i16>> [#uses=2] - %30 = call <8 x i16> @llvm.x86.sse2.pcmpeq.w(<8 x i16> %29, <8 x i16> <i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384>) nounwind readnone ; <<8 x i16>> [#uses=1] - %31 = call <8 x i16> @llvm.x86.sse2.psrli.w(<8 x i16> zeroinitializer, i32 14) nounwind readnone ; <<8 x i16>> [#uses=1] - %32 = call <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16> %31, <8 x i16> zeroinitializer) nounwind readnone ; <<8 x i16>> [#uses=1] - %tmp.i.i4 = add <8 x i16> %29, %32 ; <<8 x i16>> [#uses=1] - %33 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> undef, <8 x i16> %30) nounwind readnone ; <<8 x i16>> [#uses=1] - %34 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %tmp.i.i4, <8 x i16> %33) nounwind readnone ; <<8 x i16>> [#uses=1] - %35 = call <8 x i16> @llvm.x86.sse2.pmulh.w(<8 x i16> <i16 23170, i16 23170, i16 23170, i16 23170, i16 23170, i16 23170, i16 23170, i16 23170>, <8 x i16> %20) nounwind readnone ; <<8 x i16>> [#uses=2] - %tmp.i2.i1 = mul <8 x i16> %20, <i16 23170, i16 23170, i16 23170, i16 23170, i16 23170, i16 23170, i16 23170, i16 23170> ; <<8 x i16>> [#uses=1] - %36 = call <8 x i16> @llvm.x86.sse2.pcmpeq.w(<8 x i16> %35, <8 x i16> <i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384>) nounwind readnone ; <<8 x i16>> [#uses=1] - %37 = call <8 x i16> @llvm.x86.sse2.psrli.w(<8 x i16> %tmp.i2.i1, i32 14) nounwind readnone ; <<8 x i16>> [#uses=1] - %38 = call <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16> %37, <8 x i16> zeroinitializer) nounwind readnone ; <<8 x i16>> [#uses=1] - %tmp.i.i2 = add <8 x i16> %35, %38 ; <<8 x i16>> [#uses=1] - %39 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %19, <8 x i16> %36) nounwind readnone ; <<8 x i16>> [#uses=1] - %40 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %tmp.i.i2, <8 x i16> %39) nounwind readnone ; <<8 x i16>> [#uses=1] - %41 = call <8 x i16> @llvm.x86.sse2.pmulh.w(<8 x i16> <i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170>, <8 x i16> %20) nounwind readnone ; <<8 x i16>> [#uses=2] - %tmp.i2.i = mul <8 x i16> %20, <i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170, i16 -23170> ; <<8 x i16>> [#uses=1] - %42 = call <8 x i16> @llvm.x86.sse2.pcmpeq.w(<8 x i16> %41, <8 x i16> <i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384, i16 16384>) nounwind readnone ; <<8 x i16>> [#uses=1] - %43 = call <8 x i16> @llvm.x86.sse2.psrli.w(<8 x i16> %tmp.i2.i, i32 14) nounwind readnone ; <<8 x i16>> [#uses=1] - %44 = call <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16> %43, <8 x i16> zeroinitializer) nounwind readnone ; <<8 x i16>> [#uses=1] - %tmp.i.i = add <8 x i16> %41, %44 ; <<8 x i16>> [#uses=1] - %45 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %19, <8 x i16> %42) nounwind readnone ; <<8 x i16>> [#uses=1] - %46 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %tmp.i.i, <8 x i16> %45) nounwind readnone ; <<8 x i16>> [#uses=1] - %47 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %18, <8 x i16> %16) nounwind readnone ; <<8 x i16>> [#uses=1] - %48 = bitcast <8 x i16> %47 to <2 x i64> ; <<2 x i64>> [#uses=1] - %49 = bitcast <8 x i16> %28 to <2 x i64> ; <<2 x i64>> [#uses=1] - %50 = getelementptr i16* %out8x8, i64 8 ; <i16*> [#uses=1] - %51 = bitcast i16* %50 to <2 x i64>* ; <<2 x i64>*> [#uses=1] - store <2 x i64> %49, <2 x i64>* %51, align 16 - %52 = bitcast <8 x i16> %40 to <2 x i64> ; <<2 x i64>> [#uses=1] - %53 = getelementptr i16* %out8x8, i64 16 ; <i16*> [#uses=1] - %54 = bitcast i16* %53 to <2 x i64>* ; <<2 x i64>*> [#uses=1] - store <2 x i64> %52, <2 x i64>* %54, align 16 - %55 = getelementptr i16* %out8x8, i64 24 ; <i16*> [#uses=1] - %56 = bitcast i16* %55 to <2 x i64>* ; <<2 x i64>*> [#uses=1] - store <2 x i64> %48, <2 x i64>* %56, align 16 - %57 = bitcast <8 x i16> %46 to <2 x i64> ; <<2 x i64>> [#uses=1] - %58 = getelementptr i16* %out8x8, i64 40 ; <i16*> [#uses=1] - %59 = bitcast i16* %58 to <2 x i64>* ; <<2 x i64>*> [#uses=1] - store <2 x i64> %57, <2 x i64>* %59, align 16 - %60 = bitcast <8 x i16> %34 to <2 x i64> ; <<2 x i64>> [#uses=1] - %61 = getelementptr i16* %out8x8, i64 48 ; <i16*> [#uses=1] - %62 = bitcast i16* %61 to <2 x i64>* ; <<2 x i64>*> [#uses=1] - store <2 x i64> %60, <2 x i64>* %62, align 16 - %63 = getelementptr i16* %out8x8, i64 56 ; <i16*> [#uses=1] - %64 = bitcast i16* %63 to <2 x i64>* ; <<2 x i64>*> [#uses=1] - store <2 x i64> %22, <2 x i64>* %64, align 16 - ret void -} - -declare <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16>, <8 x i16>) nounwind readnone - -declare <8 x i16> @llvm.x86.sse2.pmulh.w(<8 x i16>, <8 x i16>) nounwind readnone - -declare <8 x i16> @llvm.x86.sse2.pcmpeq.w(<8 x i16>, <8 x i16>) nounwind readnone - -declare <8 x i16> @llvm.x86.sse2.pavg.w(<8 x i16>, <8 x i16>) nounwind readnone - -declare <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16>, <8 x i16>) nounwind readnone - -declare <8 x i16> @llvm.x86.sse2.psrli.w(<8 x i16>, i32) nounwind readnone diff --git a/test/CodeGen/X86/2009-09-07-CoalescerBug.ll b/test/CodeGen/X86/2009-09-07-CoalescerBug.ll index a5b4a79..41b4bc0 100644 --- a/test/CodeGen/X86/2009-09-07-CoalescerBug.ll +++ b/test/CodeGen/X86/2009-09-07-CoalescerBug.ll @@ -8,8 +8,7 @@ define i64 @hammer_time(i64 %modulep, i64 %physfree) nounwind ssp noredzone noimplicitfloat { ; CHECK: hammer_time: ; CHECK: movq $Xrsvd, %rax -; CHECK: movq $Xrsvd, %rsi -; CHECK: movq $Xrsvd, %rdi +; CHECK: movq $Xrsvd, %rcx entry: br i1 undef, label %if.then, label %if.end diff --git a/test/CodeGen/X86/2010-02-11-NonTemporal.ll b/test/CodeGen/X86/2010-02-11-NonTemporal.ll new file mode 100644 index 0000000..5789a0b --- /dev/null +++ b/test/CodeGen/X86/2010-02-11-NonTemporal.ll @@ -0,0 +1,22 @@ +; RUN: llc < %s -march=x86-64 | FileCheck %s +; CHECK: movnt +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" +target triple = "x86_64-unknown-linux-gnu" + +!0 = metadata !{ i32 1 } + +define void @sub_(i32* noalias %n) { +"file movnt.f90, line 2, bb1": + %n1 = alloca i32*, align 8 + %i = alloca i32, align 4 + %"$LCS_0" = alloca i64, align 8 + %"$LCS_S2" = alloca <2 x double>, align 16 + %r9 = load <2 x double>* %"$LCS_S2", align 8 + %r10 = load i64* %"$LCS_0", align 8 + %r11 = inttoptr i64 %r10 to <2 x double>* + store <2 x double> %r9, <2 x double>* %r11, align 16, !nontemporal !0 + br label %"file movnt.f90, line 18, bb5" + +"file movnt.f90, line 18, bb5": + ret void +} diff --git a/test/CodeGen/X86/2010-02-19-TailCallRetAddrBug.ll b/test/CodeGen/X86/2010-02-19-TailCallRetAddrBug.ll new file mode 100644 index 0000000..eb21dc2 --- /dev/null +++ b/test/CodeGen/X86/2010-02-19-TailCallRetAddrBug.ll @@ -0,0 +1,55 @@ +; RUN: llc -mtriple=i386-apple-darwin -tailcallopt < %s | FileCheck %s +; Check that lowered argumens do not overwrite the return address before it is moved. +; Bug 6225 +; +; If a call is a fastcc tail call and tail call optimization is enabled, the +; caller frame is replaced by the callee frame. This can require that arguments are +; placed on the former return address stack slot. Special care needs to be taken +; taken that the return address is moved / or stored in a register before +; lowering of arguments potentially overwrites the value. +; +; Move return address (76(%esp)) to a temporary register (%ebp) +; CHECK: movl 76(%esp), %ebp +; Overwrite return addresss +; CHECK: movl %ecx, 76(%esp) +; Move return address from temporary register (%ebp) to new stack location (60(%esp)) +; CHECK: movl %ebp, 60(%esp) + +%tupl_p = type [9 x i32]* + +declare fastcc void @l297(i32 %r10, i32 %r9, i32 %r8, i32 %r7, i32 %r6, i32 %r5, i32 %r3, i32 %r2) noreturn nounwind +declare fastcc void @l298(i32 %r10, i32 %r9, i32 %r4) noreturn nounwind + +define fastcc void @l186(%tupl_p %r1) noreturn nounwind { +entry: + %ptr1 = getelementptr %tupl_p %r1, i32 0, i32 0 + %r2 = load i32* %ptr1 + %ptr3 = getelementptr %tupl_p %r1, i32 0, i32 1 + %r3 = load i32* %ptr3 + %ptr5 = getelementptr %tupl_p %r1, i32 0, i32 2 + %r4 = load i32* %ptr5 + %ptr7 = getelementptr %tupl_p %r1, i32 0, i32 3 + %r5 = load i32* %ptr7 + %ptr9 = getelementptr %tupl_p %r1, i32 0, i32 4 + %r6 = load i32* %ptr9 + %ptr11 = getelementptr %tupl_p %r1, i32 0, i32 5 + %r7 = load i32* %ptr11 + %ptr13 = getelementptr %tupl_p %r1, i32 0, i32 6 + %r8 = load i32* %ptr13 + %ptr15 = getelementptr %tupl_p %r1, i32 0, i32 7 + %r9 = load i32* %ptr15 + %ptr17 = getelementptr %tupl_p %r1, i32 0, i32 8 + %r10 = load i32* %ptr17 + %cond = icmp eq i32 %r10, 3 + br i1 %cond, label %true, label %false + +true: + tail call fastcc void @l297(i32 %r10, i32 %r9, i32 %r8, i32 %r7, i32 %r6, i32 %r5, i32 %r3, i32 %r2) noreturn nounwind + ret void + +false: + tail call fastcc void @l298(i32 %r10, i32 %r9, i32 %r4) noreturn nounwind + ret void +} + + diff --git a/test/CodeGen/X86/2010-02-23-DAGCombineBug.ll b/test/CodeGen/X86/2010-02-23-DAGCombineBug.ll new file mode 100644 index 0000000..6a58e9e --- /dev/null +++ b/test/CodeGen/X86/2010-02-23-DAGCombineBug.ll @@ -0,0 +1,18 @@ +; RUN: llc < %s -march=x86 | FileCheck %s + +define i32* @t() nounwind optsize ssp { +entry: +; CHECK: t: +; CHECK: testl %eax, %eax +; CHECK: js + %cmp = icmp slt i32 undef, 0 ; <i1> [#uses=1] + %outsearch.0 = select i1 %cmp, i1 false, i1 true ; <i1> [#uses=1] + br i1 %outsearch.0, label %if.then27, label %if.else29 + +if.then27: ; preds = %entry + ret i32* undef + +if.else29: ; preds = %entry + unreachable +} + diff --git a/test/CodeGen/X86/2010-02-23-DIV8rDefinesAX.ll b/test/CodeGen/X86/2010-02-23-DIV8rDefinesAX.ll new file mode 100644 index 0000000..8543c80 --- /dev/null +++ b/test/CodeGen/X86/2010-02-23-DIV8rDefinesAX.ll @@ -0,0 +1,20 @@ +; RUN: llc < %s +; PR6374 +; +; This test produces a DIV8r instruction and uses %AX instead of %AH and %AL. +; The DIV8r must have the right imp-defs for that to work. +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" +target triple = "x86_64-apple-darwin10.0.0" + +%struct._i386_state = type { %union.anon } +%union.anon = type { [0 x i8] } + +define void @i386_aam(%struct._i386_state* nocapture %cpustate) nounwind ssp { +entry: + %call = tail call fastcc signext i8 @FETCH() ; <i8> [#uses=1] + %rem = urem i8 0, %call ; <i8> [#uses=1] + store i8 %rem, i8* undef + ret void +} + +declare fastcc signext i8 @FETCH() nounwind readnone ssp diff --git a/test/CodeGen/X86/2010-02-23-RematImplicitSubreg.ll b/test/CodeGen/X86/2010-02-23-RematImplicitSubreg.ll new file mode 100644 index 0000000..4a26ba0 --- /dev/null +++ b/test/CodeGen/X86/2010-02-23-RematImplicitSubreg.ll @@ -0,0 +1,49 @@ +; RUN: llc < %s +; PR6372 +; +; This test produces a move instruction with an implicitly defined super-register: +; +; %DL<def> = MOV8rr %reg1038<kill>, %RDX<imp-def> +; +; When %DL is rematerialized, we must remember to update live intervals for +; sub-registers %DX and %EDX. + +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" +target triple = "x86_64-apple-darwin10.0.0" + +define noalias i8* @foo() nounwind ssp { +entry: + br i1 undef, label %for.end, label %for.body + +for.body: ; preds = %if.end40, %entry + %tmp6 = load i8* undef, align 2 ; <i8> [#uses=3] + %conv11 = sext i8 %tmp6 to i64 ; <i64> [#uses=1] + %cmp15 = icmp slt i64 %conv11, undef ; <i1> [#uses=1] + br i1 %cmp15, label %if.end, label %if.then + +if.then: ; preds = %for.body + %conv18 = sext i8 %tmp6 to i32 ; <i32> [#uses=1] + %call = tail call i32 (...)* @invalid(i32 0, i32 0, i32 %conv18) nounwind ; <i32> [#uses=0] + br label %if.end + +if.end: ; preds = %if.then, %for.body + %index.0 = phi i8 [ 0, %if.then ], [ %tmp6, %for.body ] ; <i8> [#uses=1] + store i8 %index.0, i8* undef + %tmp24 = load i8* undef ; <i8> [#uses=2] + br i1 undef, label %if.end40, label %if.then36 + +if.then36: ; preds = %if.end + %conv38 = sext i8 %tmp24 to i32 ; <i32> [#uses=1] + %call39 = tail call i32 (...)* @invalid(i32 0, i32 0, i32 %conv38) nounwind ; <i32> [#uses=0] + br label %if.end40 + +if.end40: ; preds = %if.then36, %if.end + %index.1 = phi i8 [ 0, %if.then36 ], [ %tmp24, %if.end ] ; <i8> [#uses=1] + store i8 %index.1, i8* undef + br i1 false, label %for.body, label %for.end + +for.end: ; preds = %if.end40, %entry + ret i8* undef +} + +declare i32 @invalid(...) diff --git a/test/CodeGen/X86/2010-02-23-SingleDefPhiJoin.ll b/test/CodeGen/X86/2010-02-23-SingleDefPhiJoin.ll new file mode 100644 index 0000000..aeed401 --- /dev/null +++ b/test/CodeGen/X86/2010-02-23-SingleDefPhiJoin.ll @@ -0,0 +1,146 @@ +; RUN: llc < %s +; PR6363 +; +; This test case creates a phi join register with a single definition. The other +; predecessor blocks are implicit-def. +; +; If LiveIntervalAnalysis fails to recognize this as a phi join, the coalescer +; will detect an infinity valno loop. +; +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" +target triple = "x86_64-unknown-linux-gnu" + +define i32 @decode(i8* nocapture %input, i32 %offset, i8* nocapture %output) nounwind { +entry: + br i1 undef, label %meshBB86, label %meshBB102 + +bb: ; preds = %meshBB106, %meshBB102 + br i1 false, label %bb9, label %meshBB90 + +bb.nph: ; preds = %meshBB90 + br label %meshBB114 + +bb.nph.fragment: ; preds = %meshBB114 + br label %meshBB118 + +bb1.fragment: ; preds = %meshBB118 + br i1 false, label %bb2, label %bb3 + +bb2: ; preds = %bb1.fragment + br label %meshBB74 + +bb2.fragment15: ; preds = %meshBB74 + br label %meshBB98 + +bb3: ; preds = %bb1.fragment + br i1 undef, label %meshBB, label %meshBB102 + +bb4: ; preds = %meshBB + br label %meshBB118 + +bb4.fragment: ; preds = %meshBB118 + br label %meshBB82 + +bb5: ; preds = %meshBB102, %meshBB82 + br i1 false, label %bb6, label %bb7 + +bb6: ; preds = %bb5 + br label %bb7 + +bb7: ; preds = %meshBB98, %bb6, %bb5 + br label %meshBB114 + +bb7.fragment: ; preds = %meshBB114 + br i1 undef, label %meshBB74, label %bb9 + +bb9: ; preds = %bb7.fragment, %bb + br label %bb1.i23 + +bb1.i23: ; preds = %meshBB110, %bb9 + br i1 undef, label %meshBB106, label %meshBB110 + +skip_to_newline.exit26: ; preds = %meshBB106 + br label %meshBB86 + +skip_to_newline.exit26.fragment: ; preds = %meshBB86 + br i1 false, label %meshBB90, label %meshBB106 + +bb11.fragment: ; preds = %meshBB90, %meshBB86 + br label %meshBB122 + +bb1.i: ; preds = %meshBB122, %meshBB + %ooffset.2.lcssa.phi.SV.phi203 = phi i32 [ 0, %meshBB122 ], [ %ooffset.2.lcssa.phi.SV.phi233, %meshBB ] ; <i32> [#uses=1] + br label %meshBB98 + +bb1.i.fragment: ; preds = %meshBB98 + br i1 undef, label %meshBB78, label %meshBB + +skip_to_newline.exit: ; preds = %meshBB78 + br i1 undef, label %bb12, label %meshBB110 + +bb12: ; preds = %skip_to_newline.exit + br label %meshBB94 + +bb12.fragment: ; preds = %meshBB94 + br i1 false, label %bb13, label %meshBB78 + +bb13: ; preds = %bb12.fragment + br label %meshBB82 + +bb13.fragment: ; preds = %meshBB82 + br i1 undef, label %meshBB94, label %meshBB122 + +bb14: ; preds = %meshBB94 + ret i32 %ooffset.2.lcssa.phi.SV.phi250 + +bb15: ; preds = %meshBB122, %meshBB110, %meshBB78 + unreachable + +meshBB: ; preds = %bb1.i.fragment, %bb3 + %ooffset.2.lcssa.phi.SV.phi233 = phi i32 [ undef, %bb3 ], [ %ooffset.2.lcssa.phi.SV.phi209, %bb1.i.fragment ] ; <i32> [#uses=1] + br i1 undef, label %bb1.i, label %bb4 + +meshBB74: ; preds = %bb7.fragment, %bb2 + br i1 false, label %meshBB118, label %bb2.fragment15 + +meshBB78: ; preds = %bb12.fragment, %bb1.i.fragment + %ooffset.2.lcssa.phi.SV.phi239 = phi i32 [ %ooffset.2.lcssa.phi.SV.phi209, %bb1.i.fragment ], [ %ooffset.2.lcssa.phi.SV.phi250, %bb12.fragment ] ; <i32> [#uses=1] + br i1 false, label %bb15, label %skip_to_newline.exit + +meshBB82: ; preds = %bb13, %bb4.fragment + br i1 false, label %bb5, label %bb13.fragment + +meshBB86: ; preds = %skip_to_newline.exit26, %entry + br i1 undef, label %skip_to_newline.exit26.fragment, label %bb11.fragment + +meshBB90: ; preds = %skip_to_newline.exit26.fragment, %bb + br i1 false, label %bb11.fragment, label %bb.nph + +meshBB94: ; preds = %bb13.fragment, %bb12 + %ooffset.2.lcssa.phi.SV.phi250 = phi i32 [ 0, %bb13.fragment ], [ %ooffset.2.lcssa.phi.SV.phi239, %bb12 ] ; <i32> [#uses=2] + br i1 false, label %bb12.fragment, label %bb14 + +meshBB98: ; preds = %bb1.i, %bb2.fragment15 + %ooffset.2.lcssa.phi.SV.phi209 = phi i32 [ undef, %bb2.fragment15 ], [ %ooffset.2.lcssa.phi.SV.phi203, %bb1.i ] ; <i32> [#uses=2] + br i1 undef, label %bb1.i.fragment, label %bb7 + +meshBB102: ; preds = %bb3, %entry + br i1 undef, label %bb5, label %bb + +meshBB106: ; preds = %skip_to_newline.exit26.fragment, %bb1.i23 + br i1 undef, label %bb, label %skip_to_newline.exit26 + +meshBB110: ; preds = %skip_to_newline.exit, %bb1.i23 + br i1 false, label %bb15, label %bb1.i23 + +meshBB114: ; preds = %bb7, %bb.nph + %meshStackVariable115.phi = phi i32 [ 19, %bb7 ], [ 8, %bb.nph ] ; <i32> [#uses=0] + br i1 undef, label %bb.nph.fragment, label %bb7.fragment + +meshBB118: ; preds = %meshBB74, %bb4, %bb.nph.fragment + %meshCmp121 = icmp eq i32 undef, 10 ; <i1> [#uses=1] + br i1 %meshCmp121, label %bb4.fragment, label %bb1.fragment + +meshBB122: ; preds = %bb13.fragment, %bb11.fragment + br i1 false, label %bb1.i, label %bb15 +} diff --git a/test/CodeGen/X86/and-or-fold.ll b/test/CodeGen/X86/and-or-fold.ll index 7733b8a..836b5f1 100644 --- a/test/CodeGen/X86/and-or-fold.ll +++ b/test/CodeGen/X86/and-or-fold.ll @@ -1,14 +1,26 @@ -; RUN: llc < %s -march=x86 | grep and | count 1 +; RUN: llc < %s -mtriple=i686-apple-darwin | FileCheck -check-prefix=DARWIN %s +; RUN: opt < %s -O2 | llc -mtriple=x86_64-apple-darwin | FileCheck -check-prefix=DARWIN-OPT %s ; The dag combiner should fold together (x&127)|(y&16711680) -> (x|y)&c1 ; in this case. -define i32 @test6(i32 %x, i16 %y) { - %tmp1 = zext i16 %y to i32 ; <i32> [#uses=1] - %tmp2 = and i32 %tmp1, 127 ; <i32> [#uses=1] - %tmp4 = shl i32 %x, 16 ; <i32> [#uses=1] - %tmp5 = and i32 %tmp4, 16711680 ; <i32> [#uses=1] - %tmp6 = or i32 %tmp2, %tmp5 ; <i32> [#uses=1] - ret i32 %tmp6 +define i32 @test1(i32 %x, i16 %y) { + %tmp1 = zext i16 %y to i32 + %tmp2 = and i32 %tmp1, 127 + %tmp4 = shl i32 %x, 16 + %tmp5 = and i32 %tmp4, 16711680 + %tmp6 = or i32 %tmp2, %tmp5 + ret i32 %tmp6 +; DARWIN: andl $16711807, %eax } +; <rdar://problem/7529774> The optimizer shouldn't fold this into (and (or, C), D) +; if (C & D) == 0 +define i64 @test2(i64 %x) nounwind readnone ssp { +entry: + %tmp1 = and i64 %x, 123127 + %tmp2 = or i64 %tmp1, 3 + ret i64 %tmp2 +; DARWIN-OPT: andq $123124 +; DARWIN-OPT-NEXT: leaq 3 +} diff --git a/test/CodeGen/X86/code_placement_eh.ll b/test/CodeGen/X86/code_placement_eh.ll new file mode 100644 index 0000000..172d591 --- /dev/null +++ b/test/CodeGen/X86/code_placement_eh.ll @@ -0,0 +1,45 @@ +; RUN: llc < %s + +; CodePlacementOpt shouldn't try to modify this loop because +; it involves EH edges. + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32" +target triple = "i386-apple-darwin10.0" + +define void @foo() { +invcont5: + br label %bb15 + +.noexc3: ; preds = %bb15 + br i1 undef, label %bb18.i5.i, label %bb15 + +.noexc6.i.i: ; preds = %bb18.i5.i + %tmp2021 = invoke float @cosf(float 0.000000e+00) readonly + to label %bb18.i5.i unwind label %lpad.i.i ; <float> [#uses=0] + +bb18.i5.i: ; preds = %.noexc6.i.i, %bb51.i + %tmp2019 = invoke float @sinf(float 0.000000e+00) readonly + to label %.noexc6.i.i unwind label %lpad.i.i ; <float> [#uses=0] + +lpad.i.i: ; preds = %bb18.i5.i, %.noexc6.i.i + %eh_ptr.i.i = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] + unreachable + +lpad59.i: ; preds = %bb15 + %eh_ptr60.i = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] + unreachable + +bb15: ; preds = %.noexc3, %invcont5 + invoke fastcc void @_ZN28btHashedOverlappingPairCacheC2Ev() + to label %.noexc3 unwind label %lpad59.i +} + +declare i8* @llvm.eh.exception() nounwind readonly + +declare i32 @llvm.eh.selector(i8*, i8*, ...) nounwind + +declare float @sinf(float) readonly + +declare float @cosf(float) readonly + +declare fastcc void @_ZN28btHashedOverlappingPairCacheC2Ev() align 2 diff --git a/test/CodeGen/X86/critical-edge-split.ll b/test/CodeGen/X86/critical-edge-split.ll index 4fe554d..f29cbf3 100644 --- a/test/CodeGen/X86/critical-edge-split.ll +++ b/test/CodeGen/X86/critical-edge-split.ll @@ -1,4 +1,4 @@ -; RUN: llc < %s -mtriple=i386-apple-darwin -tailcallopt=false -stats -info-output-file - | grep asm-printer | grep 31 +; RUN: llc < %s -mtriple=i386-apple-darwin -stats -info-output-file - | grep asm-printer | grep 29 %CC = type { %Register } %II = type { %"struct.XX::II::$_74" } diff --git a/test/CodeGen/X86/ins_subreg_coalesce-3.ll b/test/CodeGen/X86/ins_subreg_coalesce-3.ll index 627edc5..8c1c409 100644 --- a/test/CodeGen/X86/ins_subreg_coalesce-3.ll +++ b/test/CodeGen/X86/ins_subreg_coalesce-3.ll @@ -1,4 +1,4 @@ -; RUN: llc < %s -march=x86-64 | grep mov | count 5 +; RUN: llc < %s -march=x86-64 | grep mov | count 3 %struct.COMPOSITE = type { i8, i16, i16 } %struct.FILE = type { i8*, i32, i32, i16, i16, %struct.__sbuf, i32, i8*, i32 (i8*)*, i32 (i8*, i8*, i32)*, i64 (i8*, i64, i32)*, i32 (i8*, i8*, i32)*, %struct.__sbuf, %struct.__sFILEX*, i32, [3 x i8], [1 x i8], %struct.__sbuf, i32, i64 } diff --git a/test/CodeGen/X86/licm-symbol.ll b/test/CodeGen/X86/licm-symbol.ll new file mode 100644 index 0000000..d61bbfc --- /dev/null +++ b/test/CodeGen/X86/licm-symbol.ll @@ -0,0 +1,39 @@ +; RUN: llc < %s | FileCheck %s + +; MachineLICM should be able to hoist the sF reference out of the loop. + +; CHECK: pushl %esi +; CHECK: subl $8, %esp +; CHECK: movl $176, %esi +; CHECK: addl L___sF$non_lazy_ptr, %esi +; CHECK: .align 4, 0x90 + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32" +target triple = "i386-apple-darwin8" + +%struct.FILE = type { i8*, i32, i32, i16, i16, %struct.__sbuf, i32, i8*, i32 (i8*)*, i32 (i8*, i8*, i32)*, i64 (i8*, i64, i32)*, i32 (i8*, i8*, i32)*, %struct.__sbuf, %struct.__sFILEX*, i32, [3 x i8], [1 x i8], %struct.__sbuf, i32, i64 } +%struct.__sFILEX = type opaque +%struct.__sbuf = type { i8*, i32 } +%struct.gcov_ctr_summary = type { i32, i32, i64, i64, i64 } +%struct.gcov_summary = type { i32, [1 x %struct.gcov_ctr_summary] } + +@__sF = external global [0 x %struct.FILE] ; <[0 x %struct.FILE]*> [#uses=1] + +declare i32 @fprintf(%struct.FILE* nocapture) nounwind + +define void @gcov_exit() nounwind { +entry: + br label %bb151 + +bb151: ; preds = %bb59, %bb56, %bb14 + br i1 undef, label %bb56, label %bb59 + +bb56: ; preds = %bb151 + %t0 = call i32 (%struct.FILE*)* @fprintf(%struct.FILE* getelementptr inbounds ([0 x %struct.FILE]* @__sF, i32 0, i32 2)) nounwind + br label %bb151 + +bb59: ; preds = %bb151 + %t1 = call i32 (%struct.FILE*)* @fprintf(%struct.FILE* getelementptr inbounds ([0 x %struct.FILE]* @__sF, i32 0, i32 2)) nounwind + br label %bb151 +} + diff --git a/test/CodeGen/X86/lsr-overflow.ll b/test/CodeGen/X86/lsr-overflow.ll new file mode 100644 index 0000000..0b0214c --- /dev/null +++ b/test/CodeGen/X86/lsr-overflow.ll @@ -0,0 +1,26 @@ +; RUN: llc < %s -march=x86-64 | FileCheck %s + +; The comparison uses the pre-inc value, which could lead LSR to +; try to compute -INT64_MIN. + +; CHECK: movabsq $-9223372036854775808, %rax +; CHECK: cmpq %rax, %rbx +; CHECK: sete %al + +declare i64 @bar() + +define i1 @foo() nounwind { +entry: + br label %for.cond.i + +for.cond.i: + %indvar = phi i64 [ 0, %entry ], [ %indvar.next, %for.cond.i ] + %t = call i64 @bar() + %indvar.next = add i64 %indvar, 1 + %s = icmp ne i64 %indvar.next, %t + br i1 %s, label %for.cond.i, label %__ABContainsLabel.exit + +__ABContainsLabel.exit: + %cmp = icmp eq i64 %indvar, 9223372036854775807 + ret i1 %cmp +} diff --git a/test/CodeGen/X86/lsr-reuse-trunc.ll b/test/CodeGen/X86/lsr-reuse-trunc.ll index d1d7144..a663a22 100644 --- a/test/CodeGen/X86/lsr-reuse-trunc.ll +++ b/test/CodeGen/X86/lsr-reuse-trunc.ll @@ -1,10 +1,19 @@ -; RUN: llc < %s -march=x86-64 | FileCheck %s +; RUN: llc < %s -mtriple=x86_64-apple-darwin | FileCheck %s ; Full strength reduction wouldn't reduce register pressure, so LSR should ; stick with indexing here. -; CHECK: movaps (%rsi,%rax,4), %xmm3 -; CHECK: movaps %xmm3, (%rdi,%rax,4) +; Also checks andps and andnps shares the same constantpool. Previously llvm +; will codegen two andps, one using 0x80000000, the other 0x7fffffff. +; rdar://7323335 + +; CHECK: movaps LCPI1_0 +; CHECK: movaps LCPI1_1 +; CHECK-NOT: movaps LCPI1_2 +; CHECK: movaps (%rsi,%rax,4), %xmm2 +; CHECK: andps +; CHECK: andnps +; CHECK: movaps %xmm2, (%rdi,%rax,4) ; CHECK: addq $4, %rax ; CHECK: cmpl %eax, (%rdx) ; CHECK-NEXT: jg diff --git a/test/CodeGen/X86/lsr-reuse.ll b/test/CodeGen/X86/lsr-reuse.ll index 7f2b8cc..2f6fb3f 100644 --- a/test/CodeGen/X86/lsr-reuse.ll +++ b/test/CodeGen/X86/lsr-reuse.ll @@ -1,4 +1,4 @@ -; RUN: llc < %s -march=x86-64 -O3 | FileCheck %s +; RUN: llc < %s -march=x86-64 -O3 -asm-verbose=false | FileCheck %s target datalayout = "e-p:64:64:64" target triple = "x86_64-unknown-unknown" @@ -384,3 +384,59 @@ loop: return: ret void } + +; LSR should use only one indvar for the inner loop. +; rdar://7657764 + +; CHECK: asd: +; CHECK: BB10_5: +; CHECK-NEXT: addl (%r{{[^,]*}},%rdi,4), %e +; CHECK-NEXT: incq %rdi +; CHECK-NEXT: cmpq %rdi, %r{{[^,]*}} +; CHECK-NEXT: jg + +%struct.anon = type { i32, [4200 x i32] } + +@bars = common global [123123 x %struct.anon] zeroinitializer, align 32 ; <[123123 x %struct.anon]*> [#uses=2] + +define i32 @asd(i32 %n) nounwind readonly { +entry: + %0 = icmp sgt i32 %n, 0 ; <i1> [#uses=1] + br i1 %0, label %bb.nph14, label %bb5 + +bb.nph14: ; preds = %entry + %tmp18 = zext i32 %n to i64 ; <i64> [#uses=1] + br label %bb + +bb: ; preds = %bb3, %bb.nph14 + %indvar16 = phi i64 [ 0, %bb.nph14 ], [ %indvar.next17, %bb3 ] ; <i64> [#uses=3] + %s.113 = phi i32 [ 0, %bb.nph14 ], [ %s.0.lcssa, %bb3 ] ; <i32> [#uses=2] + %scevgep2526 = getelementptr [123123 x %struct.anon]* @bars, i64 0, i64 %indvar16, i32 0 ; <i32*> [#uses=1] + %1 = load i32* %scevgep2526, align 4 ; <i32> [#uses=2] + %2 = icmp sgt i32 %1, 0 ; <i1> [#uses=1] + br i1 %2, label %bb.nph, label %bb3 + +bb.nph: ; preds = %bb + %tmp23 = sext i32 %1 to i64 ; <i64> [#uses=1] + br label %bb1 + +bb1: ; preds = %bb.nph, %bb1 + %indvar = phi i64 [ 0, %bb.nph ], [ %tmp19, %bb1 ] ; <i64> [#uses=2] + %s.07 = phi i32 [ %s.113, %bb.nph ], [ %4, %bb1 ] ; <i32> [#uses=1] + %c.08 = getelementptr [123123 x %struct.anon]* @bars, i64 0, i64 %indvar16, i32 1, i64 %indvar ; <i32*> [#uses=1] + %3 = load i32* %c.08, align 4 ; <i32> [#uses=1] + %4 = add nsw i32 %3, %s.07 ; <i32> [#uses=2] + %tmp19 = add i64 %indvar, 1 ; <i64> [#uses=2] + %5 = icmp sgt i64 %tmp23, %tmp19 ; <i1> [#uses=1] + br i1 %5, label %bb1, label %bb3 + +bb3: ; preds = %bb1, %bb + %s.0.lcssa = phi i32 [ %s.113, %bb ], [ %4, %bb1 ] ; <i32> [#uses=2] + %indvar.next17 = add i64 %indvar16, 1 ; <i64> [#uses=2] + %exitcond = icmp eq i64 %indvar.next17, %tmp18 ; <i1> [#uses=1] + br i1 %exitcond, label %bb5, label %bb + +bb5: ; preds = %bb3, %entry + %s.1.lcssa = phi i32 [ 0, %entry ], [ %s.0.lcssa, %bb3 ] ; <i32> [#uses=1] + ret i32 %s.1.lcssa +} diff --git a/test/CodeGen/X86/lsr-wrap.ll b/test/CodeGen/X86/lsr-wrap.ll new file mode 100644 index 0000000..ec8db50 --- /dev/null +++ b/test/CodeGen/X86/lsr-wrap.ll @@ -0,0 +1,37 @@ +; RUN: llc -march=x86-64 < %s | FileCheck %s + +; LSR would like to use a single IV for both of these, however it's +; not safe due to wraparound. + +; CHECK: addb $-4, %r +; CHECK: decw % + +@g_19 = common global i32 0 ; <i32*> [#uses=2] + +declare i32 @func_8(i8 zeroext) nounwind + +declare i32 @func_3(i8 signext) nounwind + +define void @func_1() nounwind { +entry: + br label %bb + +bb: ; preds = %bb, %entry + %indvar = phi i16 [ 0, %entry ], [ %indvar.next, %bb ] ; <i16> [#uses=2] + %tmp = sub i16 0, %indvar ; <i16> [#uses=1] + %tmp27 = trunc i16 %tmp to i8 ; <i8> [#uses=1] + %tmp1 = load i32* @g_19, align 4 ; <i32> [#uses=2] + %tmp2 = add i32 %tmp1, 1 ; <i32> [#uses=1] + store i32 %tmp2, i32* @g_19, align 4 + %tmp3 = trunc i32 %tmp1 to i8 ; <i8> [#uses=1] + %tmp4 = tail call i32 @func_8(i8 zeroext %tmp3) nounwind ; <i32> [#uses=0] + %tmp5 = shl i8 %tmp27, 2 ; <i8> [#uses=1] + %tmp6 = add i8 %tmp5, -112 ; <i8> [#uses=1] + %tmp7 = tail call i32 @func_3(i8 signext %tmp6) nounwind ; <i32> [#uses=0] + %indvar.next = add i16 %indvar, 1 ; <i16> [#uses=2] + %exitcond = icmp eq i16 %indvar.next, -28 ; <i1> [#uses=1] + br i1 %exitcond, label %return, label %bb + +return: ; preds = %bb + ret void +} diff --git a/test/CodeGen/X86/omit-label.ll b/test/CodeGen/X86/omit-label.ll deleted file mode 100644 index 0ec03eb..0000000 --- a/test/CodeGen/X86/omit-label.ll +++ /dev/null @@ -1,57 +0,0 @@ -; RUN: llc < %s -asm-verbose=false -mtriple=x86_64-linux-gnu | FileCheck %s -; PR4126 -; PR4732 - -; Don't omit these labels' definitions. - -; CHECK: bux: -; CHECK: LBB1_1: - -define void @bux(i32 %p_53) nounwind optsize { -entry: - %0 = icmp eq i32 %p_53, 0 ; <i1> [#uses=1] - %1 = icmp sgt i32 %p_53, 0 ; <i1> [#uses=1] - %or.cond = and i1 %0, %1 ; <i1> [#uses=1] - br i1 %or.cond, label %bb.i, label %bb3 - -bb.i: ; preds = %entry - %2 = add i32 %p_53, 1 ; <i32> [#uses=1] - %3 = icmp slt i32 %2, 0 ; <i1> [#uses=0] - br label %bb3 - -bb3: ; preds = %bb.i, %entry - %4 = tail call i32 (...)* @baz(i32 0) nounwind ; <i32> [#uses=0] - ret void -} - -declare i32 @baz(...) - -; Don't omit this label in the assembly output. -; CHECK: int321: -; CHECK: LBB2_1 -; CHECK: LBB2_1 -; CHECK: LBB2_1: - -define void @int321(i8 signext %p_103, i32 %uint8p_104) nounwind readnone { -entry: - %tobool = icmp eq i8 %p_103, 0 ; <i1> [#uses=1] - %cmp.i = icmp sgt i8 %p_103, 0 ; <i1> [#uses=1] - %or.cond = and i1 %tobool, %cmp.i ; <i1> [#uses=1] - br i1 %or.cond, label %land.end.i, label %for.cond.preheader - -land.end.i: ; preds = %entry - %conv3.i = sext i8 %p_103 to i32 ; <i32> [#uses=1] - %div.i = sdiv i32 1, %conv3.i ; <i32> [#uses=1] - %tobool.i = icmp eq i32 %div.i, -2147483647 ; <i1> [#uses=0] - br label %for.cond.preheader - -for.cond.preheader: ; preds = %land.end.i, %entry - %cmp = icmp sgt i8 %p_103, 1 ; <i1> [#uses=1] - br i1 %cmp, label %for.end.split, label %for.cond - -for.cond: ; preds = %for.cond.preheader, %for.cond - br label %for.cond - -for.end.split: ; preds = %for.cond.preheader - ret void -} diff --git a/test/CodeGen/X86/pr3495-2.ll b/test/CodeGen/X86/pr3495-2.ll index 71aa5a0..98c064a 100644 --- a/test/CodeGen/X86/pr3495-2.ll +++ b/test/CodeGen/X86/pr3495-2.ll @@ -1,4 +1,8 @@ -; RUN: llc < %s -march=x86 -relocation-model=pic -disable-fp-elim -stats |& grep {Number of reloads omited} +; RUN: llc < %s -march=x86 -relocation-model=pic -disable-fp-elim -stats |& grep {Number of loads added} | grep 1 +; PR3495 +; +; This test may not be testing what it was supposed to test. +; It used to have two spills and four reloads, but not it only has one spill and one reload. target datalayout = "e-p:32:32:32" target triple = "i386-apple-darwin9.6" diff --git a/test/CodeGen/X86/ptrtoint-constexpr.ll b/test/CodeGen/X86/ptrtoint-constexpr.ll index dd97905..d1cb34b 100644 --- a/test/CodeGen/X86/ptrtoint-constexpr.ll +++ b/test/CodeGen/X86/ptrtoint-constexpr.ll @@ -9,6 +9,6 @@ ; CHECK: .globl x ; CHECK: x: -; CHECK: .quad 3 +; CHECK: .quad ((0+1)&4294967295)*3 @x = global i64 mul (i64 3, i64 ptrtoint (i2* getelementptr (i2* null, i64 1) to i64)) diff --git a/test/CodeGen/X86/sink-hoist.ll b/test/CodeGen/X86/sink-hoist.ll index 01d7373..e1d0fe7 100644 --- a/test/CodeGen/X86/sink-hoist.ll +++ b/test/CodeGen/X86/sink-hoist.ll @@ -63,7 +63,6 @@ entry: ; CHECK: vv: ; CHECK: LCPI4_0(%rip), %xmm0 ; CHECK: LCPI4_1(%rip), %xmm1 -; CHECK: LCPI4_2(%rip), %xmm2 ; CHECK: align ; CHECK-NOT: LCPI ; CHECK: ret diff --git a/test/CodeGen/X86/sse-minmax.ll b/test/CodeGen/X86/sse-minmax.ll index 17ffb5e..19fbed0 100644 --- a/test/CodeGen/X86/sse-minmax.ll +++ b/test/CodeGen/X86/sse-minmax.ll @@ -1,17 +1,26 @@ ; RUN: llc < %s -march=x86-64 -asm-verbose=false | FileCheck %s +; RUN: llc < %s -march=x86-64 -asm-verbose=false -enable-unsafe-fp-math | FileCheck -check-prefix=UNSAFE %s +; RUN: llc < %s -march=x86-64 -asm-verbose=false -enable-finite-only-fp-math | FileCheck -check-prefix=FINITE %s ; Some of these patterns can be matched as SSE min or max. Some of ; then can be matched provided that the operands are swapped. ; Some of them can't be matched at all and require a comparison ; and a conditional branch. -; The naming convention is {,x_}{o,u}{gt,lt,ge,le}{,_inverse} +; The naming convention is {,x_,y_}{o,u}{gt,lt,ge,le}{,_inverse} ; x_ : use 0.0 instead of %y +; y_ : use -0.0 instead of %y ; _inverse : swap the arms of the select. ; CHECK: ogt: ; CHECK-NEXT: maxsd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: ogt: +; UNSAFE-NEXT: maxsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: ogt: +; FINITE-NEXT: maxsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @ogt(double %x, double %y) nounwind { %c = fcmp ogt double %x, %y %d = select i1 %c, double %x, double %y @@ -21,6 +30,12 @@ define double @ogt(double %x, double %y) nounwind { ; CHECK: olt: ; CHECK-NEXT: minsd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: olt: +; UNSAFE-NEXT: minsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: olt: +; FINITE-NEXT: minsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @olt(double %x, double %y) nounwind { %c = fcmp olt double %x, %y %d = select i1 %c, double %x, double %y @@ -31,6 +46,14 @@ define double @olt(double %x, double %y) nounwind { ; CHECK-NEXT: minsd %xmm0, %xmm1 ; CHECK-NEXT: movapd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: ogt_inverse: +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: ogt_inverse: +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @ogt_inverse(double %x, double %y) nounwind { %c = fcmp ogt double %x, %y %d = select i1 %c, double %y, double %x @@ -41,6 +64,14 @@ define double @ogt_inverse(double %x, double %y) nounwind { ; CHECK-NEXT: maxsd %xmm0, %xmm1 ; CHECK-NEXT: movapd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: olt_inverse: +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: olt_inverse: +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @olt_inverse(double %x, double %y) nounwind { %c = fcmp olt double %x, %y %d = select i1 %c, double %y, double %x @@ -49,6 +80,12 @@ define double @olt_inverse(double %x, double %y) nounwind { ; CHECK: oge: ; CHECK-NEXT: ucomisd %xmm1, %xmm0 +; UNSAFE: oge: +; UNSAFE-NEXT: maxsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: oge: +; FINITE-NEXT: maxsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @oge(double %x, double %y) nounwind { %c = fcmp oge double %x, %y %d = select i1 %c, double %x, double %y @@ -57,6 +94,10 @@ define double @oge(double %x, double %y) nounwind { ; CHECK: ole: ; CHECK-NEXT: ucomisd %xmm0, %xmm1 +; UNSAFE: ole: +; UNSAFE-NEXT: minsd %xmm1, %xmm0 +; FINITE: ole: +; FINITE-NEXT: minsd %xmm1, %xmm0 define double @ole(double %x, double %y) nounwind { %c = fcmp ole double %x, %y %d = select i1 %c, double %x, double %y @@ -65,6 +106,14 @@ define double @ole(double %x, double %y) nounwind { ; CHECK: oge_inverse: ; CHECK-NEXT: ucomisd %xmm1, %xmm0 +; UNSAFE: oge_inverse: +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: oge_inverse: +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @oge_inverse(double %x, double %y) nounwind { %c = fcmp oge double %x, %y %d = select i1 %c, double %y, double %x @@ -73,6 +122,14 @@ define double @oge_inverse(double %x, double %y) nounwind { ; CHECK: ole_inverse: ; CHECK-NEXT: ucomisd %xmm0, %xmm1 +; UNSAFE: ole_inverse: +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: ole_inverse: +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @ole_inverse(double %x, double %y) nounwind { %c = fcmp ole double %x, %y %d = select i1 %c, double %y, double %x @@ -83,6 +140,14 @@ define double @ole_inverse(double %x, double %y) nounwind { ; CHECK-NEXT: pxor %xmm1, %xmm1 ; CHECK-NEXT: maxsd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: x_ogt: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: maxsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ogt: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: maxsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ogt(double %x) nounwind { %c = fcmp ogt double %x, 0.000000e+00 %d = select i1 %c, double %x, double 0.000000e+00 @@ -93,6 +158,14 @@ define double @x_ogt(double %x) nounwind { ; CHECK-NEXT: pxor %xmm1, %xmm1 ; CHECK-NEXT: minsd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: x_olt: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: minsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_olt: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: minsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_olt(double %x) nounwind { %c = fcmp olt double %x, 0.000000e+00 %d = select i1 %c, double %x, double 0.000000e+00 @@ -104,6 +177,16 @@ define double @x_olt(double %x) nounwind { ; CHECK-NEXT: minsd %xmm0, %xmm1 ; CHECK-NEXT: movapd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: x_ogt_inverse: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ogt_inverse: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ogt_inverse(double %x) nounwind { %c = fcmp ogt double %x, 0.000000e+00 %d = select i1 %c, double 0.000000e+00, double %x @@ -115,6 +198,16 @@ define double @x_ogt_inverse(double %x) nounwind { ; CHECK-NEXT: maxsd %xmm0, %xmm1 ; CHECK-NEXT: movapd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: x_olt_inverse: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_olt_inverse: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_olt_inverse(double %x) nounwind { %c = fcmp olt double %x, 0.000000e+00 %d = select i1 %c, double 0.000000e+00, double %x @@ -122,9 +215,15 @@ define double @x_olt_inverse(double %x) nounwind { } ; CHECK: x_oge: -; CHECK-NEXT: pxor %xmm1, %xmm1 -; CHECK-NEXT: maxsd %xmm1, %xmm0 -; CHECK-NEXT: ret +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: x_oge: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: maxsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_oge: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: maxsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_oge(double %x) nounwind { %c = fcmp oge double %x, 0.000000e+00 %d = select i1 %c, double %x, double 0.000000e+00 @@ -132,9 +231,15 @@ define double @x_oge(double %x) nounwind { } ; CHECK: x_ole: -; CHECK-NEXT: pxor %xmm1, %xmm1 -; CHECK-NEXT: minsd %xmm1, %xmm0 -; CHECK-NEXT: ret +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: x_ole: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: minsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ole: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: minsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ole(double %x) nounwind { %c = fcmp ole double %x, 0.000000e+00 %d = select i1 %c, double %x, double 0.000000e+00 @@ -142,10 +247,17 @@ define double @x_ole(double %x) nounwind { } ; CHECK: x_oge_inverse: -; CHECK-NEXT: pxor %xmm1, %xmm1 -; CHECK-NEXT: minsd %xmm0, %xmm1 -; CHECK-NEXT: movapd %xmm1, %xmm0 -; CHECK-NEXT: ret +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: x_oge_inverse: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_oge_inverse: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_oge_inverse(double %x) nounwind { %c = fcmp oge double %x, 0.000000e+00 %d = select i1 %c, double 0.000000e+00, double %x @@ -153,10 +265,17 @@ define double @x_oge_inverse(double %x) nounwind { } ; CHECK: x_ole_inverse: -; CHECK-NEXT: pxor %xmm1, %xmm1 -; CHECK-NEXT: maxsd %xmm0, %xmm1 -; CHECK-NEXT: movapd %xmm1, %xmm0 -; CHECK-NEXT: ret +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: x_ole_inverse: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ole_inverse: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ole_inverse(double %x) nounwind { %c = fcmp ole double %x, 0.000000e+00 %d = select i1 %c, double 0.000000e+00, double %x @@ -164,7 +283,13 @@ define double @x_ole_inverse(double %x) nounwind { } ; CHECK: ugt: -; CHECK-NEXT: ucomisd %xmm0, %xmm1 +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: ugt: +; UNSAFE-NEXT: maxsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: ugt: +; FINITE-NEXT: maxsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @ugt(double %x, double %y) nounwind { %c = fcmp ugt double %x, %y %d = select i1 %c, double %x, double %y @@ -172,7 +297,13 @@ define double @ugt(double %x, double %y) nounwind { } ; CHECK: ult: -; CHECK-NEXT: ucomisd %xmm1, %xmm0 +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: ult: +; UNSAFE-NEXT: minsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: ult: +; FINITE-NEXT: minsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @ult(double %x, double %y) nounwind { %c = fcmp ult double %x, %y %d = select i1 %c, double %x, double %y @@ -180,7 +311,15 @@ define double @ult(double %x, double %y) nounwind { } ; CHECK: ugt_inverse: -; CHECK-NEXT: ucomisd %xmm0, %xmm1 +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: ugt_inverse: +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: ugt_inverse: +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @ugt_inverse(double %x, double %y) nounwind { %c = fcmp ugt double %x, %y %d = select i1 %c, double %y, double %x @@ -188,7 +327,15 @@ define double @ugt_inverse(double %x, double %y) nounwind { } ; CHECK: ult_inverse: -; CHECK-NEXT: ucomisd %xmm1, %xmm0 +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: ult_inverse: +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: ult_inverse: +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @ult_inverse(double %x, double %y) nounwind { %c = fcmp ult double %x, %y %d = select i1 %c, double %y, double %x @@ -196,9 +343,15 @@ define double @ult_inverse(double %x, double %y) nounwind { } ; CHECK: uge: -; CHECK-NEXT: maxsd %xmm0, %xmm1 -; CHECK-NEXT: movapd %xmm1, %xmm0 +; CHECK-NEXT: maxsd %xmm0, %xmm1 +; CHECK-NEXT: movapd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: uge: +; UNSAFE-NEXT: maxsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: uge: +; FINITE-NEXT: maxsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @uge(double %x, double %y) nounwind { %c = fcmp uge double %x, %y %d = select i1 %c, double %x, double %y @@ -209,6 +362,12 @@ define double @uge(double %x, double %y) nounwind { ; CHECK-NEXT: minsd %xmm0, %xmm1 ; CHECK-NEXT: movapd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: ule: +; UNSAFE-NEXT: minsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: ule: +; FINITE-NEXT: minsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @ule(double %x, double %y) nounwind { %c = fcmp ule double %x, %y %d = select i1 %c, double %x, double %y @@ -218,6 +377,14 @@ define double @ule(double %x, double %y) nounwind { ; CHECK: uge_inverse: ; CHECK-NEXT: minsd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: uge_inverse: +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: uge_inverse: +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @uge_inverse(double %x, double %y) nounwind { %c = fcmp uge double %x, %y %d = select i1 %c, double %y, double %x @@ -227,6 +394,14 @@ define double @uge_inverse(double %x, double %y) nounwind { ; CHECK: ule_inverse: ; CHECK-NEXT: maxsd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: ule_inverse: +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: ule_inverse: +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @ule_inverse(double %x, double %y) nounwind { %c = fcmp ule double %x, %y %d = select i1 %c, double %y, double %x @@ -234,10 +409,15 @@ define double @ule_inverse(double %x, double %y) nounwind { } ; CHECK: x_ugt: -; CHECK-NEXT: pxor %xmm1, %xmm1 -; CHECK-NEXT: maxsd %xmm0, %xmm1 -; CHECK-NEXT: movapd %xmm1, %xmm0 -; CHECK-NEXT: ret +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: x_ugt: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: maxsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ugt: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: maxsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ugt(double %x) nounwind { %c = fcmp ugt double %x, 0.000000e+00 %d = select i1 %c, double %x, double 0.000000e+00 @@ -245,10 +425,15 @@ define double @x_ugt(double %x) nounwind { } ; CHECK: x_ult: -; CHECK-NEXT: pxor %xmm1, %xmm1 -; CHECK-NEXT: minsd %xmm0, %xmm1 -; CHECK-NEXT: movapd %xmm1, %xmm0 -; CHECK-NEXT: ret +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: x_ult: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: minsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ult: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: minsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ult(double %x) nounwind { %c = fcmp ult double %x, 0.000000e+00 %d = select i1 %c, double %x, double 0.000000e+00 @@ -256,9 +441,17 @@ define double @x_ult(double %x) nounwind { } ; CHECK: x_ugt_inverse: -; CHECK-NEXT: pxor %xmm1, %xmm1 -; CHECK-NEXT: minsd %xmm1, %xmm0 -; CHECK-NEXT: ret +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: x_ugt_inverse: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ugt_inverse: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ugt_inverse(double %x) nounwind { %c = fcmp ugt double %x, 0.000000e+00 %d = select i1 %c, double 0.000000e+00, double %x @@ -266,9 +459,17 @@ define double @x_ugt_inverse(double %x) nounwind { } ; CHECK: x_ult_inverse: -; CHECK-NEXT: pxor %xmm1, %xmm1 -; CHECK-NEXT: maxsd %xmm1, %xmm0 -; CHECK-NEXT: ret +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: x_ult_inverse: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ult_inverse: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ult_inverse(double %x) nounwind { %c = fcmp ult double %x, 0.000000e+00 %d = select i1 %c, double 0.000000e+00, double %x @@ -280,6 +481,14 @@ define double @x_ult_inverse(double %x) nounwind { ; CHECK-NEXT: maxsd %xmm0, %xmm1 ; CHECK-NEXT: movapd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: x_uge: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: maxsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_uge: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: maxsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_uge(double %x) nounwind { %c = fcmp uge double %x, 0.000000e+00 %d = select i1 %c, double %x, double 0.000000e+00 @@ -291,6 +500,14 @@ define double @x_uge(double %x) nounwind { ; CHECK-NEXT: minsd %xmm0, %xmm1 ; CHECK-NEXT: movapd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: x_ule: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: minsd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ule: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: minsd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ule(double %x) nounwind { %c = fcmp ule double %x, 0.000000e+00 %d = select i1 %c, double %x, double 0.000000e+00 @@ -301,6 +518,16 @@ define double @x_ule(double %x) nounwind { ; CHECK-NEXT: pxor %xmm1, %xmm1 ; CHECK-NEXT: minsd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: x_uge_inverse: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_uge_inverse: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_uge_inverse(double %x) nounwind { %c = fcmp uge double %x, 0.000000e+00 %d = select i1 %c, double 0.000000e+00, double %x @@ -311,16 +538,301 @@ define double @x_uge_inverse(double %x) nounwind { ; CHECK-NEXT: pxor %xmm1, %xmm1 ; CHECK-NEXT: maxsd %xmm1, %xmm0 ; CHECK-NEXT: ret +; UNSAFE: x_ule_inverse: +; UNSAFE-NEXT: pxor %xmm1, %xmm1 +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: x_ule_inverse: +; FINITE-NEXT: pxor %xmm1, %xmm1 +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret define double @x_ule_inverse(double %x) nounwind { %c = fcmp ule double %x, 0.000000e+00 %d = select i1 %c, double 0.000000e+00, double %x ret double %d } +; CHECK: y_ogt: +; CHECK-NEXT: maxsd {{[^,]*}}, %xmm0 +; CHECK-NEXT: ret +; UNSAFE: y_ogt: +; UNSAFE-NEXT: maxsd {{[^,]*}}, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ogt: +; FINITE-NEXT: maxsd {{[^,]*}}, %xmm0 +; FINITE-NEXT: ret +define double @y_ogt(double %x) nounwind { + %c = fcmp ogt double %x, -0.000000e+00 + %d = select i1 %c, double %x, double -0.000000e+00 + ret double %d +} + +; CHECK: y_olt: +; CHECK-NEXT: minsd {{[^,]*}}, %xmm0 +; CHECK-NEXT: ret +; UNSAFE: y_olt: +; UNSAFE-NEXT: minsd {{[^,]*}}, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_olt: +; FINITE-NEXT: minsd {{[^,]*}}, %xmm0 +; FINITE-NEXT: ret +define double @y_olt(double %x) nounwind { + %c = fcmp olt double %x, -0.000000e+00 + %d = select i1 %c, double %x, double -0.000000e+00 + ret double %d +} + +; CHECK: y_ogt_inverse: +; CHECK-NEXT: movsd {{[^,]*}}, %xmm1 +; CHECK-NEXT: minsd %xmm0, %xmm1 +; CHECK-NEXT: movapd %xmm1, %xmm0 +; CHECK-NEXT: ret +; UNSAFE: y_ogt_inverse: +; UNSAFE-NEXT: movsd {{[^,]*}}, %xmm1 +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ogt_inverse: +; FINITE-NEXT: movsd {{[^,]*}}, %xmm1 +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret +define double @y_ogt_inverse(double %x) nounwind { + %c = fcmp ogt double %x, -0.000000e+00 + %d = select i1 %c, double -0.000000e+00, double %x + ret double %d +} + +; CHECK: y_olt_inverse: +; CHECK-NEXT: movsd {{[^,]*}}, %xmm1 +; CHECK-NEXT: maxsd %xmm0, %xmm1 +; CHECK-NEXT: movapd %xmm1, %xmm0 +; CHECK-NEXT: ret +; UNSAFE: y_olt_inverse: +; UNSAFE-NEXT: movsd {{[^,]*}}, %xmm1 +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_olt_inverse: +; FINITE-NEXT: movsd {{[^,]*}}, %xmm1 +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret +define double @y_olt_inverse(double %x) nounwind { + %c = fcmp olt double %x, -0.000000e+00 + %d = select i1 %c, double -0.000000e+00, double %x + ret double %d +} + +; CHECK: y_oge: +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: y_oge: +; UNSAFE-NEXT: maxsd {{[^,]*}}, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_oge: +; FINITE-NEXT: maxsd {{[^,]*}}, %xmm0 +; FINITE-NEXT: ret +define double @y_oge(double %x) nounwind { + %c = fcmp oge double %x, -0.000000e+00 + %d = select i1 %c, double %x, double -0.000000e+00 + ret double %d +} + +; CHECK: y_ole: +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: y_ole: +; UNSAFE-NEXT: minsd {{[^,]*}}, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ole: +; FINITE-NEXT: minsd {{[^,]*}}, %xmm0 +; FINITE-NEXT: ret +define double @y_ole(double %x) nounwind { + %c = fcmp ole double %x, -0.000000e+00 + %d = select i1 %c, double %x, double -0.000000e+00 + ret double %d +} + +; CHECK: y_oge_inverse: +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: y_oge_inverse: +; UNSAFE-NEXT: movsd {{[^,]*}}, %xmm1 +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_oge_inverse: +; FINITE-NEXT: movsd {{[^,]*}}, %xmm1 +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret +define double @y_oge_inverse(double %x) nounwind { + %c = fcmp oge double %x, -0.000000e+00 + %d = select i1 %c, double -0.000000e+00, double %x + ret double %d +} + +; CHECK: y_ole_inverse: +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: y_ole_inverse: +; UNSAFE-NEXT: movsd {{[^,]*}}, %xmm1 +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ole_inverse: +; FINITE-NEXT: movsd {{[^,]*}}, %xmm1 +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret +define double @y_ole_inverse(double %x) nounwind { + %c = fcmp ole double %x, -0.000000e+00 + %d = select i1 %c, double -0.000000e+00, double %x + ret double %d +} + +; CHECK: y_ugt: +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: y_ugt: +; UNSAFE-NEXT: maxsd {{[^,]*}}, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ugt: +; FINITE-NEXT: maxsd {{[^,]*}}, %xmm0 +; FINITE-NEXT: ret +define double @y_ugt(double %x) nounwind { + %c = fcmp ugt double %x, -0.000000e+00 + %d = select i1 %c, double %x, double -0.000000e+00 + ret double %d +} + +; CHECK: y_ult: +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: y_ult: +; UNSAFE-NEXT: minsd {{[^,]*}}, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ult: +; FINITE-NEXT: minsd {{[^,]*}}, %xmm0 +; FINITE-NEXT: ret +define double @y_ult(double %x) nounwind { + %c = fcmp ult double %x, -0.000000e+00 + %d = select i1 %c, double %x, double -0.000000e+00 + ret double %d +} + +; CHECK: y_ugt_inverse: +; CHECK: ucomisd %xmm0, %xmm1 +; UNSAFE: y_ugt_inverse: +; UNSAFE-NEXT: movsd {{[^,]*}}, %xmm1 +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ugt_inverse: +; FINITE-NEXT: movsd {{[^,]*}}, %xmm1 +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret +define double @y_ugt_inverse(double %x) nounwind { + %c = fcmp ugt double %x, -0.000000e+00 + %d = select i1 %c, double -0.000000e+00, double %x + ret double %d +} + +; CHECK: y_ult_inverse: +; CHECK: ucomisd %xmm1, %xmm0 +; UNSAFE: y_ult_inverse: +; UNSAFE-NEXT: movsd {{[^,]*}}, %xmm1 +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ult_inverse: +; FINITE-NEXT: movsd {{[^,]*}}, %xmm1 +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret +define double @y_ult_inverse(double %x) nounwind { + %c = fcmp ult double %x, -0.000000e+00 + %d = select i1 %c, double -0.000000e+00, double %x + ret double %d +} + +; CHECK: y_uge: +; CHECK-NEXT: movsd {{[^,]*}}, %xmm1 +; CHECK-NEXT: maxsd %xmm0, %xmm1 +; CHECK-NEXT: movapd %xmm1, %xmm0 +; CHECK-NEXT: ret +; UNSAFE: y_uge: +; UNSAFE-NEXT: maxsd {{[^,]*}}, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_uge: +; FINITE-NEXT: maxsd {{[^,]*}}, %xmm0 +; FINITE-NEXT: ret +define double @y_uge(double %x) nounwind { + %c = fcmp uge double %x, -0.000000e+00 + %d = select i1 %c, double %x, double -0.000000e+00 + ret double %d +} + +; CHECK: y_ule: +; CHECK-NEXT: movsd {{[^,]*}}, %xmm1 +; CHECK-NEXT: minsd %xmm0, %xmm1 +; CHECK-NEXT: movapd %xmm1, %xmm0 +; CHECK-NEXT: ret +; UNSAFE: y_ule: +; UNSAFE-NEXT: minsd {{[^,]*}}, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ule: +; FINITE-NEXT: minsd {{[^,]*}}, %xmm0 +; FINITE-NEXT: ret +define double @y_ule(double %x) nounwind { + %c = fcmp ule double %x, -0.000000e+00 + %d = select i1 %c, double %x, double -0.000000e+00 + ret double %d +} + +; CHECK: y_uge_inverse: +; CHECK-NEXT: minsd {{[^,]*}}, %xmm0 +; CHECK-NEXT: ret +; UNSAFE: y_uge_inverse: +; UNSAFE-NEXT: movsd {{[^,]*}}, %xmm1 +; UNSAFE-NEXT: minsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_uge_inverse: +; FINITE-NEXT: movsd {{[^,]*}}, %xmm1 +; FINITE-NEXT: minsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret +define double @y_uge_inverse(double %x) nounwind { + %c = fcmp uge double %x, -0.000000e+00 + %d = select i1 %c, double -0.000000e+00, double %x + ret double %d +} + +; CHECK: y_ule_inverse: +; CHECK-NEXT: maxsd {{[^,]*}}, %xmm0 +; CHECK-NEXT: ret +; UNSAFE: y_ule_inverse: +; UNSAFE-NEXT: movsd {{[^,]*}}, %xmm1 +; UNSAFE-NEXT: maxsd %xmm0, %xmm1 +; UNSAFE-NEXT: movapd %xmm1, %xmm0 +; UNSAFE-NEXT: ret +; FINITE: y_ule_inverse: +; FINITE-NEXT: movsd {{[^,]*}}, %xmm1 +; FINITE-NEXT: maxsd %xmm0, %xmm1 +; FINITE-NEXT: movapd %xmm1, %xmm0 +; FINITE-NEXT: ret +define double @y_ule_inverse(double %x) nounwind { + %c = fcmp ule double %x, -0.000000e+00 + %d = select i1 %c, double -0.000000e+00, double %x + ret double %d +} ; Test a few more misc. cases. ; CHECK: clampTo3k_a: ; CHECK: minsd +; UNSAFE: clampTo3k_a: +; UNSAFE: minsd +; FINITE: clampTo3k_a: +; FINITE: minsd define double @clampTo3k_a(double %x) nounwind readnone { entry: %0 = fcmp ogt double %x, 3.000000e+03 ; <i1> [#uses=1] @@ -330,6 +842,10 @@ entry: ; CHECK: clampTo3k_b: ; CHECK: minsd +; UNSAFE: clampTo3k_b: +; UNSAFE: minsd +; FINITE: clampTo3k_b: +; FINITE: minsd define double @clampTo3k_b(double %x) nounwind readnone { entry: %0 = fcmp uge double %x, 3.000000e+03 ; <i1> [#uses=1] @@ -339,6 +855,10 @@ entry: ; CHECK: clampTo3k_c: ; CHECK: maxsd +; UNSAFE: clampTo3k_c: +; UNSAFE: maxsd +; FINITE: clampTo3k_c: +; FINITE: maxsd define double @clampTo3k_c(double %x) nounwind readnone { entry: %0 = fcmp olt double %x, 3.000000e+03 ; <i1> [#uses=1] @@ -348,6 +868,10 @@ entry: ; CHECK: clampTo3k_d: ; CHECK: maxsd +; UNSAFE: clampTo3k_d: +; UNSAFE: maxsd +; FINITE: clampTo3k_d: +; FINITE: maxsd define double @clampTo3k_d(double %x) nounwind readnone { entry: %0 = fcmp ule double %x, 3.000000e+03 ; <i1> [#uses=1] @@ -357,6 +881,10 @@ entry: ; CHECK: clampTo3k_e: ; CHECK: maxsd +; UNSAFE: clampTo3k_e: +; UNSAFE: maxsd +; FINITE: clampTo3k_e: +; FINITE: maxsd define double @clampTo3k_e(double %x) nounwind readnone { entry: %0 = fcmp olt double %x, 3.000000e+03 ; <i1> [#uses=1] @@ -366,6 +894,10 @@ entry: ; CHECK: clampTo3k_f: ; CHECK: maxsd +; UNSAFE: clampTo3k_f: +; UNSAFE: maxsd +; FINITE: clampTo3k_f: +; FINITE: maxsd define double @clampTo3k_f(double %x) nounwind readnone { entry: %0 = fcmp ule double %x, 3.000000e+03 ; <i1> [#uses=1] @@ -375,6 +907,10 @@ entry: ; CHECK: clampTo3k_g: ; CHECK: minsd +; UNSAFE: clampTo3k_g: +; UNSAFE: minsd +; FINITE: clampTo3k_g: +; FINITE: minsd define double @clampTo3k_g(double %x) nounwind readnone { entry: %0 = fcmp ogt double %x, 3.000000e+03 ; <i1> [#uses=1] @@ -384,6 +920,10 @@ entry: ; CHECK: clampTo3k_h: ; CHECK: minsd +; UNSAFE: clampTo3k_h: +; UNSAFE: minsd +; FINITE: clampTo3k_h: +; FINITE: minsd define double @clampTo3k_h(double %x) nounwind readnone { entry: %0 = fcmp uge double %x, 3.000000e+03 ; <i1> [#uses=1] diff --git a/test/CodeGen/X86/sse3.ll b/test/CodeGen/X86/sse3.ll index b2af7c9..921161e 100644 --- a/test/CodeGen/X86/sse3.ll +++ b/test/CodeGen/X86/sse3.ll @@ -144,10 +144,9 @@ define void @t9(<4 x float>* %r, <2 x i32>* %A) nounwind { store <4 x float> %tmp13, <4 x float>* %r ret void ; X64: t9: -; X64: movsd (%rsi), %xmm0 -; X64: movaps (%rdi), %xmm1 -; X64: movlhps %xmm0, %xmm1 -; X64: movaps %xmm1, (%rdi) +; X64: movaps (%rdi), %xmm0 +; X64: movhps (%rsi), %xmm0 +; X64: movaps %xmm0, (%rdi) ; X64: ret } diff --git a/test/CodeGen/X86/stack-align.ll b/test/CodeGen/X86/stack-align.ll index cb65e9b..e971ef7 100644 --- a/test/CodeGen/X86/stack-align.ll +++ b/test/CodeGen/X86/stack-align.ll @@ -1,4 +1,4 @@ -; RUN: llc < %s -relocation-model=static -mcpu=yonah | grep {andpd.*4(%esp), %xmm} +; RUN: llc < %s -relocation-model=static -realign-stack=1 -mcpu=yonah | FileCheck %s ; The double argument is at 4(esp) which is 16-byte aligned, allowing us to ; fold the load into the andpd. @@ -12,6 +12,7 @@ entry: %tmp = getelementptr { double, double }* %z, i32 0, i32 0 ; <double*> [#uses=1] %tmp1 = load double* %tmp, align 8 ; <double> [#uses=1] %tmp2 = tail call double @fabs( double %tmp1 ) ; <double> [#uses=1] + ; CHECK: andpd{{.*}}4(%esp), %xmm %tmp3 = load double* @G, align 16 ; <double> [#uses=1] %tmp4 = tail call double @fabs( double %tmp3 ) ; <double> [#uses=1] %tmp6 = fadd double %tmp4, %tmp2 ; <double> [#uses=1] @@ -19,4 +20,20 @@ entry: ret void } +define void @test2() alignstack(16) { +entry: + ; CHECK: andl{{.*}}$-16, %esp + ret void +} + +; Use a call to force a spill. +define <2 x double> @test3(<2 x double> %x, <2 x double> %y) alignstack(32) { +entry: + ; CHECK: andl{{.*}}$-32, %esp + call void @test2() + %A = mul <2 x double> %x, %y + ret <2 x double> %A +} + declare double @fabs(double) + diff --git a/test/CodeGen/X86/store_op_load_fold.ll b/test/CodeGen/X86/store_op_load_fold.ll index 66d0e47..6e47eb3 100644 --- a/test/CodeGen/X86/store_op_load_fold.ll +++ b/test/CodeGen/X86/store_op_load_fold.ll @@ -4,7 +4,7 @@ @X = internal global i16 0 ; <i16*> [#uses=2] -define void @foo() { +define void @foo() nounwind { %tmp.0 = load i16* @X ; <i16> [#uses=1] %tmp.3 = add i16 %tmp.0, 329 ; <i16> [#uses=1] store i16 %tmp.3, i16* @X diff --git a/test/CodeGen/X86/store_op_load_fold2.ll b/test/CodeGen/X86/store_op_load_fold2.ll index 0ccfe47..46e59e9 100644 --- a/test/CodeGen/X86/store_op_load_fold2.ll +++ b/test/CodeGen/X86/store_op_load_fold2.ll @@ -1,21 +1,12 @@ -; RUN: llc < %s -march=x86 -x86-asm-syntax=intel | \ -; RUN: grep {and DWORD PTR} | count 2 +; RUN: llc < %s -march=x86 -x86-asm-syntax=intel | FileCheck %s target datalayout = "e-p:32:32" %struct.Macroblock = type { i32, i32, i32, i32, i32, [8 x i32], %struct.Macroblock*, %struct.Macroblock*, i32, [2 x [4 x [4 x [2 x i32]]]], [16 x i8], [16 x i8], i32, i64, [4 x i32], [4 x i32], i64, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i16, double, i32, i32, i32, i32, i32, i32, i32, i32, i32 } -define internal fastcc i32 @dct_chroma(i32 %uv, i32 %cr_cbp) { -entry: - br i1 true, label %cond_true2732.preheader, label %cond_true129 -cond_true129: ; preds = %entry - ret i32 0 +define internal fastcc i32 @dct_chroma(i32 %uv, i32 %cr_cbp) nounwind { cond_true2732.preheader: ; preds = %entry %tmp2666 = getelementptr %struct.Macroblock* null, i32 0, i32 13 ; <i64*> [#uses=2] %tmp2674 = trunc i32 0 to i8 ; <i8> [#uses=1] - br i1 true, label %cond_true2732.preheader.split.us, label %cond_true2732.preheader.split -cond_true2732.preheader.split.us: ; preds = %cond_true2732.preheader - br i1 true, label %cond_true2732.outer.us.us, label %cond_true2732.outer.us -cond_true2732.outer.us.us: ; preds = %cond_true2732.preheader.split.us %tmp2667.us.us = load i64* %tmp2666 ; <i64> [#uses=1] %tmp2670.us.us = load i64* null ; <i64> [#uses=1] %shift.upgrd.1 = zext i8 %tmp2674 to i64 ; <i64> [#uses=1] @@ -24,11 +15,10 @@ cond_true2732.outer.us.us: ; preds = %cond_true2732.preheader.split %tmp2676.us.us = and i64 %tmp2667.us.us, %tmp2675not.us.us ; <i64> [#uses=1] store i64 %tmp2676.us.us, i64* %tmp2666 ret i32 0 -cond_true2732.outer.us: ; preds = %cond_true2732.preheader.split.us - ret i32 0 -cond_true2732.preheader.split: ; preds = %cond_true2732.preheader - ret i32 0 -cond_next2752: ; No predecessors! - ret i32 0 + +; CHECK: and {{E..}}, DWORD PTR [360] +; CHECK: and DWORD PTR [356], {{E..}} +; CHECK: mov DWORD PTR [360], {{E..}} + } diff --git a/test/CodeGen/X86/trunc-to-bool.ll b/test/CodeGen/X86/trunc-to-bool.ll index bfab1ae..6062084 100644 --- a/test/CodeGen/X86/trunc-to-bool.ll +++ b/test/CodeGen/X86/trunc-to-bool.ll @@ -3,13 +3,14 @@ ; value and as the operand of a branch. ; RUN: llc < %s -march=x86 | FileCheck %s -define i1 @test1(i32 %X) zeroext { +define i1 @test1(i32 %X) zeroext nounwind { %Y = trunc i32 %X to i1 ret i1 %Y } +; CHECK: test1: ; CHECK: andl $1, %eax -define i1 @test2(i32 %val, i32 %mask) { +define i1 @test2(i32 %val, i32 %mask) nounwind { entry: %shifted = ashr i32 %val, %mask %anded = and i32 %shifted, 1 @@ -20,9 +21,10 @@ ret_true: ret_false: ret i1 false } -; CHECK: testb $1, %al +; CHECK: test2: +; CHECK: btl %eax -define i32 @test3(i8* %ptr) { +define i32 @test3(i8* %ptr) nounwind { %val = load i8* %ptr %tmp = trunc i8 %val to i1 br i1 %tmp, label %cond_true, label %cond_false @@ -31,9 +33,10 @@ cond_true: cond_false: ret i32 42 } -; CHECK: testb $1, %al +; CHECK: test3: +; CHECK: testb $1, (%eax) -define i32 @test4(i8* %ptr) { +define i32 @test4(i8* %ptr) nounwind { %tmp = ptrtoint i8* %ptr to i1 br i1 %tmp, label %cond_true, label %cond_false cond_true: @@ -41,9 +44,10 @@ cond_true: cond_false: ret i32 42 } -; CHECK: testb $1, %al +; CHECK: test4: +; CHECK: testb $1, 4(%esp) -define i32 @test6(double %d) { +define i32 @test5(double %d) nounwind { %tmp = fptosi double %d to i1 br i1 %tmp, label %cond_true, label %cond_false cond_true: @@ -51,4 +55,5 @@ cond_true: cond_false: ret i32 42 } +; CHECK: test5: ; CHECK: testb $1 diff --git a/test/CodeGen/X86/vec_cast.ll b/test/CodeGen/X86/vec_cast.ll index 1f899b3..6f18d13 100644 --- a/test/CodeGen/X86/vec_cast.ll +++ b/test/CodeGen/X86/vec_cast.ll @@ -31,11 +31,10 @@ define <1 x i32> @f(<1 x i16> %a) nounwind { ret <1 x i32> %c } -; TODO: Legalize doesn't yet handle this. -;define <8 x i16> @g(<8 x i32> %a) nounwind { -; %c = trunc <8 x i32> %a to <8 x i16> -; ret <8 x i16> %c -;} +define <8 x i16> @g(<8 x i32> %a) nounwind { + %c = trunc <8 x i32> %a to <8 x i16> + ret <8 x i16> %c +} define <3 x i16> @h(<3 x i32> %a) nounwind { %c = trunc <3 x i32> %a to <3 x i16> @@ -46,3 +45,12 @@ define <1 x i16> @i(<1 x i32> %a) nounwind { %c = trunc <1 x i32> %a to <1 x i16> ret <1 x i16> %c } + +; PR6438 +define void @__OpenCL_math_kernel4_kernel() nounwind { + %tmp12.i = and <4 x i32> zeroinitializer, <i32 2139095040, i32 2139095040, i32 2139095040, i32 2139095040> ; <<4 x i32>> [#uses=1] + %cmp13.i = icmp eq <4 x i32> %tmp12.i, <i32 2139095040, i32 2139095040, i32 2139095040, i32 2139095040> ; <<4 x i1>> [#uses=2] + %cmp.ext14.i = sext <4 x i1> %cmp13.i to <4 x i32> ; <<4 x i32>> [#uses=0] + %tmp2110.i = and <4 x i1> %cmp13.i, zeroinitializer ; <<4 x i1>> [#uses=0] + ret void +} diff --git a/test/CodeGen/X86/vec_insert.ll b/test/CodeGen/X86/vec_insert.ll index a7274a9..4e5d445 100644 --- a/test/CodeGen/X86/vec_insert.ll +++ b/test/CodeGen/X86/vec_insert.ll @@ -1,7 +1,7 @@ ; RUN: llc < %s -march=x86 -mattr=+sse2,-sse41 | grep movss | count 1 ; RUN: llc < %s -march=x86 -mattr=+sse2,-sse41 | not grep pinsrw -define void @test(<4 x float>* %F, i32 %I) { +define void @test(<4 x float>* %F, i32 %I) nounwind { %tmp = load <4 x float>* %F ; <<4 x float>> [#uses=1] %f = sitofp i32 %I to float ; <float> [#uses=1] %tmp1 = insertelement <4 x float> %tmp, float %f, i32 0 ; <<4 x float>> [#uses=2] @@ -10,7 +10,7 @@ define void @test(<4 x float>* %F, i32 %I) { ret void } -define void @test2(<4 x float>* %F, i32 %I, float %g) { +define void @test2(<4 x float>* %F, i32 %I, float %g) nounwind { %tmp = load <4 x float>* %F ; <<4 x float>> [#uses=1] %f = sitofp i32 %I to float ; <float> [#uses=1] %tmp1 = insertelement <4 x float> %tmp, float %f, i32 2 ; <<4 x float>> [#uses=1] diff --git a/test/CodeGen/X86/vec_shuffle-36.ll b/test/CodeGen/X86/vec_shuffle-36.ll index 8a93a7e..1ea37c8 100644 --- a/test/CodeGen/X86/vec_shuffle-36.ll +++ b/test/CodeGen/X86/vec_shuffle-36.ll @@ -1,9 +1,16 @@ -; RUN: llc < %s -march=x86 -mattr=sse41 -o %t -; RUN: grep pshufb %t | count 1 - +; RUN: llc < %s -march=x86-64 -mattr=sse41 | FileCheck %s define <8 x i16> @shuf6(<8 x i16> %T0, <8 x i16> %T1) nounwind readnone { +; CHECK: pshufb +; CHECK-NOT: pshufb +; CHECK: ret entry: - %tmp9 = shufflevector <8 x i16> %T0, <8 x i16> %T1, <8 x i32> < i32 3, i32 2, i32 0, i32 2, i32 1, i32 5, i32 6 , i32 undef > - ret <8 x i16> %tmp9 + %tmp9 = shufflevector <8 x i16> %T0, <8 x i16> %T1, <8 x i32> < i32 3, i32 2, i32 0, i32 2, i32 1, i32 5, i32 6 , i32 undef > + ret <8 x i16> %tmp9 } + +define <8 x i16> @shuf7(<8 x i16> %t0) { +; CHECK: pshufd + %tmp10 = shufflevector <8 x i16> %t0, <8 x i16> undef, <8 x i32> < i32 undef, i32 2, i32 2, i32 2, i32 2, i32 2, i32 undef, i32 undef > + ret <8 x i16> %tmp10 +}
\ No newline at end of file diff --git a/test/CodeGen/X86/vec_ss_load_fold.ll b/test/CodeGen/X86/vec_ss_load_fold.ll index b1613fb..c8b2927 100644 --- a/test/CodeGen/X86/vec_ss_load_fold.ll +++ b/test/CodeGen/X86/vec_ss_load_fold.ll @@ -1,6 +1,4 @@ -; RUN: llc < %s -march=x86 -mattr=+sse,+sse2 -o %t -; RUN: grep minss %t | grep CPI | count 2 -; RUN: grep CPI %t | not grep movss +; RUN: llc < %s -march=x86 -mattr=+sse,+sse2,+sse41 | FileCheck %s target datalayout = "e-p:32:32" target triple = "i686-apple-darwin8.7.2" @@ -17,6 +15,10 @@ define i16 @test1(float %f) nounwind { %tmp.upgrd.1 = tail call i32 @llvm.x86.sse.cvttss2si( <4 x float> %tmp59 ) ; <i32> [#uses=1] %tmp69 = trunc i32 %tmp.upgrd.1 to i16 ; <i16> [#uses=1] ret i16 %tmp69 +; CHECK: test1: +; CHECK: subss LCPI1_ +; CHECK: mulss LCPI1_ +; CHECK: minss LCPI1_ } define i16 @test2(float %f) nounwind { @@ -28,6 +30,10 @@ define i16 @test2(float %f) nounwind { %tmp = tail call i32 @llvm.x86.sse.cvttss2si( <4 x float> %tmp59 ) ; <i32> [#uses=1] %tmp69 = trunc i32 %tmp to i16 ; <i16> [#uses=1] ret i16 %tmp69 +; CHECK: test2: +; CHECK: addss LCPI2_ +; CHECK: mulss LCPI2_ +; CHECK: minss LCPI2_ } declare <4 x float> @llvm.x86.sse.sub.ss(<4 x float>, <4 x float>) @@ -39,3 +45,28 @@ declare <4 x float> @llvm.x86.sse.min.ss(<4 x float>, <4 x float>) declare <4 x float> @llvm.x86.sse.max.ss(<4 x float>, <4 x float>) declare i32 @llvm.x86.sse.cvttss2si(<4 x float>) + + +declare <4 x float> @llvm.x86.sse41.round.ss(<4 x float>, <4 x float>, i32) +declare <4 x float> @f() + +define <4 x float> @test3(<4 x float> %A, float *%b, i32 %C) nounwind { + %a = load float *%b + %B = insertelement <4 x float> undef, float %a, i32 0 + %X = call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %A, <4 x float> %B, i32 4) + ret <4 x float> %X +; CHECK: test3: +; CHECK: roundss $4, (%eax), %xmm0 +} + +define <4 x float> @test4(<4 x float> %A, float *%b, i32 %C) nounwind { + %a = load float *%b + %B = insertelement <4 x float> undef, float %a, i32 0 + %q = call <4 x float> @f() + %X = call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %q, <4 x float> %B, i32 4) + ret <4 x float> %X +; CHECK: test4: +; CHECK: movss (%eax), %xmm +; CHECK: call +; CHECK: roundss $4, %xmm{{.*}}, %xmm0 +} diff --git a/test/CodeGen/X86/xor-icmp.ll b/test/CodeGen/X86/xor-icmp.ll index a6bdb13..2d75c5d 100644 --- a/test/CodeGen/X86/xor-icmp.ll +++ b/test/CodeGen/X86/xor-icmp.ll @@ -1,5 +1,6 @@ ; RUN: llc < %s -march=x86 | FileCheck %s -check-prefix=X32 ; RUN: llc < %s -march=x86-64 | FileCheck %s -check-prefix=X64 +; rdar://7367229 define i32 @t(i32 %a, i32 %b) nounwind ssp { entry: @@ -34,3 +35,33 @@ bb1: ; preds = %entry declare i32 @foo(...) declare i32 @bar(...) + +define i32 @t2(i32 %x, i32 %y) nounwind ssp { +; X32: t2: +; X32: cmpl +; X32: sete +; X32: cmpl +; X32: sete +; X32-NOT: xor +; X32: je + +; X64: t2: +; X64: testl +; X64: sete +; X64: testl +; X64: sete +; X64-NOT: xor +; X64: je +entry: + %0 = icmp eq i32 %x, 0 ; <i1> [#uses=1] + %1 = icmp eq i32 %y, 0 ; <i1> [#uses=1] + %2 = xor i1 %1, %0 ; <i1> [#uses=1] + br i1 %2, label %bb, label %return + +bb: ; preds = %entry + %3 = tail call i32 (...)* @foo() nounwind ; <i32> [#uses=0] + ret i32 undef + +return: ; preds = %entry + ret i32 undef +} diff --git a/test/CodeGen/XCore/2010-02-25-LSR-Crash.ll b/test/CodeGen/XCore/2010-02-25-LSR-Crash.ll new file mode 100644 index 0000000..6ad9a73 --- /dev/null +++ b/test/CodeGen/XCore/2010-02-25-LSR-Crash.ll @@ -0,0 +1,26 @@ +; RUN: llc < %s -march=xcore +target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:64:64-v128:128:128-a0:0:32-n32" +target triple = "xcore-xmos-elf" + +%0 = type { i32 } +%struct.dwarf_fde = type <{ i32, i32, [0 x i8] }> +%struct.object = type { i8*, i8*, i8*, %union.anon, %0, %struct.object* } +%union.anon = type { %struct.dwarf_fde* } + +define %struct.dwarf_fde* @search_object(%struct.object* %ob, i8* %pc) { +entry: + br i1 undef, label %bb3.i15.i.i, label %bb2 + +bb3.i15.i.i: ; preds = %bb3.i15.i.i, %entry + %indvar.i.i.i = phi i32 [ %indvar.next.i.i.i, %bb3.i15.i.i ], [ 0, %entry ] ; <i32> [#uses=2] + %tmp137 = sub i32 0, %indvar.i.i.i ; <i32> [#uses=1] + %scevgep13.i.i.i = getelementptr i32* undef, i32 %tmp137 ; <i32*> [#uses=2] + %scevgep1314.i.i.i = bitcast i32* %scevgep13.i.i.i to %struct.dwarf_fde** ; <%struct.dwarf_fde**> [#uses=1] + %0 = load %struct.dwarf_fde** %scevgep1314.i.i.i, align 4 ; <%struct.dwarf_fde*> [#uses=0] + store i32 undef, i32* %scevgep13.i.i.i + %indvar.next.i.i.i = add i32 %indvar.i.i.i, 1 ; <i32> [#uses=1] + br label %bb3.i15.i.i + +bb2: ; preds = %entry + ret %struct.dwarf_fde* undef +} diff --git a/test/CodeGen/XCore/switch.ll b/test/CodeGen/XCore/switch.ll new file mode 100644 index 0000000..9cc27f2 --- /dev/null +++ b/test/CodeGen/XCore/switch.ll @@ -0,0 +1,24 @@ +; RUN: llc -march=xcore < %s | FileCheck %s + +define i32 @switch(i32 %i) { +entry: + switch i32 %i, label %default [ + i32 0, label %bb0 + i32 1, label %bb1 + i32 2, label %bb2 + i32 3, label %bb3 + ] +; CHECK-NOT: shl +; CHECK: bru +; CHECK: .jmptable +bb0: + ret i32 0 +bb1: + ret i32 1 +bb2: + ret i32 2 +bb3: + ret i32 3 +default: + ret i32 4 +} diff --git a/test/CodeGen/XCore/switch_long.ll b/test/CodeGen/XCore/switch_long.ll new file mode 100644 index 0000000..30c9e3d --- /dev/null +++ b/test/CodeGen/XCore/switch_long.ll @@ -0,0 +1,132 @@ +; RUN: llc -march=xcore < %s | FileCheck %s + +define i32 @switch(i32 %i) { +entry: + switch i32 %i, label %default [ + i32 0, label %bb0 + i32 1, label %bb1 + i32 2, label %bb2 + i32 3, label %bb3 + i32 4, label %bb4 + i32 5, label %bb5 + i32 6, label %bb6 + i32 7, label %bb7 + i32 8, label %bb8 + i32 9, label %bb9 + i32 10, label %bb10 + i32 11, label %bb11 + i32 12, label %bb12 + i32 13, label %bb13 + i32 14, label %bb14 + i32 15, label %bb15 + i32 16, label %bb16 + i32 17, label %bb17 + i32 18, label %bb18 + i32 19, label %bb19 + i32 20, label %bb20 + i32 21, label %bb21 + i32 22, label %bb22 + i32 23, label %bb23 + i32 24, label %bb24 + i32 25, label %bb25 + i32 26, label %bb26 + i32 27, label %bb27 + i32 28, label %bb28 + i32 29, label %bb29 + i32 30, label %bb30 + i32 31, label %bb31 + i32 32, label %bb32 + i32 33, label %bb33 + i32 34, label %bb34 + i32 35, label %bb35 + i32 36, label %bb36 + i32 37, label %bb37 + i32 38, label %bb38 + i32 39, label %bb39 + ] +; CHECK: shl +; CHECK: bru +; CHECK: .jmptable +bb0: + ret i32 0 +bb1: + ret i32 1 +bb2: + ret i32 2 +bb3: + ret i32 3 +bb4: + ret i32 4 +bb5: + ret i32 5 +bb6: + ret i32 6 +bb7: + ret i32 7 +bb8: + ret i32 8 +bb9: + ret i32 9 +bb10: + ret i32 0 +bb11: + ret i32 1 +bb12: + ret i32 2 +bb13: + ret i32 3 +bb14: + ret i32 4 +bb15: + ret i32 5 +bb16: + ret i32 6 +bb17: + ret i32 7 +bb18: + ret i32 8 +bb19: + ret i32 9 +bb20: + ret i32 0 +bb21: + ret i32 1 +bb22: + ret i32 2 +bb23: + ret i32 3 +bb24: + ret i32 4 +bb25: + ret i32 5 +bb26: + ret i32 6 +bb27: + ret i32 7 +bb28: + ret i32 8 +bb29: + ret i32 9 +bb30: + ret i32 0 +bb31: + ret i32 1 +bb32: + ret i32 2 +bb33: + ret i32 3 +bb34: + ret i32 4 +bb35: + ret i32 5 +bb36: + ret i32 6 +bb37: + ret i32 7 +bb38: + ret i32 8 +bb39: + ret i32 9 +default: + ret i32 0 +} diff --git a/test/DebugInfo/2009-02-27-licm.ll b/test/DebugInfo/2009-02-27-licm.ll deleted file mode 100644 index b490a28..0000000 --- a/test/DebugInfo/2009-02-27-licm.ll +++ /dev/null @@ -1,83 +0,0 @@ -;RUN: opt < %s -licm -S | grep {load } | count 4 -; ModuleID = '2009-02-27-licm.bc' -target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" -target triple = "i386-pc-linux-gnu" - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, { }*, i32, i8*, i8*, i8*, i1, i1, i8*, i32 } - %llvm.dbg.composite.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, { }*, { }*, i32 } - %llvm.dbg.subprogram.type = type { i32, { }*, { }*, i8*, i8*, i8*, { }*, i32, { }*, i1, i1 } -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str = internal constant [12 x i8] c"mt19937ar.c\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@.str1 = internal constant [58 x i8] c"/developer2/home5/youxiangc/work/project/pr965/test-licm/\00", section "llvm.metadata" ; <[58 x i8]*> [#uses=1] -@.str2 = internal constant [52 x i8] c"4.2.1 (Based on Apple Inc. build 5641) (LLVM build)\00", section "llvm.metadata" ; <[52 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { i32 458769, { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to { }*), i32 1, i8* getelementptr ([12 x i8]* @.str, i32 0, i32 0), i8* getelementptr ([58 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 true, i1 false, i8* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@mti = internal global i32 625 ; <i32*> [#uses=7] -@.str5 = internal constant [18 x i8] c"long unsigned int\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.basictype6 = internal constant %llvm.dbg.basictype.type { i32 458788, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* getelementptr ([18 x i8]* @.str5, i32 0, i32 0), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 7 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array = internal constant [2 x { }*] [{ }* null, { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to { }*)], section "llvm.metadata" ; <[2 x { }*]*> [#uses=1] -@llvm.dbg.composite = internal constant %llvm.dbg.composite.type { i32 458773, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* null, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 0, i64 0, i64 0, i64 0, i32 0, { }* null, { }* bitcast ([2 x { }*]* @llvm.dbg.array to { }*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str7 = internal constant [13 x i8] c"init_genrand\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.subprogram = internal constant %llvm.dbg.subprogram.type { i32 458798, { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to { }*), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* getelementptr ([13 x i8]* @.str7, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str7, i32 0, i32 0), i8* null, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 13, { }* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite to { }*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@mt = internal global [624 x i32] zeroinitializer, align 32 ; <[624 x i32]*> [#uses=4] - -define void @init_genrand(i32 %s) nounwind { -entry: - tail call void @llvm.dbg.func.start({ }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*)) - tail call void @llvm.dbg.stoppoint(i32 14, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - store i32 %s, i32* getelementptr ([624 x i32]* @mt, i32 0, i32 0), align 32 - tail call void @llvm.dbg.stoppoint(i32 15, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - store i32 1, i32* @mti - tail call void @llvm.dbg.stoppoint(i32 15, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - %0 = load i32* @mti, align 4 ; <i32> [#uses=1] - %1 = icmp sgt i32 %0, 623 ; <i1> [#uses=1] - br i1 %1, label %return, label %bb.nph - -bb.nph: ; preds = %entry - br label %bb - -bb: ; preds = %bb1, %bb.nph - %storemerge1 = phi i32 [ %16, %bb1 ], [ 1, %bb.nph ] ; <i32> [#uses=0] - tail call void @llvm.dbg.stoppoint(i32 16, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - %2 = load i32* @mti, align 4 ; <i32> [#uses=3] - %3 = add i32 %2, -1 ; <i32> [#uses=1] - %4 = getelementptr [624 x i32]* @mt, i32 0, i32 %3 ; <i32*> [#uses=1] - %5 = load i32* %4, align 4 ; <i32> [#uses=1] - %6 = add i32 %2, -1 ; <i32> [#uses=1] - %7 = getelementptr [624 x i32]* @mt, i32 0, i32 %6 ; <i32*> [#uses=1] - %8 = load i32* %7, align 4 ; <i32> [#uses=1] - %9 = lshr i32 %8, 30 ; <i32> [#uses=1] - %10 = xor i32 %9, %5 ; <i32> [#uses=1] - %11 = mul i32 %10, 1812433253 ; <i32> [#uses=1] - %12 = load i32* @mti, align 4 ; <i32> [#uses=1] - %13 = add i32 %11, %12 ; <i32> [#uses=1] - %14 = getelementptr [624 x i32]* @mt, i32 0, i32 %2 ; <i32*> [#uses=1] - store i32 %13, i32* %14, align 4 - tail call void @llvm.dbg.stoppoint(i32 15, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - %15 = load i32* @mti, align 4 ; <i32> [#uses=1] - %16 = add i32 %15, 1 ; <i32> [#uses=2] - br label %bb1 - -bb1: ; preds = %bb - %storemerge = phi i32 [ %16, %bb ] ; <i32> [#uses=1] - store i32 %storemerge, i32* @mti - tail call void @llvm.dbg.stoppoint(i32 15, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - %17 = load i32* @mti, align 4 ; <i32> [#uses=1] - %18 = icmp sgt i32 %17, 623 ; <i1> [#uses=1] - br i1 %18, label %bb1.return_crit_edge, label %bb - -bb1.return_crit_edge: ; preds = %bb1 - br label %return - -return: ; preds = %bb1.return_crit_edge, %entry - tail call void @llvm.dbg.stoppoint(i32 25, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - tail call void @llvm.dbg.region.end({ }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*)) - ret void -} - -declare void @llvm.dbg.func.start({ }*) nounwind - -declare void @llvm.dbg.stoppoint(i32, i32, { }*) nounwind - -declare void @llvm.dbg.region.end({ }*) nounwind diff --git a/test/DebugInfo/2009-03-03-cheapdse.ll b/test/DebugInfo/2009-03-03-cheapdse.ll deleted file mode 100644 index 9f47f16..0000000 --- a/test/DebugInfo/2009-03-03-cheapdse.ll +++ /dev/null @@ -1,80 +0,0 @@ -; RUN: opt < %s -instcombine -S | grep store | count 5 -; ModuleID = '<stdin>' -target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" -target triple = "i386-apple-darwin9.6" - type { } ; type %0 - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, %0*, i32, i8*, i8*, i8*, i1, i1, i8*, i32 } - %llvm.dbg.composite.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, %0*, %0*, i32 } - %llvm.dbg.derivedtype.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, %0* } - %llvm.dbg.subprogram.type = type { i32, %0*, %0*, i8*, i8*, i8*, %0*, i32, %0*, i1, i1 } - %struct.Matrix = type { float*, i32, i32, i32, i32 } -@llvm.dbg.compile_units = internal constant %llvm.dbg.anchor.type { i32 458752, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str = internal constant [15 x i8] c"himenobmtxpa.c\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@.str1 = internal constant [74 x i8] c"/Volumes/MacOS9/gcc/llvm/projects/llvm-test/SingleSource/Benchmarks/Misc/\00", section "llvm.metadata" ; <[74 x i8]*> [#uses=1] -@.str2 = internal constant [52 x i8] c"4.2.1 (Based on Apple Inc. build 5641) (LLVM build)\00", section "llvm.metadata" ; <[52 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 1, i8* getelementptr ([15 x i8]* @.str, i32 0, i32 0), i8* getelementptr ([74 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 true, i1 false, i8* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str3 = internal constant [6 x i8] c"float\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str3, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 4 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@.str5 = internal constant [4 x i8] c"int\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.basictype6 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str5, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 5 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.subprograms = internal constant %llvm.dbg.anchor.type { i32 458752, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str90 = internal constant [4 x i8] c"Mat\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.derivedtype92 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str93 = internal constant [2 x i8] c"m\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.derivedtype94 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([2 x i8]* @.str93, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 46, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype92 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str95 = internal constant [6 x i8] c"mnums\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.derivedtype96 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str95, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 47, i64 32, i64 32, i64 32, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str97 = internal constant [6 x i8] c"mrows\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.derivedtype98 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str97, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 48, i64 32, i64 32, i64 64, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str99 = internal constant [6 x i8] c"mcols\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.derivedtype100 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str99, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 49, i64 32, i64 32, i64 96, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str101 = internal constant [6 x i8] c"mdeps\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.derivedtype102 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str101, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 50, i64 32, i64 32, i64 128, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array103 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype94 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype96 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype98 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype100 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype102 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite104 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str90, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 45, i64 160, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array103 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str105 = internal constant [7 x i8] c"Matrix\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@llvm.dbg.derivedtype106 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str105, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 54, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite104 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype107 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype106 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array108 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype107 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite109 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array108 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str110 = internal constant [7 x i8] c"newMat\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@llvm.dbg.subprogram111 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str110, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str110, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 195, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite109 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.used = appending global [1 x i8*] [i8* bitcast (i32 (%struct.Matrix*, i32, i32, i32, i32)* @newMat to i8*)], section "llvm.metadata" ; <[1 x i8*]*> [#uses=0] - -define i32 @newMat(%struct.Matrix* %Mat, i32 %mnums, i32 %mrows, i32 %mcols, i32 %mdeps) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram111 to %0*)) - call void @llvm.dbg.stoppoint(i32 196, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %struct.Matrix* %Mat, i32 0, i32 1 ; <i32*> [#uses=1] - store i32 %mnums, i32* %0, align 4 - call void @llvm.dbg.stoppoint(i32 197, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %1 = getelementptr %struct.Matrix* %Mat, i32 0, i32 2 ; <i32*> [#uses=1] - store i32 %mrows, i32* %1, align 4 - call void @llvm.dbg.stoppoint(i32 198, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %2 = getelementptr %struct.Matrix* %Mat, i32 0, i32 3 ; <i32*> [#uses=1] - store i32 %mcols, i32* %2, align 4 - call void @llvm.dbg.stoppoint(i32 199, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %3 = getelementptr %struct.Matrix* %Mat, i32 0, i32 4 ; <i32*> [#uses=1] - store i32 %mdeps, i32* %3, align 4 - call void @llvm.dbg.stoppoint(i32 201, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %4 = mul i32 %mnums, %mrows ; <i32> [#uses=1] - %5 = mul i32 %4, %mcols ; <i32> [#uses=1] - %6 = mul i32 %5, %mdeps ; <i32> [#uses=1] - %7 = malloc float, i32 %6 ; <float*> [#uses=2] - %8 = getelementptr %struct.Matrix* %Mat, i32 0, i32 0 ; <float**> [#uses=1] - store float* %7, float** %8, align 4 - call void @llvm.dbg.stoppoint(i32 204, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %9 = icmp ne float* %7, null ; <i1> [#uses=1] - %10 = zext i1 %9 to i32 ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 204, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram111 to %0*)) - ret i32 %10 -} - -declare void @llvm.dbg.func.start(%0*) nounwind readnone - -declare void @llvm.dbg.stoppoint(i32, i32, %0*) nounwind readnone - -declare void @llvm.dbg.region.end(%0*) nounwind readnone diff --git a/test/DebugInfo/2009-03-05-gvn.ll b/test/DebugInfo/2009-03-05-gvn.ll deleted file mode 100644 index f363132..0000000 --- a/test/DebugInfo/2009-03-05-gvn.ll +++ /dev/null @@ -1,125 +0,0 @@ -; RUN: opt < %s -gvn -S | grep {load } | count 1 -; ModuleID = 'db2-before.bc' -target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" -target triple = "i386-pc-linux-gnu" - type { } ; type %0 - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, %0*, i32, i8*, i8*, i8*, i1, i1, i8*, i32 } - %llvm.dbg.composite.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, %0*, %0*, i32 } - %llvm.dbg.derivedtype.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, %0* } - %llvm.dbg.subprogram.type = type { i32, %0*, %0*, i8*, i8*, i8*, %0*, i32, %0*, i1, i1 } -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str = internal constant [12 x i8] c"mt19937ar.c\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@.str1 = internal constant [34 x i8] c"/developer/home2/zsth/test/debug/\00", section "llvm.metadata" ; <[34 x i8]*> [#uses=1] -@.str2 = internal constant [52 x i8] c"4.2.1 (Based on Apple Inc. build 5641) (LLVM build)\00", section "llvm.metadata" ; <[52 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 1, i8* getelementptr ([12 x i8]* @.str, i32 0, i32 0), i8* getelementptr ([34 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 true, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@mti = internal global i32 625 ; <i32*> [#uses=14] -@.str3 = internal constant [4 x i8] c"int\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str3, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 5 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@.str5 = internal constant [18 x i8] c"long unsigned int\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.basictype6 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str5, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 7 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str7 = internal constant [13 x i8] c"init_genrand\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.subprogram = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str7, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str7, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 58, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@mt = internal global [624 x i32] zeroinitializer, align 32 ; <[624 x i32]*> [#uses=29] -@llvm.dbg.derivedtype = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array9 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite10 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array9 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str11 = internal constant [14 x i8] c"init_by_array\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram12 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str11, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str11, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 77, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite10 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array23 = internal constant [1 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype6 to %0*)], section "llvm.metadata" ; <[1 x %0*]*> [#uses=1] -@llvm.dbg.composite24 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([1 x %0*]* @llvm.dbg.array23 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str25 = internal constant [14 x i8] c"genrand_int32\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram26 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str25, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str25, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 103, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite24 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@mag01.1661 = internal constant [2 x i32] [i32 0, i32 -1727483681] ; <[2 x i32]*> [#uses=3] -@.str35 = internal constant [9 x i8] c"long int\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@llvm.dbg.basictype36 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str35, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 5 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array37 = internal constant [1 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype36 to %0*)], section "llvm.metadata" ; <[1 x %0*]*> [#uses=1] -@llvm.dbg.composite38 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([1 x %0*]* @llvm.dbg.array37 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str39 = internal constant [14 x i8] c"genrand_int31\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram40 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str39, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str39, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 141, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite38 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str41 = internal constant [7 x i8] c"double\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@llvm.dbg.basictype42 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str41, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 64, i64 64, i64 0, i32 0, i32 4 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array43 = internal constant [1 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype42 to %0*)], section "llvm.metadata" ; <[1 x %0*]*> [#uses=1] -@llvm.dbg.composite44 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([1 x %0*]* @llvm.dbg.array43 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str45 = internal constant [14 x i8] c"genrand_real1\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram46 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str45, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str45, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 147, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite44 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str47 = internal constant [14 x i8] c"genrand_real2\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram48 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str47, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str47, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 154, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite44 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str49 = internal constant [14 x i8] c"genrand_real3\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram50 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str49, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str49, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 161, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite44 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str51 = internal constant [14 x i8] c"genrand_res53\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram52 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str51, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str51, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 168, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite44 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array57 = internal constant [1 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*)], section "llvm.metadata" ; <[1 x %0*]*> [#uses=1] -@llvm.dbg.composite58 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([1 x %0*]* @llvm.dbg.array57 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str59 = internal constant [5 x i8] c"main\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.subprogram60 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str59, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str59, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 175, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite58 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str69 = internal constant [32 x i8] c"1000 outputs of genrand_int32()\00" ; <[32 x i8]*> [#uses=1] -@.str70 = internal constant [7 x i8] c"%10lu \00" ; <[7 x i8]*> [#uses=1] -@.str71 = internal constant [33 x i8] c"\0A1000 outputs of genrand_real2()\00" ; <[33 x i8]*> [#uses=1] -@.str72 = internal constant [8 x i8] c"%10.8f \00" ; <[8 x i8]*> [#uses=1] - -define void @init_genrand(i32 %s) nounwind { -entry: - tail call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to %0*)) - tail call void @llvm.dbg.stoppoint(i32 59, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - store i32 %s, i32* getelementptr ([624 x i32]* @mt, i32 0, i32 0), align 32 - tail call void @llvm.dbg.stoppoint(i32 60, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - store i32 1, i32* @mti - tail call void @llvm.dbg.stoppoint(i32 60, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br i1 false, label %return, label %bb.nph - -bb.nph: ; preds = %entry - %mti.promoted = load i32* @mti ; <i32> [#uses=1] - br label %bb - -bb: ; preds = %bb1, %bb.nph - %indvar = phi i32 [ 0, %bb.nph ], [ %indvar.next, %bb1 ] ; <i32> [#uses=2] - %mti.tmp.0 = add i32 %indvar, %mti.promoted ; <i32> [#uses=5] - tail call void @llvm.dbg.stoppoint(i32 61, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = add i32 %mti.tmp.0, -1 ; <i32> [#uses=1] - %1 = getelementptr [624 x i32]* @mt, i32 0, i32 %0 ; <i32*> [#uses=1] - %2 = load i32* %1, align 4 ; <i32> [#uses=1] - %3 = add i32 %mti.tmp.0, -1 ; <i32> [#uses=1] - %4 = getelementptr [624 x i32]* @mt, i32 0, i32 %3 ; <i32*> [#uses=1] - %5 = load i32* %4, align 4 ; <i32> [#uses=1] - %6 = lshr i32 %5, 30 ; <i32> [#uses=1] - %7 = xor i32 %6, %2 ; <i32> [#uses=1] - %8 = mul i32 %7, 1812433253 ; <i32> [#uses=1] - %9 = add i32 %8, %mti.tmp.0 ; <i32> [#uses=1] - %10 = getelementptr [624 x i32]* @mt, i32 0, i32 %mti.tmp.0 ; <i32*> [#uses=1] - store i32 %9, i32* %10, align 4 - tail call void @llvm.dbg.stoppoint(i32 60, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %11 = add i32 %mti.tmp.0, 1 ; <i32> [#uses=2] - br label %bb1 - -bb1: ; preds = %bb - tail call void @llvm.dbg.stoppoint(i32 60, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %12 = icmp sgt i32 %11, 623 ; <i1> [#uses=1] - %indvar.next = add i32 %indvar, 1 ; <i32> [#uses=1] - br i1 %12, label %bb1.return_crit_edge, label %bb - -bb1.return_crit_edge: ; preds = %bb1 - store i32 %11, i32* @mti - br label %return - -return: ; preds = %bb1.return_crit_edge, %entry - tail call void @llvm.dbg.stoppoint(i32 70, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - tail call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to %0*)) - ret void -} - -declare void @llvm.dbg.func.start(%0*) nounwind - -declare void @llvm.dbg.stoppoint(i32, i32, %0*) nounwind - -declare void @llvm.dbg.region.end(%0*) nounwind - -declare i32 @puts(i8* nocapture) nounwind - -declare i32 @printf(i8* noalias nocapture, ...) nounwind - -declare i32 @putchar(i32) nounwind diff --git a/test/DebugInfo/deaddebuglabel.ll b/test/DebugInfo/deaddebuglabel.ll deleted file mode 100644 index a9af12b..0000000 --- a/test/DebugInfo/deaddebuglabel.ll +++ /dev/null @@ -1,62 +0,0 @@ -; RUN: llc %s -o - -O0 | grep "label" | count 8 -; PR2614 -; XFAIL: * - -target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-f80:32:32-v64:64:64-v128:128:128-a0:0:64" -target triple = "i686-pc-linux-gnu" - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, { }*, i32, i8*, i8*, i8* } - %llvm.dbg.compositetype.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, { }*, { }* } - %llvm.dbg.derivedtype.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, { }* } - %llvm.dbg.global_variable.type = type { i32, { }*, { }*, i8*, i8*, i8*, { }*, i32, { }*, i1, i1, { }* } - %llvm.dbg.subprogram.type = type { i32, { }*, { }*, i8*, i8*, i8*, { }*, i32, { }*, i1, i1 } - %llvm.dbg.variable.type = type { i32, { }*, i8*, { }*, i32, { }* } -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 393216, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@llvm.dbg.global_variables = linkonce constant %llvm.dbg.anchor.type { i32 393216, i32 52 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=0] -@llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { i32 393216, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str = internal constant [17 x i8] c"deaddebuglabel.d\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@.str1 = internal constant [50 x i8] c"/home/kamm/eigenes/projekte/llvmdc/llvmdc/mytests\00", section "llvm.metadata" ; <[50 x i8]*> [#uses=1] -@.str2 = internal constant [48 x i8] c"LLVMDC (http://www.dsource.org/projects/llvmdc)\00", section "llvm.metadata" ; <[48 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { - i32 393233, - { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to { }*), - i32 2, - i8* getelementptr ([17 x i8]* @.str, i32 0, i32 0), - i8* getelementptr ([50 x i8]* @.str1, i32 0, i32 0), - i8* getelementptr ([48 x i8]* @.str2, i32 0, i32 0) }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str5 = internal constant [20 x i8] c"deaddebuglabel.main\00", section "llvm.metadata" ; <[20 x i8]*> [#uses=1] -@.str6 = internal constant [5 x i8] c"main\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.subprogram7 = internal constant %llvm.dbg.subprogram.type { - i32 393262, - { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to { }*), - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i8* getelementptr ([20 x i8]* @.str5, i32 0, i32 0), - i8* getelementptr ([20 x i8]* @.str5, i32 0, i32 0), - i8* getelementptr ([5 x i8]* @.str6, i32 0, i32 0), - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i32 3, - { }* null, - i1 false, - i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] - -declare void @llvm.dbg.func.start({ }*) nounwind - -declare void @llvm.dbg.stoppoint(i32, i32, { }*) nounwind - -declare void @llvm.dbg.region.end({ }*) nounwind - -define fastcc i32 @main() { -entry.main: - call void @llvm.dbg.func.start( { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram7 to { }*) ) - br i1 true, label %reachable, label %unreachable - -reachable: ; preds = %entry.main - call void @llvm.dbg.region.end( { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram7 to { }*) ) - ret i32 1 - -unreachable: ; preds = %entry.main - call void @llvm.dbg.stoppoint( i32 7, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - call void @llvm.dbg.region.end( { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram7 to { }*) ) - ret i32 0 -} diff --git a/test/DebugInfo/funccall.ll b/test/DebugInfo/funccall.ll deleted file mode 100644 index e44b029..0000000 --- a/test/DebugInfo/funccall.ll +++ /dev/null @@ -1,147 +0,0 @@ -;; RUN: llc < %s - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, { }*, i32, i8*, i8*, i8* } - %llvm.dbg.global_variable.type = type { i32, { }*, { }*, i8*, i8*, i8*, { }*, i32, { }*, i1, i1, { }* } - %llvm.dbg.subprogram.type = type { i32, { }*, { }*, i8*, i8*, i8*, { }*, i32, { }*, i1, i1 } - %llvm.dbg.variable.type = type { i32, { }*, i8*, { }*, i32, { }* } -@llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { i32 393216, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 393216, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@llvm.dbg.global_variables = linkonce constant %llvm.dbg.anchor.type { i32 393216, i32 52 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@llvm.dbg.subprogram = internal constant %llvm.dbg.subprogram.type { - i32 393262, - { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to { }*), - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i8* getelementptr ([4 x i8]* @str, i32 0, i32 0), - i8* getelementptr ([4 x i8]* @str, i32 0, i32 0), - i8* null, - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i32 4, - { }* null, - i1 false, - i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@str = internal constant [4 x i8] c"foo\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { - i32 393233, - { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to { }*), - i32 1, - i8* getelementptr ([11 x i8]* @str1, i32 0, i32 0), - i8* getelementptr ([50 x i8]* @str2, i32 0, i32 0), - i8* getelementptr ([45 x i8]* @str3, i32 0, i32 0) }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@str1 = internal constant [11 x i8] c"funccall.c\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@str2 = internal constant [50 x i8] c"/Volumes/Big2/llvm/llvm/test/Regression/Debugger/\00", section "llvm.metadata" ; <[50 x i8]*> [#uses=1] -@str3 = internal constant [45 x i8] c"4.0.1 LLVM (Apple Computer, Inc. build 5421)\00", section "llvm.metadata" ; <[45 x i8]*> [#uses=1] -@llvm.dbg.variable = internal constant %llvm.dbg.variable.type { - i32 393472, - { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*), - i8* getelementptr ([2 x i8]* @str4, i32 0, i32 0), - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i32 5, - { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@str4 = internal constant [2 x i8] c"t\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type { - i32 393252, - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i8* getelementptr ([4 x i8]* @str15, i32 0, i32 0), - { }* null, - i32 0, - i64 32, - i64 32, - i64 0, - i32 0, - i32 5 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@str15 = internal constant [4 x i8] c"int\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.subprogram2 = internal constant %llvm.dbg.subprogram.type { - i32 393262, - { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to { }*), - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i8* getelementptr ([5 x i8]* @str6, i32 0, i32 0), - i8* getelementptr ([5 x i8]* @str6, i32 0, i32 0), - i8* null, - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i32 8, - { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*), - i1 false, - i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@str6 = internal constant [5 x i8] c"main\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.variable3 = internal constant %llvm.dbg.variable.type { - i32 393474, - { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram2 to { }*), - i8* getelementptr ([7 x i8]* @str7, i32 0, i32 0), - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i32 8, - { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@str7 = internal constant [7 x i8] c"retval\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@llvm.dbg.global_variable = internal constant %llvm.dbg.global_variable.type { - i32 393268, - { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.global_variables to { }*), - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i8* getelementptr ([2 x i8]* @str4, i32 0, i32 0), - i8* getelementptr ([2 x i8]* @str4, i32 0, i32 0), - i8* null, - { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), - i32 2, - { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*), - i1 true, - i1 true, - { }* bitcast (i32* @q to { }*) }, section "llvm.metadata" ; <%llvm.dbg.global_variable.type*> [#uses=0] -@str4.upgrd.1 = internal constant [2 x i8] c"q\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=0] -@q = internal global i32 0 ; <i32*> [#uses=7] - -declare void @llvm.dbg.func.start({ }*) - -declare void @llvm.dbg.stoppoint(i32, i32, { }*) - -declare void @llvm.dbg.declare({ }*, { }*) - -declare void @llvm.dbg.region.start({ }*) - -declare void @llvm.dbg.region.end({ }*) - -define void @foo() { -entry: - %t = alloca i32, align 4 ; <i32*> [#uses=3] - %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] - call void @llvm.dbg.func.start( { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*) ) - call void @llvm.dbg.stoppoint( i32 4, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - %t.upgrd.2 = bitcast i32* %t to { }* ; <{ }*> [#uses=1] - call void @llvm.dbg.declare( { }* %t.upgrd.2, { }* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable to { }*) ) - call void @llvm.dbg.stoppoint( i32 5, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - %tmp = load i32* @q ; <i32> [#uses=1] - store i32 %tmp, i32* %t - call void @llvm.dbg.stoppoint( i32 6, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - %tmp1 = load i32* %t ; <i32> [#uses=1] - %tmp2 = add i32 %tmp1, 1 ; <i32> [#uses=1] - store i32 %tmp2, i32* @q - call void @llvm.dbg.stoppoint( i32 7, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - call void @llvm.dbg.region.end( { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*) ) - ret void -} - -define i32 @main() { -entry: - %retval = alloca i32, align 4 ; <i32*> [#uses=3] - %tmp = alloca i32, align 4 ; <i32*> [#uses=2] - %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] - call void @llvm.dbg.func.start( { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram2 to { }*) ) - call void @llvm.dbg.stoppoint( i32 8, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - %retval.upgrd.3 = bitcast i32* %retval to { }* ; <{ }*> [#uses=1] - call void @llvm.dbg.declare( { }* %retval.upgrd.3, { }* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable3 to { }*) ) - call void @llvm.dbg.stoppoint( i32 9, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - store i32 0, i32* @q - call void @llvm.dbg.stoppoint( i32 10, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - call void (...)* bitcast (void ()* @foo to void (...)*)( ) - call void @llvm.dbg.stoppoint( i32 11, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - %tmp.upgrd.4 = load i32* @q ; <i32> [#uses=1] - %tmp1 = sub i32 %tmp.upgrd.4, 1 ; <i32> [#uses=1] - store i32 %tmp1, i32* @q - call void @llvm.dbg.stoppoint( i32 13, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - %tmp2 = load i32* @q ; <i32> [#uses=1] - store i32 %tmp2, i32* %tmp - %tmp3 = load i32* %tmp ; <i32> [#uses=1] - store i32 %tmp3, i32* %retval - %retval.upgrd.5 = load i32* %retval ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint( i32 14, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*) ) - call void @llvm.dbg.region.end( { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram2 to { }*) ) - ret i32 %retval.upgrd.5 -} diff --git a/test/DebugInfo/globalGetElementPtr.ll b/test/DebugInfo/globalGetElementPtr.ll deleted file mode 100644 index 155deb7..0000000 --- a/test/DebugInfo/globalGetElementPtr.ll +++ /dev/null @@ -1,264 +0,0 @@ -; RUN: llc < %s -; ModuleID = 'foo.c' -target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" -target triple = "i686-apple-darwin8" - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, %struct.anon*, i8*, %struct.anon*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, %struct.anon*, i32, i8*, i8*, i8* } - %llvm.dbg.compositetype.type = type { i32, %struct.anon*, i8*, %struct.anon*, i32, i64, i64, i64, i32, %struct.anon*, %struct.anon* } - %llvm.dbg.derivedtype.type = type { i32, %struct.anon*, i8*, %struct.anon*, i32, i64, i64, i64, i32, %struct.anon* } - %llvm.dbg.global_variable.type = type { i32, %struct.anon*, %struct.anon*, i8*, i8*, i8*, %struct.anon*, i32, %struct.anon*, i1, i1, %struct.anon* } - %llvm.dbg.subprogram.type = type { i32, %struct.anon*, %struct.anon*, i8*, i8*, i8*, %struct.anon*, i32, %struct.anon*, i1, i1 } - %llvm.dbg.subrange.type = type { i32, i64, i64 } - %llvm.dbg.variable.type = type { i32, %struct.anon*, i8*, %struct.anon*, i32, %struct.anon* } - %struct.S271 = type { [0 x %struct.anon], %struct.anon } - %struct.anon = type { } -@llvm.dbg.subprogram = internal constant %llvm.dbg.subprogram.type { - i32 393262, - %struct.anon* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %struct.anon*), - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* getelementptr ([4 x i8]* @.str3, i32 0, i32 0), - i8* getelementptr ([4 x i8]* @.str3, i32 0, i32 0), - i8* null, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 2, - %struct.anon* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype to %struct.anon*), - i1 false, - i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { i32 393216, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { - i32 393233, - %struct.anon* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %struct.anon*), - i32 1, - i8* getelementptr ([6 x i8]* @.str, i32 0, i32 0), - i8* getelementptr ([23 x i8]* @.str1, i32 0, i32 0), - i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0) }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 393216, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str = internal constant [6 x i8] c"foo.c\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str1 = internal constant [23 x i8] c"/Volumes/MacOS9/tests/\00", section "llvm.metadata" ; <[23 x i8]*> [#uses=1] -@.str2 = internal constant [52 x i8] c"4.2.1 (Based on Apple Inc. build 5546) (LLVM build)\00", section "llvm.metadata" ; <[52 x i8]*> [#uses=1] -@.str3 = internal constant [4 x i8] c"var\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.derivedtype = internal constant %llvm.dbg.derivedtype.type { - i32 393231, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* null, - %struct.anon* null, - i32 0, - i64 32, - i64 32, - i64 0, - i32 0, - %struct.anon* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type { - i32 393252, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* getelementptr ([5 x i8]* @.str4, i32 0, i32 0), - %struct.anon* null, - i32 0, - i64 8, - i64 8, - i64 0, - i32 0, - i32 6 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@.str4 = internal constant [5 x i8] c"char\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.variable = internal constant %llvm.dbg.variable.type { - i32 393474, - %struct.anon* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to %struct.anon*), - i8* getelementptr ([7 x i8]* @.str5, i32 0, i32 0), - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 2, - %struct.anon* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str5 = internal constant [7 x i8] c"retval\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@a271 = weak global [0 x %struct.S271] zeroinitializer ; <[0 x %struct.S271]*> [#uses=3] -@llvm.dbg.subprogram6 = internal constant %llvm.dbg.subprogram.type { - i32 393262, - %struct.anon* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %struct.anon*), - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* getelementptr ([5 x i8]* @.str7, i32 0, i32 0), - i8* getelementptr ([5 x i8]* @.str7, i32 0, i32 0), - i8* null, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 3, - %struct.anon* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype8 to %struct.anon*), - i1 false, - i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str7 = internal constant [5 x i8] c"main\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.basictype8 = internal constant %llvm.dbg.basictype.type { - i32 393252, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* getelementptr ([4 x i8]* @.str9, i32 0, i32 0), - %struct.anon* null, - i32 0, - i64 32, - i64 32, - i64 0, - i32 0, - i32 5 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@.str9 = internal constant [4 x i8] c"int\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.variable10 = internal constant %llvm.dbg.variable.type { - i32 393474, - %struct.anon* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram6 to %struct.anon*), - i8* getelementptr ([7 x i8]* @.str5, i32 0, i32 0), - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 3, - %struct.anon* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype8 to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.global_variable = internal constant %llvm.dbg.global_variable.type { - i32 393268, - %struct.anon* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.global_variables to %struct.anon*), - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* getelementptr ([5 x i8]* @.str11, i32 0, i32 0), - i8* getelementptr ([5 x i8]* @.str11, i32 0, i32 0), - i8* null, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 1, - %struct.anon* bitcast (%llvm.dbg.compositetype.type* @llvm.dbg.compositetype to %struct.anon*), - i1 false, - i1 true, - %struct.anon* getelementptr ([0 x %struct.S271]* @a271, i32 0, i32 0, i32 0, i32 0) }, section "llvm.metadata" ; <%llvm.dbg.global_variable.type*> [#uses=0] -@llvm.dbg.global_variables = linkonce constant %llvm.dbg.anchor.type { i32 393216, i32 52 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str11 = internal constant [5 x i8] c"a271\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.compositetype = internal constant %llvm.dbg.compositetype.type { - i32 393217, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* null, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 0, - i64 0, - i64 8, - i64 0, - i32 0, - %struct.anon* bitcast (%llvm.dbg.compositetype.type* @llvm.dbg.compositetype12 to %struct.anon*), - %struct.anon* bitcast ([1 x %struct.anon*]* @llvm.dbg.array25 to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.compositetype.type*> [#uses=1] -@llvm.dbg.compositetype12 = internal constant %llvm.dbg.compositetype.type { - i32 393235, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* getelementptr ([5 x i8]* @.str13, i32 0, i32 0), - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 1, - i64 0, - i64 8, - i64 0, - i32 0, - %struct.anon* null, - %struct.anon* bitcast ([2 x %struct.anon*]* @llvm.dbg.array23 to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.compositetype.type*> [#uses=1] -@.str13 = internal constant [5 x i8] c"S271\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.derivedtype14 = internal constant %llvm.dbg.derivedtype.type { - i32 393229, - %struct.anon* null, - i8* getelementptr ([2 x i8]* @.str15, i32 0, i32 0), - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 1, - i64 0, - i64 8, - i64 0, - i32 0, - %struct.anon* bitcast (%llvm.dbg.compositetype.type* @llvm.dbg.compositetype16 to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str15 = internal constant [2 x i8] c"a\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.compositetype16 = internal constant %llvm.dbg.compositetype.type { - i32 393217, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* null, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 0, - i64 0, - i64 8, - i64 0, - i32 0, - %struct.anon* bitcast (%llvm.dbg.compositetype.type* @llvm.dbg.compositetype17 to %struct.anon*), - %struct.anon* bitcast ([1 x %struct.anon*]* @llvm.dbg.array18 to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.compositetype.type*> [#uses=1] -@llvm.dbg.compositetype17 = internal constant %llvm.dbg.compositetype.type { - i32 393235, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* null, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 1, - i64 0, - i64 8, - i64 0, - i32 0, - %struct.anon* null, - %struct.anon* bitcast ([0 x %struct.anon*]* @llvm.dbg.array to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.compositetype.type*> [#uses=1] -@llvm.dbg.array = internal constant [0 x %struct.anon*] zeroinitializer, section "llvm.metadata" ; <[0 x %struct.anon*]*> [#uses=1] -@llvm.dbg.subrange = internal constant %llvm.dbg.subrange.type { - i32 393249, - i64 0, - i64 4 }, section "llvm.metadata" ; <%llvm.dbg.subrange.type*> [#uses=1] -@llvm.dbg.array18 = internal constant [1 x %struct.anon*] [ %struct.anon* bitcast (%llvm.dbg.subrange.type* @llvm.dbg.subrange to %struct.anon*) ], section "llvm.metadata" ; <[1 x %struct.anon*]*> [#uses=1] -@llvm.dbg.derivedtype19 = internal constant %llvm.dbg.derivedtype.type { - i32 393229, - %struct.anon* null, - i8* getelementptr ([2 x i8]* @.str20, i32 0, i32 0), - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 1, - i64 0, - i64 8, - i64 0, - i32 0, - %struct.anon* bitcast (%llvm.dbg.compositetype.type* @llvm.dbg.compositetype21 to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str20 = internal constant [2 x i8] c"b\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.compositetype21 = internal constant %llvm.dbg.compositetype.type { - i32 393235, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i8* null, - %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*), - i32 1, - i64 0, - i64 8, - i64 0, - i32 0, - %struct.anon* null, - %struct.anon* bitcast ([0 x %struct.anon*]* @llvm.dbg.array22 to %struct.anon*) }, section "llvm.metadata" ; <%llvm.dbg.compositetype.type*> [#uses=1] -@llvm.dbg.array22 = internal constant [0 x %struct.anon*] zeroinitializer, section "llvm.metadata" ; <[0 x %struct.anon*]*> [#uses=1] -@llvm.dbg.array23 = internal constant [2 x %struct.anon*] [ %struct.anon* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype14 to %struct.anon*), %struct.anon* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype19 to %struct.anon*) ], section "llvm.metadata" ; <[2 x %struct.anon*]*> [#uses=1] -@llvm.dbg.subrange24 = internal constant %llvm.dbg.subrange.type { - i32 393249, - i64 0, - i64 4 }, section "llvm.metadata" ; <%llvm.dbg.subrange.type*> [#uses=1] -@llvm.dbg.array25 = internal constant [1 x %struct.anon*] [ %struct.anon* bitcast (%llvm.dbg.subrange.type* @llvm.dbg.subrange24 to %struct.anon*) ], section "llvm.metadata" ; <[1 x %struct.anon*]*> [#uses=1] - -define i8* @var() { -entry: - %retval = alloca i8* ; <i8**> [#uses=3] - %tmp = alloca i8* ; <i8**> [#uses=2] - %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] - call void @llvm.dbg.func.start( %struct.anon* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to %struct.anon*) ) - call void @llvm.dbg.stoppoint( i32 2, i32 0, %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*) ) - %retval1 = bitcast i8** %retval to %struct.anon* ; <%struct.anon*> [#uses=1] - call void @llvm.dbg.declare( %struct.anon* %retval1, %struct.anon* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable to %struct.anon*) ) - bitcast %struct.S271* getelementptr ([0 x %struct.S271]* @a271, i32 0, i32 0) to i8* ; <i8*>:0 [#uses=0] - store i8* bitcast ([0 x %struct.S271]* @a271 to i8*), i8** %tmp, align 4 - %tmp2 = load i8** %tmp, align 4 ; <i8*> [#uses=1] - store i8* %tmp2, i8** %retval, align 4 - br label %return - -return: ; preds = %entry - %retval3 = load i8** %retval ; <i8*> [#uses=1] - call void @llvm.dbg.stoppoint( i32 2, i32 0, %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*) ) - call void @llvm.dbg.region.end( %struct.anon* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to %struct.anon*) ) - ret i8* %retval3 -} - -declare void @llvm.dbg.func.start(%struct.anon*) nounwind - -declare void @llvm.dbg.stoppoint(i32, i32, %struct.anon*) nounwind - -declare void @llvm.dbg.declare(%struct.anon*, %struct.anon*) nounwind - -declare void @llvm.dbg.region.end(%struct.anon*) nounwind - -define i32 @main() { -entry: - %retval = alloca i32 ; <i32*> [#uses=2] - %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] - call void @llvm.dbg.func.start( %struct.anon* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram6 to %struct.anon*) ) - call void @llvm.dbg.stoppoint( i32 3, i32 0, %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*) ) - %retval1 = bitcast i32* %retval to %struct.anon* ; <%struct.anon*> [#uses=1] - call void @llvm.dbg.declare( %struct.anon* %retval1, %struct.anon* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable10 to %struct.anon*) ) - br label %return - -return: ; preds = %entry - %retval2 = load i32* %retval ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint( i32 3, i32 0, %struct.anon* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %struct.anon*) ) - call void @llvm.dbg.region.end( %struct.anon* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram6 to %struct.anon*) ) - ret i32 %retval2 -} diff --git a/test/DebugInfo/inheritance.ll b/test/DebugInfo/inheritance.ll new file mode 100644 index 0000000..a689cb2 --- /dev/null +++ b/test/DebugInfo/inheritance.ll @@ -0,0 +1,151 @@ +; RUN: llc %s -o /dev/null +; PR 2613. + +%struct.__class_type_info_pseudo = type { %struct.__type_info_pseudo } +%struct.__type_info_pseudo = type { i8*, i8* } +%struct.test1 = type { i32 (...)** } + +@_ZTV5test1 = weak_odr constant [4 x i32 (...)*] [i32 (...)* null, i32 (...)* bitcast (%struct.__class_type_info_pseudo* @_ZTI5test1 to i32 (...)*), i32 (...)* bitcast (void (%struct.test1*)* @_ZN5test1D1Ev to i32 (...)*), i32 (...)* bitcast (void (%struct.test1*)* @_ZN5test1D0Ev to i32 (...)*)], align 32 ; <[4 x i32 (...)*]*> [#uses=1] +@_ZTI5test1 = weak_odr constant %struct.__class_type_info_pseudo { %struct.__type_info_pseudo { i8* inttoptr (i64 add (i64 ptrtoint ([0 x i32 (...)*]* @_ZTVN10__cxxabiv117__class_type_infoE to i64), i64 16) to i8*), i8* getelementptr inbounds ([7 x i8]* @_ZTS5test1, i64 0, i64 0) } }, align 16 ; <%struct.__class_type_info_pseudo*> [#uses=1] +@_ZTVN10__cxxabiv117__class_type_infoE = external constant [0 x i32 (...)*] ; <[0 x i32 (...)*]*> [#uses=1] +@_ZTS5test1 = weak_odr constant [7 x i8] c"5test1\00" ; <[7 x i8]*> [#uses=2] + +define i32 @main() nounwind ssp { +entry: + %retval = alloca i32 ; <i32*> [#uses=2] + %0 = alloca i32 ; <i32*> [#uses=2] + %tst = alloca %struct.test1 ; <%struct.test1*> [#uses=1] + %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] + call void @llvm.dbg.declare(metadata !{%struct.test1* %tst}, metadata !0), !dbg !21 + call void @_ZN5test1C1Ev(%struct.test1* %tst) nounwind, !dbg !22 + store i32 0, i32* %0, align 4, !dbg !23 + %1 = load i32* %0, align 4, !dbg !23 ; <i32> [#uses=1] + store i32 %1, i32* %retval, align 4, !dbg !23 + br label %return, !dbg !23 + +return: ; preds = %entry + %retval1 = load i32* %retval, !dbg !23 ; <i32> [#uses=1] + ret i32 %retval1, !dbg !23 +} + +define linkonce_odr void @_ZN5test1C1Ev(%struct.test1* %this) nounwind ssp align 2 { +entry: + %this_addr = alloca %struct.test1* ; <%struct.test1**> [#uses=2] + %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] + call void @llvm.dbg.declare(metadata !{%struct.test1** %this_addr}, metadata !24), !dbg !28 + store %struct.test1* %this, %struct.test1** %this_addr + %0 = load %struct.test1** %this_addr, align 8, !dbg !28 ; <%struct.test1*> [#uses=1] + %1 = getelementptr inbounds %struct.test1* %0, i32 0, i32 0, !dbg !28 ; <i32 (...)***> [#uses=1] + store i32 (...)** getelementptr inbounds ([4 x i32 (...)*]* @_ZTV5test1, i64 0, i64 2), i32 (...)*** %1, align 8, !dbg !28 + br label %return, !dbg !28 + +return: ; preds = %entry + ret void, !dbg !29 +} + +declare void @llvm.dbg.declare(metadata, metadata) nounwind readnone + +define linkonce_odr void @_ZN5test1D1Ev(%struct.test1* %this) nounwind ssp align 2 { +entry: + %this_addr = alloca %struct.test1* ; <%struct.test1**> [#uses=3] + %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] + call void @llvm.dbg.declare(metadata !{%struct.test1** %this_addr}, metadata !32), !dbg !34 + store %struct.test1* %this, %struct.test1** %this_addr + %0 = load %struct.test1** %this_addr, align 8, !dbg !35 ; <%struct.test1*> [#uses=1] + %1 = getelementptr inbounds %struct.test1* %0, i32 0, i32 0, !dbg !35 ; <i32 (...)***> [#uses=1] + store i32 (...)** getelementptr inbounds ([4 x i32 (...)*]* @_ZTV5test1, i64 0, i64 2), i32 (...)*** %1, align 8, !dbg !35 + br label %bb, !dbg !37 + +bb: ; preds = %entry + %2 = trunc i32 0 to i8, !dbg !37 ; <i8> [#uses=1] + %toBool = icmp ne i8 %2, 0, !dbg !37 ; <i1> [#uses=1] + br i1 %toBool, label %bb1, label %bb2, !dbg !37 + +bb1: ; preds = %bb + %3 = load %struct.test1** %this_addr, align 8, !dbg !37 ; <%struct.test1*> [#uses=1] + %4 = bitcast %struct.test1* %3 to i8*, !dbg !37 ; <i8*> [#uses=1] + call void @_ZdlPv(i8* %4) nounwind, !dbg !37 + br label %bb2, !dbg !37 + +bb2: ; preds = %bb1, %bb + br label %return, !dbg !37 + +return: ; preds = %bb2 + ret void, !dbg !37 +} + +define linkonce_odr void @_ZN5test1D0Ev(%struct.test1* %this) nounwind ssp align 2 { +entry: + %this_addr = alloca %struct.test1* ; <%struct.test1**> [#uses=3] + %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] + call void @llvm.dbg.declare(metadata !{%struct.test1** %this_addr}, metadata !38), !dbg !40 + store %struct.test1* %this, %struct.test1** %this_addr + %0 = load %struct.test1** %this_addr, align 8, !dbg !41 ; <%struct.test1*> [#uses=1] + %1 = getelementptr inbounds %struct.test1* %0, i32 0, i32 0, !dbg !41 ; <i32 (...)***> [#uses=1] + store i32 (...)** getelementptr inbounds ([4 x i32 (...)*]* @_ZTV5test1, i64 0, i64 2), i32 (...)*** %1, align 8, !dbg !41 + br label %bb, !dbg !43 + +bb: ; preds = %entry + %2 = trunc i32 1 to i8, !dbg !43 ; <i8> [#uses=1] + %toBool = icmp ne i8 %2, 0, !dbg !43 ; <i1> [#uses=1] + br i1 %toBool, label %bb1, label %bb2, !dbg !43 + +bb1: ; preds = %bb + %3 = load %struct.test1** %this_addr, align 8, !dbg !43 ; <%struct.test1*> [#uses=1] + %4 = bitcast %struct.test1* %3 to i8*, !dbg !43 ; <i8*> [#uses=1] + call void @_ZdlPv(i8* %4) nounwind, !dbg !43 + br label %bb2, !dbg !43 + +bb2: ; preds = %bb1, %bb + br label %return, !dbg !43 + +return: ; preds = %bb2 + ret void, !dbg !43 +} + +declare void @_ZdlPv(i8*) nounwind + +!0 = metadata !{i32 459008, metadata !1, metadata !"tst", metadata !4, i32 13, metadata !8} ; [ DW_TAG_auto_variable ] +!1 = metadata !{i32 458763, metadata !2, i32 0, i32 0} ; [ DW_TAG_lexical_block ] +!2 = metadata !{i32 458763, metadata !3, i32 0, i32 0} ; [ DW_TAG_lexical_block ] +!3 = metadata !{i32 458798, i32 0, metadata !4, metadata !"main", metadata !"main", metadata !"main", metadata !4, i32 11, metadata !5, i1 false, i1 true, i32 0, i32 0, null, i1 false} ; [ DW_TAG_subprogram ] +!4 = metadata !{i32 458769, i32 0, i32 4, metadata !"inheritance.cpp", metadata !"/tmp/", metadata !"4.2.1 (Based on Apple Inc. build 5658) (LLVM build)", i1 true, i1 false, metadata !"", i32 0} ; [ DW_TAG_compile_unit ] +!5 = metadata !{i32 458773, metadata !4, metadata !"", metadata !4, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !6, i32 0, null} ; [ DW_TAG_subroutine_type ] +!6 = metadata !{metadata !7} +!7 = metadata !{i32 458788, metadata !4, metadata !"int", metadata !4, i32 0, i64 32, i64 32, i64 0, i32 0, i32 5} ; [ DW_TAG_base_type ] +!8 = metadata !{i32 458771, metadata !4, metadata !"test1", metadata !4, i32 1, i64 64, i64 64, i64 0, i32 0, null, metadata !9, i32 0, metadata !8} ; [ DW_TAG_structure_type ] +!9 = metadata !{metadata !10, metadata !14, metadata !18} +!10 = metadata !{i32 458765, metadata !8, metadata !"_vptr$test1", metadata !4, i32 1, i64 64, i64 64, i64 0, i32 0, metadata !11} ; [ DW_TAG_member ] +!11 = metadata !{i32 458767, metadata !4, metadata !"", metadata !4, i32 0, i64 64, i64 64, i64 0, i32 0, metadata !12} ; [ DW_TAG_pointer_type ] +!12 = metadata !{i32 458767, metadata !4, metadata !"__vtbl_ptr_type", metadata !13, i32 0, i64 0, i64 0, i64 0, i32 0, metadata !5} ; [ DW_TAG_pointer_type ] +!13 = metadata !{i32 458769, i32 0, i32 4, metadata !"<built-in>", metadata !"/tmp/", metadata !"4.2.1 (Based on Apple Inc. build 5658) (LLVM build)", i1 false, i1 false, metadata !"", i32 0} ; [ DW_TAG_compile_unit ] +!14 = metadata !{i32 458798, i32 0, metadata !8, metadata !"test1", metadata !"test1", metadata !"", metadata !4, i32 1, metadata !15, i1 false, i1 false, i32 0, i32 0, null, i1 true} ; [ DW_TAG_subprogram ] +!15 = metadata !{i32 458773, metadata !4, metadata !"", metadata !4, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !16, i32 0, null} ; [ DW_TAG_subroutine_type ] +!16 = metadata !{null, metadata !17} +!17 = metadata !{i32 458767, metadata !4, metadata !"", metadata !4, i32 0, i64 64, i64 64, i64 0, i32 64, metadata !8} ; [ DW_TAG_pointer_type ] +!18 = metadata !{i32 458798, i32 0, metadata !8, metadata !"~test1", metadata !"~test1", metadata !"", metadata !4, i32 4, metadata !19, i1 false, i1 false, i32 1, i32 0, metadata !8, i1 false} ; [ DW_TAG_subprogram ] +!19 = metadata !{i32 458773, metadata !4, metadata !"", metadata !4, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !20, i32 0, null} ; [ DW_TAG_subroutine_type ] +!20 = metadata !{null, metadata !17, metadata !7} +!21 = metadata !{i32 11, i32 0, metadata !1, null} +!22 = metadata !{i32 13, i32 0, metadata !1, null} +!23 = metadata !{i32 14, i32 0, metadata !1, null} +!24 = metadata !{i32 459009, metadata !25, metadata !"this", metadata !4, i32 13, metadata !26} ; [ DW_TAG_arg_variable ] +!25 = metadata !{i32 458798, i32 0, metadata !4, metadata !"test1", metadata !"test1", metadata !"_ZN5test1C1Ev", metadata !4, i32 1, metadata !15, i1 false, i1 true, i32 0, i32 0, null, i1 false} ; [ DW_TAG_subprogram ] +!26 = metadata !{i32 458790, metadata !4, metadata !"", metadata !4, i32 0, i64 64, i64 64, i64 0, i32 64, metadata !27} ; [ DW_TAG_const_type ] +!27 = metadata !{i32 458767, metadata !4, metadata !"", metadata !4, i32 0, i64 64, i64 64, i64 0, i32 0, metadata !8} ; [ DW_TAG_pointer_type ] +!28 = metadata !{i32 1, i32 0, metadata !25, null} +!29 = metadata !{i32 1, i32 0, metadata !30, null} +!30 = metadata !{i32 458763, metadata !31, i32 0, i32 0} ; [ DW_TAG_lexical_block ] +!31 = metadata !{i32 458763, metadata !25, i32 0, i32 0} ; [ DW_TAG_lexical_block ] +!32 = metadata !{i32 459009, metadata !33, metadata !"this", metadata !4, i32 4, metadata !26} ; [ DW_TAG_arg_variable ] +!33 = metadata !{i32 458798, i32 0, metadata !8, metadata !"~test1", metadata !"~test1", metadata !"_ZN5test1D1Ev", metadata !4, i32 4, metadata !15, i1 false, i1 true, i32 1, i32 0, metadata !8, i1 false} ; [ DW_TAG_subprogram ] +!34 = metadata !{i32 4, i32 0, metadata !33, null} +!35 = metadata !{i32 5, i32 0, metadata !36, null} +!36 = metadata !{i32 458763, metadata !33, i32 0, i32 0} ; [ DW_TAG_lexical_block ] +!37 = metadata !{i32 6, i32 0, metadata !36, null} +!38 = metadata !{i32 459009, metadata !39, metadata !"this", metadata !4, i32 4, metadata !26} ; [ DW_TAG_arg_variable ] +!39 = metadata !{i32 458798, i32 0, metadata !8, metadata !"~test1", metadata !"~test1", metadata !"_ZN5test1D0Ev", metadata !4, i32 4, metadata !15, i1 false, i1 true, i32 1, i32 1, metadata !8, i1 false} ; [ DW_TAG_subprogram ] +!40 = metadata !{i32 4, i32 0, metadata !39, null} +!41 = metadata !{i32 5, i32 0, metadata !42, null} +!42 = metadata !{i32 458763, metadata !39, i32 0, i32 0} ; [ DW_TAG_lexical_block ] +!43 = metadata !{i32 6, i32 0, metadata !42, null} diff --git a/test/ExecutionEngine/2010-01-15-UndefValue.ll b/test/ExecutionEngine/2010-01-15-UndefValue.ll index 7d646eb..33ca63a 100644 --- a/test/ExecutionEngine/2010-01-15-UndefValue.ll +++ b/test/ExecutionEngine/2010-01-15-UndefValue.ll @@ -3,7 +3,7 @@ define i32 @main() { %a = add i32 0, undef - %b = add float 0.0, undef - %c = add double 0.0, undef + %b = fadd float 0.0, undef + %c = fadd double 0.0, undef ret i32 0 } diff --git a/test/FrontendC++/2010-02-08-NamespaceVar.cpp b/test/FrontendC++/2010-02-08-NamespaceVar.cpp deleted file mode 100644 index cd8247a..0000000 --- a/test/FrontendC++/2010-02-08-NamespaceVar.cpp +++ /dev/null @@ -1,16 +0,0 @@ -// RUN: %llvmgxx -S %s -o - | grep cX - -namespace C { - int c = 1; - namespace { - int cX = 6; - void marker2() { - cX; - } - } -} - -int main() { - C::marker2(); - return 0; -} diff --git a/test/FrontendC++/2010-02-17-DbgArtificialArg.cpp b/test/FrontendC++/2010-02-17-DbgArtificialArg.cpp new file mode 100644 index 0000000..4c8e0e5 --- /dev/null +++ b/test/FrontendC++/2010-02-17-DbgArtificialArg.cpp @@ -0,0 +1,14 @@ +// RUN: %llvmgcc -g -S %s -o - | grep DW_TAG_pointer_type | grep "i32 458767, metadata .., metadata ..., metadata .., i32 ., i64 .., i64 .., i64 0, i32 64, metadata ..." +// Here, second to last argument "i32 64" indicates that artificial type is set. +// Test to artificial attribute attahed to "this" pointer type. +// Radar 7655792 and 7655002 + +class A { +public: + int fn1(int i) const { return i + 2; }; +}; + +int foo() { + A a; + return a.fn1(1); +} diff --git a/test/FrontendC/2003-12-14-ExternInlineSupport.c b/test/FrontendC/2003-12-14-ExternInlineSupport.c index 510d1f8..fb92ec7 100644 --- a/test/FrontendC/2003-12-14-ExternInlineSupport.c +++ b/test/FrontendC/2003-12-14-ExternInlineSupport.c @@ -1,3 +1,3 @@ -// RUN: %llvmgcc -Os -xc %s -c -o - | llvm-dis | not grep dead_function +// RUN: %llvmgcc -xc %s -c -o - | llvm-dis | not grep dead_function extern __inline__ void dead_function() {} diff --git a/test/FrontendC/2007-09-17-WeakRef.c b/test/FrontendC/2007-09-17-WeakRef.c index 3cdd47e..6c420ea 100644 --- a/test/FrontendC/2007-09-17-WeakRef.c +++ b/test/FrontendC/2007-09-17-WeakRef.c @@ -1,6 +1,6 @@ // RUN: %llvmgcc -O1 -S %s -o - | grep icmp // PR1678 -// XFAIL: llvmgcc4.0.1 + extern void B (void); static __typeof(B) A __attribute__ ((__weakref__("B"))); int active (void) diff --git a/test/FrontendC/2010-02-15-DbgStaticVar.c b/test/FrontendC/2010-02-15-DbgStaticVar.c new file mode 100644 index 0000000..7827d96 --- /dev/null +++ b/test/FrontendC/2010-02-15-DbgStaticVar.c @@ -0,0 +1,13 @@ +// RUN: %llvmgcc -g -S %s -o - | grep "metadata ..b., metadata ..b., metadata ...," +// Test to check intentionally empty linkage name for a static variable. +// Radar 7651244. +static int foo(int a) +{ + static int b = 1; + return b+a; +} + +int main() { + int j = foo(1); + return 0; +} diff --git a/test/FrontendC/2010-02-16-DbgVarScope.c b/test/FrontendC/2010-02-16-DbgVarScope.c new file mode 100644 index 0000000..1d912d0f --- /dev/null +++ b/test/FrontendC/2010-02-16-DbgVarScope.c @@ -0,0 +1,30 @@ +// RUN: %llvmgcc -S -O0 -g %s -o - | \ +// RUN: llc --disable-fp-elim -o %t.s -O0 -relocation-model=pic +// RUN: %compile_c %t.s -o %t.o +// RUN: %link %t.o -o %t.exe +// RUN: echo {break 24\nrun\np loc\n} > %t.in +// RN: gdb -q -batch -n -x %t.in %t.exe | tee %t.out | \ +// RN: grep {$1 = 1} + +int g1 = 1; +int g2 = 2; + +int __attribute__((always_inline)) bar() { + return g2 - g1; +} +void foobar() {} + +void foo(int s) { + unsigned loc = 0; + if (s) { + loc = 1; + foobar(); + } else { + loc = bar(); + foobar(); + } +} + +int main() { + foo(0); +} diff --git a/test/FrontendC/2010-02-18-Dbg-VectorType.c b/test/FrontendC/2010-02-18-Dbg-VectorType.c new file mode 100644 index 0000000..d34031f --- /dev/null +++ b/test/FrontendC/2010-02-18-Dbg-VectorType.c @@ -0,0 +1,9 @@ +// RUN: %llvmgcc -S -O0 -g %s -o - | grep DW_TAG_typedef | grep float4 +typedef float float4 __attribute__((vector_size(16))); + +int main(){ + volatile float4 x = (float4) { 0.0f, 1.0f, 2.0f, 3.0f }; + x += x; + return 0; +} + diff --git a/test/FrontendObjC/2010-02-23-DbgInheritance.m b/test/FrontendObjC/2010-02-23-DbgInheritance.m new file mode 100644 index 0000000..7e1cf67 --- /dev/null +++ b/test/FrontendObjC/2010-02-23-DbgInheritance.m @@ -0,0 +1,9 @@ +// RUN: %llvmgcc %s -S -g -o - | grep -v DW_TAG_member +// Interface P should not be a member of interface I in debug info. +@interface P +@end + +@interface I : P +@end + +void fn(I *iptr) {} diff --git a/test/LLVMC/AppendCmdHook.td b/test/LLVMC/AppendCmdHook.td index 4a9d391..539a93f 100644 --- a/test/LLVMC/AppendCmdHook.td +++ b/test/LLVMC/AppendCmdHook.td @@ -13,14 +13,14 @@ def OptList : OptionList<[ ]>; def dummy_tool : Tool<[ -(cmd_line "dummy_cmd $INFILE"), +(command "dummy_cmd"), (in_language "dummy_lang"), (out_language "dummy_lang"), (actions (case - // CHECK: push_back("-arg1") - // CHECK: push_back("-arg2") + // CHECK: , "-arg1")); + // CHECK: , "-arg2")); (switch_on "dummy1"), (append_cmd "-arg1 -arg2"), - // CHECK: push_back("-arg3") + // CHECK: , "-arg3")); // CHECK: hooks::MyHook() (switch_on "dummy2"), (append_cmd "-arg3 $CALL(MyHook)"))) ]>; diff --git a/test/LLVMC/EnvParentheses.td b/test/LLVMC/EnvParentheses.td index 77aab95..c563171 100644 --- a/test/LLVMC/EnvParentheses.td +++ b/test/LLVMC/EnvParentheses.td @@ -1,13 +1,13 @@ // Check the fix for PR4157. // http://llvm.org/bugs/show_bug.cgi?id=4157 // RUN: tblgen -I %p/../../include --gen-llvmc %s -o %t -// RUN: not grep {)));} %t +// RUN: not grep {FOO")));} %t // RUN: %compile_cxx -fexceptions -x c++ %t include "llvm/CompilerDriver/Common.td" def dummy_tool : Tool<[ -(cmd_line "gcc -o $OUTFILE $INFILE $ENV(FOO)/bar"), +(command "gcc $ENV(FOO)/bar"), (in_language "dummy"), (out_language "dummy") ]>; diff --git a/test/LLVMC/ExternOptions.td b/test/LLVMC/ExternOptions.td index a05f2ca..77cb4bf 100644 --- a/test/LLVMC/ExternOptions.td +++ b/test/LLVMC/ExternOptions.td @@ -13,7 +13,7 @@ def OptList : OptionList<[(switch_option "Wall", (extern)), (prefix_list_option "L", (extern))]>; def dummy_tool : Tool<[ -(cmd_line "dummy_cmd $INFILE"), +(command "dummy_cmd"), (in_language "dummy"), (out_language "dummy"), (actions (case diff --git a/test/LLVMC/ForwardAs.td b/test/LLVMC/ForwardAs.td index ce6fbb0..7c3bd17 100644 --- a/test/LLVMC/ForwardAs.td +++ b/test/LLVMC/ForwardAs.td @@ -9,11 +9,11 @@ include "llvm/CompilerDriver/Common.td" def OptList : OptionList<[(parameter_option "dummy", (extern))]>; def dummy_tool : Tool<[ -(cmd_line "dummy_cmd $INFILE"), +(command "dummy_cmd"), (in_language "dummy"), (out_language "dummy"), (actions (case - // CHECK: vec.push_back("unique_name") + // CHECK: "unique_name")); (not_empty "dummy"), (forward_as "dummy", "unique_name"))) ]>; diff --git a/test/LLVMC/ForwardTransformedValue.td b/test/LLVMC/ForwardTransformedValue.td index e263453..2caef6c 100644 --- a/test/LLVMC/ForwardTransformedValue.td +++ b/test/LLVMC/ForwardTransformedValue.td @@ -13,7 +13,7 @@ def OptList : OptionList<[(parameter_option "a", (extern)), // CHECK: std::string HookB def dummy_tool : Tool<[ -(cmd_line "dummy_cmd $INFILE"), +(command "dummy_cmd"), (in_language "dummy"), (out_language "dummy"), (actions (case diff --git a/test/LLVMC/ForwardValue.td b/test/LLVMC/ForwardValue.td index 31e395e..463235c 100644 --- a/test/LLVMC/ForwardValue.td +++ b/test/LLVMC/ForwardValue.td @@ -10,13 +10,13 @@ def OptList : OptionList<[(parameter_option "a", (extern)), (prefix_list_option "b", (extern))]>; def dummy_tool : Tool<[ -(cmd_line "dummy_cmd $INFILE"), +(command "dummy_cmd"), (in_language "dummy"), (out_language "dummy"), (actions (case - // CHECK: vec.push_back(AutoGeneratedParameter_a) + // CHECK: , AutoGeneratedParameter_a)); (not_empty "a"), (forward_value "a"), - // CHECK: std::copy(AutoGeneratedList_b.begin() + // CHECK: B = AutoGeneratedList_b.begin() (not_empty "b"), (forward_value "b"))) ]>; diff --git a/test/LLVMC/HookWithArguments.td b/test/LLVMC/HookWithArguments.td index ba0bbe1..312fa9c 100644 --- a/test/LLVMC/HookWithArguments.td +++ b/test/LLVMC/HookWithArguments.td @@ -6,12 +6,12 @@ include "llvm/CompilerDriver/Common.td" // CHECK: Hook(const char* Arg0, const char* Arg1, const char* Arg2); +// CHECK: "/path" // CHECK: std::getenv("VARIABLE") // CHECK: "/2path" -// CHECK: "/path" def dummy_tool : Tool<[ -(cmd_line "$CALL(Hook, 'Arg1', 'Arg2', 'Arg3 Arg3Cont')/path arg1 $ENV(VARIABLE)/2path arg2 $INFILE"), +(command "$CALL(Hook, 'Arg1', 'Arg2', 'Arg3 Arg3Cont')/path arg1 $ENV(VARIABLE)/2path arg2"), (in_language "dummy"), (out_language "dummy") ]>; diff --git a/test/LLVMC/HookWithInFile.td b/test/LLVMC/HookWithInFile.td index e15e43c..f58e3f4 100644 --- a/test/LLVMC/HookWithInFile.td +++ b/test/LLVMC/HookWithInFile.td @@ -7,7 +7,7 @@ include "llvm/CompilerDriver/Common.td" def dummy_tool : Tool<[ // CHECK: Hook(inFile.c_str()) -(cmd_line "$CALL(Hook, '$INFILE')/path $INFILE"), +(command "$CALL(Hook, '$INFILE')/path"), (in_language "dummy"), (out_language "dummy") ]>; diff --git a/test/LLVMC/Init.td b/test/LLVMC/Init.td index 355d83f..ff9a0d8 100644 --- a/test/LLVMC/Init.td +++ b/test/LLVMC/Init.td @@ -13,7 +13,7 @@ def OptList : OptionList<[ ]>; def dummy_tool : Tool<[ -(cmd_line "dummy_cmd $INFILE"), +(command "dummy_cmd"), (in_language "dummy_lang"), (out_language "dummy_lang"), (actions (case diff --git a/test/LLVMC/MultiValuedOption.td b/test/LLVMC/MultiValuedOption.td index 8cb1878..b52af57 100644 --- a/test/LLVMC/MultiValuedOption.td +++ b/test/LLVMC/MultiValuedOption.td @@ -12,7 +12,7 @@ def OptList : OptionList<[ (parameter_list_option "baz", (multi_val 2), (extern))]>; def dummy_tool : Tool<[ -(cmd_line "dummy_cmd $INFILE"), +(command "dummy_cmd"), (in_language "dummy"), (out_language "dummy"), (actions (case diff --git a/test/LLVMC/MultiplePluginPriorities.td b/test/LLVMC/MultiplePluginPriorities.td index f108641..2fe0645 100644 --- a/test/LLVMC/MultiplePluginPriorities.td +++ b/test/LLVMC/MultiplePluginPriorities.td @@ -4,6 +4,10 @@ // Disable for Darwin PPC: <rdar://problem/7598390> // XFAIL: powerpc-apple-darwin +// Generally XFAIL'ed for now, this is (sometimes?) failing on x86_64-apple-darwin10. +// RUN: false +// XFAIL: * + include "llvm/CompilerDriver/Common.td" def Graph : CompilationGraph<[]>; diff --git a/test/LLVMC/NoActions.td b/test/LLVMC/NoActions.td index 9c2d45a..015bfdd 100644 --- a/test/LLVMC/NoActions.td +++ b/test/LLVMC/NoActions.td @@ -7,7 +7,7 @@ include "llvm/CompilerDriver/Common.td" // CHECK: class dummy_tool : public Tool { def dummy_tool : Tool<[ -(cmd_line "dummy_cmd $INFILE"), +(command "dummy_cmd"), (in_language "dummy"), (out_language "dummy") ]>; diff --git a/test/LLVMC/OneOrMore.td b/test/LLVMC/OneOrMore.td index ddf7cd1..42ec693 100644 --- a/test/LLVMC/OneOrMore.td +++ b/test/LLVMC/OneOrMore.td @@ -13,7 +13,7 @@ def OptList : OptionList<[ (parameter_list_option "baz", (optional))]>; def dummy_tool : Tool<[ -(cmd_line "dummy_cmd $INFILE"), +(command "dummy_cmd"), (in_language "dummy"), (out_language "dummy"), (actions (case diff --git a/test/LLVMC/OptionPreprocessor.td b/test/LLVMC/OptionPreprocessor.td index 8d748ee..8a31481 100644 --- a/test/LLVMC/OptionPreprocessor.td +++ b/test/LLVMC/OptionPreprocessor.td @@ -52,7 +52,7 @@ def dummy : Tool< [(in_language "dummy"), (out_language "dummy"), (output_suffix "d"), - (cmd_line "dummy $INFILE -o $OUTFILE"), + (command "dummy"), (actions (case (switch_on "foo"), (error), (switch_on "bar"), (error), (switch_on "baz"), (error), diff --git a/test/MC/MachO/Darwin/optimal_nop.s b/test/MC/MachO/Darwin/optimal_nop.s new file mode 100644 index 0000000..29cb073 --- /dev/null +++ b/test/MC/MachO/Darwin/optimal_nop.s @@ -0,0 +1,156 @@ +// Validate that we can assemble this file exactly like the platform +// assembler. +// +// RUN: llvm-mc -filetype=obj -triple i386-apple-darwin10 -o %t.mc.o %s +// RUN: as -arch i386 -o %t.as.o %s +// RUN: diff %t.mc.o %t.as.o + +# 1 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + # nop + # 0x90 + .align 1, 0x90 + ret +# 2 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + # xchg %ax,%ax + # 0x66, 0x90 + .align 2, 0x90 + ret +# 3 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + # nopl (%[re]ax) + # 0x0f, 0x1f, 0x00 + .align 2, 0x90 + ret +# 4 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + ret + ret + # nopl 0(%[re]ax) + # 0x0f, 0x1f, 0x40, 0x00 + .align 3, 0x90 + ret +# 5 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + ret + # nopl 0(%[re]ax,%[re]ax,1) + # 0x0f, 0x1f, 0x44, 0x00, 0x00 + .align 3, 0x90 + ret +# 6 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + # nopw 0(%[re]ax,%[re]ax,1) + # 0x66, 0x0f, 0x1f, 0x44, 0x00, 0x00 + .align 3, 0x90 + ret +# 7 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + # nopl 0L(%[re]ax) + # 0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00 + .align 3, 0x90 + ret +# 8 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + ret + ret + ret + ret + ret + ret + # nopl 0L(%[re]ax,%[re]ax,1) + # 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00 + .align 3, 0x90 + ret +# 9 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + ret + ret + ret + ret + ret + # nopw 0L(%[re]ax,%[re]ax,1) + # 0x66, 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00 + .align 4, 0x90 + ret +# 10 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + ret + ret + ret + ret + ret + # nopw %cs:0L(%[re]ax,%[re]ax,1) + # 0x66, 0x2e, 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00 + .align 4, 0x90 + ret +# 11 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + ret + ret + ret + # nopw %cs:0L(%[re]ax,%[re]ax,1) + # 0x66, 0x2e, 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00 + .align 4, 0x90 + ret +# 12 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + ret + ret + # nopw 0(%[re]ax,%[re]ax,1) + # nopw 0(%[re]ax,%[re]ax,1) + # 0x66, 0x0f, 0x1f, 0x44, 0x00, 0x00, + # 0x66, 0x0f, 0x1f, 0x44, 0x00, 0x00 + .align 4, 0x90 + ret +# 13 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + ret + # nopw 0(%[re]ax,%[re]ax,1) + # nopl 0L(%[re]ax) + # 0x66, 0x0f, 0x1f, 0x44, 0x00, 0x00, + # 0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00 + .align 4, 0x90 + ret +# 14 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + ret + # nopl 0L(%[re]ax) + # nopl 0L(%[re]ax) + # 0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00, + # 0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00 + .align 4, 0x90 + ret +# 15 byte nop test + .align 4, 0 # start with 16 byte alignment filled with zeros + ret + # nopl 0L(%[re]ax) + # nopl 0L(%[re]ax,%[re]ax,1) + # 0x0f, 0x1f, 0x80, 0x00, 0x00, 0x00, 0x00, + # 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00 + .align 4, 0x90 + ret diff --git a/test/Makefile b/test/Makefile index e7776f8..3750fdb 100644 --- a/test/Makefile +++ b/test/Makefile @@ -155,7 +155,6 @@ site.exp: FORCE @echo 'set target_triplet "$(TARGET_TRIPLE)"' >> site.tmp @echo 'set TARGETS_TO_BUILD "$(TARGETS_TO_BUILD)"' >> site.tmp @echo 'set llvmgcc_langs "$(LLVMGCC_LANGS)"' >> site.tmp - @echo 'set llvmgcc_version "$(LLVMGCC_VERSION)"' >> site.tmp @echo 'set llvmtoolsdir "$(ToolDir)"' >>site.tmp @echo 'set llvmlibsdir "$(LibDir)"' >>site.tmp @echo 'set llvm_bindings "$(BINDINGS_TO_BUILD)"' >> site.tmp @@ -170,7 +169,6 @@ site.exp: FORCE @echo 'set link "' $(CXX) $(CPP.Flags) $(CXX.Flags) $(TargetCommonOpts) $(CompileCommonOpts) $(LD.Flags) '"' >>site.tmp @echo 'set llvmgcc "$(LLVMGCC) $(TargetCommonOpts) $(EXTRA_OPTIONS)"' >> site.tmp @echo 'set llvmgxx "$(LLVMGCC) $(TargetCommonOpts) $(EXTRA_OPTIONS)"' >> site.tmp - @echo 'set llvmgccmajvers "$(LLVMGCC_MAJVERS)"' >> site.tmp @echo 'set bugpoint_topts $(BUGPOINT_TOPTS)' >> site.tmp @echo 'set shlibext "$(SHLIBEXT)"' >> site.tmp @echo 'set ocamlopt "$(OCAMLOPT) -cc \"$(CXX_FOR_OCAMLOPT)\" -I $(LibDir)/ocaml"' >> site.tmp @@ -200,4 +198,6 @@ Unit/lit.site.cfg: $(PROJ_OBJ_DIR)/Unit/.dir FORCE -e "s#@LLVM_TOOLS_DIR@#$(ToolDir)#g" \ -e "s#@LLVMGCCDIR@#$(LLVMGCCDIR)#g" \ -e "s#@LLVM_BUILD_MODE@#$(BuildMode)#g" \ + -e "s#@ENABLE_SHARED@#$(ENABLE_SHARED)#g" \ + -e "s#@SHLIBPATH_VAR@#$(SHLIBPATH_VAR)#g" \ $(PROJ_SRC_DIR)/Unit/lit.site.cfg.in > $@ diff --git a/test/Makefile.tests b/test/Makefile.tests index 90e9f2c..aeb5871 100644 --- a/test/Makefile.tests +++ b/test/Makefile.tests @@ -49,15 +49,15 @@ clean:: # Compile from X.c to Output/X.ll Output/%.ll: %.c $(LCC1) Output/.dir $(INCLUDES) - -$(LLVMGCCWITHPATH) $(CPPFLAGS) $(LCCFLAGS) -S $< -o $@ + -$(LLVMCC) $(CPPFLAGS) $(LCCFLAGS) -S $< -o $@ # Compile from X.cpp to Output/X.ll Output/%.ll: %.cpp $(LCC1XX) Output/.dir $(INCLUDES) - -$(LLVMGXXWITHPATH) $(CPPFLAGS) $(LCXXFLAGS) -S $< -o $@ + -$(LLVMCXX) $(CPPFLAGS) $(LCXXFLAGS) -S $< -o $@ # Compile from X.cc to Output/X.ll Output/%.ll: %.cc $(LCC1XX) Output/.dir $(INCLUDES) - -$(LLVMGXXWITHPATH) $(CPPFLAGS) $(LCXXFLAGS) -S $< -o $@ + -$(LLVMCXX) $(CPPFLAGS) $(LCXXFLAGS) -S $< -o $@ # LLVM Assemble from Output/X.ll to Output/X.bc. Output/X.ll must have come # from GCC output, so use GCCAS. diff --git a/test/Other/2008-03-19-PassManager.ll b/test/Other/2008-03-19-PassManager.ll deleted file mode 100644 index e208222..0000000 --- a/test/Other/2008-03-19-PassManager.ll +++ /dev/null @@ -1,58 +0,0 @@ -; PR 2034 -; RUN: opt < %s -anders-aa -instcombine -gvn -disable-output - %struct.FULL = type { i32, i32, [1000 x float*] } - -define i32 @sgesl(%struct.FULL* %a, i32* %ipvt, float* %b, i32 %job) { -entry: - %a_addr = alloca %struct.FULL* ; <%struct.FULL**> [#uses=1] - %ipvt_addr = alloca i32* ; <i32**> [#uses=1] - %b_addr = alloca float* ; <float**> [#uses=1] - %job_addr = alloca i32 ; <i32*> [#uses=1] - %akk = alloca float* ; <float**> [#uses=2] - %k = alloca i32 ; <i32*> [#uses=1] - %l = alloca i32 ; <i32*> [#uses=1] - %n = alloca i32 ; <i32*> [#uses=1] - %nm1 = alloca i32 ; <i32*> [#uses=1] - %tmp5 = load i32* %job_addr, align 4 ; <i32> [#uses=1] - %tmp6 = icmp eq i32 %tmp5, 0 ; <i1> [#uses=1] - %tmp67 = zext i1 %tmp6 to i8 ; <i8> [#uses=1] - %toBool = icmp ne i8 %tmp67, 0 ; <i1> [#uses=1] - br i1 %toBool, label %cond_true, label %cond_next137 - -cond_true: ; preds = %entry - %tmp732 = load i32* %nm1, align 4 ; <i32> [#uses=1] - %tmp743 = icmp slt i32 0, %tmp732 ; <i1> [#uses=1] - %tmp74754 = zext i1 %tmp743 to i8 ; <i8> [#uses=1] - %toBool765 = icmp ne i8 %tmp74754, 0 ; <i1> [#uses=1] - br i1 %toBool765, label %bb, label %bb77 - -bb: ; preds = %cond_true - %tmp9 = load %struct.FULL** %a_addr, align 4 ; <%struct.FULL*> [#uses=1] - %tmp10 = getelementptr %struct.FULL* %tmp9, i32 0, i32 2 ; <[1000 x float*]*> [#uses=1] - %tmp11 = getelementptr [1000 x float*]* %tmp10, i32 0, i32 0 ; <float**> [#uses=1] - %tmp12 = load float** %tmp11, align 4 ; <float*> [#uses=1] - %tmp13 = load i32* %k, align 4 ; <i32> [#uses=1] - %tmp14 = getelementptr float* %tmp12, i32 %tmp13 ; <float*> [#uses=1] - store float* %tmp14, float** %akk, align 4 - %tmp17 = load float** %b_addr, align 4 ; <float*> [#uses=0] - %tmp18 = load i32* %l, align 4 ; <i32> [#uses=0] - ret i32 0 - -bb77: ; preds = %cond_true - ret i32 0 - -cond_next137: ; preds = %entry - %tmp18922 = load i32* %n, align 4 ; <i32> [#uses=1] - %tmp19023 = icmp slt i32 0, %tmp18922 ; <i1> [#uses=1] - %tmp19019124 = zext i1 %tmp19023 to i8 ; <i8> [#uses=1] - %toBool19225 = icmp ne i8 %tmp19019124, 0 ; <i1> [#uses=1] - br i1 %toBool19225, label %bb138, label %bb193 - -bb138: ; preds = %cond_next137 - store float* null, float** %akk, align 4 - ret i32 0 - -bb193: ; preds = %cond_next137 - %tmp196 = load i32** %ipvt_addr, align 4 ; <i32*> [#uses=0] - ret i32 0 -} diff --git a/test/Other/constant-fold-gep.ll b/test/Other/constant-fold-gep.ll index 2888b3d..5358e1f 100644 --- a/test/Other/constant-fold-gep.ll +++ b/test/Other/constant-fold-gep.ll @@ -69,6 +69,8 @@ ; PLAIN: @g = constant i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) ; PLAIN: @h = constant i64 ptrtoint (i1** getelementptr (i1** null, i32 1) to i64) ; PLAIN: @i = constant i64 ptrtoint (i1** getelementptr (%2* null, i64 0, i32 1) to i64) +; PLAIN: @j = constant i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) +; PLAIN: @k = constant i64 ptrtoint (double* getelementptr (double* null, i32 1) to i64) ; OPT: @a = constant i64 mul (i64 ptrtoint (double* getelementptr (double* null, i32 1) to i64), i64 2310) ; OPT: @b = constant i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) ; OPT: @c = constant i64 mul (i64 ptrtoint (double* getelementptr (double* null, i32 1) to i64), i64 2) @@ -78,6 +80,8 @@ ; OPT: @g = constant i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) ; OPT: @h = constant i64 ptrtoint (i1** getelementptr (i1** null, i32 1) to i64) ; OPT: @i = constant i64 ptrtoint (i1** getelementptr (%2* null, i64 0, i32 1) to i64) +; OPT: @j = constant i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) +; OPT: @k = constant i64 ptrtoint (double* getelementptr (double* null, i32 1) to i64) ; TO: @a = constant i64 18480 ; TO: @b = constant i64 8 ; TO: @c = constant i64 16 @@ -87,6 +91,8 @@ ; TO: @g = constant i64 8 ; TO: @h = constant i64 8 ; TO: @i = constant i64 8 +; TO: @j = constant i64 8 +; TO: @k = constant i64 8 @a = constant i64 mul (i64 3, i64 mul (i64 ptrtoint ({[7 x double], [7 x double]}* getelementptr ({[7 x double], [7 x double]}* null, i64 11) to i64), i64 5)) @b = constant i64 ptrtoint ([13 x double]* getelementptr ({i1, [13 x double]}* null, i64 0, i32 1) to i64) @@ -97,6 +103,8 @@ @g = constant i64 ptrtoint ({double, double}* getelementptr ({i1, {double, double}}* null, i64 0, i32 1) to i64) @h = constant i64 ptrtoint (double** getelementptr (double** null, i64 1) to i64) @i = constant i64 ptrtoint (double** getelementptr ({i1, double*}* null, i64 0, i32 1) to i64) +@j = constant i64 ptrtoint (union {double, double}* getelementptr ({i1, union {double, double}}* null, i64 0, i32 1) to i64) +@k = constant i64 ptrtoint (union {double, double}* getelementptr (union {double, double}* null, i64 1) to i64) ; The target-dependent folder should cast GEP indices to integer-sized pointers. @@ -244,15 +252,23 @@ define i1* @hoo1() nounwind { ; PLAIN: ret i64 %t ; PLAIN: } ; PLAIN: define i64 @fg() nounwind { -; PLAIN: %t = bitcast i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) +; PLAIN: %t = bitcast i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) to i64 ; PLAIN: ret i64 %t ; PLAIN: } ; PLAIN: define i64 @fh() nounwind { -; PLAIN: %t = bitcast i64 ptrtoint (i1** getelementptr (i1** null, i32 1) to i64) +; PLAIN: %t = bitcast i64 ptrtoint (i1** getelementptr (i1** null, i32 1) to i64) to i64 ; PLAIN: ret i64 %t ; PLAIN: } ; PLAIN: define i64 @fi() nounwind { -; PLAIN: %t = bitcast i64 ptrtoint (i1** getelementptr (%2* null, i64 0, i32 1) to i64) +; PLAIN: %t = bitcast i64 ptrtoint (i1** getelementptr (%2* null, i64 0, i32 1) to i64) to i64 +; PLAIN: ret i64 %t +; PLAIN: } +; PLAIN: define i64 @fj() nounwind { +; PLAIN: %t = bitcast i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) to i64 +; PLAIN: ret i64 %t +; PLAIN: } +; PLAIN: define i64 @fk() nounwind { +; PLAIN: %t = bitcast i64 ptrtoint (double* getelementptr (double* null, i32 1) to i64) to i64 ; PLAIN: ret i64 %t ; PLAIN: } ; OPT: define i64 @fa() nounwind { @@ -282,6 +298,12 @@ define i1* @hoo1() nounwind { ; OPT: define i64 @fi() nounwind { ; OPT: ret i64 ptrtoint (i1** getelementptr (%2* null, i64 0, i32 1) to i64) ; OPT: } +; OPT: define i64 @fj() nounwind { +; OPT: ret i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) +; OPT: } +; OPT: define i64 @fk() nounwind { +; OPT: ret i64 ptrtoint (double* getelementptr (double* null, i32 1) to i64) +; OPT: } ; TO: define i64 @fa() nounwind { ; TO: ret i64 18480 ; TO: } @@ -309,6 +331,12 @@ define i1* @hoo1() nounwind { ; TO: define i64 @fi() nounwind { ; TO: ret i64 8 ; TO: } +; TO: define i64 @fj() nounwind { +; TO: ret i64 8 +; TO: } +; TO: define i64 @fk() nounwind { +; TO: ret i64 8 +; TO: } ; SCEV: Classifying expressions for: @fa ; SCEV: %t = bitcast i64 mul (i64 ptrtoint (double* getelementptr (double* null, i32 1) to i64), i64 2310) to i64 ; SCEV: --> (2310 * sizeof(double)) @@ -328,14 +356,20 @@ define i1* @hoo1() nounwind { ; SCEV: %t = bitcast i64 1 to i64 ; SCEV: --> 1 ; SCEV: Classifying expressions for: @fg -; SCEV: %t = bitcast i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) +; SCEV: %t = bitcast i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) to i64 ; SCEV: --> alignof(double) ; SCEV: Classifying expressions for: @fh -; SCEV: %t = bitcast i64 ptrtoint (i1** getelementptr (i1** null, i32 1) to i64) +; SCEV: %t = bitcast i64 ptrtoint (i1** getelementptr (i1** null, i32 1) to i64) to i64 ; SCEV: --> sizeof(i1*) ; SCEV: Classifying expressions for: @fi -; SCEV: %t = bitcast i64 ptrtoint (i1** getelementptr (%2* null, i64 0, i32 1) to i64) +; SCEV: %t = bitcast i64 ptrtoint (i1** getelementptr (%2* null, i64 0, i32 1) to i64) to i64 ; SCEV: --> alignof(i1*) +; SCEV: Classifying expressions for: @fj +; SCEV: %t = bitcast i64 ptrtoint (double* getelementptr (%0* null, i64 0, i32 1) to i64) to i64 +; SCEV: --> alignof(double) +; SCEV: Classifying expressions for: @fk +; SCEV: %t = bitcast i64 ptrtoint (double* getelementptr (double* null, i32 1) to i64) to i64 +; SCEV: --> sizeof(double) define i64 @fa() nounwind { %t = bitcast i64 mul (i64 3, i64 mul (i64 ptrtoint ({[7 x double], [7 x double]}* getelementptr ({[7 x double], [7 x double]}* null, i64 11) to i64), i64 5)) to i64 @@ -373,6 +407,14 @@ define i64 @fi() nounwind { %t = bitcast i64 ptrtoint (double** getelementptr ({i1, double*}* null, i64 0, i32 1) to i64) to i64 ret i64 %t } +define i64 @fj() nounwind { + %t = bitcast i64 ptrtoint (union {double, double}* getelementptr ({i1, union {double, double}}* null, i64 0, i32 1) to i64) to i64 + ret i64 %t +} +define i64 @fk() nounwind { + %t = bitcast i64 ptrtoint (union {double, double}* getelementptr (union {double, double}* null, i64 1) to i64) to i64 + ret i64 %t +} ; PLAIN: define i64* @fM() nounwind { ; PLAIN: %t = bitcast i64* getelementptr (i64* null, i32 1) to i64* diff --git a/test/Transforms/GVN/2008-02-13-NewPHI.ll b/test/Transforms/GVN/2008-02-13-NewPHI.ll index 54998db..80b519d 100644 --- a/test/Transforms/GVN/2008-02-13-NewPHI.ll +++ b/test/Transforms/GVN/2008-02-13-NewPHI.ll @@ -1,4 +1,4 @@ -; RUN: opt < %s -anders-aa -gvn +; RUN: opt < %s -gvn ; PR2032 define i32 @sscal(i32 %n, double %sa1, float* %sx, i32 %incx) { diff --git a/test/Transforms/GVN/2009-03-05-dbg.ll b/test/Transforms/GVN/2009-03-05-dbg.ll deleted file mode 100644 index cad3312..0000000 --- a/test/Transforms/GVN/2009-03-05-dbg.ll +++ /dev/null @@ -1,66 +0,0 @@ -; RUN: opt < %s -gvn -disable-output - %llvm.dbg.compile_unit.type = type { i32, { }*, i32, i8*, i8*, i8*, i1, i1, i8*, i32 } -@llvm.dbg.compile_unit298 = external constant %llvm.dbg.compile_unit.type ; <%llvm.dbg.compile_unit.type*> [#uses=1] - -declare void @llvm.dbg.stoppoint(i32, i32, { }*) nounwind - -define i8* @__deregister_frame_info_bases(i8* %begin) { -entry: - br i1 false, label %bb17, label %bb - -bb: ; preds = %entry - br i1 false, label %bb17, label %bb6.preheader - -bb6.preheader: ; preds = %bb - br label %bb6 - -bb3: ; preds = %bb6 - br i1 false, label %bb4, label %bb6 - -bb4: ; preds = %bb3 - br label %out - -bb6: ; preds = %bb3, %bb6.preheader - br i1 false, label %bb14.loopexit, label %bb3 - -bb8: ; preds = %bb14 - br i1 false, label %bb9, label %bb11 - -bb9: ; preds = %bb8 - br i1 false, label %bb10, label %bb13 - -bb10: ; preds = %bb9 - br label %out - -bb11: ; preds = %bb8 - br i1 false, label %bb12, label %bb13 - -bb12: ; preds = %bb11 - br label %out - -bb13: ; preds = %bb11, %bb9 - br label %bb14 - -bb14.loopexit: ; preds = %bb6 - br label %bb14 - -bb14: ; preds = %bb14.loopexit, %bb13 - br i1 false, label %bb15.loopexit, label %bb8 - -out: ; preds = %bb12, %bb10, %bb4 - tail call void @llvm.dbg.stoppoint(i32 217, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit298 to { }*)) - br i1 false, label %bb15, label %bb16 - -bb15.loopexit: ; preds = %bb14 - br label %bb15 - -bb15: ; preds = %bb15.loopexit, %out - tail call void @llvm.dbg.stoppoint(i32 217, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit298 to { }*)) - unreachable - -bb16: ; preds = %out - ret i8* null - -bb17: ; preds = %bb, %entry - ret i8* null -} diff --git a/test/Transforms/GVN/pre-load.ll b/test/Transforms/GVN/pre-load.ll index 7047d4e..d40a467 100644 --- a/test/Transforms/GVN/pre-load.ll +++ b/test/Transforms/GVN/pre-load.ll @@ -362,3 +362,30 @@ bb: return: ret void } + +; Test critical edge splitting. +define i32 @test11(i32* %p, i1 %C, i32 %N) { +; CHECK: @test11 +block1: + br i1 %C, label %block2, label %block3 + +block2: + %cond = icmp sgt i32 %N, 1 + br i1 %cond, label %block4, label %block5 +; CHECK: load i32* %p +; CHECK-NEXT: br label %block4 + +block3: + store i32 0, i32* %p + br label %block4 + +block4: + %PRE = load i32* %p + br label %block5 + +block5: + %ret = phi i32 [ 0, %block2 ], [ %PRE, %block4 ] + ret i32 %ret +; CHECK: block4: +; CHECK-NEXT: phi i32 +} diff --git a/test/Transforms/GlobalOpt/2009-03-03-dbg.ll b/test/Transforms/GlobalOpt/2009-03-03-dbg.ll deleted file mode 100644 index 070f89f..0000000 --- a/test/Transforms/GlobalOpt/2009-03-03-dbg.ll +++ /dev/null @@ -1,54 +0,0 @@ -; RUN: opt < %s -globalopt -S | not grep global_variable42 -; XFAIL: * - - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, { }*, i32, i8*, i8*, i8*, i1, i1, i8*, i32 } - %llvm.dbg.composite.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, { }*, { }*, i32 } - %llvm.dbg.derivedtype.type = type { i32, { }*, i8*, { }*, i32, i64, i64, i64, i32, { }* } - %llvm.dbg.global_variable.type = type { i32, { }*, { }*, i8*, i8*, i8*, { }*, i32, { }*, i1, i1, { }* } - %llvm.dbg.subprogram.type = type { i32, { }*, { }*, i8*, i8*, i8*, { }*, i32, { }*, i1, i1 } - %llvm.dbg.subrange.type = type { i32, i64, i64 } - %llvm.dbg.variable.type = type { i32, { }*, i8*, { }*, i32, { }* } -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str = internal constant [4 x i8] c"a.c\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@.str1 = internal constant [5 x i8] c"/tmp\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str2 = internal constant [57 x i8] c"4.2.1 (Based on Apple Inc. build 5636) (LLVM build 2099)\00", section "llvm.metadata" ; <[57 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { i32 458769, { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to { }*), i32 1, i8* getelementptr ([4 x i8]* @.str, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([57 x i8]* @.str2, i32 0, i32 0), i1 true, i1 false, i8* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str3 = internal constant [4 x i8] c"int\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type { i32 458788, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* getelementptr ([4 x i8]* @.str3, i32 0, i32 0), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 5 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array = internal constant [1 x { }*] [ { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*) ], section "llvm.metadata" ; <[1 x { }*]*> [#uses=1] -@llvm.dbg.composite = internal constant %llvm.dbg.composite.type { i32 458773, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* null, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 0, i64 0, i64 0, i64 0, i32 0, { }* null, { }* bitcast ([1 x { }*]* @llvm.dbg.array to { }*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str4 = internal constant [5 x i8] c"main\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.subprogram = internal constant %llvm.dbg.subprogram.type { i32 458798, { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to { }*), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* getelementptr ([5 x i8]* @.str4, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str4, i32 0, i32 0), i8* null, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 3, { }* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite to { }*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype = internal constant %llvm.dbg.derivedtype.type { i32 458767, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* null, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 0, i64 32, i64 32, i64 0, i32 0, { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str5 = internal constant [6 x i8] c"i_ptr\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.variable = internal constant %llvm.dbg.variable.type { i32 459008, { }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*), i8* getelementptr ([6 x i8]* @.str5, i32 0, i32 0), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 5, { }* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype to { }*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=0] -@sillyArray.1433 = internal global [8 x i32] [ i32 2, i32 3, i32 5, i32 7, i32 11, i32 13, i32 17, i32 19 ] ; <[8 x i32]*> [#uses=1] -@llvm.dbg.subrange = internal constant %llvm.dbg.subrange.type { i32 458785, i64 0, i64 7 }, section "llvm.metadata" ; <%llvm.dbg.subrange.type*> [#uses=1] -@llvm.dbg.array6 = internal constant [1 x { }*] [ { }* bitcast (%llvm.dbg.subrange.type* @llvm.dbg.subrange to { }*) ], section "llvm.metadata" ; <[1 x { }*]*> [#uses=1] -@llvm.dbg.composite7 = internal constant %llvm.dbg.composite.type { i32 458753, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* null, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 0, i64 256, i64 32, i64 0, i32 0, { }* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to { }*), { }* bitcast ([1 x { }*]* @llvm.dbg.array6 to { }*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.global_variables = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 52 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str8 = internal constant [16 x i8] c"sillyArray.1433\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@.str9 = internal constant [11 x i8] c"sillyArray\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.global_variable42 = internal constant %llvm.dbg.global_variable.type { i32 458804, { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.global_variables to { }*), { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i8* getelementptr ([16 x i8]* @.str8, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str9, i32 0, i32 0), i8* null, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*), i32 4, { }* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite7 to { }*), i1 true, i1 true, { }* bitcast ([8 x i32]* @sillyArray.1433 to { }*) }, section "llvm.metadata" ; <%llvm.dbg.global_variable.type*> [#uses=0] - -define i32 @main() nounwind { -entry: - %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] - call void @llvm.dbg.func.start({ }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*)) - call void @llvm.dbg.stoppoint(i32 5, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - call void @llvm.dbg.stoppoint(i32 6, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - call void @llvm.dbg.stoppoint(i32 6, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - call void @llvm.dbg.region.end({ }* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to { }*)) - ret i32 0 -} - -declare void @llvm.dbg.func.start({ }*) nounwind - -declare void @llvm.dbg.declare({ }*, { }*) nounwind - -declare void @llvm.dbg.stoppoint(i32, i32, { }*) nounwind - -declare void @llvm.dbg.region.end({ }*) nounwind diff --git a/test/Transforms/GlobalOpt/2009-03-05-dbg.ll b/test/Transforms/GlobalOpt/2009-03-05-dbg.ll index a5f9ed3..3154856 100644 --- a/test/Transforms/GlobalOpt/2009-03-05-dbg.ll +++ b/test/Transforms/GlobalOpt/2009-03-05-dbg.ll @@ -1,67 +1,76 @@ ; RUN: opt < %s -globalopt -stats -disable-output |& grep "1 globalopt - Number of global vars shrunk to booleans" -; XFAIL: * - type { } ; type %0 - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, %0*, i32, i8*, i8*, i8*, i1, i1, i8*, i32 } - %llvm.dbg.composite.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, %0*, %0*, i32 } - %llvm.dbg.global_variable.type = type { i32, %0*, %0*, i8*, i8*, i8*, %0*, i32, %0*, i1, i1, %0* } - %llvm.dbg.subprogram.type = type { i32, %0*, %0*, i8*, i8*, i8*, %0*, i32, %0*, i1, i1 } - %llvm.dbg.variable.type = type { i32, %0*, i8*, %0*, i32, %0* } -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str = internal constant [5 x i8] c"gs.c\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str1 = internal constant [6 x i8] c"/tmp/\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str2 = internal constant [55 x i8] c"4.2.1 (Based on Apple Inc. build 5641) (LLVM build 00)\00", section "llvm.metadata" ; <[55 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 1, i8* getelementptr ([5 x i8]* @.str, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([55 x i8]* @.str2, i32 0, i32 0), i1 true, i1 false, i8* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str3 = internal constant [4 x i8] c"int\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str3, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 5 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str4 = internal constant [4 x i8] c"foo\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.subprogram = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str4, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str4, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 4, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str5 = internal constant [2 x i8] c"i\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.variable = internal constant %llvm.dbg.variable.type { i32 459009, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to %0*), i8* getelementptr ([2 x i8]* @.str5, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 4, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@Stop = internal global i32 0 ; <i32*> [#uses=4] -@llvm.dbg.global_variables = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 52 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str6 = internal constant [5 x i8] c"Stop\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.global_variable = internal constant %llvm.dbg.global_variable.type { i32 458804, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.global_variables to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str6, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str6, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 2, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*), i1 true, i1 true, %0* bitcast (i32* @Stop to %0*) }, section "llvm.metadata" ; <%llvm.dbg.global_variable.type*> [#uses=0] +@Stop = internal global i32 0 ; <i32*> [#uses=3] -define i32 @foo(i32 %i) nounwind { +define i32 @foo(i32 %i) nounwind ssp { entry: - %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to %0*)) - call void @llvm.dbg.stoppoint(i32 5, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = load i32* @Stop, align 4 ; <i32> [#uses=1] - %1 = icmp eq i32 %0, 1 ; <i1> [#uses=1] - br i1 %1, label %bb, label %bb1 + %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] + call void @llvm.dbg.value(metadata !{i32 %i}, i64 0, metadata !3) + %0 = icmp eq i32 %i, 1, !dbg !7 ; <i1> [#uses=1] + br i1 %0, label %bb, label %bb1, !dbg !7 -bb: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 6, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - store i32 0, i32* @Stop, align 4 - call void @llvm.dbg.stoppoint(i32 7, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %2 = mul i32 %i, 42 ; <i32> [#uses=1] - br label %bb2 +bb: ; preds = %entry + store i32 0, i32* @Stop, align 4, !dbg !9 + %1 = mul nsw i32 %i, 42, !dbg !10 ; <i32> [#uses=1] + call void @llvm.dbg.value(metadata !{i32 %1}, i64 0, metadata !3), !dbg !10 + br label %bb2, !dbg !10 -bb1: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 9, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - store i32 1, i32* @Stop, align 4 - call void @llvm.dbg.stoppoint(i32 10, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb2 +bb1: ; preds = %entry + store i32 1, i32* @Stop, align 4, !dbg !11 + br label %bb2, !dbg !11 -bb2: ; preds = %bb1, %bb - %.0 = phi i32 [ %i, %bb1 ], [ %2, %bb ] ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 10, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.stoppoint(i32 10, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to %0*)) - ret i32 %.0 +bb2: ; preds = %bb1, %bb + %i_addr.0 = phi i32 [ %1, %bb ], [ %i, %bb1 ] ; <i32> [#uses=1] + br label %return, !dbg !12 + +return: ; preds = %bb2 + ret i32 %i_addr.0, !dbg !12 } -declare void @llvm.dbg.func.start(%0*) nounwind readnone +declare void @llvm.dbg.declare(metadata, metadata) nounwind readnone + +define i32 @bar() nounwind ssp { +entry: + %"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0] + %0 = load i32* @Stop, align 4, !dbg !13 ; <i32> [#uses=1] + %1 = icmp eq i32 %0, 1, !dbg !13 ; <i1> [#uses=1] + br i1 %1, label %bb, label %bb1, !dbg !13 + +bb: ; preds = %entry + br label %bb2, !dbg !18 + +bb1: ; preds = %entry + br label %bb2, !dbg !19 + +bb2: ; preds = %bb1, %bb + %.0 = phi i32 [ 0, %bb ], [ 1, %bb1 ] ; <i32> [#uses=1] + br label %return, !dbg !19 + +return: ; preds = %bb2 + ret i32 %.0, !dbg !19 +} -declare void @llvm.dbg.declare(%0*, %0*) nounwind readnone +declare void @llvm.dbg.value(metadata, i64, metadata) nounwind readnone -declare void @llvm.dbg.stoppoint(i32, i32, %0*) nounwind readnone +!llvm.dbg.gv = !{!0} -declare void @llvm.dbg.region.end(%0*) nounwind readnone +!0 = metadata !{i32 458804, i32 0, metadata !1, metadata !"Stop", metadata !"Stop", metadata !"", metadata !1, i32 2, metadata !2, i1 true, i1 true, i32* @Stop} ; [ DW_TAG_variable ] +!1 = metadata !{i32 458769, i32 0, i32 1, metadata !"g.c", metadata !"/tmp", metadata !"4.2.1 (Based on Apple Inc. build 5658) (LLVM build)", i1 true, i1 false, metadata !"", i32 0} ; [ DW_TAG_compile_unit ] +!2 = metadata !{i32 458788, metadata !1, metadata !"int", metadata !1, i32 0, i64 32, i64 32, i64 0, i32 0, i32 5} ; [ DW_TAG_base_type ] +!3 = metadata !{i32 459009, metadata !4, metadata !"i", metadata !1, i32 4, metadata !2} ; [ DW_TAG_arg_variable ] +!4 = metadata !{i32 458798, i32 0, metadata !1, metadata !"foo", metadata !"foo", metadata !"foo", metadata !1, i32 4, metadata !5, i1 false, i1 true, i32 0, i32 0, null, i1 false} ; [ DW_TAG_subprogram ] +!5 = metadata !{i32 458773, metadata !1, metadata !"", metadata !1, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !6, i32 0, null} ; [ DW_TAG_subroutine_type ] +!6 = metadata !{metadata !2, metadata !2} +!7 = metadata !{i32 5, i32 0, metadata !8, null} +!8 = metadata !{i32 458763, metadata !4, i32 0, i32 0} ; [ DW_TAG_lexical_block ] +!9 = metadata !{i32 6, i32 0, metadata !8, null} +!10 = metadata !{i32 7, i32 0, metadata !8, null} +!11 = metadata !{i32 9, i32 0, metadata !8, null} +!12 = metadata !{i32 11, i32 0, metadata !8, null} +!13 = metadata !{i32 14, i32 0, metadata !14, null} +!14 = metadata !{i32 458763, metadata !15, i32 0, i32 0} ; [ DW_TAG_lexical_block ] +!15 = metadata !{i32 458798, i32 0, metadata !1, metadata !"bar", metadata !"bar", metadata !"bar", metadata !1, i32 13, metadata !16, i1 false, i1 true, i32 0, i32 0, null, i1 false} ; [ DW_TAG_subprogram ] +!16 = metadata !{i32 458773, metadata !1, metadata !"", metadata !1, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !17, i32 0, null} ; [ DW_TAG_subroutine_type ] +!17 = metadata !{metadata !2} +!18 = metadata !{i32 15, i32 0, metadata !14, null} +!19 = metadata !{i32 16, i32 0, metadata !14, null} diff --git a/test/Transforms/GlobalOpt/2010-02-25-MallocPromote.ll b/test/Transforms/GlobalOpt/2010-02-25-MallocPromote.ll new file mode 100644 index 0000000..27352fa --- /dev/null +++ b/test/Transforms/GlobalOpt/2010-02-25-MallocPromote.ll @@ -0,0 +1,18 @@ +; PR6422 +; RUN: opt -globalopt -S %s +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" +target triple = "x86_64-unknown-linux-gnu" + +@fixLRBT = internal global i32* null ; <i32**> [#uses=2] + +declare noalias i8* @malloc(i32) + +define i32 @parser() nounwind { +bb918: + %malloccall.i10 = call i8* @malloc(i32 16) nounwind ; <i8*> [#uses=1] + %0 = bitcast i8* %malloccall.i10 to i32* ; <i32*> [#uses=1] + store i32* %0, i32** @fixLRBT, align 8 + %1 = load i32** @fixLRBT, align 8 ; <i32*> [#uses=0] + %A = load i32* %1 + ret i32 %A +} diff --git a/test/Transforms/GlobalOpt/2010-02-26-MallocSROA.ll b/test/Transforms/GlobalOpt/2010-02-26-MallocSROA.ll new file mode 100644 index 0000000..6f1996a --- /dev/null +++ b/test/Transforms/GlobalOpt/2010-02-26-MallocSROA.ll @@ -0,0 +1,27 @@ +; RUN: opt -globalopt -S %s +; PR6435 +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" +target triple = "x86_64-unknown-linux-gnu" + +%struct.xyz = type { double, i32 } + +@Y = internal global %struct.xyz* null ; <%struct.xyz**> [#uses=2] +@numf2s = external global i32 ; <i32*> [#uses=1] + +define fastcc void @init_net() nounwind { +entry: + %0 = load i32* @numf2s, align 4 ; <i32> [#uses=1] + %mallocsize2 = shl i32 %0, 4 ; <i32> [#uses=1] + %malloccall3 = tail call i8* @malloc(i32 %mallocsize2) nounwind ; <i8*> [#uses=1] + %1 = bitcast i8* %malloccall3 to %struct.xyz* ; <%struct.xyz*> [#uses=1] + store %struct.xyz* %1, %struct.xyz** @Y, align 8 + ret void +} + +define fastcc void @load_train(i8* %trainfile, i32 %mode, i32 %objects) nounwind { +entry: + %0 = load %struct.xyz** @Y, align 8 ; <%struct.xyz*> [#uses=0] + ret void +} + +declare noalias i8* @malloc(i32) diff --git a/test/Transforms/GlobalOpt/ctor-list-opt-dbg.ll b/test/Transforms/GlobalOpt/ctor-list-opt-dbg.ll deleted file mode 100644 index f794e9f..0000000 --- a/test/Transforms/GlobalOpt/ctor-list-opt-dbg.ll +++ /dev/null @@ -1,98 +0,0 @@ -; RUN: opt < %s -globalopt -S | not grep CTOR -@llvm.global_ctors = appending global [10 x { i32, void ()* }] [ { i32, void ()* } { i32 65535, void ()* @CTOR1 }, { i32, void ()* } { i32 65535, void ()* @CTOR1 }, { i32, void ()* } { i32 65535, void ()* @CTOR2 }, { i32, void ()* } { i32 65535, void ()* @CTOR3 }, { i32, void ()* } { i32 65535, void ()* @CTOR4 }, { i32, void ()* } { i32 65535, void ()* @CTOR5 }, { i32, void ()* } { i32 65535, void ()* @CTOR6 }, { i32, void ()* } { i32 65535, void ()* @CTOR7 }, { i32, void ()* } { i32 65535, void ()* @CTOR8 }, { i32, void ()* } { i32 2147483647, void ()* null } ] ; <[10 x { i32, void ()* }]*> [#uses=0] -@G = global i32 0 ; <i32*> [#uses=1] -@G2 = global i32 0 ; <i32*> [#uses=1] -@G3 = global i32 -123 ; <i32*> [#uses=2] -@X = global { i32, [2 x i32] } { i32 0, [2 x i32] [ i32 17, i32 21 ] } ; <{ i32, [2 x i32] }*> [#uses=2] -@Y = global i32 -1 ; <i32*> [#uses=2] -@Z = global i32 123 ; <i32*> [#uses=1] -@D = global double 0.000000e+00 ; <double*> [#uses=1] -@CTORGV = internal global i1 false ; <i1*> [#uses=2] - - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, { }*, i32, i8*, i8*, i8*, i1, i1, i8* } - -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 17 }, section "llvm.metadata" - -@.str = internal constant [4 x i8] c"a.c\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@.str1 = internal constant [6 x i8] c"/tmp/\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str2 = internal constant [55 x i8] c"4.2.1 (Based on Apple Inc. build 5636) (LLVM build 00)\00", section "llvm.metadata" ; <[55 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { i32 458769, { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to { }*), i32 1, i8* getelementptr ([4 x i8]* @.str, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([55 x i8]* @.str2, i32 0, i32 0), i1 true, i1 false, i8* null }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] - -declare void @llvm.dbg.stoppoint(i32, i32, { }*) nounwind - -define internal void @CTOR1() { - ret void -} - -define internal void @CTOR2() { - %A = add i32 1, 23 ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 5, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - store i32 %A, i32* @G - store i1 true, i1* @CTORGV - ret void -} - -define internal void @CTOR3() { - %X = or i1 true, false ; <i1> [#uses=1] - br label %Cont - -Cont: ; preds = %0 - br i1 %X, label %S, label %T - -S: ; preds = %Cont - store i32 24, i32* @G2 - ret void - -T: ; preds = %Cont - ret void -} - -define internal void @CTOR4() { - %X = load i32* @G3 ; <i32> [#uses=1] - %Y = add i32 %X, 123 ; <i32> [#uses=1] - store i32 %Y, i32* @G3 - ret void -} - -define internal void @CTOR5() { - %X.2p = getelementptr inbounds { i32, [2 x i32] }* @X, i32 0, i32 1, i32 0 ; <i32*> [#uses=2] - %X.2 = load i32* %X.2p ; <i32> [#uses=1] - %X.1p = getelementptr inbounds { i32, [2 x i32] }* @X, i32 0, i32 0 ; <i32*> [#uses=1] - store i32 %X.2, i32* %X.1p - store i32 42, i32* %X.2p - ret void -} - -define internal void @CTOR6() { - %A = alloca i32 ; <i32*> [#uses=2] - %y = load i32* @Y ; <i32> [#uses=1] - store i32 %y, i32* %A - %Av = load i32* %A ; <i32> [#uses=1] - %Av1 = add i32 %Av, 1 ; <i32> [#uses=1] - store i32 %Av1, i32* @Y - ret void -} - -define internal void @CTOR7() { - call void @setto( i32* @Z, i32 0 ) - ret void -} - -define void @setto(i32* %P, i32 %V) { - store i32 %V, i32* %P - ret void -} - -declare double @cos(double) - -define internal void @CTOR8() { - %X = call double @cos( double 1.000000e+00 ) ; <double> [#uses=1] - store double %X, double* @D - ret void -} - -define i1 @accessor() { - %V = load i1* @CTORGV ; <i1> [#uses=1] - ret i1 %V -} diff --git a/test/Transforms/IndVarSimplify/2003-09-12-MultiplePred.ll b/test/Transforms/IndVarSimplify/2003-09-12-MultiplePred.ll index 36d5006..ecd5086 100644 --- a/test/Transforms/IndVarSimplify/2003-09-12-MultiplePred.ll +++ b/test/Transforms/IndVarSimplify/2003-09-12-MultiplePred.ll @@ -7,7 +7,7 @@ define i32 @test() { LoopHead: ; preds = %LoopHead, %0, %0 %A = phi i32 [ 7, %0 ], [ 7, %0 ], [ %B, %LoopHead ] ; <i32> [#uses=1] %B = add i32 %A, 1 ; <i32> [#uses=2] - br i1 false, label %LoopHead, label %Out + br i1 true, label %LoopHead, label %Out Out: ; preds = %LoopHead ret i32 %B diff --git a/test/Transforms/IndVarSimplify/2003-12-10-RemoveInstrCrash.ll b/test/Transforms/IndVarSimplify/2003-12-10-RemoveInstrCrash.ll index 70ea11e..7536331 100644 --- a/test/Transforms/IndVarSimplify/2003-12-10-RemoveInstrCrash.ll +++ b/test/Transforms/IndVarSimplify/2003-12-10-RemoveInstrCrash.ll @@ -10,7 +10,7 @@ no_exit: ; preds = %no_exit, %entry %k.0.pn = phi i32 [ %inc.4, %no_exit ], [ 1, %entry ] ; <i32> [#uses=1] %inc.3 = add i32 %j.0.pn, 1 ; <i32> [#uses=1] %inc.4 = add i32 %k.0.pn, 1 ; <i32> [#uses=1] - br i1 false, label %no_exit, label %loopexit + br i1 undef, label %no_exit, label %loopexit loopexit: ; preds = %no_exit, %entry ret void diff --git a/test/Transforms/IndVarSimplify/2003-12-15-Crash.ll b/test/Transforms/IndVarSimplify/2003-12-15-Crash.ll index 5aa2d90..662828c 100644 --- a/test/Transforms/IndVarSimplify/2003-12-15-Crash.ll +++ b/test/Transforms/IndVarSimplify/2003-12-15-Crash.ll @@ -8,7 +8,7 @@ cond_continue.3: ; preds = %entry loopexit.14: ; preds = %entry %tmp.738 = sub i32 0, 0 ; <i32> [#uses=1] - br i1 false, label %no_exit.15.preheader, label %loopexit.15 + br i1 undef, label %no_exit.15.preheader, label %loopexit.15 no_exit.15.preheader: ; preds = %loopexit.14 br label %no_exit.15 @@ -16,7 +16,7 @@ no_exit.15.preheader: ; preds = %loopexit.14 no_exit.15: ; preds = %no_exit.15, %no_exit.15.preheader %highC.0 = phi i32 [ %tmp.738, %no_exit.15.preheader ], [ %dec.0, %no_exit.15 ] ; <i32> [#uses=1] %dec.0 = add i32 %highC.0, -1 ; <i32> [#uses=1] - br i1 false, label %no_exit.15, label %loopexit.15 + br i1 undef, label %no_exit.15, label %loopexit.15 loopexit.15: ; preds = %no_exit.15, %loopexit.14 ret void diff --git a/test/Transforms/IndVarSimplify/2005-11-18-Crash.ll b/test/Transforms/IndVarSimplify/2005-11-18-Crash.ll index f9a3fe6..7202c7a 100644 --- a/test/Transforms/IndVarSimplify/2005-11-18-Crash.ll +++ b/test/Transforms/IndVarSimplify/2005-11-18-Crash.ll @@ -9,7 +9,7 @@ entry: no_exit.0: ; preds = %no_exit.0, %entry %p.0.0 = phi i32* [ getelementptr ([29 x [29 x [2 x i32]]]* @fixtab, i32 0, i32 0, i32 0, i32 0), %entry ], [ %inc.0, %no_exit.0 ] ; <i32*> [#uses=1] %inc.0 = getelementptr i32* %p.0.0, i32 1 ; <i32*> [#uses=1] - br i1 false, label %no_exit.0, label %no_exit.1 + br i1 undef, label %no_exit.0, label %no_exit.1 no_exit.1: ; preds = %no_exit.0 ret void diff --git a/test/Transforms/IndVarSimplify/2006-12-10-BitCast.ll b/test/Transforms/IndVarSimplify/2006-12-10-BitCast.ll index 79ac1f0..80c9ebf 100644 --- a/test/Transforms/IndVarSimplify/2006-12-10-BitCast.ll +++ b/test/Transforms/IndVarSimplify/2006-12-10-BitCast.ll @@ -20,7 +20,7 @@ cond_next182.i: ; preds = %cond_next182.i, %cond_true52 %tmp194.i53 = bitcast i32 %decay.i.0 to float ; <float> [#uses=1] %tmp195.i = fsub float %tmp194.i53, 8.000000e+00 ; <float> [#uses=1] %tmp195.i.upgrd.1 = bitcast float %tmp195.i to i32 ; <i32> [#uses=1] - br i1 false, label %cond_next182.i, label %bb418.i.preheader + br i1 undef, label %cond_next182.i, label %bb418.i.preheader bb418.i.preheader: ; preds = %cond_next182.i ret void diff --git a/test/Transforms/IndVarSimplify/2009-05-24-useafterfree.ll b/test/Transforms/IndVarSimplify/2009-05-24-useafterfree.ll index 9ad8691..d73eee8 100644 --- a/test/Transforms/IndVarSimplify/2009-05-24-useafterfree.ll +++ b/test/Transforms/IndVarSimplify/2009-05-24-useafterfree.ll @@ -12,7 +12,7 @@ bb.nph1.preheader: ; preds = %4 br label %bb.nph1 bb.nph1: ; preds = %.outer, %bb.nph1.preheader - br i1 false, label %bb.nph3.preheader, label %.outer + br i1 undef, label %bb.nph3.preheader, label %.outer bb.nph3.preheader: ; preds = %bb.nph1 br label %bb.nph3 @@ -31,7 +31,7 @@ bb.nph3: ; preds = %bb.nph3, %bb.nph3.preheader br label %.outer .outer: ; preds = %.outer.loopexit, %bb.nph1 - br i1 false, label %bb.nph1, label %.outer._crit_edge.loopexit + br i1 undef, label %bb.nph1, label %.outer._crit_edge.loopexit .outer._crit_edge.loopexit: ; preds = %.outer br label %.outer._crit_edge diff --git a/test/Transforms/IndVarSimplify/avoid-i0.ll b/test/Transforms/IndVarSimplify/avoid-i0.ll index d110a8a..59661fa 100644 --- a/test/Transforms/IndVarSimplify/avoid-i0.ll +++ b/test/Transforms/IndVarSimplify/avoid-i0.ll @@ -10,13 +10,13 @@ entry: bb: ; preds = %bb6, %entry %p_71_addr.0 = phi i8 [ %p_71, %entry ], [ %0, %bb6 ] ; <i8> [#uses=0] - br i1 false, label %bb4, label %bb1 + br i1 undef, label %bb4, label %bb1 bb1: ; preds = %bb ret i32 0 bb4: ; preds = %bb4, %bb - br i1 false, label %bb6, label %bb4 + br i1 undef, label %bb6, label %bb4 bb6: ; preds = %bb4 %0 = and i8 0, 0 ; <i8> [#uses=1] diff --git a/test/Transforms/IndVarSimplify/max-pointer.ll b/test/Transforms/IndVarSimplify/max-pointer.ll index 71bc720..f18f968 100644 --- a/test/Transforms/IndVarSimplify/max-pointer.ll +++ b/test/Transforms/IndVarSimplify/max-pointer.ll @@ -16,7 +16,7 @@ bb2: ; preds = %bb2, %entry %str2Ptr_addr.1 = phi i8* [ %str2Ptr_addr.0, %entry ], [ %1, %bb2 ] ; <i8*> [#uses=1] %1 = getelementptr i8* %str2Ptr_addr.1, i64 1 ; <i8*> [#uses=2] %2 = icmp ult i8* %1, %inLastBytePtr ; <i1> [#uses=0] - br i1 false, label %bb2, label %return + br i1 undef, label %bb2, label %return return: ; preds = %bb2 ret void @@ -32,7 +32,7 @@ bb2: ; preds = %bb2, %entry %str2Ptr_addr.1 = phi i8* [ %str2Ptr_addr.0, %entry ], [ %1, %bb2 ] ; <i8*> [#uses=1] %1 = getelementptr i8* %str2Ptr_addr.1, i64 1 ; <i8*> [#uses=2] %2 = icmp slt i8* %1, %inLastBytePtr ; <i1> [#uses=0] - br i1 false, label %bb2, label %return + br i1 undef, label %bb2, label %return return: ; preds = %bb2 ret void diff --git a/test/Transforms/InstCombine/2009-01-19-fmod-constant-float-specials.ll b/test/Transforms/InstCombine/2009-01-19-fmod-constant-float-specials.ll index 79a2f1f..1421347 100644 --- a/test/Transforms/InstCombine/2009-01-19-fmod-constant-float-specials.ll +++ b/test/Transforms/InstCombine/2009-01-19-fmod-constant-float-specials.ll @@ -1,5 +1,4 @@ -; RUN: opt < %s -simplifycfg -instcombine -S | grep 0x7FF8000000000000 | count 7 -; RUN: opt < %s -simplifycfg -instcombine -S | grep 0x7FF00000FFFFFFFF | count 5 +; RUN: opt < %s -simplifycfg -instcombine -S | grep 0x7FF8000000000000 | count 12 ; RUN: opt < %s -simplifycfg -instcombine -S | grep {0\\.0} | count 3 ; RUN: opt < %s -simplifycfg -instcombine -S | grep {3\\.5} | count 1 ; diff --git a/test/Transforms/InstCombine/bswap-fold.ll b/test/Transforms/InstCombine/bswap-fold.ll index 034c70e..a6b30c0 100644 --- a/test/Transforms/InstCombine/bswap-fold.ll +++ b/test/Transforms/InstCombine/bswap-fold.ll @@ -67,3 +67,9 @@ define i16 @test8(i64 %A) { %D = tail call i16 @llvm.bswap.i16(i16 %C) nounwind ret i16 %D } + +; Misc: Fold bswap(undef) to undef. +define i64 @foo() { + %a = call i64 @llvm.bswap.i64(i64 undef) + ret i64 %a +} diff --git a/test/Transforms/InstCombine/constant-fold-ptr-casts.ll b/test/Transforms/InstCombine/constant-fold-ptr-casts.ll deleted file mode 100644 index 9b6c6c3..0000000 --- a/test/Transforms/InstCombine/constant-fold-ptr-casts.ll +++ /dev/null @@ -1,18 +0,0 @@ -; RUN: opt < %s -instcombine -S | grep {ret i32 2143034560} - -; Instcombine should be able to completely fold this code. - -target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" -target triple = "i686-apple-darwin8" - -@bar = constant [3 x i64] [i64 9220983451228067448, i64 9220983451228067449, i64 9220983450959631991], align 8 - -define i32 @foo() nounwind { -entry: - %tmp87.2 = load i64* inttoptr (i32 add (i32 16, i32 ptrtoint ([3 x i64]* @bar to i32)) to i64*), align 8 - %t0 = bitcast i64 %tmp87.2 to double - %tmp9192.2 = fptrunc double %t0 to float - %t1 = bitcast float %tmp9192.2 to i32 - ret i32 %t1 -} - diff --git a/test/Transforms/InstCombine/fcmp-select.ll b/test/Transforms/InstCombine/fcmp-select.ll new file mode 100644 index 0000000..e04ab3e --- /dev/null +++ b/test/Transforms/InstCombine/fcmp-select.ll @@ -0,0 +1,53 @@ +; RUN: opt < %s -instcombine -S | FileCheck %s + +; x != y ? x : y -> x if it's the right kind of != and at least +; one of x and y is not negative zero. + +; CHECK: f0 +; CHECK: ret double %x +define double @f0(double %x) nounwind readnone { +entry: + %cmp = fcmp une double %x, -1.0 + %cond = select i1 %cmp, double %x, double -1.0 + ret double %cond +} +; CHECK: f1 +; CHECK: ret double -1.000000e+00 +define double @f1(double %x) nounwind readnone { +entry: + %cmp = fcmp une double %x, -1.0 + %cond = select i1 %cmp, double -1.0, double %x + ret double %cond +} +; CHECK: f2 +; CHECK: ret double %cond +define double @f2(double %x, double %y) nounwind readnone { +entry: + %cmp = fcmp une double %x, %y + %cond = select i1 %cmp, double %x, double %y + ret double %cond +} +; CHECK: f3 +; CHECK: ret double %cond +define double @f3(double %x, double %y) nounwind readnone { +entry: + %cmp = fcmp une double %x, %y + %cond = select i1 %cmp, double %y, double %x + ret double %cond +} +; CHECK: f4 +; CHECK: ret double %cond +define double @f4(double %x) nounwind readnone { +entry: + %cmp = fcmp one double %x, -1.0 + %cond = select i1 %cmp, double %x, double -1.0 + ret double %cond +} +; CHECK: f5 +; CHECK: ret double %cond +define double @f5(double %x) nounwind readnone { +entry: + %cmp = fcmp one double %x, -1.0 + %cond = select i1 %cmp, double -1.0, double %x + ret double %cond +} diff --git a/test/Transforms/InstCombine/fcmp-special.ll b/test/Transforms/InstCombine/fcmp-special.ll new file mode 100644 index 0000000..a39021e --- /dev/null +++ b/test/Transforms/InstCombine/fcmp-special.ll @@ -0,0 +1,155 @@ +; RUN: opt < %s -instcombine -S | FileCheck %s + +; Infinity + +; CHECK: inf0 +; CHECK: ret i1 false +define i1 @inf0(double %arg) nounwind readnone { + %tmp = fcmp ogt double %arg, 0x7FF0000000000000 + ret i1 %tmp +} + +; CHECK: inf1 +; CHECK: ret i1 true +define i1 @inf1(double %arg) nounwind readnone { + %tmp = fcmp ule double %arg, 0x7FF0000000000000 + ret i1 %tmp +} + +; Negative infinity + +; CHECK: ninf0 +; CHECK: ret i1 false +define i1 @ninf0(double %arg) nounwind readnone { + %tmp = fcmp olt double %arg, 0xFFF0000000000000 + ret i1 %tmp +} + +; CHECK: ninf1 +; CHECK: ret i1 true +define i1 @ninf1(double %arg) nounwind readnone { + %tmp = fcmp uge double %arg, 0xFFF0000000000000 + ret i1 %tmp +} + +; NaNs + +; CHECK: nan0 +; CHECK: ret i1 false +define i1 @nan0(double %arg) nounwind readnone { + %tmp = fcmp ord double %arg, 0x7FF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nan1 +; CHECK: ret i1 false +define i1 @nan1(double %arg) nounwind readnone { + %tmp = fcmp oeq double %arg, 0x7FF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nan2 +; CHECK: ret i1 false +define i1 @nan2(double %arg) nounwind readnone { + %tmp = fcmp olt double %arg, 0x7FF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nan3 +; CHECK: ret i1 true +define i1 @nan3(double %arg) nounwind readnone { + %tmp = fcmp uno double %arg, 0x7FF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nan4 +; CHECK: ret i1 true +define i1 @nan4(double %arg) nounwind readnone { + %tmp = fcmp une double %arg, 0x7FF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nan5 +; CHECK: ret i1 true +define i1 @nan5(double %arg) nounwind readnone { + %tmp = fcmp ult double %arg, 0x7FF00000FFFFFFFF + ret i1 %tmp +} + +; Negative NaN. + +; CHECK: nnan0 +; CHECK: ret i1 false +define i1 @nnan0(double %arg) nounwind readnone { + %tmp = fcmp ord double %arg, 0xFFF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nnan1 +; CHECK: ret i1 false +define i1 @nnan1(double %arg) nounwind readnone { + %tmp = fcmp oeq double %arg, 0xFFF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nnan2 +; CHECK: ret i1 false +define i1 @nnan2(double %arg) nounwind readnone { + %tmp = fcmp olt double %arg, 0xFFF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nnan3 +; CHECK: ret i1 true +define i1 @nnan3(double %arg) nounwind readnone { + %tmp = fcmp uno double %arg, 0xFFF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nnan4 +; CHECK: ret i1 true +define i1 @nnan4(double %arg) nounwind readnone { + %tmp = fcmp une double %arg, 0xFFF00000FFFFFFFF + ret i1 %tmp +} + +; CHECK: nnan5 +; CHECK: ret i1 true +define i1 @nnan5(double %arg) nounwind readnone { + %tmp = fcmp ult double %arg, 0xFFF00000FFFFFFFF + ret i1 %tmp +} + +; Negative zero. + +; CHECK: nzero0 +; CHECK: ret i1 true +define i1 @nzero0() { + %tmp = fcmp oeq double 0.0, -0.0 + ret i1 %tmp +} + +; CHECK: nzero1 +; CHECK: ret i1 false +define i1 @nzero1() { + %tmp = fcmp ogt double 0.0, -0.0 + ret i1 %tmp +} + +; Misc. + +; CHECK: misc0 +; CHECK: %tmp = fcmp ord double %arg, 0.000000e+00 +; CHECK: ret i1 %tmp +define i1 @misc0(double %arg) { + %tmp = fcmp oeq double %arg, %arg + ret i1 %tmp +} + +; CHECK: misc1 +; CHECK: ret i1 false +define i1 @misc1(double %arg) { + %tmp = fcmp one double %arg, %arg + ret i1 %tmp +} + diff --git a/test/Transforms/InstCombine/multi-use-or.ll b/test/Transforms/InstCombine/multi-use-or.ll index 9bbef23..8c6a0e0 100644 --- a/test/Transforms/InstCombine/multi-use-or.ll +++ b/test/Transforms/InstCombine/multi-use-or.ll @@ -1,4 +1,4 @@ -; RUN: opt < %s -instcombine -S | grep {add double .sx, .sy} +; RUN: opt < %s -instcombine -S | grep {fadd double .sx, .sy} ; The 'or' has multiple uses, make sure that this doesn't prevent instcombine ; from propagating the extends to the truncs. diff --git a/test/Transforms/InstCombine/objsize.ll b/test/Transforms/InstCombine/objsize.ll index dbc7d31..9df1224 100644 --- a/test/Transforms/InstCombine/objsize.ll +++ b/test/Transforms/InstCombine/objsize.ll @@ -23,12 +23,12 @@ entry: br i1 %cmp, label %cond.true, label %cond.false cond.true: - %1 = load i8** %retval; - ret i8* %1; + %1 = load i8** %retval + ret i8* %1 cond.false: - %2 = load i8** %retval; - ret i8* %2; + %2 = load i8** %retval + ret i8* %2 } define i32 @f() nounwind { @@ -64,7 +64,7 @@ entry: } @.str5 = private constant [9 x i32] [i32 97, i32 98, i32 99, i32 100, i32 0, i32 - 101, i32 102, i32 103, i32 0], align 4 ; + 101, i32 102, i32 103, i32 0], align 4 define i32 @test2() nounwind { ; CHECK: @test2 ; CHECK-NEXT: ret i32 34 @@ -72,4 +72,34 @@ define i32 @test2() nounwind { ret i32 %1 } -declare i32 @llvm.objectsize.i32(i8*, i1) nounwind readonly
\ No newline at end of file +; rdar://7674946 +@array = internal global [480 x float] zeroinitializer ; <[480 x float]*> [#uses=1] + +declare i8* @__memcpy_chk(i8*, i8*, i32, i32) nounwind + +declare i32 @llvm.objectsize.i32(i8*, i1) nounwind readonly + +declare i8* @__inline_memcpy_chk(i8*, i8*, i32) nounwind inlinehint + +define void @test3() nounwind { +; CHECK: @test3 +entry: + br i1 undef, label %bb11, label %bb12 + +bb11: + %0 = getelementptr inbounds float* getelementptr inbounds ([480 x float]* @array, i32 0, i32 128), i32 -127 ; <float*> [#uses=1] + %1 = bitcast float* %0 to i8* ; <i8*> [#uses=1] + %2 = call i32 @llvm.objectsize.i32(i8* %1, i1 false) ; <i32> [#uses=1] + %3 = call i8* @__memcpy_chk(i8* undef, i8* undef, i32 512, i32 %2) nounwind ; <i8*> [#uses=0] +; CHECK: @__memcpy_chk + unreachable + +bb12: + %4 = getelementptr inbounds float* getelementptr inbounds ([480 x float]* @array, i32 0, i32 128), i32 -127 ; <float*> [#uses=1] + %5 = bitcast float* %4 to i8* ; <i8*> [#uses=1] + %6 = call i8* @__inline_memcpy_chk(i8* %5, i8* undef, i32 512) nounwind inlinehint ; <i8*> [#uses=0] +; CHECK: @__inline_memcpy_chk + unreachable +} + +declare i32 @llvm.objectsize.i32(i8*, i1) nounwind readonly diff --git a/test/Transforms/InstCombine/ptr-int-cast.ll b/test/Transforms/InstCombine/ptr-int-cast.ll index c7ae689..ad11e43 100644 --- a/test/Transforms/InstCombine/ptr-int-cast.ll +++ b/test/Transforms/InstCombine/ptr-int-cast.ll @@ -1,17 +1,29 @@ -; RUN: opt < %s -instcombine -S > %t +; RUN: opt < %s -instcombine -S | FileCheck %s target datalayout = "E-p:64:64:64-a0:0:8-f32:32:32-f64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-v64:64:64-v128:128:128" define i1 @test1(i32 *%x) nounwind { entry: -; RUN: grep {ptrtoint i32\\* %x to i64} %t +; CHECK: test1 +; CHECK: ptrtoint i32* %x to i64 %tmp = ptrtoint i32* %x to i1 ret i1 %tmp } define i32* @test2(i128 %x) nounwind { entry: -; RUN: grep {inttoptr i64 %.mp1 to i32\\*} %t +; CHECK: test2 +; CHECK: inttoptr i64 %tmp1 to i32* %tmp = inttoptr i128 %x to i32* ret i32* %tmp } +; PR3574 +; CHECK: f0 +; CHECK: %tmp = zext i32 %a0 to i64 +; CHECK: ret i64 %tmp +define i64 @f0(i32 %a0) nounwind { + %t0 = inttoptr i32 %a0 to i8* + %t1 = ptrtoint i8* %t0 to i64 + ret i64 %t1 +} + diff --git a/test/Transforms/InstCombine/vec_narrow.ll b/test/Transforms/InstCombine/vec_narrow.ll index daf7bcf..c05c802 100644 --- a/test/Transforms/InstCombine/vec_narrow.ll +++ b/test/Transforms/InstCombine/vec_narrow.ll @@ -1,5 +1,5 @@ ; RUN: opt < %s -instcombine -S | \ -; RUN: grep {add float} +; RUN: grep {fadd float} %V = type <4 x float> diff --git a/test/Transforms/InstCombine/xor2.ll b/test/Transforms/InstCombine/xor2.ll index de3d65d..67f05ef 100644 --- a/test/Transforms/InstCombine/xor2.ll +++ b/test/Transforms/InstCombine/xor2.ll @@ -22,8 +22,8 @@ define i1 @test1(i32 %A) { ; PR1014 define i32 @test2(i32 %tmp1) { ; CHECK: @test2 -; CHECK-NEXT: or i32 %tmp1, 8 -; CHECK-NEXT: and i32 +; CHECK-NEXT: and i32 %tmp1, 32 +; CHECK-NEXT: or i32 %ovm, 8 ; CHECK-NEXT: ret i32 %ovm = and i32 %tmp1, 32 %ov3 = add i32 %ovm, 145 @@ -33,8 +33,8 @@ define i32 @test2(i32 %tmp1) { define i32 @test3(i32 %tmp1) { ; CHECK: @test3 -; CHECK-NEXT: or i32 %tmp1, 8 -; CHECK-NEXT: and i32 +; CHECK-NEXT: and i32 %tmp1, 32 +; CHECK-NEXT: or i32 %tmp, 8 ; CHECK-NEXT: ret i32 %ovm = or i32 %tmp1, 145 %ov31 = and i32 %ovm, 177 diff --git a/test/Transforms/LoopDeletion/simplify-then-delete.ll b/test/Transforms/LoopDeletion/simplify-then-delete.ll new file mode 100644 index 0000000..5a21672 --- /dev/null +++ b/test/Transforms/LoopDeletion/simplify-then-delete.ll @@ -0,0 +1,65 @@ +; RUN: opt < %s -S -indvars -loop-deletion -simplifycfg | FileCheck %s +; PR5794 + +; Indvars and loop deletion should be able to eliminate all looping +; in this testcase. + +; CHECK: define i32 @pmat(i32 %m, i32 %n, double* %y) nounwind { +; CHECK-NEXT: entry: +; CHECK-NEXT: ret i32 0 +; CHECK-NEXT: } + +target datalayout = "e-p:64:64:64" + +define i32 @pmat(i32 %m, i32 %n, double* %y) nounwind { +entry: + %cmp4 = icmp sgt i32 %m, 0 + br i1 %cmp4, label %bb.n10, label %w.e12 + +w.c: + %cmp = icmp slt i32 %inc11, %m + br i1 %cmp, label %w.c2.p, label %w.c.w.e12c + +w.c.w.e12c: + br label %w.c.w.e12c.s + +w.c.w.e12c.s: + br label %w.e12 + +bb.n10: + %cmp51 = icmp sgt i32 %n, 0 + br i1 %cmp51, label %bb.n10.w.c.w.e12c.sc, label %bb.n10.bb.n10.sc + +bb.n10.bb.n10.sc: + br label %bb.n10.s + +bb.n10.w.c.w.e12c.sc: + br label %w.c.w.e12c.s + +bb.n10.s: + br label %w.c2.p + +w.c2.p: + %i.05 = phi i32 [ 0, %bb.n10.s ], [ %inc11, %w.c ] + br i1 false, label %bb.n, label %w.e + +w.c2: + br i1 undef, label %w.b6, label %w.c2.w.ec + +w.c2.w.ec: + br label %w.e + +bb.n: + br label %w.b6 + +w.b6: + br label %w.c2 + +w.e: + %i.08 = phi i32 [ undef, %w.c2.w.ec ], [ %i.05, %w.c2.p ] + %inc11 = add nsw i32 %i.08, 1 + br label %w.c + +w.e12: + ret i32 0 +} diff --git a/test/Transforms/LoopIndexSplit/SplitValue-2007-08-24-dbg.ll b/test/Transforms/LoopIndexSplit/SplitValue-2007-08-24-dbg.ll deleted file mode 100644 index 4ab95fc..0000000 --- a/test/Transforms/LoopIndexSplit/SplitValue-2007-08-24-dbg.ll +++ /dev/null @@ -1,71 +0,0 @@ -; Split loop. Save last value. Split value is off by one in this example. -; RUN: opt < %s -loop-index-split -disable-output -stats |& \ -; RUN: grep "loop-index-split" | count 1 - - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, { }*, i32, i8*, i8*, i8*, i1, i1, i8* } - -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 17 }, section "llvm.metadata" - -@.str = internal constant [4 x i8] c"a.c\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@.str1 = internal constant [6 x i8] c"/tmp/\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str2 = internal constant [55 x i8] c"4.2.1 (Based on Apple Inc. build 5636) (LLVM build 00)\00", section "llvm.metadata" ; <[55 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { i32 458769, { }* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to { }*), i32 1, i8* getelementptr ([4 x i8]* @.str, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([55 x i8]* @.str2, i32 0, i32 0), i1 true, i1 false, i8* null }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] - -declare void @llvm.dbg.stoppoint(i32, i32, { }*) nounwind - - -@k = external global i32 ; <i32*> [#uses=2] - -define void @foobar(i32 %a, i32 %b) { -entry: - br label %bb - -bb: ; preds = %cond_next16, %entry - %i.01.0 = phi i32 [ 0, %entry ], [ %tmp18, %cond_next16 ] ; <i32> [#uses=5] - %tsum.18.0 = phi i32 [ 42, %entry ], [ %tsum.013.1, %cond_next16 ] ; <i32> [#uses=3] - %tmp1 = icmp sgt i32 %i.01.0, 50 ; <i1> [#uses=1] -call void @llvm.dbg.stoppoint(i32 5, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - br i1 %tmp1, label %cond_true, label %cond_false - -cond_true: ; preds = %bb - %tmp4 = tail call i32 @foo( i32 %i.01.0 ) ; <i32> [#uses=1] - %tmp6 = add i32 %tmp4, %tsum.18.0 ; <i32> [#uses=2] - %tmp914 = load i32* @k, align 4 ; <i32> [#uses=1] - %tmp1015 = icmp eq i32 %tmp914, 0 ; <i1> [#uses=1] -call void @llvm.dbg.stoppoint(i32 5, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - br i1 %tmp1015, label %cond_next16, label %cond_true13 - -cond_false: ; preds = %bb - %tmp8 = tail call i32 @bar( i32 %i.01.0 ) ; <i32> [#uses=0] - %tmp9 = load i32* @k, align 4 ; <i32> [#uses=1] - %tmp10 = icmp eq i32 %tmp9, 0 ; <i1> [#uses=1] -call void @llvm.dbg.stoppoint(i32 5, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - br i1 %tmp10, label %cond_next16, label %cond_true13 - -cond_true13: ; preds = %cond_false, %cond_true - %tsum.013.0 = phi i32 [ %tmp6, %cond_true ], [ %tsum.18.0, %cond_false ] ; <i32> [#uses=1] - %tmp15 = tail call i32 @bar( i32 %i.01.0 ) ; <i32> [#uses=0] -call void @llvm.dbg.stoppoint(i32 5, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - br label %cond_next16 - -cond_next16: ; preds = %cond_false, %cond_true, %cond_true13 - %tsum.013.1 = phi i32 [ %tsum.013.0, %cond_true13 ], [ %tmp6, %cond_true ], [ %tsum.18.0, %cond_false ] ; <i32> [#uses=2] - %tmp18 = add i32 %i.01.0, 1 ; <i32> [#uses=3] - %tmp21 = icmp slt i32 %tmp18, 100 ; <i1> [#uses=1] -call void @llvm.dbg.stoppoint(i32 5, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - br i1 %tmp21, label %bb, label %bb24 - -bb24: ; preds = %cond_next16 - %tmp18.lcssa = phi i32 [ %tmp18, %cond_next16 ] ; <i32> [#uses=1] - %tsum.013.1.lcssa = phi i32 [ %tsum.013.1, %cond_next16 ] ; <i32> [#uses=1] - %tmp27 = tail call i32 @t( i32 %tmp18.lcssa, i32 %tsum.013.1.lcssa ) ; <i32> [#uses=0] -call void @llvm.dbg.stoppoint(i32 5, i32 0, { }* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to { }*)) - ret void -} - -declare i32 @foo(i32) - -declare i32 @bar(i32) - -declare i32 @t(i32, i32) diff --git a/test/Transforms/LoopStrengthReduce/nonlinear-postinc.ll b/test/Transforms/LoopStrengthReduce/nonlinear-postinc.ll new file mode 100644 index 0000000..1e63770 --- /dev/null +++ b/test/Transforms/LoopStrengthReduce/nonlinear-postinc.ll @@ -0,0 +1,44 @@ +; RUN: opt < %s -loop-reduce +; PR6453 + +target datalayout = "e-p:64:64:64" + +define void @_ZNK15PolynomialSpaceILi3EE13compute_indexEjRA3_j() nounwind { +entry: + br label %bb6 + +bb6: + %t4 = phi i32 [ 0, %entry ], [ %t3, %bb5 ] + %t16 = sub i32 undef, %t4 + %t25 = sub i32 undef, %t4 + %t26 = add i32 undef, %t25 + br label %bb4 + +bb4: + %t2 = phi i32 [ %t1, %bb3 ], [ 0, %bb6 ] + %t17 = mul i32 %t2, %t16 + %t18 = zext i32 %t2 to i33 + %t19 = add i32 %t2, -1 + %t20 = zext i32 %t19 to i33 + %t21 = mul i33 %t18, %t20 + %t22 = lshr i33 %t21, 1 + %t23 = trunc i33 %t22 to i32 + %t24 = sub i32 %t17, %t23 + %t27 = add i32 %t24, %t26 + br i1 false, label %bb1, label %bb5 + +bb1: + %t = icmp ugt i32 %t27, undef + br i1 %t, label %bb2, label %bb3 + +bb3: + %t1 = add i32 %t2, 1 + br label %bb4 + +bb5: + %t3 = add i32 %t4, 1 + br label %bb6 + +bb2: + ret void +} diff --git a/test/Transforms/SCCP/retvalue-undef.ll b/test/Transforms/SCCP/retvalue-undef.ll new file mode 100644 index 0000000..389561f --- /dev/null +++ b/test/Transforms/SCCP/retvalue-undef.ll @@ -0,0 +1,32 @@ +; RUN: opt -ipsccp -S %s | FileCheck %s +; PR6414 +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" +target triple = "x86_64-unknown-linux-gnu" + +define internal i32 ()* @f() { + ret i32 ()* @g +} + +define internal i32 @g() { + ret i32 8 +} + +; CHECK: internal i32 @g() +; CHECK-NEXT: ret i32 8 + +define internal void @outer_mod() { + %1 = call i32 ()* ()* @f() ; <i32 ()*> [#uses=1] + %2 = call i32 %1() ; <i32> [#uses=0] + ret void +} + +define internal void @module_init() { + call void @register_outer_mod(void ()* @outer_mod) + ret void +} + +declare void @register_outer_mod(void ()*) + +define i32 @main() { + ret i32 0 +} diff --git a/test/Transforms/ScalarRepl/2009-03-17-CleanUp.ll b/test/Transforms/ScalarRepl/2009-03-17-CleanUp.ll deleted file mode 100644 index 9c70aae..0000000 --- a/test/Transforms/ScalarRepl/2009-03-17-CleanUp.ll +++ /dev/null @@ -1,3961 +0,0 @@ -; RUN: opt < %s -scalarrepl -S | grep store | not grep undef - -; ModuleID = '<stdin>' -target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" -target triple = "i386-pc-linux-gnu" - type { } ; type %0 - type { double, double } ; type %1 - type { i32, void ()* } ; type %2 - %llvm.dbg.anchor.type = type { i32, i32 } - %llvm.dbg.basictype.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, i32 } - %llvm.dbg.compile_unit.type = type { i32, %0*, i32, i8*, i8*, i8*, i1, i1, i8*, i32 } - %llvm.dbg.composite.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, %0*, %0*, i32 } - %llvm.dbg.derivedtype.type = type { i32, %0*, i8*, %0*, i32, i64, i64, i64, i32, %0* } - %llvm.dbg.global_variable.type = type { i32, %0*, %0*, i8*, i8*, i8*, %0*, i32, %0*, i1, i1, %0* } - %llvm.dbg.subprogram.type = type { i32, %0*, %0*, i8*, i8*, i8*, %0*, i32, %0*, i1, i1 } - %llvm.dbg.subrange.type = type { i32, i64, i64 } - %llvm.dbg.variable.type = type { i32, %0*, i8*, %0*, i32, %0* } - %struct..0._50 = type { i32 } - %struct..1__pthread_mutex_s = type { i32, i32, i32, i32, i32, %struct..0._50 } - %struct.__class_type_info_pseudo = type { %struct.__type_info_pseudo } - %struct.__locale_struct = type { [13 x %struct.locale_data*], i16*, i32*, i32*, [13 x i8*] } - %struct.__pthread_slist_t = type { %struct.__pthread_slist_t* } - %struct.__si_class_type_info_pseudo = type { %struct.__type_info_pseudo, %"struct.std::type_info"* } - %struct.__type_info_pseudo = type { i8*, i8* } - %struct.locale_data = type opaque - %"struct.polynomial<double>" = type { i32 (...)**, double*, i32 } - %"struct.polynomial<std::complex<double> >" = type { i32 (...)**, %"struct.std::complex<double>"*, i32 } - %struct.pthread_attr_t = type { i32, [32 x i8] } - %struct.pthread_mutex_t = type { %struct..1__pthread_mutex_s } - %struct.pthread_mutexattr_t = type { i32 } - %"struct.std::allocator<char>" = type <{ i8 }> - %"struct.std::basic_ios<char,std::char_traits<char> >" = type { %"struct.std::ios_base", %"struct.std::basic_ostream<char,std::char_traits<char> >"*, i8, i8, %"struct.std::basic_streambuf<char,std::char_traits<char> >"*, %"struct.std::ctype<char>"*, %"struct.std::num_get<char,std::istreambuf_iterator<char, std::char_traits<char> > >"*, %"struct.std::num_get<char,std::istreambuf_iterator<char, std::char_traits<char> > >"* } - %"struct.std::basic_ostream<char,std::char_traits<char> >" = type { i32 (...)**, %"struct.std::basic_ios<char,std::char_traits<char> >" } - %"struct.std::basic_streambuf<char,std::char_traits<char> >" = type { i32 (...)**, i8*, i8*, i8*, i8*, i8*, i8*, %"struct.std::locale" } - %"struct.std::basic_string<char,std::char_traits<char>,std::allocator<char> >::_Alloc_hider" = type { i8* } - %"struct.std::complex<double>" = type { %1 } - %"struct.std::ctype<char>" = type { %"struct.std::locale::facet", %struct.__locale_struct*, i8, i32*, i32*, i16*, i8, [256 x i8], [256 x i8], i8 } - %"struct.std::exception" = type { i32 (...)** } - %"struct.std::ios_base" = type { i32 (...)**, i32, i32, i32, i32, i32, %"struct.std::ios_base::_Callback_list"*, %"struct.std::ios_base::_Words", [8 x %"struct.std::ios_base::_Words"], i32, %"struct.std::ios_base::_Words"*, %"struct.std::locale" } - %"struct.std::ios_base::Init" = type <{ i8 }> - %"struct.std::ios_base::_Callback_list" = type { %"struct.std::ios_base::_Callback_list"*, void (i32, %"struct.std::ios_base"*, i32)*, i32, i32 } - %"struct.std::ios_base::_Words" = type { i8*, i32 } - %"struct.std::locale" = type { %"struct.std::locale::_Impl"* } - %"struct.std::locale::_Impl" = type { i32, %"struct.std::locale::facet"**, i32, %"struct.std::locale::facet"**, i8** } - %"struct.std::locale::facet" = type { i32 (...)**, i32 } - %"struct.std::num_get<char,std::istreambuf_iterator<char, std::char_traits<char> > >" = type { %"struct.std::locale::facet" } - %"struct.std::num_put<char,std::ostreambuf_iterator<char, std::char_traits<char> > >" = type { %"struct.std::locale::facet" } - %"struct.std::overflow_error" = type { %"struct.std::runtime_error" } - %"struct.std::runtime_error" = type { %"struct.std::exception", %"struct.std::string" } - %"struct.std::string" = type { %"struct.std::basic_string<char,std::char_traits<char>,std::allocator<char> >::_Alloc_hider" } - %"struct.std::type_info" = type { i32 (...)**, i8* } -@llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 17 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@.str = internal constant [13 x i8] c"fftbench.cpp\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@.str1 = internal constant [42 x i8] c"/developer/home2/zsth/test/debug/tmp3/X3/\00", section "llvm.metadata" ; <[42 x i8]*> [#uses=1] -@.str2 = internal constant [52 x i8] c"4.2.1 (Based on Apple Inc. build 5641) (LLVM build)\00", section "llvm.metadata" ; <[52 x i8]*> [#uses=1] -@llvm.dbg.compile_unit = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([13 x i8]* @.str, i32 0, i32 0), i8* getelementptr ([42 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 true, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str3 = internal constant [8 x i8] c"complex\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str4 = internal constant [110 x i8] c"/developer/home2/zsth/projects/llvm.org/install/lib/gcc/i686-pc-linux-gnu/4.2.1/../../../../include/c++/4.2.1\00", section "llvm.metadata" ; <[110 x i8]*> [#uses=1] -@llvm.dbg.compile_unit5 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([110 x i8]* @.str4, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str6 = internal constant [16 x i8] c"complex<double>\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@.str7 = internal constant [15 x i8] c"complex double\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str7, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 128, i64 64, i64 0, i32 0, i32 3 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@.str8 = internal constant [9 x i8] c"_M_value\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@llvm.dbg.derivedtype = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str8, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1195, i64 128, i64 64, i64 0, i32 1, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype9 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype9 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite10 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { i32 458752, i32 46 }, section "llvm.metadata" ; <%llvm.dbg.anchor.type*> [#uses=1] -@llvm.dbg.subprogram = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1161, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite10 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str11 = internal constant [7 x i8] c"double\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@llvm.dbg.basictype12 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str11, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 64, i64 64, i64 0, i32 0, i32 4 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array13 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype9 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype12 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype12 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite14 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array13 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram15 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1215, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite14 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str16 = internal constant [15 x i8] c"complex<float>\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@.str18 = internal constant [14 x i8] c"complex float\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.basictype19 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str18, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 64, i64 32, i64 0, i32 0, i32 3 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.derivedtype20 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str8, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1042, i64 64, i64 32, i64 0, i32 1, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype19 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype21 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite171 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array22 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype21 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype19 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite23 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array22 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram24 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1007, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite23 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str25 = internal constant [6 x i8] c"float\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.basictype26 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str25, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 4 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array27 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype21 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype26 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype26 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite28 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array27 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram29 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1062, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite28 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype30 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 128, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype31 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype30 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array32 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype21 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype31 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite33 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array32 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram34 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1464, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite33 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str35 = internal constant [21 x i8] c"complex<long double>\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@.str37 = internal constant [20 x i8] c"complex long double\00", section "llvm.metadata" ; <[20 x i8]*> [#uses=1] -@llvm.dbg.basictype38 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([20 x i8]* @.str37, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 192, i64 32, i64 0, i32 0, i32 3 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.derivedtype39 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str8, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1348, i64 192, i64 32, i64 0, i32 1, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype38 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype40 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite122 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array41 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype40 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype38 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite42 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array41 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram43 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1314, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite42 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str44 = internal constant [12 x i8] c"long double\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.basictype45 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str44, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 96, i64 32, i64 0, i32 0, i32 4 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array46 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype40 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype45 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype45 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite47 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array46 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram48 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1352, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite47 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype49 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 64, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite171 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype50 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype49 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array51 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype40 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype50 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite52 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array51 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram53 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1480, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite52 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array54 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype40 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype31 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite55 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array54 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram56 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1484, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite55 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype57 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype45 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array58 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype57 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype40 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite59 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array58 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str60 = internal constant [5 x i8] c"real\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str61 = internal constant [24 x i8] c"_ZNSt7complexIeE4realEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram62 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str61, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1359, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite59 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str63 = internal constant [9 x i8] c"stddef.h\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@.str64 = internal constant [88 x i8] c"/developer/home2/zsth/projects/llvm.org/install/lib/gcc/i686-pc-linux-gnu/4.2.1/include\00", section "llvm.metadata" ; <[88 x i8]*> [#uses=1] -@llvm.dbg.compile_unit65 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([9 x i8]* @.str63, i32 0, i32 0), i8* getelementptr ([88 x i8]* @.str64, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str66 = internal constant [8 x i8] c"float_t\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.derivedtype67 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str66, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit65 to %0*), i32 214, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype45 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str68 = internal constant [10 x i8] c"mathdef.h\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str69 = internal constant [18 x i8] c"/usr/include/bits\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.compile_unit70 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([10 x i8]* @.str68, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str69, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str71 = internal constant [9 x i8] c"double_t\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@llvm.dbg.derivedtype72 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str71, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit70 to %0*), i32 36, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype67 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype73 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 96, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype72 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype74 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype73 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype75 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 192, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite122 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype76 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype75 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array77 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype74 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype76 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite78 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array77 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str79 = internal constant [25 x i8] c"_ZNKSt7complexIeE4realEv\00", section "llvm.metadata" ; <[25 x i8]*> [#uses=1] -@llvm.dbg.subprogram80 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str79, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1363, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite78 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str81 = internal constant [5 x i8] c"imag\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str82 = internal constant [24 x i8] c"_ZNSt7complexIeE4imagEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram83 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str82, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1367, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite59 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str84 = internal constant [25 x i8] c"_ZNKSt7complexIeE4imagEv\00", section "llvm.metadata" ; <[25 x i8]*> [#uses=1] -@llvm.dbg.subprogram85 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str84, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1371, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite78 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype86 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite122 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array87 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype86 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype40 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype45 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite88 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array87 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str89 = internal constant [10 x i8] c"operator=\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str90 = internal constant [21 x i8] c"_ZNSt7complexIeEaSEe\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram91 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str90, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1375, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite88 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str92 = internal constant [11 x i8] c"operator+=\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str93 = internal constant [21 x i8] c"_ZNSt7complexIeEpLEe\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram94 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str93, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1383, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite88 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str95 = internal constant [11 x i8] c"operator-=\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str96 = internal constant [21 x i8] c"_ZNSt7complexIeEmIEe\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram97 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str96, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1390, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite88 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str98 = internal constant [11 x i8] c"operator*=\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str99 = internal constant [21 x i8] c"_ZNSt7complexIeEmLEe\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram100 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str99, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1397, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite88 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str101 = internal constant [11 x i8] c"operator/=\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str102 = internal constant [21 x i8] c"_ZNSt7complexIeEdVEe\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram103 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str102, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1404, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite88 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.composite104 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 55, i64 0, i64 0, i64 0, i32 4, %0* null, %0* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype105 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 8, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite104 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype106 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype105 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array107 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype86 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype40 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype106 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite108 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array107 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram109 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1335, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite108 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram110 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1337, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite108 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram111 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1339, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite108 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram112 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1341, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite108 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram113 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1343, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite108 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype114 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 192, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype38 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype115 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype114 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array116 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype115 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype76 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite117 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array116 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str118 = internal constant [6 x i8] c"__rep\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str119 = internal constant [26 x i8] c"_ZNKSt7complexIeE5__repEv\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.subprogram120 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str119, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1345, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite117 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array121 = internal constant [20 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype39 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram43 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram48 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram53 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram56 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram62 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram80 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram83 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram85 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram91 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram94 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram97 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram100 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram103 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram109 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram110 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram111 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram112 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram113 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram120 to %0*)], section "llvm.metadata" ; <[20 x %0*]*> [#uses=1] -@llvm.dbg.composite122 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([21 x i8]* @.str35, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1310, i64 192, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([20 x %0*]* @llvm.dbg.array121 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype123 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 192, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite122 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype124 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype123 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array125 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype21 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype124 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite126 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array125 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram127 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1468, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite126 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype128 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype26 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array129 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype128 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype21 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite130 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array129 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str131 = internal constant [24 x i8] c"_ZNSt7complexIfE4realEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram132 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str131, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1046, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite130 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype133 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype26 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype134 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype133 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype135 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype49 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array136 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype134 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype135 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite137 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array136 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str138 = internal constant [25 x i8] c"_ZNKSt7complexIfE4realEv\00", section "llvm.metadata" ; <[25 x i8]*> [#uses=1] -@llvm.dbg.subprogram139 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str138, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1050, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite137 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str140 = internal constant [24 x i8] c"_ZNSt7complexIfE4imagEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram141 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str140, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1054, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite130 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str142 = internal constant [25 x i8] c"_ZNKSt7complexIfE4imagEv\00", section "llvm.metadata" ; <[25 x i8]*> [#uses=1] -@llvm.dbg.subprogram143 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str142, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1058, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite137 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype144 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite171 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array145 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype144 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype21 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype26 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite146 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array145 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str147 = internal constant [21 x i8] c"_ZNSt7complexIfEaSEf\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram148 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str147, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1069, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite146 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str149 = internal constant [21 x i8] c"_ZNSt7complexIfEpLEf\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram150 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str149, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1077, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite146 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str151 = internal constant [21 x i8] c"_ZNSt7complexIfEmIEf\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram152 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str151, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1084, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite146 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str153 = internal constant [21 x i8] c"_ZNSt7complexIfEmLEf\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram154 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str153, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1091, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite146 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str155 = internal constant [21 x i8] c"_ZNSt7complexIfEdVEf\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram156 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str155, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1098, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite146 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array157 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype144 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype21 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype106 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite158 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array157 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram159 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1029, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite158 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram160 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1031, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite158 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram161 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1033, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite158 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram162 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1035, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite158 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram163 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1037, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite158 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype164 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 64, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype19 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype165 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype164 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array166 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype165 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype135 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite167 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array166 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str168 = internal constant [26 x i8] c"_ZNKSt7complexIfE5__repEv\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.subprogram169 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str168, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1039, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite167 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array170 = internal constant [20 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype20 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram24 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram29 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram34 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram127 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram132 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram139 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram141 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram143 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram148 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram150 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram152 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram154 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram156 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram159 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram160 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram161 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram162 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram163 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram169 to %0*)], section "llvm.metadata" ; <[20 x %0*]*> [#uses=1] -@llvm.dbg.composite171 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str16, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1003, i64 64, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([20 x %0*]* @llvm.dbg.array170 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype172 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 64, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite171 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype173 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype172 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array174 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype9 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype173 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite175 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array174 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram176 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1472, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite175 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array177 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype9 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype124 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite178 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array177 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram179 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1476, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite178 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype180 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype12 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array181 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype180 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype9 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite182 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array181 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str183 = internal constant [24 x i8] c"_ZNSt7complexIdE4realEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram184 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str183, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1199, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite182 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype185 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 64, i64 64, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype12 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype186 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype185 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype187 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype30 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array188 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype186 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype187 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite189 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array188 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str190 = internal constant [25 x i8] c"_ZNKSt7complexIdE4realEv\00", section "llvm.metadata" ; <[25 x i8]*> [#uses=1] -@llvm.dbg.subprogram191 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str190, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1203, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite189 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str192 = internal constant [24 x i8] c"_ZNSt7complexIdE4imagEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram193 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str192, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1207, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite182 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str194 = internal constant [25 x i8] c"_ZNKSt7complexIdE4imagEv\00", section "llvm.metadata" ; <[25 x i8]*> [#uses=1] -@llvm.dbg.subprogram195 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str194, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1211, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite189 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype196 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array197 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype196 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype9 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype12 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite198 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array197 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str199 = internal constant [21 x i8] c"_ZNSt7complexIdEaSEd\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram200 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str199, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1222, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite198 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str201 = internal constant [21 x i8] c"_ZNSt7complexIdEpLEd\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram202 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str201, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1230, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite198 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str203 = internal constant [21 x i8] c"_ZNSt7complexIdEmIEd\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram204 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str203, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1237, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite198 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str205 = internal constant [21 x i8] c"_ZNSt7complexIdEmLEd\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram206 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str205, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1244, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite198 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str207 = internal constant [21 x i8] c"_ZNSt7complexIdEdVEd\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram208 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str207, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1251, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite198 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array209 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype196 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype9 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype106 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite210 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array209 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram211 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1182, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite210 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram212 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1184, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite210 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram213 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1186, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite210 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram214 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1188, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite210 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram215 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1190, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite210 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype216 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 128, i64 64, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype217 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype216 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array218 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype217 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype187 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite219 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array218 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str220 = internal constant [26 x i8] c"_ZNKSt7complexIdE5__repEv\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.subprogram221 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str220, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1192, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite219 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array222 = internal constant [20 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram15 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram176 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram179 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram184 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram191 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram193 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram195 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram200 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram202 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram204 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram206 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram208 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram211 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram212 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram213 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram214 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram215 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram221 to %0*)], section "llvm.metadata" ; <[20 x %0*]*> [#uses=1] -@llvm.dbg.composite223 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([16 x i8]* @.str6, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1157, i64 128, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([20 x %0*]* @llvm.dbg.array222 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype224 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array225 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype224 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array225 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str226 = internal constant [22 x i8] c"_ZNSt7complexIdEC1ECd\00", section "llvm.metadata" ; <[22 x i8]*> [#uses=1] -@llvm.dbg.subprogram227 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([22 x i8]* @.str226, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1161, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str230 = internal constant [4 x i8] c"__z\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.variable231 = internal constant %llvm.dbg.variable.type { i32 459009, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram227 to %0*), i8* getelementptr ([4 x i8]* @.str230, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1161, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.subprogram232 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str118, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str220, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1192, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite219 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram235 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str183, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1199, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite182 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram237 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str60, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str190, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1203, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite189 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram239 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str81, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str194, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1211, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite189 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str241 = internal constant [22 x i8] c"_ZNSt7complexIdEC1Edd\00", section "llvm.metadata" ; <[22 x i8]*> [#uses=1] -@llvm.dbg.subprogram242 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str3, i32 0, i32 0), i8* getelementptr ([22 x i8]* @.str241, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1215, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite14 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str244 = internal constant [4 x i8] c"__r\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.subprogram248 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str199, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1222, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite198 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram252 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str101, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str207, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1251, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite198 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array255 = internal constant [1 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype12 to %0*)], section "llvm.metadata" ; <[1 x %0*]*> [#uses=1] -@llvm.dbg.composite256 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([1 x %0*]* @llvm.dbg.array255 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str257 = internal constant [14 x i8] c"random_double\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram258 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str257, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str257, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 55, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite256 to %0*), i1 true, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=0] -@.str259 = internal constant [7 x i8] c"result\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@_ZZL13random_doublevE4seed = internal global i32 1325 ; <i32*> [#uses=14] -@.str266 = internal constant [19 x i8] c"polynomial<double>\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@.str268 = internal constant [4 x i8] c"int\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.basictype269 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str268, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 5 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.array270 = internal constant [1 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[1 x %0*]*> [#uses=1] -@llvm.dbg.composite271 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([1 x %0*]* @llvm.dbg.array270 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype272 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite271 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype273 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype272 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str274 = internal constant [17 x i8] c"_vptr.polynomial\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.derivedtype275 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([17 x i8]* @.str274, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 84, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype273 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype276 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype12 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str277 = internal constant [8 x i8] c"m_coeff\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.derivedtype278 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str277, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 164, i64 32, i64 32, i64 32, i32 2, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype276 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str279 = internal constant [13 x i8] c"unsigned int\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.basictype280 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str279, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, i32 7 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@.str281 = internal constant [7 x i8] c"size_t\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@llvm.dbg.derivedtype282 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str281, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit65 to %0*), i32 152, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str283 = internal constant [9 x i8] c"m_degree\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@llvm.dbg.derivedtype284 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str283, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 167, i64 32, i64 32, i64 64, i32 2, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype285 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array286 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite287 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array286 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str288 = internal constant [11 x i8] c"polynomial\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.subprogram289 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 211, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite287 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype290 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype185 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array291 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype290 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite292 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array291 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram293 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 220, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite292 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array294 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype186 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite295 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array294 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram296 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 232, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite295 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype297 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 96, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype298 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype297 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array299 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype298 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite300 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array299 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram301 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 242, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite300 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array302 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite303 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array302 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str304 = internal constant [12 x i8] c"~polynomial\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.subprogram305 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 252, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite303 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype306 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array307 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype306 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype298 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite308 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array307 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str309 = internal constant [27 x i8] c"_ZN10polynomialIdEaSERKS0_\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram310 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str309, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 259, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite308 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array311 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype186 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite312 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array311 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str313 = internal constant [11 x i8] c"initialize\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str314 = internal constant [35 x i8] c"_ZN10polynomialIdE10initializeERKd\00", section "llvm.metadata" ; <[35 x i8]*> [#uses=1] -@llvm.dbg.subprogram315 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str313, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str313, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str314, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 203, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite312 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array316 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype306 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite317 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array316 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str318 = internal constant [8 x i8] c"stretch\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str319 = internal constant [29 x i8] c"_ZN10polynomialIdE7stretchEj\00", section "llvm.metadata" ; <[29 x i8]*> [#uses=1] -@llvm.dbg.subprogram320 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str318, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str318, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str319, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 276, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite317 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype321 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype297 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array322 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype321 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite323 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array322 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str324 = internal constant [7 x i8] c"degree\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@.str325 = internal constant [29 x i8] c"_ZNK10polynomialIdE6degreeEv\00", section "llvm.metadata" ; <[29 x i8]*> [#uses=1] -@llvm.dbg.subprogram326 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str324, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str324, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str325, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 111, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite323 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array327 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype12 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype321 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite328 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array327 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str329 = internal constant [4 x i8] c"get\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@.str330 = internal constant [26 x i8] c"_ZNK10polynomialIdE3getEj\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.subprogram331 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str329, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str329, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str330, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 300, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite328 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array332 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype180 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite333 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array332 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str334 = internal constant [11 x i8] c"operator[]\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str335 = internal constant [23 x i8] c"_ZN10polynomialIdEixEj\00", section "llvm.metadata" ; <[23 x i8]*> [#uses=1] -@llvm.dbg.subprogram336 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([23 x i8]* @.str335, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 306, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite333 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array337 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype12 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype321 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype186 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite338 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array337 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str339 = internal constant [11 x i8] c"operator()\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str340 = internal constant [26 x i8] c"_ZNK10polynomialIdEclERKd\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.subprogram341 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str339, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str339, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str340, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 313, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite338 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array342 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype321 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite343 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array342 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str344 = internal constant [10 x i8] c"operator-\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str345 = internal constant [24 x i8] c"_ZNK10polynomialIdEngEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram346 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str345, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 335, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite343 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str347 = internal constant [10 x i8] c"operator+\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str348 = internal constant [24 x i8] c"_ZNK10polynomialIdEpsEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram349 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str348, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 346, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite343 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array350 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype321 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype298 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite351 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array350 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str352 = internal constant [28 x i8] c"_ZNK10polynomialIdEplERKS0_\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram353 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str352, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 353, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite351 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str354 = internal constant [28 x i8] c"_ZNK10polynomialIdEmiERKS0_\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram355 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str354, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 376, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite351 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array356 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite357 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array356 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str358 = internal constant [5 x i8] c"log2\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str359 = internal constant [26 x i8] c"_ZN10polynomialIdE4log2Ej\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.subprogram360 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str358, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str358, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str359, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 404, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite357 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array361 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite362 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array361 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str363 = internal constant [10 x i8] c"flip_bits\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str364 = internal constant [32 x i8] c"_ZN10polynomialIdE9flip_bitsEjj\00", section "llvm.metadata" ; <[32 x i8]*> [#uses=1] -@llvm.dbg.subprogram365 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str363, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str363, i32 0, i32 0), i8* getelementptr ([32 x i8]* @.str364, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 423, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite362 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array366 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite367 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array366 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str368 = internal constant [12 x i8] c"stretch_fft\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@.str369 = internal constant [34 x i8] c"_ZN10polynomialIdE11stretch_fftEv\00", section "llvm.metadata" ; <[34 x i8]*> [#uses=1] -@llvm.dbg.subprogram370 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str368, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str368, i32 0, i32 0), i8* getelementptr ([34 x i8]* @.str369, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 443, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite367 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str371 = internal constant [34 x i8] c"polynomial<std::complex<double> >\00", section "llvm.metadata" ; <[34 x i8]*> [#uses=1] -@llvm.dbg.derivedtype373 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([17 x i8]* @.str274, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 84, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype273 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype374 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str277, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 164, i64 32, i64 32, i64 32, i32 2, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype224 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype375 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str283, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 167, i64 32, i64 32, i64 64, i32 2, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype376 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array377 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite378 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array377 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram379 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 211, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite378 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array380 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype187 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite381 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array380 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram382 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 220, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite381 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array383 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype31 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite384 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array383 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram385 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 232, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite384 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype386 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 96, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype387 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype386 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array388 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype387 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite389 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array388 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram390 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 242, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite389 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array391 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite392 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array391 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram393 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 252, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite392 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype394 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array395 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype394 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype387 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite396 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array395 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str397 = internal constant [39 x i8] c"_ZN10polynomialISt7complexIdEEaSERKS2_\00", section "llvm.metadata" ; <[39 x i8]*> [#uses=1] -@llvm.dbg.subprogram398 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([39 x i8]* @.str397, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 259, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite396 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array399 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype31 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite400 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array399 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str401 = internal constant [49 x i8] c"_ZN10polynomialISt7complexIdEE10initializeERKS1_\00", section "llvm.metadata" ; <[49 x i8]*> [#uses=1] -@llvm.dbg.subprogram402 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str313, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str313, i32 0, i32 0), i8* getelementptr ([49 x i8]* @.str401, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 203, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite400 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array403 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype394 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite404 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array403 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str405 = internal constant [41 x i8] c"_ZN10polynomialISt7complexIdEE7stretchEj\00", section "llvm.metadata" ; <[41 x i8]*> [#uses=1] -@llvm.dbg.subprogram406 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str318, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str318, i32 0, i32 0), i8* getelementptr ([41 x i8]* @.str405, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 276, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite404 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype407 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype386 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array408 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype407 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite409 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array408 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str410 = internal constant [41 x i8] c"_ZNK10polynomialISt7complexIdEE6degreeEv\00", section "llvm.metadata" ; <[41 x i8]*> [#uses=1] -@llvm.dbg.subprogram411 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str324, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str324, i32 0, i32 0), i8* getelementptr ([41 x i8]* @.str410, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 111, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite409 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array412 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype407 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite413 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array412 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str414 = internal constant [38 x i8] c"_ZNK10polynomialISt7complexIdEE3getEj\00", section "llvm.metadata" ; <[38 x i8]*> [#uses=1] -@llvm.dbg.subprogram415 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str329, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str329, i32 0, i32 0), i8* getelementptr ([38 x i8]* @.str414, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 300, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite413 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array416 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype196 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite417 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array416 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str418 = internal constant [35 x i8] c"_ZN10polynomialISt7complexIdEEixEj\00", section "llvm.metadata" ; <[35 x i8]*> [#uses=1] -@llvm.dbg.subprogram419 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str418, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 306, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite417 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array420 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype407 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype31 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite421 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array420 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str422 = internal constant [40 x i8] c"_ZNK10polynomialISt7complexIdEEclERKS1_\00", section "llvm.metadata" ; <[40 x i8]*> [#uses=1] -@llvm.dbg.subprogram423 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str339, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str339, i32 0, i32 0), i8* getelementptr ([40 x i8]* @.str422, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 313, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite421 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array424 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype407 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite425 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array424 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str426 = internal constant [36 x i8] c"_ZNK10polynomialISt7complexIdEEngEv\00", section "llvm.metadata" ; <[36 x i8]*> [#uses=1] -@llvm.dbg.subprogram427 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([36 x i8]* @.str426, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 335, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite425 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str428 = internal constant [36 x i8] c"_ZNK10polynomialISt7complexIdEEpsEv\00", section "llvm.metadata" ; <[36 x i8]*> [#uses=1] -@llvm.dbg.subprogram429 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([36 x i8]* @.str428, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 346, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite425 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array430 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype407 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype387 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite431 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array430 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str432 = internal constant [40 x i8] c"_ZNK10polynomialISt7complexIdEEplERKS2_\00", section "llvm.metadata" ; <[40 x i8]*> [#uses=1] -@llvm.dbg.subprogram433 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([40 x i8]* @.str432, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 353, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite431 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str434 = internal constant [40 x i8] c"_ZNK10polynomialISt7complexIdEEmiERKS2_\00", section "llvm.metadata" ; <[40 x i8]*> [#uses=1] -@llvm.dbg.subprogram435 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([40 x i8]* @.str434, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 376, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite431 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str436 = internal constant [38 x i8] c"_ZN10polynomialISt7complexIdEE4log2Ej\00", section "llvm.metadata" ; <[38 x i8]*> [#uses=1] -@llvm.dbg.subprogram437 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str358, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str358, i32 0, i32 0), i8* getelementptr ([38 x i8]* @.str436, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 404, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite357 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str438 = internal constant [44 x i8] c"_ZN10polynomialISt7complexIdEE9flip_bitsEjj\00", section "llvm.metadata" ; <[44 x i8]*> [#uses=1] -@llvm.dbg.subprogram439 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str363, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str363, i32 0, i32 0), i8* getelementptr ([44 x i8]* @.str438, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 423, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite362 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array440 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite441 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array440 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str442 = internal constant [46 x i8] c"_ZN10polynomialISt7complexIdEE11stretch_fftEv\00", section "llvm.metadata" ; <[46 x i8]*> [#uses=1] -@llvm.dbg.subprogram443 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str368, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str368, i32 0, i32 0), i8* getelementptr ([46 x i8]* @.str442, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 443, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite441 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str444 = internal constant [49 x i8] c"polynomial<std::complex<std::complex<double> > >\00", section "llvm.metadata" ; <[49 x i8]*> [#uses=1] -@llvm.dbg.composite445 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([49 x i8]* @.str444, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 84, i64 0, i64 0, i64 0, i32 4, %0* null, %0* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array446 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite445 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype387 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite447 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array446 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str448 = internal constant [12 x i8] c"bit_reverse\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@.str449 = internal constant [50 x i8] c"_ZN10polynomialISt7complexIdEE11bit_reverseERKS2_\00", section "llvm.metadata" ; <[50 x i8]*> [#uses=1] -@llvm.dbg.subprogram450 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([50 x i8]* @.str449, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 469, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite447 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype451 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 8, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite445 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype452 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype451 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array453 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite445 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype452 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite454 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array453 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str455 = internal constant [59 x i8] c"_ZN10polynomialISt7complexIdEE11bit_reverseERKS_IS0_IS1_EE\00", section "llvm.metadata" ; <[59 x i8]*> [#uses=1] -@llvm.dbg.subprogram456 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([59 x i8]* @.str455, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 483, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite454 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str457 = internal constant [4 x i8] c"fft\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@.str458 = internal constant [41 x i8] c"_ZN10polynomialISt7complexIdEE3fftERKS2_\00", section "llvm.metadata" ; <[41 x i8]*> [#uses=1] -@llvm.dbg.subprogram459 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str457, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str457, i32 0, i32 0), i8* getelementptr ([41 x i8]* @.str458, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 497, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite447 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str460 = internal constant [12 x i8] c"inverse_fft\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@.str461 = internal constant [59 x i8] c"_ZN10polynomialISt7complexIdEE11inverse_fftERKS_IS0_IS1_EE\00", section "llvm.metadata" ; <[59 x i8]*> [#uses=1] -@llvm.dbg.subprogram462 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str460, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str460, i32 0, i32 0), i8* getelementptr ([59 x i8]* @.str461, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 535, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite454 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str463 = internal constant [10 x i8] c"operator*\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str464 = internal constant [40 x i8] c"_ZNK10polynomialISt7complexIdEEmlERKS2_\00", section "llvm.metadata" ; <[40 x i8]*> [#uses=1] -@llvm.dbg.subprogram465 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str463, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str463, i32 0, i32 0), i8* getelementptr ([40 x i8]* @.str464, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 576, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite431 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str466 = internal constant [39 x i8] c"_ZN10polynomialISt7complexIdEEpLERKS2_\00", section "llvm.metadata" ; <[39 x i8]*> [#uses=1] -@llvm.dbg.subprogram467 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([39 x i8]* @.str466, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 625, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite396 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str468 = internal constant [39 x i8] c"_ZN10polynomialISt7complexIdEEmIERKS2_\00", section "llvm.metadata" ; <[39 x i8]*> [#uses=1] -@llvm.dbg.subprogram469 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([39 x i8]* @.str468, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 636, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite396 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str470 = internal constant [39 x i8] c"_ZN10polynomialISt7complexIdEEmLERKS2_\00", section "llvm.metadata" ; <[39 x i8]*> [#uses=1] -@llvm.dbg.subprogram471 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([39 x i8]* @.str470, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 647, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite396 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array472 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite473 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array472 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str474 = internal constant [8 x i8] c"acquire\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str475 = internal constant [41 x i8] c"_ZN10polynomialISt7complexIdEE7acquireEv\00", section "llvm.metadata" ; <[41 x i8]*> [#uses=1] -@llvm.dbg.subprogram476 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str474, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str474, i32 0, i32 0), i8* getelementptr ([41 x i8]* @.str475, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 181, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite473 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str477 = internal constant [8 x i8] c"release\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str478 = internal constant [41 x i8] c"_ZN10polynomialISt7complexIdEE7releaseEv\00", section "llvm.metadata" ; <[41 x i8]*> [#uses=1] -@llvm.dbg.subprogram479 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str477, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str477, i32 0, i32 0), i8* getelementptr ([41 x i8]* @.str478, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 188, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite473 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array480 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype376 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype187 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite481 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array480 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str482 = internal constant [10 x i8] c"deep_copy\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str483 = internal constant [47 x i8] c"_ZN10polynomialISt7complexIdEE9deep_copyEPKS1_\00", section "llvm.metadata" ; <[47 x i8]*> [#uses=1] -@llvm.dbg.subprogram484 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([47 x i8]* @.str483, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 195, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite481 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array485 = internal constant [33 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype373 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype374 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype375 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram379 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram382 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram385 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram390 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram393 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram398 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram402 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram406 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram411 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram415 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram419 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram423 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram427 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram429 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram433 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram435 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram437 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram439 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram443 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram450 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram456 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram459 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram462 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram465 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram467 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram469 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram471 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram476 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram479 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram484 to %0*)], section "llvm.metadata" ; <[33 x %0*]*> [#uses=1] -@llvm.dbg.composite486 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([34 x i8]* @.str371, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 84, i64 96, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([33 x %0*]* @llvm.dbg.array485 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array487 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype298 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite488 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array487 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str489 = internal constant [38 x i8] c"_ZN10polynomialIdE11bit_reverseERKS0_\00", section "llvm.metadata" ; <[38 x i8]*> [#uses=1] -@llvm.dbg.subprogram490 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([38 x i8]* @.str489, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 469, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite488 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array491 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype387 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite492 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array491 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str493 = internal constant [52 x i8] c"_ZN10polynomialIdE11bit_reverseERKS_ISt7complexIdEE\00", section "llvm.metadata" ; <[52 x i8]*> [#uses=1] -@llvm.dbg.subprogram494 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str493, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 483, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite492 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str495 = internal constant [29 x i8] c"_ZN10polynomialIdE3fftERKS0_\00", section "llvm.metadata" ; <[29 x i8]*> [#uses=1] -@llvm.dbg.subprogram496 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str457, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str457, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str495, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 497, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite488 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str497 = internal constant [52 x i8] c"_ZN10polynomialIdE11inverse_fftERKS_ISt7complexIdEE\00", section "llvm.metadata" ; <[52 x i8]*> [#uses=1] -@llvm.dbg.subprogram498 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str460, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str460, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str497, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 535, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite492 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str499 = internal constant [28 x i8] c"_ZNK10polynomialIdEmlERKS0_\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram500 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str463, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str463, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str499, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 576, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite351 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str501 = internal constant [27 x i8] c"_ZN10polynomialIdEpLERKS0_\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram502 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str501, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 625, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite308 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str503 = internal constant [27 x i8] c"_ZN10polynomialIdEmIERKS0_\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram504 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str503, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 636, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite308 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str505 = internal constant [27 x i8] c"_ZN10polynomialIdEmLERKS0_\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram506 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str98, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str505, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 647, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite308 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array507 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite508 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array507 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str509 = internal constant [29 x i8] c"_ZN10polynomialIdE7acquireEv\00", section "llvm.metadata" ; <[29 x i8]*> [#uses=1] -@llvm.dbg.subprogram510 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str474, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str474, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str509, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 181, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite508 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str511 = internal constant [29 x i8] c"_ZN10polynomialIdE7releaseEv\00", section "llvm.metadata" ; <[29 x i8]*> [#uses=1] -@llvm.dbg.subprogram512 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str477, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str477, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str511, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 188, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite508 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array513 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype290 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite514 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array513 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str515 = internal constant [33 x i8] c"_ZN10polynomialIdE9deep_copyEPKd\00", section "llvm.metadata" ; <[33 x i8]*> [#uses=1] -@llvm.dbg.subprogram516 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([33 x i8]* @.str515, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 195, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite514 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array517 = internal constant [33 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype275 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype278 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype284 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram289 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram293 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram296 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram301 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram305 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram310 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram315 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram320 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram326 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram331 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram336 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram341 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram346 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram349 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram353 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram355 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram360 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram365 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram370 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram490 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram494 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram496 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram498 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram500 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram502 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram504 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram506 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram510 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram512 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram516 to %0*)], section "llvm.metadata" ; <[33 x %0*]*> [#uses=1] -@llvm.dbg.composite518 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([19 x i8]* @.str266, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 84, i64 96, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([33 x %0*]* @llvm.dbg.array517 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype519 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array520 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype180 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype519 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite521 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array520 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram522 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([23 x i8]* @.str335, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 306, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite521 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram527 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str324, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str324, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str325, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 111, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite323 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram530 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str418, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 306, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite417 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array534 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype196 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype224 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype31 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite535 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array534 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str536 = internal constant [19 x i8] c"operator*=<double>\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@.str537 = internal constant [35 x i8] c"_ZNSt7complexIdEmLIdEERS0_RKS_IT_E\00", section "llvm.metadata" ; <[35 x i8]*> [#uses=1] -@llvm.dbg.subprogram538 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([19 x i8]* @.str536, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str536, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str537, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1286, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite535 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str541 = internal constant [4 x i8] c"__t\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@llvm.dbg.variable542 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram538 to %0*), i8* getelementptr ([4 x i8]* @.str541, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1288, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.subprogram543 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([33 x i8]* @.str515, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 195, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite514 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram549 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str358, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str358, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str359, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 404, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite357 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array555 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype31 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype31 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite556 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array555 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str557 = internal constant [18 x i8] c"operator*<double>\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@.str558 = internal constant [32 x i8] c"_ZStmlIdESt7complexIT_ERKS2_S4_\00", section "llvm.metadata" ; <[32 x i8]*> [#uses=1] -@llvm.dbg.subprogram559 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str557, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str557, i32 0, i32 0), i8* getelementptr ([32 x i8]* @.str558, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 378, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite556 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable564 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram559 to %0*), i8* getelementptr ([4 x i8]* @.str244, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 380, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.subprogram565 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str482, i32 0, i32 0), i8* getelementptr ([47 x i8]* @.str483, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 195, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite481 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram569 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str324, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str324, i32 0, i32 0), i8* getelementptr ([41 x i8]* @.str410, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 111, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite409 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array572 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype31 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite573 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array572 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str574 = internal constant [18 x i8] c"operator-<double>\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@.str575 = internal constant [29 x i8] c"_ZStngIdESt7complexIT_ERKS2_\00", section "llvm.metadata" ; <[29 x i8]*> [#uses=1] -@llvm.dbg.subprogram576 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str574, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str574, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str575, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 443, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite573 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram578 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str329, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str329, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str330, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 300, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite328 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram581 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str363, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str363, i32 0, i32 0), i8* getelementptr ([32 x i8]* @.str364, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 423, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite362 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str591 = internal constant [19 x i8] c"operator/=<double>\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@.str592 = internal constant [35 x i8] c"_ZNSt7complexIdEdVIdEERS0_RKS_IT_E\00", section "llvm.metadata" ; <[35 x i8]*> [#uses=1] -@llvm.dbg.subprogram593 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([19 x i8]* @.str591, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str591, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str592, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1297, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite535 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable596 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram593 to %0*), i8* getelementptr ([4 x i8]* @.str541, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1299, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str597 = internal constant [18 x i8] c"operator/<double>\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@.str598 = internal constant [32 x i8] c"_ZStdvIdESt7complexIT_ERKS2_S4_\00", section "llvm.metadata" ; <[32 x i8]*> [#uses=1] -@llvm.dbg.subprogram599 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str597, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str597, i32 0, i32 0), i8* getelementptr ([32 x i8]* @.str598, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 408, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite556 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable602 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram599 to %0*), i8* getelementptr ([4 x i8]* @.str244, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 410, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str603 = internal constant [19 x i8] c"operator+=<double>\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@.str604 = internal constant [35 x i8] c"_ZNSt7complexIdEpLIdEERS0_RKS_IT_E\00", section "llvm.metadata" ; <[35 x i8]*> [#uses=1] -@llvm.dbg.subprogram605 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([19 x i8]* @.str603, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str603, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str604, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1268, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite535 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str608 = internal constant [18 x i8] c"operator+<double>\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@.str609 = internal constant [32 x i8] c"_ZStplIdESt7complexIT_ERKS2_S4_\00", section "llvm.metadata" ; <[32 x i8]*> [#uses=1] -@llvm.dbg.subprogram610 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str608, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str608, i32 0, i32 0), i8* getelementptr ([32 x i8]* @.str609, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 318, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite556 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable613 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram610 to %0*), i8* getelementptr ([4 x i8]* @.str244, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 320, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str614 = internal constant [19 x i8] c"operator-=<double>\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@.str615 = internal constant [35 x i8] c"_ZNSt7complexIdEmIIdEERS0_RKS_IT_E\00", section "llvm.metadata" ; <[35 x i8]*> [#uses=1] -@llvm.dbg.subprogram616 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([19 x i8]* @.str614, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str614, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str615, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 1277, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite535 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str619 = internal constant [32 x i8] c"_ZStmiIdESt7complexIT_ERKS2_S4_\00", section "llvm.metadata" ; <[32 x i8]*> [#uses=1] -@llvm.dbg.subprogram620 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str574, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str574, i32 0, i32 0), i8* getelementptr ([32 x i8]* @.str619, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 348, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite556 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable623 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram620 to %0*), i8* getelementptr ([4 x i8]* @.str244, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 350, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.subprogram624 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str329, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str329, i32 0, i32 0), i8* getelementptr ([38 x i8]* @.str414, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 300, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite413 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array627 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite628 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array627 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str629 = internal constant [42 x i8] c"__static_initialization_and_destruction_0\00", section "llvm.metadata" ; <[42 x i8]*> [#uses=1] -@llvm.dbg.subprogram630 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([42 x i8]* @.str629, i32 0, i32 0), i8* getelementptr ([42 x i8]* @.str629, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 703, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite628 to %0*), i1 true, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=0] -@.str635 = internal constant [9 x i8] c"iostream\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@llvm.dbg.compile_unit636 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([9 x i8]* @.str635, i32 0, i32 0), i8* getelementptr ([110 x i8]* @.str4, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@_ZStL8__ioinit = internal global %"struct.std::allocator<char>" zeroinitializer ; <%"struct.std::allocator<char>"*> [#uses=2] -@.str638 = internal constant [115 x i8] c"/developer/home2/zsth/projects/llvm.org/install/lib/gcc/i686-pc-linux-gnu/4.2.1/../../../../include/c++/4.2.1/bits\00", section "llvm.metadata" ; <[115 x i8]*> [#uses=1] -@__dso_handle = external global i8* ; <i8**> [#uses=1] -@_ZGVN10polynomialIdE4PI2IE = weak global i64 0, align 8 ; <i64*> [#uses=1] -@_ZN10polynomialIdE4PI2IE = weak global %"struct.std::complex<double>" zeroinitializer ; <%"struct.std::complex<double>"*> [#uses=3] -@llvm.dbg.array654 = internal constant [1 x %0*] zeroinitializer, section "llvm.metadata" ; <[1 x %0*]*> [#uses=1] -@llvm.dbg.composite655 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([1 x %0*]* @llvm.dbg.array654 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str656 = internal constant [16 x i8] c"_GLOBAL__I_main\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram657 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([16 x i8]* @.str656, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str656, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 704, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite655 to %0*), i1 true, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype658 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* null }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array659 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype658 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite660 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array659 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str661 = internal constant [8 x i8] c"__tcf_0\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.subprogram662 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str661, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str661, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit636 to %0*), i32 77, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite660 to %0*), i1 true, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram665 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str477, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str477, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str511, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 188, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite508 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str667 = internal constant [23 x i8] c"_ZN10polynomialIdED0Ev\00", section "llvm.metadata" ; <[23 x i8]*> [#uses=1] -@llvm.dbg.subprogram668 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([23 x i8]* @.str667, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 252, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite508 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@_ZTV10polynomialIdE = weak constant [4 x i32 (...)*] [i32 (...)* null, i32 (...)* bitcast (%struct.__class_type_info_pseudo* @_ZTI10polynomialIdE to i32 (...)*), i32 (...)* bitcast (void (%"struct.polynomial<double>"*)* @_ZN10polynomialIdED1Ev to i32 (...)*), i32 (...)* bitcast (void (%"struct.polynomial<double>"*)* @_ZN10polynomialIdED0Ev to i32 (...)*)], align 8 ; <[4 x i32 (...)*]*> [#uses=1] -@_ZTI10polynomialIdE = weak constant %struct.__class_type_info_pseudo { %struct.__type_info_pseudo { i8* inttoptr (i32 add (i32 ptrtoint ([0 x i32 (...)*]* @_ZTVN10__cxxabiv117__class_type_infoE to i32), i32 8) to i8*), i8* getelementptr ([16 x i8]* @_ZTS10polynomialIdE, i32 0, i32 0) } } ; <%struct.__class_type_info_pseudo*> [#uses=1] -@_ZTVN10__cxxabiv117__class_type_infoE = external constant [0 x i32 (...)*] ; <[0 x i32 (...)*]*> [#uses=1] -@_ZTS10polynomialIdE = weak constant [16 x i8] c"10polynomialIdE\00" ; <[16 x i8]*> [#uses=1] -@.str671 = internal constant [5 x i8] c"char\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.basictype672 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str671, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 8, i64 8, i64 0, i32 0, i32 6 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@.str676 = internal constant [11 x i8] c"<built-in>\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.compile_unit677 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([11 x i8]* @.str676, i32 0, i32 0), i8* getelementptr ([42 x i8]* @.str1, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@llvm.dbg.array680 = internal constant [0 x %0*] zeroinitializer, section "llvm.metadata" ; <[0 x %0*]*> [#uses=1] -@.str690 = internal constant [23 x i8] c"_ZN10polynomialIdED1Ev\00", section "llvm.metadata" ; <[23 x i8]*> [#uses=1] -@llvm.dbg.subprogram691 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([23 x i8]* @.str690, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 252, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite508 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram693 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str477, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str477, i32 0, i32 0), i8* getelementptr ([41 x i8]* @.str478, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 188, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite473 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str695 = internal constant [35 x i8] c"_ZN10polynomialISt7complexIdEED0Ev\00", section "llvm.metadata" ; <[35 x i8]*> [#uses=1] -@llvm.dbg.subprogram696 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str695, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 252, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite473 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@_ZTV10polynomialISt7complexIdEE = weak constant [4 x i32 (...)*] [i32 (...)* null, i32 (...)* bitcast (%struct.__class_type_info_pseudo* @_ZTI10polynomialISt7complexIdEE to i32 (...)*), i32 (...)* bitcast (void (%"struct.polynomial<std::complex<double> >"*)* @_ZN10polynomialISt7complexIdEED1Ev to i32 (...)*), i32 (...)* bitcast (void (%"struct.polynomial<std::complex<double> >"*)* @_ZN10polynomialISt7complexIdEED0Ev to i32 (...)*)], align 8 ; <[4 x i32 (...)*]*> [#uses=1] -@_ZTI10polynomialISt7complexIdEE = weak constant %struct.__class_type_info_pseudo { %struct.__type_info_pseudo { i8* inttoptr (i32 add (i32 ptrtoint ([0 x i32 (...)*]* @_ZTVN10__cxxabiv117__class_type_infoE to i32), i32 8) to i8*), i8* getelementptr ([28 x i8]* @_ZTS10polynomialISt7complexIdEE, i32 0, i32 0) } } ; <%struct.__class_type_info_pseudo*> [#uses=1] -@_ZTS10polynomialISt7complexIdEE = weak constant [28 x i8] c"10polynomialISt7complexIdEE\00" ; <[28 x i8]*> [#uses=1] -@.str707 = internal constant [35 x i8] c"_ZN10polynomialISt7complexIdEED1Ev\00", section "llvm.metadata" ; <[35 x i8]*> [#uses=1] -@llvm.dbg.subprogram708 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str304, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str707, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 252, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite473 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram710 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str474, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str474, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str509, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 181, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite508 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str712 = internal constant [23 x i8] c"_ZN10polynomialIdEC1Ej\00", section "llvm.metadata" ; <[23 x i8]*> [#uses=1] -@llvm.dbg.subprogram713 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([23 x i8]* @.str712, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 211, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite287 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str716 = internal constant [27 x i8] c"_ZN10polynomialIdEC1ERKS0_\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram717 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str716, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 242, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite300 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram720 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str318, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str318, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str319, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 276, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite317 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram729 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str309, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 259, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite308 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram732 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str474, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str474, i32 0, i32 0), i8* getelementptr ([41 x i8]* @.str475, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 181, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite473 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str734 = internal constant [35 x i8] c"_ZN10polynomialISt7complexIdEEC1Ej\00", section "llvm.metadata" ; <[35 x i8]*> [#uses=1] -@llvm.dbg.subprogram735 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str288, i32 0, i32 0), i8* getelementptr ([35 x i8]* @.str734, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 211, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite378 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram738 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([38 x i8]* @.str489, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 469, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite488 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable744 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram738 to %0*), i8* getelementptr ([7 x i8]* @.str259, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 473, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.subprogram747 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str448, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str493, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 483, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite492 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable751 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram747 to %0*), i8* getelementptr ([7 x i8]* @.str259, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 487, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.subprogram753 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([39 x i8]* @.str397, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 259, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite396 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram756 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str368, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str368, i32 0, i32 0), i8* getelementptr ([34 x i8]* @.str369, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 443, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite367 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str759 = internal constant [35 x i8] c"overflow in fft polynomial stretch\00" ; <[35 x i8]*> [#uses=1] -@_ZTISt14overflow_error = weak constant %struct.__si_class_type_info_pseudo { %struct.__type_info_pseudo { i8* inttoptr (i32 add (i32 ptrtoint ([0 x i32 (...)*]* @_ZTVN10__cxxabiv120__si_class_type_infoE to i32), i32 8) to i8*), i8* getelementptr ([19 x i8]* @_ZTSSt14overflow_error, i32 0, i32 0) }, %"struct.std::type_info"* bitcast (%struct.__si_class_type_info_pseudo* @_ZTISt13runtime_error to %"struct.std::type_info"*) } ; <%struct.__si_class_type_info_pseudo*> [#uses=2] -@_ZTVN10__cxxabiv120__si_class_type_infoE = external constant [0 x i32 (...)*] ; <[0 x i32 (...)*]*> [#uses=1] -@_ZTSSt14overflow_error = weak constant [19 x i8] c"St14overflow_error\00" ; <[19 x i8]*> [#uses=1] -@_ZTISt13runtime_error = external constant %struct.__si_class_type_info_pseudo ; <%struct.__si_class_type_info_pseudo*> [#uses=1] -@.str769 = internal constant [10 x i8] c"stdexcept\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.compile_unit770 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([10 x i8]* @.str769, i32 0, i32 0), i8* getelementptr ([110 x i8]* @.str4, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str773 = internal constant [15 x i8] c"overflow_error\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@.str775 = internal constant [14 x i8] c"runtime_error\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@.str777 = internal constant [10 x i8] c"exception\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.compile_unit778 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([10 x i8]* @.str777, i32 0, i32 0), i8* getelementptr ([110 x i8]* @.str4, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str780 = internal constant [16 x i8] c"_vptr.exception\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.derivedtype781 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([16 x i8]* @.str780, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit778 to %0*), i32 57, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype273 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype782 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite800 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array783 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype782 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite784 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array783 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram785 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str777, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str777, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit778 to %0*), i32 59, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite784 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array786 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype782 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite787 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array786 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str788 = internal constant [11 x i8] c"~exception\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.subprogram789 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str788, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str788, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit778 to %0*), i32 60, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite787 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype790 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 8, i64 8, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype791 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype790 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype792 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite800 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype793 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype792 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array794 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype793 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite795 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array794 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str796 = internal constant [5 x i8] c"what\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str797 = internal constant [24 x i8] c"_ZNKSt9exception4whatEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram798 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str796, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str796, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str797, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit778 to %0*), i32 63, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite795 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array799 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype781 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram785 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram789 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram798 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite800 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str777, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit778 to %0*), i32 57, i64 32, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array799 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype801 = internal constant %llvm.dbg.derivedtype.type { i32 458780, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite800 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str802 = internal constant [12 x i8] c"stringfwd.h\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.compile_unit803 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([12 x i8]* @.str802, i32 0, i32 0), i8* getelementptr ([115 x i8]* @.str638, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str804 = internal constant [64 x i8] c"basic_string<char,std::char_traits<char>,std::allocator<char> >\00", section "llvm.metadata" ; <[64 x i8]*> [#uses=1] -@.str806 = internal constant [15 x i8] c"basic_string.h\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.compile_unit807 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([15 x i8]* @.str806, i32 0, i32 0), i8* getelementptr ([115 x i8]* @.str638, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str808 = internal constant [13 x i8] c"_Alloc_hider\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@.str810 = internal constant [16 x i8] c"allocator<char>\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@.str812 = internal constant [16 x i8] c"new_allocator.h\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@.str813 = internal constant [114 x i8] c"/developer/home2/zsth/projects/llvm.org/install/lib/gcc/i686-pc-linux-gnu/4.2.1/../../../../include/c++/4.2.1/ext\00", section "llvm.metadata" ; <[114 x i8]*> [#uses=1] -@llvm.dbg.compile_unit814 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([16 x i8]* @.str812, i32 0, i32 0), i8* getelementptr ([114 x i8]* @.str813, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str815 = internal constant [20 x i8] c"new_allocator<char>\00", section "llvm.metadata" ; <[20 x i8]*> [#uses=1] -@llvm.dbg.derivedtype817 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite877 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array818 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype817 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite819 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array818 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str820 = internal constant [14 x i8] c"new_allocator\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram821 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str820, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str820, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 68, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite819 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype822 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 8, i64 8, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite877 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype823 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype822 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array824 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype817 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype823 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite825 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array824 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram826 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str820, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str820, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 70, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite825 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.composite827 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 54, i64 0, i64 0, i64 0, i32 4, %0* null, %0* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype828 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 8, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite827 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype829 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype828 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array830 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype817 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype829 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite831 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array830 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram832 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str820, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str820, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 73, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite831 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array833 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype817 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite834 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array833 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str835 = internal constant [15 x i8] c"~new_allocator\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram836 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str835, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str835, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 75, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite834 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype837 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype838 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype822 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype839 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array840 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype838 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype839 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite841 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array840 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str842 = internal constant [8 x i8] c"address\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str843 = internal constant [44 x i8] c"_ZNK9__gnu_cxx13new_allocatorIcE7addressERc\00", section "llvm.metadata" ; <[44 x i8]*> [#uses=1] -@llvm.dbg.subprogram844 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str842, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str842, i32 0, i32 0), i8* getelementptr ([44 x i8]* @.str843, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 78, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite841 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype845 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype790 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array846 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype838 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype845 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite847 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array846 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str848 = internal constant [45 x i8] c"_ZNK9__gnu_cxx13new_allocatorIcE7addressERKc\00", section "llvm.metadata" ; <[45 x i8]*> [#uses=1] -@llvm.dbg.subprogram849 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str842, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str842, i32 0, i32 0), i8* getelementptr ([45 x i8]* @.str848, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 81, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite847 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype850 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* null }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array851 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype817 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype850 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite852 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array851 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str853 = internal constant [9 x i8] c"allocate\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@.str854 = internal constant [46 x i8] c"_ZN9__gnu_cxx13new_allocatorIcE8allocateEjPKv\00", section "llvm.metadata" ; <[46 x i8]*> [#uses=1] -@llvm.dbg.subprogram855 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str853, i32 0, i32 0), i8* getelementptr ([9 x i8]* @.str853, i32 0, i32 0), i8* getelementptr ([46 x i8]* @.str854, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 86, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite852 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array856 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype817 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite857 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array856 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str858 = internal constant [11 x i8] c"deallocate\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str859 = internal constant [48 x i8] c"_ZN9__gnu_cxx13new_allocatorIcE10deallocateEPcj\00", section "llvm.metadata" ; <[48 x i8]*> [#uses=1] -@llvm.dbg.subprogram860 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str858, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str858, i32 0, i32 0), i8* getelementptr ([48 x i8]* @.str859, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 96, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite857 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array861 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype838 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite862 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array861 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str863 = internal constant [9 x i8] c"max_size\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@.str864 = internal constant [44 x i8] c"_ZNK9__gnu_cxx13new_allocatorIcE8max_sizeEv\00", section "llvm.metadata" ; <[44 x i8]*> [#uses=1] -@llvm.dbg.subprogram865 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str863, i32 0, i32 0), i8* getelementptr ([9 x i8]* @.str863, i32 0, i32 0), i8* getelementptr ([44 x i8]* @.str864, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 100, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite862 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array866 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype817 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype845 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite867 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array866 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str868 = internal constant [10 x i8] c"construct\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str869 = internal constant [48 x i8] c"_ZN9__gnu_cxx13new_allocatorIcE9constructEPcRKc\00", section "llvm.metadata" ; <[48 x i8]*> [#uses=1] -@llvm.dbg.subprogram870 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str868, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str868, i32 0, i32 0), i8* getelementptr ([48 x i8]* @.str869, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 106, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite867 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array871 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype817 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite872 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array871 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str873 = internal constant [8 x i8] c"destroy\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str874 = internal constant [43 x i8] c"_ZN9__gnu_cxx13new_allocatorIcE7destroyEPc\00", section "llvm.metadata" ; <[43 x i8]*> [#uses=1] -@llvm.dbg.subprogram875 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str873, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str873, i32 0, i32 0), i8* getelementptr ([43 x i8]* @.str874, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 110, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite872 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array876 = internal constant [11 x %0*] [%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram821 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram826 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram832 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram836 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram844 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram849 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram855 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram860 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram865 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram870 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram875 to %0*)], section "llvm.metadata" ; <[11 x %0*]*> [#uses=1] -@llvm.dbg.composite877 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([20 x i8]* @.str815, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit814 to %0*), i32 54, i64 8, i64 8, i64 0, i32 0, %0* null, %0* bitcast ([11 x %0*]* @llvm.dbg.array876 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype878 = internal constant %llvm.dbg.derivedtype.type { i32 458780, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit803 to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite877 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype879 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite902 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array880 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype879 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite881 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array880 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str882 = internal constant [12 x i8] c"allocator.h\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.compile_unit883 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([12 x i8]* @.str882, i32 0, i32 0), i8* getelementptr ([115 x i8]* @.str638, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str884 = internal constant [10 x i8] c"allocator\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.subprogram885 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str884, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str884, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit883 to %0*), i32 100, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite881 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype886 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 8, i64 8, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite902 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype887 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype886 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array888 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype879 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite889 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array888 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram890 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str884, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str884, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit883 to %0*), i32 102, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite889 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.composite891 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit803 to %0*), i32 49, i64 0, i64 0, i64 0, i32 4, %0* null, %0* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype892 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 8, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite891 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype893 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype892 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array894 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype879 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype893 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite895 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array894 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram896 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str884, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str884, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit883 to %0*), i32 106, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite895 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array897 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype879 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite898 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array897 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str899 = internal constant [11 x i8] c"~allocator\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.subprogram900 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str899, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str899, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit883 to %0*), i32 108, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite898 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array901 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype878 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram885 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram890 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram896 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram900 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite902 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([16 x i8]* @.str810, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit803 to %0*), i32 49, i64 8, i64 8, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array901 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype903 = internal constant %llvm.dbg.derivedtype.type { i32 458780, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite902 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str904 = internal constant [5 x i8] c"_M_p\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.derivedtype905 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str904, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 264, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype906 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite911 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array907 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype906 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite908 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array907 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram909 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str808, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str808, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 261, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite908 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array910 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype903 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype905 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram909 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite911 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str808, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 260, i64 32, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array910 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str912 = internal constant [12 x i8] c"_M_dataplus\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.derivedtype913 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str912, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 276, i64 32, i64 32, i64 0, i32 1, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite911 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str914 = internal constant [7 x i8] c"string\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@llvm.dbg.derivedtype915 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str914, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit803 to %0*), i32 56, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1693 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype916 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype915 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype917 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype916 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array918 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite919 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array918 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str920 = internal constant [8 x i8] c"_M_data\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str921 = internal constant [17 x i8] c"_ZNKSs7_M_dataEv\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram922 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str920, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str920, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str921, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 279, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite919 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype923 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1693 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array924 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite925 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array924 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str926 = internal constant [17 x i8] c"_ZNSs7_M_dataEPc\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram927 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str920, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str920, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str926, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 283, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite925 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str928 = internal constant [5 x i8] c"_Rep\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str930 = internal constant [10 x i8] c"_Rep_base\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str932 = internal constant [10 x i8] c"_M_length\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.derivedtype933 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str932, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 149, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str934 = internal constant [12 x i8] c"_M_capacity\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.derivedtype935 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str934, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 150, i64 32, i64 32, i64 32, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str936 = internal constant [10 x i8] c"ptrdiff_t\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.derivedtype937 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str936, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit677 to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype938 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype937 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str939 = internal constant [8 x i8] c"types.h\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.compile_unit940 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([8 x i8]* @.str939, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str69, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str941 = internal constant [10 x i8] c"__int32_t\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.derivedtype942 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str941, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit940 to %0*), i32 43, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype938 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str943 = internal constant [8 x i8] c"__pid_t\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.derivedtype944 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str943, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit940 to %0*), i32 145, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype942 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str945 = internal constant [10 x i8] c"__daddr_t\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.derivedtype946 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str945, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit940 to %0*), i32 154, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype944 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str947 = internal constant [8 x i8] c"__key_t\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.derivedtype948 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str947, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit940 to %0*), i32 157, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype946 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str949 = internal constant [12 x i8] c"__clockid_t\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.derivedtype950 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str949, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit940 to %0*), i32 158, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype948 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str951 = internal constant [10 x i8] c"__ssize_t\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.derivedtype952 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str951, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit940 to %0*), i32 181, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype950 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str953 = internal constant [11 x i8] c"__intptr_t\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.derivedtype954 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str953, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit940 to %0*), i32 189, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype952 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str955 = internal constant [12 x i8] c"_G_config.h\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@.str956 = internal constant [13 x i8] c"/usr/include\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.compile_unit957 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([12 x i8]* @.str955, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str956, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str958 = internal constant [11 x i8] c"_G_int32_t\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.derivedtype959 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str958, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit957 to %0*), i32 55, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype954 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str960 = internal constant [11 x i8] c"nl_types.h\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.compile_unit961 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([11 x i8]* @.str960, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str956, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str962 = internal constant [8 x i8] c"nl_item\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.derivedtype963 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str962, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit961 to %0*), i32 34, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype959 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str964 = internal constant [7 x i8] c"time.h\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@llvm.dbg.compile_unit965 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([7 x i8]* @.str964, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str956, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str966 = internal constant [10 x i8] c"clockid_t\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.derivedtype967 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str966, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit965 to %0*), i32 77, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype963 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str968 = internal constant [6 x i8] c"pid_t\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.derivedtype969 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str968, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit965 to %0*), i32 105, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype967 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str970 = internal constant [15 x i8] c"__sig_atomic_t\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.derivedtype971 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str970, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit965 to %0*), i32 413, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype969 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str972 = internal constant [15 x i8] c"pthreadtypes.h\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.compile_unit973 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([15 x i8]* @.str972, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str69, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str974 = internal constant [15 x i8] c"pthread_once_t\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.derivedtype975 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str974, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit973 to %0*), i32 109, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype971 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype976 = internal constant %llvm.dbg.derivedtype.type { i32 458805, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype975 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str977 = internal constant [19 x i8] c"pthread_spinlock_t\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.derivedtype978 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([19 x i8]* @.str977, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit973 to %0*), i32 142, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype976 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str979 = internal constant [10 x i8] c"pthread.h\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.compile_unit980 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([10 x i8]* @.str979, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str956, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str981 = internal constant [8 x i8] c"ssize_t\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.derivedtype982 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str981, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit980 to %0*), i32 1101, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype978 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str983 = internal constant [9 x i8] c"unistd.h\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@llvm.dbg.compile_unit984 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([9 x i8]* @.str983, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str956, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str985 = internal constant [9 x i8] c"intptr_t\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@llvm.dbg.derivedtype986 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str985, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit984 to %0*), i32 226, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype982 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str987 = internal constant [15 x i8] c"gthr-default.h\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@.str988 = internal constant [133 x i8] c"/developer/home2/zsth/projects/llvm.org/install/lib/gcc/i686-pc-linux-gnu/4.2.1/../../../../include/c++/4.2.1/i686-pc-linux-gnu/bits\00", section "llvm.metadata" ; <[133 x i8]*> [#uses=1] -@llvm.dbg.compile_unit989 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([15 x i8]* @.str987, i32 0, i32 0), i8* getelementptr ([133 x i8]* @.str988, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str990 = internal constant [17 x i8] c"__gthread_once_t\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.derivedtype991 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([17 x i8]* @.str990, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit989 to %0*), i32 46, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype986 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype992 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str941, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit940 to %0*), i32 43, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype991 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str993 = internal constant [9 x i8] c"stdint.h\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@llvm.dbg.compile_unit994 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([9 x i8]* @.str993, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str956, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str995 = internal constant [8 x i8] c"int32_t\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.derivedtype996 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str995, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit994 to %0*), i32 38, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype992 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str997 = internal constant [14 x i8] c"int_least32_t\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.derivedtype998 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str997, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit994 to %0*), i32 67, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype996 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str999 = internal constant [13 x i8] c"int_fast16_t\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1000 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str999, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit994 to %0*), i32 91, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype998 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1001 = internal constant [13 x i8] c"int_fast32_t\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1002 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1001, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit994 to %0*), i32 97, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1000 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1003 = internal constant [11 x i8] c"postypes.h\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.compile_unit1004 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([11 x i8]* @.str1003, i32 0, i32 0), i8* getelementptr ([115 x i8]* @.str638, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str1005 = internal constant [11 x i8] c"streamsize\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1006 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1005, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1004 to %0*), i32 72, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1002 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1007 = internal constant [17 x i8] c"/usr/include/sys\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.compile_unit1008 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([8 x i8]* @.str939, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1007, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str1009 = internal constant [8 x i8] c"daddr_t\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1010 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1009, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1008 to %0*), i32 105, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1006 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1011 = internal constant [6 x i8] c"key_t\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1012 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1011, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1008 to %0*), i32 117, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1010 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1013 = internal constant [11 x i8] c"register_t\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1014 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1013, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1008 to %0*), i32 204, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1012 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1015 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str936, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit677 to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1014 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1016 = internal constant [9 x i8] c"stdlib.h\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@llvm.dbg.compile_unit1017 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([9 x i8]* @.str1016, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str956, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str1018 = internal constant [13 x i8] c"_Atomic_word\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1019 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1018, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1017 to %0*), i32 962, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1015 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1020 = internal constant [12 x i8] c"_M_refcount\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1021 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str1020, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 151, i64 32, i64 32, i64 64, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1019 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1022 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype933 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype935 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1021 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1023 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str930, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 148, i64 96, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1022 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype1024 = internal constant %llvm.dbg.derivedtype.type { i32 458780, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1023 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1025 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1091 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1026 = internal constant [1 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1025 to %0*)], section "llvm.metadata" ; <[1 x %0*]*> [#uses=1] -@llvm.dbg.composite1027 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([1 x %0*]* @llvm.dbg.array1026 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1028 = internal constant [13 x i8] c"_S_empty_rep\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@.str1029 = internal constant [27 x i8] c"_ZNSs4_Rep12_S_empty_repEv\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram1030 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1028, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1028, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str1029, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 180, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1027 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1031 = internal constant [5 x i8] c"bool\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.basictype1032 = internal constant %llvm.dbg.basictype.type { i32 458788, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1031, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 8, i64 8, i64 0, i32 0, i32 2 }, section "llvm.metadata" ; <%llvm.dbg.basictype.type*> [#uses=1] -@llvm.dbg.derivedtype1033 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 96, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1091 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1034 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1033 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1035 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype1032 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1034 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1036 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1035 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1037 = internal constant [13 x i8] c"_M_is_leaked\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@.str1038 = internal constant [28 x i8] c"_ZNKSs4_Rep12_M_is_leakedEv\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram1039 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1037, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1037, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str1038, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 190, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1036 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1040 = internal constant [13 x i8] c"_M_is_shared\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@.str1041 = internal constant [28 x i8] c"_ZNKSs4_Rep12_M_is_sharedEv\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram1042 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1040, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1040, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str1041, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 194, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1036 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype1043 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1091 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1044 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1043 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1045 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1044 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1046 = internal constant [14 x i8] c"_M_set_leaked\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@.str1047 = internal constant [28 x i8] c"_ZNSs4_Rep13_M_set_leakedEv\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram1048 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1046, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1046, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str1047, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 198, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1045 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1049 = internal constant [16 x i8] c"_M_set_sharable\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@.str1050 = internal constant [30 x i8] c"_ZNSs4_Rep15_M_set_sharableEv\00", section "llvm.metadata" ; <[30 x i8]*> [#uses=1] -@llvm.dbg.subprogram1051 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([16 x i8]* @.str1049, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1049, i32 0, i32 0), i8* getelementptr ([30 x i8]* @.str1050, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 202, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1045 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1052 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1043 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1053 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1052 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1054 = internal constant [27 x i8] c"_M_set_length_and_sharable\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@.str1055 = internal constant [41 x i8] c"_ZNSs4_Rep26_M_set_length_and_sharableEj\00", section "llvm.metadata" ; <[41 x i8]*> [#uses=1] -@llvm.dbg.subprogram1056 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([27 x i8]* @.str1054, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str1054, i32 0, i32 0), i8* getelementptr ([41 x i8]* @.str1055, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 206, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1053 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1057 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1043 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1058 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1057 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1059 = internal constant [11 x i8] c"_M_refdata\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str1060 = internal constant [25 x i8] c"_ZNSs4_Rep10_M_refdataEv\00", section "llvm.metadata" ; <[25 x i8]*> [#uses=1] -@llvm.dbg.subprogram1061 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1059, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1059, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str1060, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 216, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1058 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1062 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1043 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1063 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1062 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1064 = internal constant [8 x i8] c"_M_grab\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str1065 = internal constant [30 x i8] c"_ZNSs4_Rep7_M_grabERKSaIcES2_\00", section "llvm.metadata" ; <[30 x i8]*> [#uses=1] -@llvm.dbg.subprogram1066 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1064, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1064, i32 0, i32 0), i8* getelementptr ([30 x i8]* @.str1065, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 220, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1063 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1067 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1043 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1068 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1067 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1069 = internal constant [17 x i8] c"basic_string.tcc\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.compile_unit1070 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([17 x i8]* @.str1069, i32 0, i32 0), i8* getelementptr ([115 x i8]* @.str638, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str1071 = internal constant [10 x i8] c"_S_create\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str1072 = internal constant [31 x i8] c"_ZNSs4_Rep9_S_createEjjRKSaIcE\00", section "llvm.metadata" ; <[31 x i8]*> [#uses=1] -@llvm.dbg.subprogram1073 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str1071, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str1071, i32 0, i32 0), i8* getelementptr ([31 x i8]* @.str1072, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 529, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1068 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1074 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1043 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1075 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1074 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1076 = internal constant [11 x i8] c"_M_dispose\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str1077 = internal constant [31 x i8] c"_ZNSs4_Rep10_M_disposeERKSaIcE\00", section "llvm.metadata" ; <[31 x i8]*> [#uses=1] -@llvm.dbg.subprogram1078 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1076, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1076, i32 0, i32 0), i8* getelementptr ([31 x i8]* @.str1077, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 231, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1075 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1079 = internal constant [11 x i8] c"_M_destroy\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str1080 = internal constant [31 x i8] c"_ZNSs4_Rep10_M_destroyERKSaIcE\00", section "llvm.metadata" ; <[31 x i8]*> [#uses=1] -@llvm.dbg.subprogram1081 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1079, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1079, i32 0, i32 0), i8* getelementptr ([31 x i8]* @.str1080, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 427, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1075 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1082 = internal constant [11 x i8] c"_M_refcopy\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str1083 = internal constant [25 x i8] c"_ZNSs4_Rep10_M_refcopyEv\00", section "llvm.metadata" ; <[25 x i8]*> [#uses=1] -@llvm.dbg.subprogram1084 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1082, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1082, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str1083, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 245, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1058 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1085 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1043 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1086 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1085 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1087 = internal constant [9 x i8] c"_M_clone\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@.str1088 = internal constant [29 x i8] c"_ZNSs4_Rep8_M_cloneERKSaIcEj\00", section "llvm.metadata" ; <[29 x i8]*> [#uses=1] -@llvm.dbg.subprogram1089 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str1087, i32 0, i32 0), i8* getelementptr ([9 x i8]* @.str1087, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str1088, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 606, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1086 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1090 = internal constant [14 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1024 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1030 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1039 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1042 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1048 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1051 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1056 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1061 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1066 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1073 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1078 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1081 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1084 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1089 to %0*)], section "llvm.metadata" ; <[14 x %0*]*> [#uses=1] -@llvm.dbg.composite1091 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str928, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 155, i64 96, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([14 x %0*]* @llvm.dbg.array1090 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype1092 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1091 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1093 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1092 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1094 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1093 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1095 = internal constant [16 x i8] c"_ZNKSs6_M_repEv\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram1096 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str1087, i32 0, i32 0), i8* getelementptr ([9 x i8]* @.str1087, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1095, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 287, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1094 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1097 = internal constant [15 x i8] c"stl_iterator.h\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.compile_unit1098 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([15 x i8]* @.str1097, i32 0, i32 0), i8* getelementptr ([115 x i8]* @.str638, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str1099 = internal constant [97 x i8] c"__normal_iterator<char*,std::basic_string<char, std::char_traits<char>, std::allocator<char> > >\00", section "llvm.metadata" ; <[97 x i8]*> [#uses=1] -@.str1101 = internal constant [11 x i8] c"_M_current\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1102 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1101, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 639, i64 32, i64 32, i64 0, i32 2, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1103 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1104 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1103 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1105 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1104 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1106 = internal constant [18 x i8] c"__normal_iterator\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1107 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str1106, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1106, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 650, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1105 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1108 = internal constant [10 x i8] c"__caddr_t\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1109 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str1108, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit940 to %0*), i32 188, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1110 = internal constant [15 x i8] c"__gnuc_va_list\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1111 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str1110, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit957 to %0*), i32 58, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1109 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1112 = internal constant [8 x i8] c"libio.h\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.compile_unit1113 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([8 x i8]* @.str1112, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str956, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str1114 = internal constant [8 x i8] c"va_list\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1115 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1114, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1113 to %0*), i32 491, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1111 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1116 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1115 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1117 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1116 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1118 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1103 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1117 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1119 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1118 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1120 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str1106, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1106, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 653, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1119 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.composite1121 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 637, i64 0, i64 0, i64 0, i32 4, %0* null, %0* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype1122 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 8, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1121 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1123 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1122 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1124 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1103 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1123 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1125 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1124 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1126 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str1106, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1106, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 660, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1125 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype1127 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1128 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1127 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1129 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype839 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1128 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1130 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1129 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1131 = internal constant [44 x i8] c"_ZNK9__gnu_cxx17__normal_iteratorIPcSsEdeEv\00", section "llvm.metadata" ; <[44 x i8]*> [#uses=1] -@llvm.dbg.subprogram1132 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str463, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str463, i32 0, i32 0), i8* getelementptr ([44 x i8]* @.str1131, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 665, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1130 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1133 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1128 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1134 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1133 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1135 = internal constant [11 x i8] c"operator->\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str1136 = internal constant [44 x i8] c"_ZNK9__gnu_cxx17__normal_iteratorIPcSsEptEv\00", section "llvm.metadata" ; <[44 x i8]*> [#uses=1] -@llvm.dbg.subprogram1137 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1135, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1135, i32 0, i32 0), i8* getelementptr ([44 x i8]* @.str1136, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 669, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1134 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype1138 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1139 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1138 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1103 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1140 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1139 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1141 = internal constant [11 x i8] c"operator++\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str1142 = internal constant [43 x i8] c"_ZN9__gnu_cxx17__normal_iteratorIPcSsEppEv\00", section "llvm.metadata" ; <[43 x i8]*> [#uses=1] -@llvm.dbg.subprogram1143 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1141, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1141, i32 0, i32 0), i8* getelementptr ([43 x i8]* @.str1142, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 673, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1140 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1144 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1103 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1145 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1144 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1146 = internal constant [43 x i8] c"_ZN9__gnu_cxx17__normal_iteratorIPcSsEppEi\00", section "llvm.metadata" ; <[43 x i8]*> [#uses=1] -@llvm.dbg.subprogram1147 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1141, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1141, i32 0, i32 0), i8* getelementptr ([43 x i8]* @.str1146, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 680, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1145 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1148 = internal constant [11 x i8] c"operator--\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@.str1149 = internal constant [43 x i8] c"_ZN9__gnu_cxx17__normal_iteratorIPcSsEmmEv\00", section "llvm.metadata" ; <[43 x i8]*> [#uses=1] -@llvm.dbg.subprogram1150 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1148, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1148, i32 0, i32 0), i8* getelementptr ([43 x i8]* @.str1149, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 685, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1140 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1151 = internal constant [43 x i8] c"_ZN9__gnu_cxx17__normal_iteratorIPcSsEmmEi\00", section "llvm.metadata" ; <[43 x i8]*> [#uses=1] -@llvm.dbg.subprogram1152 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str1148, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1148, i32 0, i32 0), i8* getelementptr ([43 x i8]* @.str1151, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 692, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1145 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype1153 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1015 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1154 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype839 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1128 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1153 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1155 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1154 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1156 = internal constant [46 x i8] c"_ZNK9__gnu_cxx17__normal_iteratorIPcSsEixERKi\00", section "llvm.metadata" ; <[46 x i8]*> [#uses=1] -@llvm.dbg.subprogram1157 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([46 x i8]* @.str1156, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 697, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1155 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1158 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1138 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1103 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1153 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1159 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1158 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1160 = internal constant [45 x i8] c"_ZN9__gnu_cxx17__normal_iteratorIPcSsEpLERKi\00", section "llvm.metadata" ; <[45 x i8]*> [#uses=1] -@llvm.dbg.subprogram1161 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([45 x i8]* @.str1160, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 701, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1159 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1162 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1128 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1153 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1163 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1162 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1164 = internal constant [46 x i8] c"_ZNK9__gnu_cxx17__normal_iteratorIPcSsEplERKi\00", section "llvm.metadata" ; <[46 x i8]*> [#uses=1] -@llvm.dbg.subprogram1165 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str347, i32 0, i32 0), i8* getelementptr ([46 x i8]* @.str1164, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 705, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1163 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1166 = internal constant [45 x i8] c"_ZN9__gnu_cxx17__normal_iteratorIPcSsEmIERKi\00", section "llvm.metadata" ; <[45 x i8]*> [#uses=1] -@llvm.dbg.subprogram1167 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str95, i32 0, i32 0), i8* getelementptr ([45 x i8]* @.str1166, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 709, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1159 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1168 = internal constant [46 x i8] c"_ZNK9__gnu_cxx17__normal_iteratorIPcSsEmiERKi\00", section "llvm.metadata" ; <[46 x i8]*> [#uses=1] -@llvm.dbg.subprogram1169 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str344, i32 0, i32 0), i8* getelementptr ([46 x i8]* @.str1168, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 713, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1163 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1170 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1117 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1128 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1171 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1170 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1172 = internal constant [5 x i8] c"base\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str1173 = internal constant [47 x i8] c"_ZNK9__gnu_cxx17__normal_iteratorIPcSsE4baseEv\00", section "llvm.metadata" ; <[47 x i8]*> [#uses=1] -@llvm.dbg.subprogram1174 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1172, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1172, i32 0, i32 0), i8* getelementptr ([47 x i8]* @.str1173, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 717, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1171 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1175 = internal constant [16 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1102 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1107 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1120 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1126 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1132 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1137 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1143 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1147 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1150 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1152 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1157 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1161 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1165 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1167 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1169 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1174 to %0*)], section "llvm.metadata" ; <[16 x %0*]*> [#uses=1] -@llvm.dbg.composite1176 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([97 x i8]* @.str1099, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 637, i64 32, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([16 x %0*]* @llvm.dbg.array1175 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array1177 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1178 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1177 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1179 = internal constant [19 x i8] c"_ZNKSs9_M_ibeginEv\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.subprogram1180 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1172, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1172, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str1179, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 293, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1178 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1181 = internal constant [8 x i8] c"_M_iend\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str1182 = internal constant [17 x i8] c"_ZNKSs7_M_iendEv\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram1183 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1181, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1181, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1182, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 297, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1178 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1184 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1185 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1184 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1186 = internal constant [8 x i8] c"_M_leak\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str1187 = internal constant [16 x i8] c"_ZNSs7_M_leakEv\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram1188 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1186, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1186, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1187, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 301, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1185 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1189 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1190 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1189 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1191 = internal constant [9 x i8] c"_M_check\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@.str1192 = internal constant [21 x i8] c"_ZNKSs8_M_checkEjPKc\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram1193 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str1191, i32 0, i32 0), i8* getelementptr ([9 x i8]* @.str1191, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str1192, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 308, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1190 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1194 = internal constant [5 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1195 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1194 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1196 = internal constant [16 x i8] c"_M_check_length\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@.str1197 = internal constant [30 x i8] c"_ZNKSs15_M_check_lengthEjjPKc\00", section "llvm.metadata" ; <[30 x i8]*> [#uses=1] -@llvm.dbg.subprogram1198 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([16 x i8]* @.str1196, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1196, i32 0, i32 0), i8* getelementptr ([30 x i8]* @.str1197, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 316, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1195 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1199 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1200 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1199 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1201 = internal constant [9 x i8] c"_M_limit\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@.str1202 = internal constant [19 x i8] c"_ZNKSs8_M_limitEjj\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.subprogram1203 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str1201, i32 0, i32 0), i8* getelementptr ([9 x i8]* @.str1201, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str1202, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 324, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1200 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1204 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype1032 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1205 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1204 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1206 = internal constant [12 x i8] c"_M_disjunct\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@.str1207 = internal constant [24 x i8] c"_ZNKSs11_M_disjunctEPKc\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram1208 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str1206, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str1206, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str1207, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 332, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1205 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1209 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1210 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1209 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1211 = internal constant [8 x i8] c"_M_copy\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str1212 = internal constant [21 x i8] c"_ZNSs7_M_copyEPcPKcj\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram1213 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1211, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1211, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str1212, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 341, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1210 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1214 = internal constant [8 x i8] c"_M_move\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str1215 = internal constant [21 x i8] c"_ZNSs7_M_moveEPcPKcj\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram1216 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1214, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1214, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str1215, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 350, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1210 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1217 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1218 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1217 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1219 = internal constant [10 x i8] c"_M_assign\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str1220 = internal constant [21 x i8] c"_ZNSs9_M_assignEPcjc\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram1221 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str1219, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str1219, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str1220, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 359, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1218 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1222 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* null, %0* null], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1223 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1222 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1224 = internal constant [14 x i8] c"_S_copy_chars\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram1225 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 371, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1223 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1226 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1227 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1226 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1228 = internal constant [64 x i8] c"_ZNSs13_S_copy_charsEPcN9__gnu_cxx17__normal_iteratorIS_SsEES2_\00", section "llvm.metadata" ; <[64 x i8]*> [#uses=1] -@llvm.dbg.subprogram1229 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* getelementptr ([64 x i8]* @.str1228, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 378, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1227 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1230 = internal constant [103 x i8] c"__normal_iterator<const char*,std::basic_string<char, std::char_traits<char>, std::allocator<char> > >\00", section "llvm.metadata" ; <[103 x i8]*> [#uses=1] -@llvm.dbg.composite1231 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([103 x i8]* @.str1230, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 637, i64 0, i64 0, i64 0, i32 4, %0* null, %0* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array1232 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1231 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1231 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1233 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1232 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1234 = internal constant [65 x i8] c"_ZNSs13_S_copy_charsEPcN9__gnu_cxx17__normal_iteratorIPKcSsEES4_\00", section "llvm.metadata" ; <[65 x i8]*> [#uses=1] -@llvm.dbg.subprogram1235 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* getelementptr ([65 x i8]* @.str1234, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 382, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1233 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1236 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1237 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1236 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1238 = internal constant [28 x i8] c"_ZNSs13_S_copy_charsEPcS_S_\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram1239 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str1238, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 386, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1237 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1240 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1241 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1240 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1242 = internal constant [30 x i8] c"_ZNSs13_S_copy_charsEPcPKcS1_\00", section "llvm.metadata" ; <[30 x i8]*> [#uses=1] -@llvm.dbg.subprogram1243 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1224, i32 0, i32 0), i8* getelementptr ([30 x i8]* @.str1242, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 390, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1241 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1244 = internal constant [5 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1245 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1244 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1246 = internal constant [10 x i8] c"_M_mutate\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str1247 = internal constant [20 x i8] c"_ZNSs9_M_mutateEjjj\00", section "llvm.metadata" ; <[20 x i8]*> [#uses=1] -@llvm.dbg.subprogram1248 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str1246, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str1246, i32 0, i32 0), i8* getelementptr ([20 x i8]* @.str1247, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 451, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1245 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1249 = internal constant [13 x i8] c"_M_leak_hard\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@.str1250 = internal constant [22 x i8] c"_ZNSs12_M_leak_hardEv\00", section "llvm.metadata" ; <[22 x i8]*> [#uses=1] -@llvm.dbg.subprogram1251 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1249, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1249, i32 0, i32 0), i8* getelementptr ([22 x i8]* @.str1250, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 437, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1185 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1252 = internal constant [22 x i8] c"_ZNSs12_S_empty_repEv\00", section "llvm.metadata" ; <[22 x i8]*> [#uses=1] -@llvm.dbg.subprogram1253 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1028, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1028, i32 0, i32 0), i8* getelementptr ([22 x i8]* @.str1252, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 400, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1027 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1254 = internal constant [13 x i8] c"basic_string\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.subprogram1255 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 2055, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1185 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1256 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1257 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1256 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1258 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 191, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1257 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype1259 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype916 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1260 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1261 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1260 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1262 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 183, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1261 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1263 = internal constant [5 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1264 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1263 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1265 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 197, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1264 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1266 = internal constant [6 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1267 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1266 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1268 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 208, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1267 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1269 = internal constant [5 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1270 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1269 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1271 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 219, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1270 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1272 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1273 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1272 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1274 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 226, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1273 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1275 = internal constant [5 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1276 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1275 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1277 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 233, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1276 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1278 = internal constant [5 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* null, %0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1279 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1278 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1280 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1254, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 477, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1279 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1281 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1282 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1281 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1283 = internal constant [14 x i8] c"~basic_string\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram1284 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1283, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1283, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 482, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1282 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype1285 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1693 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1286 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1287 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1286 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1288 = internal constant [13 x i8] c"_ZNSsaSERKSs\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.subprogram1289 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1288, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 490, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1287 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1290 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1291 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1290 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1292 = internal constant [12 x i8] c"_ZNSsaSEPKc\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.subprogram1293 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str1292, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 498, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1291 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1294 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1295 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1294 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1296 = internal constant [10 x i8] c"_ZNSsaSEc\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.subprogram1297 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str89, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str1296, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 509, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1295 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1298 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1299 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1298 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1300 = internal constant [6 x i8] c"begin\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str1301 = internal constant [14 x i8] c"_ZNSs5beginEv\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram1302 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1300, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1300, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1301, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 521, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1299 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1303 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1231 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1304 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1303 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1305 = internal constant [15 x i8] c"_ZNKSs5beginEv\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram1306 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1300, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1300, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1305, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 532, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1304 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1307 = internal constant [4 x i8] c"end\00", section "llvm.metadata" ; <[4 x i8]*> [#uses=1] -@.str1308 = internal constant [12 x i8] c"_ZNSs3endEv\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.subprogram1309 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str1307, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str1307, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str1308, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 540, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1299 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1310 = internal constant [13 x i8] c"_ZNKSs3endEv\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.subprogram1311 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str1307, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str1307, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1310, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 551, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1304 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1312 = internal constant [128 x i8] c"reverse_iterator<__gnu_cxx::__normal_iterator<char*, std::basic_string<char, std::char_traits<char>, std::allocator<char> > > >\00", section "llvm.metadata" ; <[128 x i8]*> [#uses=1] -@llvm.dbg.composite1313 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([128 x i8]* @.str1312, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 100, i64 0, i64 0, i64 0, i32 4, %0* null, %0* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array1314 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1313 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1315 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1314 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1316 = internal constant [7 x i8] c"rbegin\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@.str1317 = internal constant [15 x i8] c"_ZNSs6rbeginEv\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram1318 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1316, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1316, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1317, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 560, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1315 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1319 = internal constant [134 x i8] c"reverse_iterator<__gnu_cxx::__normal_iterator<const char*, std::basic_string<char, std::char_traits<char>, std::allocator<char> > > >\00", section "llvm.metadata" ; <[134 x i8]*> [#uses=1] -@llvm.dbg.composite1320 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([134 x i8]* @.str1319, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1098 to %0*), i32 100, i64 0, i64 0, i64 0, i32 4, %0* null, %0* null, i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array1321 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1320 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1322 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1321 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1323 = internal constant [16 x i8] c"_ZNKSs6rbeginEv\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram1324 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1316, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1316, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1323, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 569, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1322 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1325 = internal constant [5 x i8] c"rend\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str1326 = internal constant [13 x i8] c"_ZNSs4rendEv\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.subprogram1327 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1325, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1325, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1326, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 578, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1315 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1328 = internal constant [14 x i8] c"_ZNKSs4rendEv\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram1329 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1325, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1325, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1328, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 587, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1322 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1330 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1331 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1330 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1332 = internal constant [5 x i8] c"size\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str1333 = internal constant [14 x i8] c"_ZNKSs4sizeEv\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram1334 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1332, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1332, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1333, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 595, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1331 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1335 = internal constant [7 x i8] c"length\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@.str1336 = internal constant [16 x i8] c"_ZNKSs6lengthEv\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram1337 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1335, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1335, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1336, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 601, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1331 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1338 = internal constant [18 x i8] c"_ZNKSs8max_sizeEv\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1339 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str863, i32 0, i32 0), i8* getelementptr ([9 x i8]* @.str863, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1338, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 606, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1331 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1340 = internal constant [4 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1341 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1340 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1342 = internal constant [7 x i8] c"resize\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@.str1343 = internal constant [16 x i8] c"_ZNSs6resizeEjc\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram1344 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1342, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1342, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1343, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 622, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1341 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1345 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1346 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1345 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1347 = internal constant [15 x i8] c"_ZNSs6resizeEj\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram1348 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1342, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1342, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1347, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 633, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1346 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1349 = internal constant [9 x i8] c"capacity\00", section "llvm.metadata" ; <[9 x i8]*> [#uses=1] -@.str1350 = internal constant [18 x i8] c"_ZNKSs8capacityEv\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1351 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([9 x i8]* @.str1349, i32 0, i32 0), i8* getelementptr ([9 x i8]* @.str1349, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1350, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 641, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1331 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.composite1352 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1345 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1353 = internal constant [8 x i8] c"reserve\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str1354 = internal constant [16 x i8] c"_ZNSs7reserveEj\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram1355 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1353, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1353, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1354, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 484, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1352 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1356 = internal constant [6 x i8] c"clear\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str1357 = internal constant [14 x i8] c"_ZNSs5clearEv\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram1358 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1356, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1356, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1357, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 668, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1185 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1359 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype1032 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1360 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1359 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1361 = internal constant [6 x i8] c"empty\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str1362 = internal constant [15 x i8] c"_ZNKSs5emptyEv\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram1363 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1361, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1361, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1362, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 675, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1360 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1364 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype845 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1365 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1364 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1366 = internal constant [11 x i8] c"_ZNKSsixEj\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.subprogram1367 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1366, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 690, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1365 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1368 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype839 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1369 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1368 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1370 = internal constant [10 x i8] c"_ZNSsixEj\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.subprogram1371 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str334, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str1370, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 707, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1369 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1372 = internal constant [3 x i8] c"at\00", section "llvm.metadata" ; <[3 x i8]*> [#uses=1] -@.str1373 = internal constant [12 x i8] c"_ZNKSs2atEj\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.subprogram1374 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([3 x i8]* @.str1372, i32 0, i32 0), i8* getelementptr ([3 x i8]* @.str1372, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str1373, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 728, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1365 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1375 = internal constant [11 x i8] c"_ZNSs2atEj\00", section "llvm.metadata" ; <[11 x i8]*> [#uses=1] -@llvm.dbg.subprogram1376 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([3 x i8]* @.str1372, i32 0, i32 0), i8* getelementptr ([3 x i8]* @.str1372, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str1375, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 747, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1369 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1377 = internal constant [13 x i8] c"_ZNSspLERKSs\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.subprogram1378 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1377, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 762, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1287 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1379 = internal constant [12 x i8] c"_ZNSspLEPKc\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.subprogram1380 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str1379, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 771, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1291 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1381 = internal constant [10 x i8] c"_ZNSspLEc\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@llvm.dbg.subprogram1382 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([11 x i8]* @.str92, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str1381, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 780, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1295 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1383 = internal constant [7 x i8] c"append\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@.str1384 = internal constant [18 x i8] c"_ZNSs6appendERKSs\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1385 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1384, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 330, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1287 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1386 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1387 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1386 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1388 = internal constant [20 x i8] c"_ZNSs6appendERKSsjj\00", section "llvm.metadata" ; <[20 x i8]*> [#uses=1] -@llvm.dbg.subprogram1389 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([20 x i8]* @.str1388, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 347, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1387 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1390 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1391 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1390 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1392 = internal constant [18 x i8] c"_ZNSs6appendEPKcj\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1393 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1392, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 303, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1391 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1394 = internal constant [17 x i8] c"_ZNSs6appendEPKc\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram1395 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1394, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 824, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1291 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1396 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1397 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1396 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1398 = internal constant [16 x i8] c"_ZNSs6appendEjc\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram1399 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1398, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 286, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1397 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1400 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* null, %0* null], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1401 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1400 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1402 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1383, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 851, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1401 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1403 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1404 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1403 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1405 = internal constant [10 x i8] c"push_back\00", section "llvm.metadata" ; <[10 x i8]*> [#uses=1] -@.str1406 = internal constant [18 x i8] c"_ZNSs9push_backEc\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1407 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str1405, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str1405, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1406, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 859, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1404 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1408 = internal constant [7 x i8] c"assign\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@.str1409 = internal constant [18 x i8] c"_ZNSs6assignERKSs\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1410 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1409, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 248, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1287 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1411 = internal constant [20 x i8] c"_ZNSs6assignERKSsjj\00", section "llvm.metadata" ; <[20 x i8]*> [#uses=1] -@llvm.dbg.subprogram1412 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([20 x i8]* @.str1411, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 889, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1387 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1413 = internal constant [18 x i8] c"_ZNSs6assignEPKcj\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1414 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1413, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 264, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1391 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1415 = internal constant [17 x i8] c"_ZNSs6assignEPKc\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram1416 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1415, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 917, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1291 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1417 = internal constant [16 x i8] c"_ZNSs6assignEjc\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram1418 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1417, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 933, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1397 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.composite1419 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1400 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1420 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1408, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 946, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1419 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1421 = internal constant [5 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1422 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1421 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1423 = internal constant [7 x i8] c"insert\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@.str1424 = internal constant [53 x i8] c"_ZNSs6insertEN9__gnu_cxx17__normal_iteratorIPcSsEEjc\00", section "llvm.metadata" ; <[53 x i8]*> [#uses=1] -@llvm.dbg.subprogram1425 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([53 x i8]* @.str1424, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 962, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1422 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1426 = internal constant [5 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* null, %0* null], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1427 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1426 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1428 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 978, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1427 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1429 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1430 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1429 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1431 = internal constant [19 x i8] c"_ZNSs6insertEjRKSs\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.subprogram1432 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str1431, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 993, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1430 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1433 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1434 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1433 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1435 = internal constant [21 x i8] c"_ZNSs6insertEjRKSsjj\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram1436 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str1435, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1016, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1434 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1437 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1438 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1437 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1439 = internal constant [19 x i8] c"_ZNSs6insertEjPKcj\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.subprogram1440 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str1439, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 365, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1438 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1441 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1442 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1441 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1443 = internal constant [18 x i8] c"_ZNSs6insertEjPKc\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1444 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1443, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1056, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1442 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1445 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1446 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1445 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1447 = internal constant [17 x i8] c"_ZNSs6insertEjjc\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram1448 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1447, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1079, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1446 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1449 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1450 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1449 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1451 = internal constant [52 x i8] c"_ZNSs6insertEN9__gnu_cxx17__normal_iteratorIPcSsEEc\00", section "llvm.metadata" ; <[52 x i8]*> [#uses=1] -@llvm.dbg.subprogram1452 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1423, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str1451, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1096, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1450 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1453 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1454 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1453 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1455 = internal constant [6 x i8] c"erase\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str1456 = internal constant [15 x i8] c"_ZNSs5eraseEjj\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram1457 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1455, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1455, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1456, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1120, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1454 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1458 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1459 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1458 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1460 = internal constant [50 x i8] c"_ZNSs5eraseEN9__gnu_cxx17__normal_iteratorIPcSsEE\00", section "llvm.metadata" ; <[50 x i8]*> [#uses=1] -@llvm.dbg.subprogram1461 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1455, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1455, i32 0, i32 0), i8* getelementptr ([50 x i8]* @.str1460, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1136, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1459 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1462 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1463 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1462 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1464 = internal constant [53 x i8] c"_ZNSs5eraseEN9__gnu_cxx17__normal_iteratorIPcSsEES2_\00", section "llvm.metadata" ; <[53 x i8]*> [#uses=1] -@llvm.dbg.subprogram1465 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1455, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1455, i32 0, i32 0), i8* getelementptr ([53 x i8]* @.str1464, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1156, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1463 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1466 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1467 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1466 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1468 = internal constant [8 x i8] c"replace\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str1469 = internal constant [21 x i8] c"_ZNSs7replaceEjjRKSs\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram1470 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str1469, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1183, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1467 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1471 = internal constant [7 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[7 x %0*]*> [#uses=1] -@llvm.dbg.composite1472 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([7 x %0*]* @llvm.dbg.array1471 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1473 = internal constant [23 x i8] c"_ZNSs7replaceEjjRKSsjj\00", section "llvm.metadata" ; <[23 x i8]*> [#uses=1] -@llvm.dbg.subprogram1474 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([23 x i8]* @.str1473, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1206, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1472 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1475 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1476 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1475 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1477 = internal constant [21 x i8] c"_ZNSs7replaceEjjPKcj\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram1478 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str1477, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 397, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1476 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1479 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1480 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1479 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1481 = internal constant [20 x i8] c"_ZNSs7replaceEjjPKc\00", section "llvm.metadata" ; <[20 x i8]*> [#uses=1] -@llvm.dbg.subprogram1482 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([20 x i8]* @.str1481, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1248, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1480 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1483 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1484 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1483 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1485 = internal constant [19 x i8] c"_ZNSs7replaceEjjjc\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.subprogram1486 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str1485, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1271, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1484 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1487 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1488 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1487 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1489 = internal constant [59 x i8] c"_ZNSs7replaceEN9__gnu_cxx17__normal_iteratorIPcSsEES2_RKSs\00", section "llvm.metadata" ; <[59 x i8]*> [#uses=1] -@llvm.dbg.subprogram1490 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([59 x i8]* @.str1489, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1289, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1488 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1491 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1492 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1491 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1493 = internal constant [59 x i8] c"_ZNSs7replaceEN9__gnu_cxx17__normal_iteratorIPcSsEES2_PKcj\00", section "llvm.metadata" ; <[59 x i8]*> [#uses=1] -@llvm.dbg.subprogram1494 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([59 x i8]* @.str1493, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1307, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1492 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1495 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1496 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1495 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1497 = internal constant [58 x i8] c"_ZNSs7replaceEN9__gnu_cxx17__normal_iteratorIPcSsEES2_PKc\00", section "llvm.metadata" ; <[58 x i8]*> [#uses=1] -@llvm.dbg.subprogram1498 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([58 x i8]* @.str1497, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1328, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1496 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1499 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1500 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1499 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1501 = internal constant [57 x i8] c"_ZNSs7replaceEN9__gnu_cxx17__normal_iteratorIPcSsEES2_jc\00", section "llvm.metadata" ; <[57 x i8]*> [#uses=1] -@llvm.dbg.subprogram1502 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([57 x i8]* @.str1501, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1349, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1500 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1503 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* null, %0* null], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1504 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1503 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1505 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1373, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1504 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1506 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1507 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1506 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1508 = internal constant [61 x i8] c"_ZNSs7replaceEN9__gnu_cxx17__normal_iteratorIPcSsEES2_S1_S1_\00", section "llvm.metadata" ; <[61 x i8]*> [#uses=1] -@llvm.dbg.subprogram1509 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([61 x i8]* @.str1508, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1385, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1507 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1510 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1511 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1510 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1512 = internal constant [61 x i8] c"_ZNSs7replaceEN9__gnu_cxx17__normal_iteratorIPcSsEES2_PKcS4_\00", section "llvm.metadata" ; <[61 x i8]*> [#uses=1] -@llvm.dbg.subprogram1513 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([61 x i8]* @.str1512, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1396, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1511 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1514 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1515 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1514 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1516 = internal constant [61 x i8] c"_ZNSs7replaceEN9__gnu_cxx17__normal_iteratorIPcSsEES2_S2_S2_\00", section "llvm.metadata" ; <[61 x i8]*> [#uses=1] -@llvm.dbg.subprogram1517 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([61 x i8]* @.str1516, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1406, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1515 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1518 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1231 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1231 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1519 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1518 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1520 = internal constant [70 x i8] c"_ZNSs7replaceEN9__gnu_cxx17__normal_iteratorIPcSsEES2_NS0_IPKcSsEES5_\00", section "llvm.metadata" ; <[70 x i8]*> [#uses=1] -@llvm.dbg.subprogram1521 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1468, i32 0, i32 0), i8* getelementptr ([70 x i8]* @.str1520, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1417, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1519 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1522 = internal constant [18 x i8] c"cpp_type_traits.h\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.compile_unit1523 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([18 x i8]* @.str1522, i32 0, i32 0), i8* getelementptr ([115 x i8]* @.str638, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str1524 = internal constant [12 x i8] c"__true_type\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@llvm.dbg.composite1526 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str1524, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1523 to %0*), i32 97, i64 8, i64 8, i64 0, i32 0, %0* null, %0* bitcast ([0 x %0*]* @llvm.dbg.array680 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array1527 = internal constant [7 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* null, %0* null, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1526 to %0*)], section "llvm.metadata" ; <[7 x %0*]*> [#uses=1] -@llvm.dbg.composite1528 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([7 x %0*]* @llvm.dbg.array1527 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1529 = internal constant [20 x i8] c"_M_replace_dispatch\00", section "llvm.metadata" ; <[20 x i8]*> [#uses=1] -@llvm.dbg.subprogram1530 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([20 x i8]* @.str1529, i32 0, i32 0), i8* getelementptr ([20 x i8]* @.str1529, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1430, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1528 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1531 = internal constant [13 x i8] c"__false_type\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.composite1533 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1531, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1523 to %0*), i32 98, i64 8, i64 8, i64 0, i32 0, %0* null, %0* bitcast ([0 x %0*]* @llvm.dbg.array680 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array1534 = internal constant [7 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1176 to %0*), %0* null, %0* null, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1533 to %0*)], section "llvm.metadata" ; <[7 x %0*]*> [#uses=1] -@llvm.dbg.composite1535 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([7 x %0*]* @llvm.dbg.array1534 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1536 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([20 x i8]* @.str1529, i32 0, i32 0), i8* getelementptr ([20 x i8]* @.str1529, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1436, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1535 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1537 = internal constant [15 x i8] c"_M_replace_aux\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@.str1538 = internal constant [27 x i8] c"_ZNSs14_M_replace_auxEjjjc\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram1539 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str1537, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1537, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str1538, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 651, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1484 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1540 = internal constant [16 x i8] c"_M_replace_safe\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@.str1541 = internal constant [30 x i8] c"_ZNSs15_M_replace_safeEjjPKcj\00", section "llvm.metadata" ; <[30 x i8]*> [#uses=1] -@llvm.dbg.subprogram1542 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([16 x i8]* @.str1540, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1540, i32 0, i32 0), i8* getelementptr ([30 x i8]* @.str1541, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 664, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1476 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1543 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* null, %0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1533 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1544 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1543 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1545 = internal constant [17 x i8] c"_S_construct_aux\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram1546 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([17 x i8]* @.str1545, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1545, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1451, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1544 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1547 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* null, %0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1526 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1548 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1547 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1549 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([17 x i8]* @.str1545, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1545, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1460, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1548 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1550 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* null, %0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1551 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1550 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1552 = internal constant [13 x i8] c"_S_construct\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@llvm.dbg.subprogram1553 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1552, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1552, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1466, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1551 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1554 = internal constant [26 x i8] c"stl_iterator_base_types.h\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.compile_unit1555 = internal constant %llvm.dbg.compile_unit.type { i32 458769, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.compile_units to %0*), i32 4, i8* getelementptr ([26 x i8]* @.str1554, i32 0, i32 0), i8* getelementptr ([115 x i8]* @.str638, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str2, i32 0, i32 0), i1 false, i1 false, i8* null, i32 -1 }, section "llvm.metadata" ; <%llvm.dbg.compile_unit.type*> [#uses=1] -@.str1556 = internal constant [19 x i8] c"input_iterator_tag\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.composite1558 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([19 x i8]* @.str1556, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1555 to %0*), i32 80, i64 8, i64 8, i64 0, i32 0, %0* null, %0* bitcast ([0 x %0*]* @llvm.dbg.array680 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array1559 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* null, %0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1558 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1560 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1559 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1561 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1552, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1552, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1476, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1560 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1562 = internal constant [21 x i8] c"forward_iterator_tag\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1564 = internal constant %llvm.dbg.derivedtype.type { i32 458780, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1555 to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1558 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1565 = internal constant [1 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1564 to %0*)], section "llvm.metadata" ; <[1 x %0*]*> [#uses=1] -@llvm.dbg.composite1566 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([21 x i8]* @.str1562, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1555 to %0*), i32 84, i64 8, i64 8, i64 0, i32 0, %0* null, %0* bitcast ([1 x %0*]* @llvm.dbg.array1565 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.array1567 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* null, %0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*), %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1566 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1568 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1567 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1569 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1552, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1552, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1483, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1568 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1570 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype887 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1571 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1570 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1572 = internal constant [30 x i8] c"_ZNSs12_S_constructEjcRKSaIcE\00", section "llvm.metadata" ; <[30 x i8]*> [#uses=1] -@llvm.dbg.subprogram1573 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1552, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1552, i32 0, i32 0), i8* getelementptr ([30 x i8]* @.str1572, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 166, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1571 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1574 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1575 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1574 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1576 = internal constant [5 x i8] c"copy\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str1577 = internal constant [17 x i8] c"_ZNKSs4copyEPcjj\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram1578 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1576, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1576, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1577, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 705, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1575 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1579 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype923 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1285 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1580 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1579 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1581 = internal constant [5 x i8] c"swap\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str1582 = internal constant [15 x i8] c"_ZNSs4swapERSs\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram1583 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1581, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1581, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1582, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 501, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1580 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1584 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1585 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1584 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1586 = internal constant [6 x i8] c"c_str\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str1587 = internal constant [15 x i8] c"_ZNKSs5c_strEv\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram1588 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1586, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1586, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1587, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1522, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1585 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1589 = internal constant [5 x i8] c"data\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str1590 = internal constant [14 x i8] c"_ZNKSs4dataEv\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@llvm.dbg.subprogram1591 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1589, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1589, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1590, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1532, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1585 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1592 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite902 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1593 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1592 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1594 = internal constant [14 x i8] c"get_allocator\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@.str1595 = internal constant [24 x i8] c"_ZNKSs13get_allocatorEv\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram1596 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1594, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1594, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str1595, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1539, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1593 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1597 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1598 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1597 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1599 = internal constant [5 x i8] c"find\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@.str1600 = internal constant [18 x i8] c"_ZNKSs4findEPKcjj\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1601 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1599, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1599, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1600, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 719, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1598 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1602 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1603 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1602 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1604 = internal constant [18 x i8] c"_ZNKSs4findERKSsj\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1605 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1599, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1599, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1604, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1567, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1603 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1606 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1607 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1606 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1608 = internal constant [17 x i8] c"_ZNKSs4findEPKcj\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram1609 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1599, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1599, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1608, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1581, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1607 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1610 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype282 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype672 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1611 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1610 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1612 = internal constant [15 x i8] c"_ZNKSs4findEcj\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram1613 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1599, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1599, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1612, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 742, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1611 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.composite1614 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1602 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1615 = internal constant [6 x i8] c"rfind\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@.str1616 = internal constant [19 x i8] c"_ZNKSs5rfindERKSsj\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.subprogram1617 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1615, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1615, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str1616, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1611, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1614 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1618 = internal constant [19 x i8] c"_ZNKSs5rfindEPKcjj\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.subprogram1619 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1615, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1615, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str1618, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 760, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1598 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.composite1620 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1606 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1621 = internal constant [18 x i8] c"_ZNKSs5rfindEPKcj\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@llvm.dbg.subprogram1622 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1615, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1615, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1621, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1639, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1620 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.composite1623 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1610 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1624 = internal constant [16 x i8] c"_ZNKSs5rfindEcj\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@llvm.dbg.subprogram1625 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([6 x i8]* @.str1615, i32 0, i32 0), i8* getelementptr ([6 x i8]* @.str1615, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1624, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 781, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1623 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1626 = internal constant [14 x i8] c"find_first_of\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@.str1627 = internal constant [28 x i8] c"_ZNKSs13find_first_ofERKSsj\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram1628 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1626, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1626, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str1627, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1669, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1603 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1629 = internal constant [28 x i8] c"_ZNKSs13find_first_ofEPKcjj\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram1630 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1626, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1626, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str1629, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 798, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1598 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1631 = internal constant [27 x i8] c"_ZNKSs13find_first_ofEPKcj\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram1632 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1626, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1626, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str1631, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1697, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1607 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1633 = internal constant [25 x i8] c"_ZNKSs13find_first_ofEcj\00", section "llvm.metadata" ; <[25 x i8]*> [#uses=1] -@llvm.dbg.subprogram1634 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1626, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1626, i32 0, i32 0), i8* getelementptr ([25 x i8]* @.str1633, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1716, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1611 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1635 = internal constant [13 x i8] c"find_last_of\00", section "llvm.metadata" ; <[13 x i8]*> [#uses=1] -@.str1636 = internal constant [27 x i8] c"_ZNKSs12find_last_ofERKSsj\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram1637 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1635, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1635, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str1636, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1730, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1614 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1638 = internal constant [27 x i8] c"_ZNKSs12find_last_ofEPKcjj\00", section "llvm.metadata" ; <[27 x i8]*> [#uses=1] -@llvm.dbg.subprogram1639 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1635, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1635, i32 0, i32 0), i8* getelementptr ([27 x i8]* @.str1638, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 813, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1598 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1640 = internal constant [26 x i8] c"_ZNKSs12find_last_ofEPKcj\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.subprogram1641 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1635, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1635, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str1640, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1758, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1620 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1642 = internal constant [24 x i8] c"_ZNKSs12find_last_ofEcj\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram1643 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([13 x i8]* @.str1635, i32 0, i32 0), i8* getelementptr ([13 x i8]* @.str1635, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str1642, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1777, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1623 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1644 = internal constant [18 x i8] c"find_first_not_of\00", section "llvm.metadata" ; <[18 x i8]*> [#uses=1] -@.str1645 = internal constant [32 x i8] c"_ZNKSs17find_first_not_ofERKSsj\00", section "llvm.metadata" ; <[32 x i8]*> [#uses=1] -@llvm.dbg.subprogram1646 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str1644, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1644, i32 0, i32 0), i8* getelementptr ([32 x i8]* @.str1645, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1791, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1603 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1647 = internal constant [32 x i8] c"_ZNKSs17find_first_not_ofEPKcjj\00", section "llvm.metadata" ; <[32 x i8]*> [#uses=1] -@llvm.dbg.subprogram1648 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str1644, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1644, i32 0, i32 0), i8* getelementptr ([32 x i8]* @.str1647, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 834, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1598 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1649 = internal constant [31 x i8] c"_ZNKSs17find_first_not_ofEPKcj\00", section "llvm.metadata" ; <[31 x i8]*> [#uses=1] -@llvm.dbg.subprogram1650 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str1644, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1644, i32 0, i32 0), i8* getelementptr ([31 x i8]* @.str1649, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1820, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1607 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1651 = internal constant [29 x i8] c"_ZNKSs17find_first_not_ofEcj\00", section "llvm.metadata" ; <[29 x i8]*> [#uses=1] -@llvm.dbg.subprogram1652 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([18 x i8]* @.str1644, i32 0, i32 0), i8* getelementptr ([18 x i8]* @.str1644, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str1651, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 846, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1611 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1653 = internal constant [17 x i8] c"find_last_not_of\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@.str1654 = internal constant [31 x i8] c"_ZNKSs16find_last_not_ofERKSsj\00", section "llvm.metadata" ; <[31 x i8]*> [#uses=1] -@llvm.dbg.subprogram1655 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([17 x i8]* @.str1653, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1653, i32 0, i32 0), i8* getelementptr ([31 x i8]* @.str1654, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1850, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1614 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1656 = internal constant [31 x i8] c"_ZNKSs16find_last_not_ofEPKcjj\00", section "llvm.metadata" ; <[31 x i8]*> [#uses=1] -@llvm.dbg.subprogram1657 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([17 x i8]* @.str1653, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1653, i32 0, i32 0), i8* getelementptr ([31 x i8]* @.str1656, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 857, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1598 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1658 = internal constant [30 x i8] c"_ZNKSs16find_last_not_ofEPKcj\00", section "llvm.metadata" ; <[30 x i8]*> [#uses=1] -@llvm.dbg.subprogram1659 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([17 x i8]* @.str1653, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1653, i32 0, i32 0), i8* getelementptr ([30 x i8]* @.str1658, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1879, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1620 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1660 = internal constant [28 x i8] c"_ZNKSs16find_last_not_ofEcj\00", section "llvm.metadata" ; <[28 x i8]*> [#uses=1] -@llvm.dbg.subprogram1661 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([17 x i8]* @.str1653, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1653, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str1660, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 878, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1623 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1662 = internal constant [4 x %0*] [%0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1693 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[4 x %0*]*> [#uses=1] -@llvm.dbg.composite1663 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([4 x %0*]* @llvm.dbg.array1662 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1664 = internal constant [7 x i8] c"substr\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@.str1665 = internal constant [17 x i8] c"_ZNKSs6substrEjj\00", section "llvm.metadata" ; <[17 x i8]*> [#uses=1] -@llvm.dbg.subprogram1666 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1664, i32 0, i32 0), i8* getelementptr ([7 x i8]* @.str1664, i32 0, i32 0), i8* getelementptr ([17 x i8]* @.str1665, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1911, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1663 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1667 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1668 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1667 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1669 = internal constant [8 x i8] c"compare\00", section "llvm.metadata" ; <[8 x i8]*> [#uses=1] -@.str1670 = internal constant [20 x i8] c"_ZNKSs7compareERKSs\00", section "llvm.metadata" ; <[20 x i8]*> [#uses=1] -@llvm.dbg.subprogram1671 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([20 x i8]* @.str1670, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit807 to %0*), i32 1929, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1668 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1672 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1673 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1672 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1674 = internal constant [22 x i8] c"_ZNKSs7compareEjjRKSs\00", section "llvm.metadata" ; <[22 x i8]*> [#uses=1] -@llvm.dbg.subprogram1675 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([22 x i8]* @.str1674, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 898, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1673 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1676 = internal constant [7 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1259 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[7 x %0*]*> [#uses=1] -@llvm.dbg.composite1677 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([7 x %0*]* @llvm.dbg.array1676 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1678 = internal constant [24 x i8] c"_ZNKSs7compareEjjRKSsjj\00", section "llvm.metadata" ; <[24 x i8]*> [#uses=1] -@llvm.dbg.subprogram1679 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([24 x i8]* @.str1678, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 914, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1677 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1680 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1681 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1680 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1682 = internal constant [19 x i8] c"_ZNKSs7compareEPKc\00", section "llvm.metadata" ; <[19 x i8]*> [#uses=1] -@llvm.dbg.subprogram1683 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([19 x i8]* @.str1682, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 931, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1681 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1684 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1685 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1684 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1686 = internal constant [21 x i8] c"_ZNKSs7compareEjjPKc\00", section "llvm.metadata" ; <[21 x i8]*> [#uses=1] -@llvm.dbg.subprogram1687 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([21 x i8]* @.str1686, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 946, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1685 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1688 = internal constant [6 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype917 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype280 to %0*)], section "llvm.metadata" ; <[6 x %0*]*> [#uses=1] -@llvm.dbg.composite1689 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([6 x %0*]* @llvm.dbg.array1688 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1690 = internal constant [22 x i8] c"_ZNKSs7compareEjjPKcj\00", section "llvm.metadata" ; <[22 x i8]*> [#uses=1] -@llvm.dbg.subprogram1691 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([8 x i8]* @.str1669, i32 0, i32 0), i8* getelementptr ([22 x i8]* @.str1690, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit1070 to %0*), i32 963, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1689 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1692 = internal constant [143 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype913 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram922 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram927 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1096 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1180 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1183 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1188 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1193 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1198 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1203 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1208 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1213 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1216 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1221 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1225 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1229 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1235 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1239 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1243 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1248 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1251 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1253 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1255 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1258 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1262 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1265 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1268 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1271 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1274 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1277 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1280 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1284 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1289 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1293 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1297 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1302 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1306 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1309 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1311 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1318 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1324 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1327 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1329 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1334 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1337 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1339 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1344 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1348 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1351 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1355 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1358 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1363 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1367 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1371 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1374 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1376 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1378 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1380 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1382 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1385 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1389 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1393 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1395 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1399 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1402 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1407 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1410 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1412 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1414 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1416 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1418 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1420 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1425 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1428 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1432 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1436 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1440 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1444 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1448 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1452 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1457 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1461 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1465 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1470 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1474 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1478 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1482 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1486 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1490 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1494 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1498 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1502 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1505 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1509 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1513 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1517 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1521 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1530 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1536 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1539 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1542 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1546 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1549 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1553 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1561 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1569 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1573 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1578 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1583 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1588 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1591 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1596 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1601 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1605 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1609 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1613 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1617 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1619 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1622 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1625 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1628 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1630 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1632 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1634 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1637 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1639 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1641 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1643 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1646 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1648 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1650 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1652 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1655 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1657 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1659 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1661 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1666 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1671 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1675 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1679 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1683 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1687 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1691 to %0*)], section "llvm.metadata" ; <[143 x %0*]*> [#uses=1] -@llvm.dbg.composite1693 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([64 x i8]* @.str804, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit803 to %0*), i32 56, i64 32, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([143 x %0*]* @llvm.dbg.array1692 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype1694 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str914, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit803 to %0*), i32 56, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1693 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@.str1695 = internal constant [7 x i8] c"_M_msg\00", section "llvm.metadata" ; <[7 x i8]*> [#uses=1] -@llvm.dbg.derivedtype1696 = internal constant %llvm.dbg.derivedtype.type { i32 458765, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str1695, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 109, i64 32, i64 32, i64 32, i32 1, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1694 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1697 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1714 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1698 = internal constant %llvm.dbg.derivedtype.type { i32 458774, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([7 x i8]* @.str914, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit803 to %0*), i32 56, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype916 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1699 = internal constant %llvm.dbg.derivedtype.type { i32 458768, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1698 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1700 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1697 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1699 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1701 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1700 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1702 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str775, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str775, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 114, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1701 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1703 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1697 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1704 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1703 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1705 = internal constant [15 x i8] c"~runtime_error\00", section "llvm.metadata" ; <[15 x i8]*> [#uses=1] -@llvm.dbg.subprogram1706 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str1705, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str1705, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 117, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1704 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.derivedtype1707 = internal constant %llvm.dbg.derivedtype.type { i32 458790, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 64, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1714 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1708 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1707 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1709 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype791 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1708 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1710 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1709 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1711 = internal constant [29 x i8] c"_ZNKSt13runtime_error4whatEv\00", section "llvm.metadata" ; <[29 x i8]*> [#uses=1] -@llvm.dbg.subprogram1712 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str796, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str796, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str1711, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 122, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1710 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1713 = internal constant [5 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype801 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1696 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1702 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1706 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1712 to %0*)], section "llvm.metadata" ; <[5 x %0*]*> [#uses=1] -@llvm.dbg.composite1714 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str775, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 108, i64 64, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([5 x %0*]* @llvm.dbg.array1713 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype1715 = internal constant %llvm.dbg.derivedtype.type { i32 458780, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1714 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.derivedtype1716 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1721 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1717 = internal constant [3 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1716 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1699 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1718 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1717 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.subprogram1719 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str773, i32 0, i32 0), i8* getelementptr ([15 x i8]* @.str773, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 136, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1718 to %0*), i1 false, i1 false }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1720 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1715 to %0*), %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1719 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1721 = internal constant %llvm.dbg.composite.type { i32 458771, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([15 x i8]* @.str773, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 134, i64 64, i64 32, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1720 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@llvm.dbg.derivedtype1722 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1721 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1723 = internal constant [2 x %0*] [%0* null, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1722 to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1724 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1723 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1725 = internal constant [16 x i8] c"~overflow_error\00", section "llvm.metadata" ; <[16 x i8]*> [#uses=1] -@.str1726 = internal constant [26 x i8] c"_ZNSt14overflow_errorD1Ev\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.subprogram1727 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([16 x i8]* @.str1725, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1725, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str1726, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*), i32 134, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1724 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@_ZTVSt14overflow_error = weak constant [5 x i32 (...)*] [i32 (...)* null, i32 (...)* bitcast (%struct.__si_class_type_info_pseudo* @_ZTISt14overflow_error to i32 (...)*), i32 (...)* bitcast (void (%"struct.std::overflow_error"*)* @_ZNSt14overflow_errorD1Ev to i32 (...)*), i32 (...)* bitcast (void (%"struct.std::overflow_error"*)* @_ZNSt14overflow_errorD0Ev to i32 (...)*), i32 (...)* bitcast (i8* (%"struct.std::runtime_error"*)* @_ZNKSt13runtime_error4whatEv to i32 (...)*)], align 8 ; <[5 x i32 (...)*]*> [#uses=1] -@.str1735 = internal constant [26 x i8] c"_ZNSt14overflow_errorD0Ev\00", section "llvm.metadata" ; <[26 x i8]*> [#uses=1] -@llvm.dbg.subprogram1736 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([16 x i8]* @.str1725, i32 0, i32 0), i8* getelementptr ([16 x i8]* @.str1725, i32 0, i32 0), i8* getelementptr ([26 x i8]* @.str1735, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 702, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1724 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.array1738 = internal constant [2 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype to %0*)], section "llvm.metadata" ; <[2 x %0*]*> [#uses=1] -@llvm.dbg.composite1739 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([2 x %0*]* @llvm.dbg.array1738 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1740 = internal constant [14 x i8] c"__complex_exp\00", section "llvm.metadata" ; <[14 x i8]*> [#uses=1] -@.str1741 = internal constant [22 x i8] c"_ZSt13__complex_expCd\00", section "llvm.metadata" ; <[22 x i8]*> [#uses=1] -@llvm.dbg.subprogram1742 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([14 x i8]* @.str1740, i32 0, i32 0), i8* getelementptr ([14 x i8]* @.str1740, i32 0, i32 0), i8* getelementptr ([22 x i8]* @.str1741, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 730, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1739 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1744 = internal constant [12 x i8] c"exp<double>\00", section "llvm.metadata" ; <[12 x i8]*> [#uses=1] -@.str1745 = internal constant [31 x i8] c"_ZSt3expIdESt7complexIT_ERKS2_\00", section "llvm.metadata" ; <[31 x i8]*> [#uses=1] -@llvm.dbg.subprogram1746 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str1744, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str1744, i32 0, i32 0), i8* getelementptr ([31 x i8]* @.str1745, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*), i32 738, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite573 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.subprogram1748 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([4 x i8]* @.str457, i32 0, i32 0), i8* getelementptr ([4 x i8]* @.str457, i32 0, i32 0), i8* getelementptr ([29 x i8]* @.str495, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 497, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite488 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable1751 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1748 to %0*), i8* getelementptr ([7 x i8]* @.str259, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 503, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1752 = internal constant [2 x i8] c"u\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.variable1753 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1748 to %0*), i8* getelementptr ([2 x i8]* @.str1752, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 501, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1754 = internal constant [2 x i8] c"t\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.variable1755 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1748 to %0*), i8* getelementptr ([2 x i8]* @.str1754, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 501, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1756 = internal constant [2 x i8] c"w\00", section "llvm.metadata" ; <[2 x i8]*> [#uses=1] -@llvm.dbg.variable1757 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1748 to %0*), i8* getelementptr ([2 x i8]* @.str1756, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 501, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1758 = internal constant [3 x i8] c"wm\00", section "llvm.metadata" ; <[3 x i8]*> [#uses=1] -@llvm.dbg.variable1759 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1748 to %0*), i8* getelementptr ([3 x i8]* @.str1758, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 501, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.subprogram1771 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([12 x i8]* @.str460, i32 0, i32 0), i8* getelementptr ([12 x i8]* @.str460, i32 0, i32 0), i8* getelementptr ([52 x i8]* @.str497, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 535, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite492 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable1774 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1771 to %0*), i8* getelementptr ([7 x i8]* @.str259, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 541, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.variable1775 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1771 to %0*), i8* getelementptr ([2 x i8]* @.str1752, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 539, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.variable1776 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1771 to %0*), i8* getelementptr ([2 x i8]* @.str1754, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 539, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.variable1777 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1771 to %0*), i8* getelementptr ([2 x i8]* @.str1756, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 539, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.variable1778 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1771 to %0*), i8* getelementptr ([3 x i8]* @.str1758, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 539, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite223 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.subprogram1785 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([10 x i8]* @.str463, i32 0, i32 0), i8* getelementptr ([10 x i8]* @.str463, i32 0, i32 0), i8* getelementptr ([28 x i8]* @.str499, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 576, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite351 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@llvm.dbg.variable1791 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1785 to %0*), i8* getelementptr ([7 x i8]* @.str259, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 614, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1794 = internal constant [5 x i8] c"dft2\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.variable1795 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1785 to %0*), i8* getelementptr ([5 x i8]* @.str1794, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 601, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1796 = internal constant [5 x i8] c"dft1\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.variable1797 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1785 to %0*), i8* getelementptr ([5 x i8]* @.str1796, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 600, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite486 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1798 = internal constant [3 x i8] c"a2\00", section "llvm.metadata" ; <[3 x i8]*> [#uses=1] -@llvm.dbg.variable1799 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1785 to %0*), i8* getelementptr ([3 x i8]* @.str1798, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 591, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1800 = internal constant [3 x i8] c"a1\00", section "llvm.metadata" ; <[3 x i8]*> [#uses=1] -@llvm.dbg.variable1801 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1785 to %0*), i8* getelementptr ([3 x i8]* @.str1800, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 590, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@llvm.dbg.derivedtype1802 = internal constant %llvm.dbg.derivedtype.type { i32 458767, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 32, i64 32, i64 0, i32 0, %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype837 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.derivedtype.type*> [#uses=1] -@llvm.dbg.array1803 = internal constant [3 x %0*] [%0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*), %0* bitcast (%llvm.dbg.basictype.type* @llvm.dbg.basictype269 to %0*), %0* bitcast (%llvm.dbg.derivedtype.type* @llvm.dbg.derivedtype1802 to %0*)], section "llvm.metadata" ; <[3 x %0*]*> [#uses=1] -@llvm.dbg.composite1804 = internal constant %llvm.dbg.composite.type { i32 458773, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 0, i64 0, i64 0, i64 0, i32 0, %0* null, %0* bitcast ([3 x %0*]* @llvm.dbg.array1803 to %0*), i32 0 }, section "llvm.metadata" ; <%llvm.dbg.composite.type*> [#uses=1] -@.str1805 = internal constant [5 x i8] c"main\00", section "llvm.metadata" ; <[5 x i8]*> [#uses=1] -@llvm.dbg.subprogram1806 = internal constant %llvm.dbg.subprogram.type { i32 458798, %0* bitcast (%llvm.dbg.anchor.type* @llvm.dbg.subprograms to %0*), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i8* getelementptr ([5 x i8]* @.str1805, i32 0, i32 0), i8* getelementptr ([5 x i8]* @.str1805, i32 0, i32 0), i8* null, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 654, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite1804 to %0*), i1 false, i1 true }, section "llvm.metadata" ; <%llvm.dbg.subprogram.type*> [#uses=1] -@.str1816 = internal constant [6 x i8] c"poly3\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.variable1817 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1806 to %0*), i8* getelementptr ([6 x i8]* @.str1816, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 674, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1818 = internal constant [6 x i8] c"poly2\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.variable1819 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1806 to %0*), i8* getelementptr ([6 x i8]* @.str1818, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 673, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1820 = internal constant [6 x i8] c"poly1\00", section "llvm.metadata" ; <[6 x i8]*> [#uses=1] -@llvm.dbg.variable1821 = internal constant %llvm.dbg.variable.type { i32 459008, %0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1806 to %0*), i8* getelementptr ([6 x i8]* @.str1820, i32 0, i32 0), %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*), i32 672, %0* bitcast (%llvm.dbg.composite.type* @llvm.dbg.composite518 to %0*) }, section "llvm.metadata" ; <%llvm.dbg.variable.type*> [#uses=1] -@.str1824 = internal constant [4 x i8] c"-ga\00", align 4 ; <[4 x i8]*> [#uses=0] -@_ZSt4cout = external global %"struct.std::basic_ostream<char,std::char_traits<char> >" ; <%"struct.std::basic_ostream<char,std::char_traits<char> >"*> [#uses=3] -@.str1825 = internal constant [32 x i8] c"\0Afftbench (Std. C++) run time: \00" ; <[32 x i8]*> [#uses=1] -@.str1826 = internal constant [3 x i8] c"\0A\0A\00" ; <[3 x i8]*> [#uses=1] -@llvm.global_ctors = appending global [1 x %2] [%2 { i32 65535, void ()* @_GLOBAL__I_main }] ; <[1 x %2]*> [#uses=0] - -@_ZL20__gthrw_pthread_oncePiPFvvE = alias weak i32 (i32*, void ()*)* @pthread_once ; <i32 (i32*, void ()*)*> [#uses=0] -@_ZL27__gthrw_pthread_getspecificj = alias weak i8* (i32)* @pthread_getspecific ; <i8* (i32)*> [#uses=0] -@_ZL27__gthrw_pthread_setspecificjPKv = alias weak i32 (i32, i8*)* @pthread_setspecific ; <i32 (i32, i8*)*> [#uses=0] -@_ZL22__gthrw_pthread_createPmPK14pthread_attr_tPFPvS3_ES3_ = alias weak i32 (i32*, %struct.pthread_attr_t*, i8* (i8*)*, i8*)* @pthread_create ; <i32 (i32*, %struct.pthread_attr_t*, i8* (i8*)*, i8*)*> [#uses=0] -@_ZL22__gthrw_pthread_cancelm = alias weak i32 (i32)* @pthread_cancel ; <i32 (i32)*> [#uses=0] -@_ZL26__gthrw_pthread_mutex_lockP15pthread_mutex_t = alias weak i32 (%struct.pthread_mutex_t*)* @pthread_mutex_lock ; <i32 (%struct.pthread_mutex_t*)*> [#uses=0] -@_ZL29__gthrw_pthread_mutex_trylockP15pthread_mutex_t = alias weak i32 (%struct.pthread_mutex_t*)* @pthread_mutex_trylock ; <i32 (%struct.pthread_mutex_t*)*> [#uses=0] -@_ZL28__gthrw_pthread_mutex_unlockP15pthread_mutex_t = alias weak i32 (%struct.pthread_mutex_t*)* @pthread_mutex_unlock ; <i32 (%struct.pthread_mutex_t*)*> [#uses=0] -@_ZL26__gthrw_pthread_mutex_initP15pthread_mutex_tPK19pthread_mutexattr_t = alias weak i32 (%struct.pthread_mutex_t*, %struct..0._50*)* @pthread_mutex_init ; <i32 (%struct.pthread_mutex_t*, %struct..0._50*)*> [#uses=0] -@_ZL26__gthrw_pthread_key_createPjPFvPvE = alias weak i32 (i32*, void (i8*)*)* @pthread_key_create ; <i32 (i32*, void (i8*)*)*> [#uses=0] -@_ZL26__gthrw_pthread_key_deletej = alias weak i32 (i32)* @pthread_key_delete ; <i32 (i32)*> [#uses=0] -@_ZL30__gthrw_pthread_mutexattr_initP19pthread_mutexattr_t = alias weak i32 (%struct..0._50*)* @pthread_mutexattr_init ; <i32 (%struct..0._50*)*> [#uses=0] -@_ZL33__gthrw_pthread_mutexattr_settypeP19pthread_mutexattr_ti = alias weak i32 (%struct..0._50*, i32)* @pthread_mutexattr_settype ; <i32 (%struct..0._50*, i32)*> [#uses=0] -@_ZL33__gthrw_pthread_mutexattr_destroyP19pthread_mutexattr_t = alias weak i32 (%struct..0._50*)* @pthread_mutexattr_destroy ; <i32 (%struct..0._50*)*> [#uses=0] - -define i32 @main(i32 %argc, i8** nocapture %argv) { -entry: - %n.0.reg2mem = alloca i32 ; <i32*> [#uses=5] - %poly3 = alloca %"struct.polynomial<double>", align 8 ; <%"struct.polynomial<double>"*> [#uses=5] - %poly2 = alloca %"struct.polynomial<double>", align 8 ; <%"struct.polynomial<double>"*> [#uses=6] - %poly1 = alloca %"struct.polynomial<double>", align 8 ; <%"struct.polynomial<double>"*> [#uses=6] - %0 = alloca %"struct.polynomial<double>", align 8 ; <%"struct.polynomial<double>"*> [#uses=4] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1806 to %0*)) - %1 = bitcast %"struct.polynomial<double>"* %poly3 to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %1, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1817 to %0*)) - %2 = bitcast %"struct.polynomial<double>"* %poly2 to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %2, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1819 to %0*)) - %3 = bitcast %"struct.polynomial<double>"* %poly1 to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %3, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1821 to %0*)) - call void @llvm.dbg.stoppoint(i32 659, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %4 = icmp sgt i32 %argc, 1 ; <i1> [#uses=1] - br i1 %4, label %bb4, label %bb5 - -bb1: ; preds = %bb4 - call void @llvm.dbg.stoppoint(i32 663, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %5 = getelementptr i8** %argv, i32 1 ; <i8**> [#uses=1] - %6 = load i8** %5, align 1 ; <i8*> [#uses=1] - %tmp = bitcast i8* %6 to i32* ; <i32*> [#uses=1] - %lhsv = load i32* %tmp, align 1 ; <i32> [#uses=1] - %7 = icmp eq i32 %lhsv, 6383405 ; <i1> [#uses=1] - br i1 %7, label %bb5, label %bb3 - -bb3: ; preds = %bb1 - call void @llvm.dbg.stoppoint(i32 661, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %8 = add i32 %i.0, 1 ; <i32> [#uses=1] - br label %bb4 - -bb4: ; preds = %bb3, %entry - %i.0 = phi i32 [ %8, %bb3 ], [ 1, %entry ] ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 661, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %9 = icmp slt i32 %i.0, %argc ; <i1> [#uses=1] - br i1 %9, label %bb1, label %bb5 - -bb5: ; preds = %bb4, %bb1, %entry - %ga_testing.0 = phi i8 [ 0, %entry ], [ 0, %bb4 ], [ 1, %bb1 ] ; <i8> [#uses=1] - call void @llvm.dbg.stoppoint(i32 672, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdEC1Ej(%"struct.polynomial<double>"* %poly1, i32 524288) - call void @llvm.dbg.stoppoint(i32 673, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdEC1Ej(%"struct.polynomial<double>"* %poly2, i32 524288) - to label %invcont unwind label %lpad - -invcont: ; preds = %bb5 - call void @llvm.dbg.stoppoint(i32 674, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdEC1Ej(%"struct.polynomial<double>"* %poly3, i32 1048575) - to label %bb8.thread unwind label %lpad47 - -bb8.thread: ; preds = %invcont - store i32 0, i32* %n.0.reg2mem - call void @llvm.dbg.stoppoint(i32 676, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb7 - -bb7: ; preds = %bb8, %bb8.thread - call void @llvm.dbg.stoppoint(i32 678, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %n.0.reload3 = load i32* %n.0.reg2mem ; <i32> [#uses=1] - %10 = call double* @_ZN10polynomialIdEixEj(%"struct.polynomial<double>"* %poly1, i32 %n.0.reload3) nounwind ; <double*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 68, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %11 = load i32* @_ZZL13random_doublevE4seed, align 4 ; <i32> [#uses=1] - %12 = xor i32 %11, 123459876 ; <i32> [#uses=3] - store i32 %12, i32* @_ZZL13random_doublevE4seed, align 4 - call void @llvm.dbg.stoppoint(i32 69, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %13 = sdiv i32 %12, 127773 ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 70, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %14 = mul i32 %13, 127773 ; <i32> [#uses=1] - %15 = sub i32 %12, %14 ; <i32> [#uses=1] - %16 = mul i32 %15, 16807 ; <i32> [#uses=1] - %17 = mul i32 %13, 2836 ; <i32> [#uses=1] - %18 = sub i32 %16, %17 ; <i32> [#uses=2] - store i32 %18, i32* @_ZZL13random_doublevE4seed, align 4 - call void @llvm.dbg.stoppoint(i32 72, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %19 = icmp slt i32 %18, 0 ; <i1> [#uses=1] - br i1 %19, label %bb.i, label %_ZL13random_doublev.exit - -bb.i: ; preds = %bb7 - call void @llvm.dbg.stoppoint(i32 73, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %20 = load i32* @_ZZL13random_doublevE4seed, align 4 ; <i32> [#uses=1] - %21 = add i32 %20, 2147483647 ; <i32> [#uses=1] - store i32 %21, i32* @_ZZL13random_doublevE4seed, align 4 - br label %_ZL13random_doublev.exit - -_ZL13random_doublev.exit: ; preds = %bb.i, %bb7 - call void @llvm.dbg.stoppoint(i32 75, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %22 = load i32* @_ZZL13random_doublevE4seed, align 4 ; <i32> [#uses=2] - %23 = sitofp i32 %22 to double ; <double> [#uses=1] - %24 = fmul double %23, 0x3E340000002813D9 ; <double> [#uses=1] - call void @llvm.dbg.stoppoint(i32 76, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %25 = xor i32 %22, 123459876 ; <i32> [#uses=1] - store i32 %25, i32* @_ZZL13random_doublevE4seed, align 4 - call void @llvm.dbg.stoppoint(i32 78, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - store double %24, double* %10, align 8 - call void @llvm.dbg.stoppoint(i32 679, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %n.0.reload4 = load i32* %n.0.reg2mem ; <i32> [#uses=1] - %26 = call double* @_ZN10polynomialIdEixEj(%"struct.polynomial<double>"* %poly2, i32 %n.0.reload4) nounwind ; <double*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 68, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %27 = load i32* @_ZZL13random_doublevE4seed, align 4 ; <i32> [#uses=1] - %28 = xor i32 %27, 123459876 ; <i32> [#uses=3] - store i32 %28, i32* @_ZZL13random_doublevE4seed, align 4 - call void @llvm.dbg.stoppoint(i32 69, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %29 = sdiv i32 %28, 127773 ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 70, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %30 = mul i32 %29, 127773 ; <i32> [#uses=1] - %31 = sub i32 %28, %30 ; <i32> [#uses=1] - %32 = mul i32 %31, 16807 ; <i32> [#uses=1] - %33 = mul i32 %29, 2836 ; <i32> [#uses=1] - %34 = sub i32 %32, %33 ; <i32> [#uses=2] - store i32 %34, i32* @_ZZL13random_doublevE4seed, align 4 - call void @llvm.dbg.stoppoint(i32 72, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %35 = icmp slt i32 %34, 0 ; <i1> [#uses=1] - br i1 %35, label %bb.i1, label %bb8 - -bb.i1: ; preds = %_ZL13random_doublev.exit - call void @llvm.dbg.stoppoint(i32 73, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %36 = load i32* @_ZZL13random_doublevE4seed, align 4 ; <i32> [#uses=1] - %37 = add i32 %36, 2147483647 ; <i32> [#uses=1] - store i32 %37, i32* @_ZZL13random_doublevE4seed, align 4 - br label %bb8 - -bb8: ; preds = %bb.i1, %_ZL13random_doublev.exit - call void @llvm.dbg.stoppoint(i32 75, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %38 = load i32* @_ZZL13random_doublevE4seed, align 4 ; <i32> [#uses=2] - %39 = sitofp i32 %38 to double ; <double> [#uses=1] - %40 = fmul double %39, 0x3E340000002813D9 ; <double> [#uses=1] - call void @llvm.dbg.stoppoint(i32 76, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %41 = xor i32 %38, 123459876 ; <i32> [#uses=1] - store i32 %41, i32* @_ZZL13random_doublevE4seed, align 4 - call void @llvm.dbg.stoppoint(i32 78, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - store double %40, double* %26, align 8 - call void @llvm.dbg.stoppoint(i32 676, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %n.0.reload5 = load i32* %n.0.reg2mem ; <i32> [#uses=1] - %42 = add i32 %n.0.reload5, 1 ; <i32> [#uses=2] - store i32 %42, i32* %n.0.reg2mem - call void @llvm.dbg.stoppoint(i32 676, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %43 = icmp sgt i32 %42, 524287 ; <i1> [#uses=1] - br i1 %43, label %bb9, label %bb7 - -bb9: ; preds = %bb8 - call void @llvm.dbg.stoppoint(i32 687, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZNK10polynomialIdEmlERKS0_(%"struct.polynomial<double>"* noalias sret %0, %"struct.polynomial<double>"* %poly1, %"struct.polynomial<double>"* %poly2) - to label %invcont10 unwind label %lpad51 - -invcont10: ; preds = %bb9 - call void @llvm.dbg.stoppoint(i32 687, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %44 = invoke %"struct.polynomial<double>"* @_ZN10polynomialIdEaSERKS0_(%"struct.polynomial<double>"* %poly3, %"struct.polynomial<double>"* %0) - to label %invcont11 unwind label %lpad55 ; <%"struct.polynomial<double>"*> [#uses=0] - -invcont11: ; preds = %invcont10 - call void @llvm.dbg.stoppoint(i32 687, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %0) - to label %bb16 unwind label %lpad51 - -bb16: ; preds = %invcont11 - call void @llvm.dbg.stoppoint(i32 695, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %toBool = icmp eq i8 %ga_testing.0, 0 ; <i1> [#uses=1] - br i1 %toBool, label %bb19, label %bb17 - -bb17: ; preds = %bb16 - call void @llvm.dbg.stoppoint(i32 696, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %45 = invoke %"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZNSolsEd(%"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZSt4cout, double 0.000000e+00) - to label %bb23 unwind label %lpad51 ; <%"struct.std::basic_ostream<char,std::char_traits<char> >"*> [#uses=0] - -bb19: ; preds = %bb16 - call void @llvm.dbg.stoppoint(i32 698, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %46 = invoke %"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc(%"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZSt4cout, i8* getelementptr ([32 x i8]* @.str1825, i32 0, i32 0)) - to label %invcont20 unwind label %lpad51 ; <%"struct.std::basic_ostream<char,std::char_traits<char> >"*> [#uses=1] - -invcont20: ; preds = %bb19 - call void @llvm.dbg.stoppoint(i32 698, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %47 = invoke %"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZNSolsEd(%"struct.std::basic_ostream<char,std::char_traits<char> >"* %46, double 0.000000e+00) - to label %invcont21 unwind label %lpad51 ; <%"struct.std::basic_ostream<char,std::char_traits<char> >"*> [#uses=1] - -invcont21: ; preds = %invcont20 - call void @llvm.dbg.stoppoint(i32 698, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %48 = invoke %"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc(%"struct.std::basic_ostream<char,std::char_traits<char> >"* %47, i8* getelementptr ([3 x i8]* @.str1826, i32 0, i32 0)) - to label %bb23 unwind label %lpad51 ; <%"struct.std::basic_ostream<char,std::char_traits<char> >"*> [#uses=0] - -bb23: ; preds = %invcont21, %bb17 - call void @llvm.dbg.stoppoint(i32 700, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %49 = invoke %"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZNSo5flushEv(%"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZSt4cout) - to label %invcont24 unwind label %lpad51 ; <%"struct.std::basic_ostream<char,std::char_traits<char> >"*> [#uses=0] - -invcont24: ; preds = %bb23 - call void @llvm.dbg.stoppoint(i32 702, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %poly3) - to label %bb31 unwind label %lpad47 - -bb31: ; preds = %invcont24 - call void @llvm.dbg.stoppoint(i32 702, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %poly2) - to label %bb38 unwind label %lpad - -bb38: ; preds = %bb31 - call void @llvm.dbg.stoppoint(i32 702, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %poly1) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1806 to %0*)) - ret i32 0 - -lpad: ; preds = %bb31, %bb5 - %eh_ptr = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select46 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - br label %ppad - -lpad47: ; preds = %invcont24, %invcont - %eh_ptr48 = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select50 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr48, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - br label %ppad75 - -lpad51: ; preds = %bb23, %invcont21, %invcont20, %bb19, %bb17, %invcont11, %bb9 - %eh_ptr52 = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select54 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr52, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - br label %ppad76 - -lpad55: ; preds = %invcont10 - %eh_ptr56 = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select58 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr56, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 687, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %0) - to label %ppad76 unwind label %lpad59 - -lpad59: ; preds = %lpad55 - %eh_ptr60 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select62 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr60, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 687, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -lpad63: ; preds = %ppad76 - %eh_ptr64 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select66 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr64, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 702, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -lpad67: ; preds = %ppad75 - %eh_ptr68 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select70 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr68, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 702, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -lpad71: ; preds = %ppad - %eh_ptr72 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select74 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr72, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 702, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -ppad: ; preds = %ppad75, %lpad - %eh_exception.2 = phi i8* [ %eh_ptr, %lpad ], [ %eh_exception.1, %ppad75 ] ; <i8*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 702, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %poly1) - to label %Unwind unwind label %lpad71 - -ppad75: ; preds = %ppad76, %lpad47 - %eh_exception.1 = phi i8* [ %eh_ptr48, %lpad47 ], [ %eh_exception.0, %ppad76 ] ; <i8*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 702, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %poly2) - to label %ppad unwind label %lpad67 - -ppad76: ; preds = %lpad55, %lpad51 - %eh_exception.0 = phi i8* [ %eh_ptr52, %lpad51 ], [ %eh_ptr56, %lpad55 ] ; <i8*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 702, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %poly3) - to label %ppad75 unwind label %lpad63 - -Unwind: ; preds = %ppad - call void @_Unwind_Resume(i8* %eh_exception.2) - unreachable -} - -define internal void @_GLOBAL__I_main() { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram657 to %0*)) - call void @llvm.dbg.stoppoint(i32 77, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit636 to %0*)) - call void @_ZNSt8ios_base4InitC1Ev(%"struct.std::allocator<char>"* @_ZStL8__ioinit) - %0 = call i32 @__cxa_atexit(void (i8*)* @__tcf_0, i8* null, i8* bitcast (i8** @__dso_handle to i8*)) nounwind ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 400, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %1 = load i8* bitcast (i64* @_ZGVN10polynomialIdE4PI2IE to i8*), align 8 ; <i8> [#uses=1] - %2 = icmp eq i8 %1, 0 ; <i1> [#uses=1] - br i1 %2, label %bb2.i, label %_Z41__static_initialization_and_destruction_0ii.exit - -bb2.i: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 400, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - store i8 1, i8* bitcast (i64* @_ZGVN10polynomialIdE4PI2IE to i8*), align 8 - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* @_ZN10polynomialIdE4PI2IE, double 0.000000e+00, double 0x401921FB54442D18) nounwind - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram657 to %0*)) - ret void - -_Z41__static_initialization_and_destruction_0ii.exit: ; preds = %entry - ret void -} - -define linkonce void @_ZNSt7complexIdEC1ECd(%"struct.std::complex<double>"* %this, %1 %__z) nounwind { -entry: - %__z_addr = alloca %1, align 8 ; <%1*> [#uses=4] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram227 to %0*)) - %0 = bitcast %1* %__z_addr to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %0, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable231 to %0*)) - store %1 %__z, %1* %__z_addr, align 8 - call void @llvm.dbg.stoppoint(i32 1161, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %1 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %2 = getelementptr %1* %__z_addr, i32 0, i32 0 ; <double*> [#uses=1] - %3 = load double* %2, align 8 ; <double> [#uses=1] - store double %3, double* %1, align 4 - %4 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %5 = getelementptr %1* %__z_addr, i32 0, i32 1 ; <double*> [#uses=1] - %6 = load double* %5, align 8 ; <double> [#uses=1] - store double %6, double* %4, align 4 - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram227 to %0*)) - ret void -} - -declare void @llvm.dbg.func.start(%0*) nounwind readnone - -declare void @llvm.dbg.declare(%0*, %0*) nounwind readnone - -declare void @llvm.dbg.stoppoint(i32, i32, %0*) nounwind readnone - -declare void @llvm.dbg.region.end(%0*) nounwind readnone - -define linkonce %1* @_ZNKSt7complexIdE5__repEv(%"struct.std::complex<double>"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram232 to %0*)) - call void @llvm.dbg.stoppoint(i32 1192, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %0 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0 ; <%1*> [#uses=1] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram232 to %0*)) - ret %1* %0 -} - -define linkonce double* @_ZNSt7complexIdE4realEv(%"struct.std::complex<double>"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram235 to %0*)) - call void @llvm.dbg.stoppoint(i32 1200, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %0 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram235 to %0*)) - ret double* %0 -} - -define linkonce double* @_ZNKSt7complexIdE4realEv(%"struct.std::complex<double>"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram237 to %0*)) - call void @llvm.dbg.stoppoint(i32 1204, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %0 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram237 to %0*)) - ret double* %0 -} - -define linkonce double* @_ZNKSt7complexIdE4imagEv(%"struct.std::complex<double>"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram239 to %0*)) - call void @llvm.dbg.stoppoint(i32 1212, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %0 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram239 to %0*)) - ret double* %0 -} - -define linkonce void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %this, double %__r, double %__i) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram242 to %0*)) - call void @llvm.dbg.stoppoint(i32 1217, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %0 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - store double %__r, double* %0, align 4 - call void @llvm.dbg.stoppoint(i32 1218, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %1 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - store double %__i, double* %1, align 4 - call void @llvm.dbg.stoppoint(i32 1219, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram242 to %0*)) - ret void -} - -define linkonce %"struct.std::complex<double>"* @_ZNSt7complexIdEaSEd(%"struct.std::complex<double>"* %this, double %__d) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram248 to %0*)) - call void @llvm.dbg.stoppoint(i32 1224, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %0 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - store double %__d, double* %0, align 4 - call void @llvm.dbg.stoppoint(i32 1225, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %1 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - store double 0.000000e+00, double* %1, align 4 - call void @llvm.dbg.stoppoint(i32 1226, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram248 to %0*)) - ret %"struct.std::complex<double>"* %this -} - -define linkonce %"struct.std::complex<double>"* @_ZNSt7complexIdEdVEd(%"struct.std::complex<double>"* %this, double %__d) nounwind { -entry: - %0 = alloca %1, align 8 ; <%1*> [#uses=4] - %1 = alloca %1, align 8 ; <%1*> [#uses=4] - %2 = alloca %1, align 8 ; <%1*> [#uses=4] - %memtmp = alloca %1, align 8 ; <%1*> [#uses=4] - %memtmp1 = alloca %1, align 8 ; <%1*> [#uses=4] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram252 to %0*)) - call void @llvm.dbg.stoppoint(i32 1253, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %3 = getelementptr %1* %2, i32 0, i32 0 ; <double*> [#uses=1] - %4 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %5 = load double* %4, align 4 ; <double> [#uses=1] - store double %5, double* %3, align 8 - %6 = getelementptr %1* %2, i32 0, i32 1 ; <double*> [#uses=1] - %7 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %8 = load double* %7, align 4 ; <double> [#uses=1] - store double %8, double* %6, align 8 - %real = getelementptr %1* %1, i32 0, i32 0 ; <double*> [#uses=1] - store double %__d, double* %real, align 8 - %imag = getelementptr %1* %1, i32 0, i32 1 ; <double*> [#uses=1] - store double 0.000000e+00, double* %imag, align 8 - %9 = getelementptr %1* %memtmp, i32 0, i32 0 ; <double*> [#uses=1] - %10 = getelementptr %1* %2, i32 0, i32 0 ; <double*> [#uses=1] - %11 = load double* %10, align 8 ; <double> [#uses=1] - store double %11, double* %9, align 8 - %12 = getelementptr %1* %memtmp, i32 0, i32 1 ; <double*> [#uses=1] - %13 = getelementptr %1* %2, i32 0, i32 1 ; <double*> [#uses=1] - %14 = load double* %13, align 8 ; <double> [#uses=1] - store double %14, double* %12, align 8 - %15 = getelementptr %1* %memtmp1, i32 0, i32 0 ; <double*> [#uses=1] - %16 = getelementptr %1* %1, i32 0, i32 0 ; <double*> [#uses=1] - %17 = load double* %16, align 8 ; <double> [#uses=1] - store double %17, double* %15, align 8 - %18 = getelementptr %1* %memtmp1, i32 0, i32 1 ; <double*> [#uses=1] - %19 = getelementptr %1* %1, i32 0, i32 1 ; <double*> [#uses=1] - %20 = load double* %19, align 8 ; <double> [#uses=1] - store double %20, double* %18, align 8 - %real2 = getelementptr %1* %memtmp, i32 0, i32 0 ; <double*> [#uses=1] - %real3 = load double* %real2, align 8 ; <double> [#uses=2] - %imag4 = getelementptr %1* %memtmp, i32 0, i32 1 ; <double*> [#uses=1] - %imag5 = load double* %imag4, align 8 ; <double> [#uses=2] - %real6 = getelementptr %1* %memtmp1, i32 0, i32 0 ; <double*> [#uses=1] - %real7 = load double* %real6, align 8 ; <double> [#uses=4] - %imag8 = getelementptr %1* %memtmp1, i32 0, i32 1 ; <double*> [#uses=1] - %imag9 = load double* %imag8, align 8 ; <double> [#uses=4] - %21 = fmul double %real3, %real7 ; <double> [#uses=1] - %22 = fmul double %imag5, %imag9 ; <double> [#uses=1] - %23 = fadd double %21, %22 ; <double> [#uses=1] - %24 = fmul double %real7, %real7 ; <double> [#uses=1] - %25 = fmul double %imag9, %imag9 ; <double> [#uses=1] - %26 = fadd double %24, %25 ; <double> [#uses=2] - %27 = fdiv double %23, %26 ; <double> [#uses=1] - %28 = fmul double %imag5, %real7 ; <double> [#uses=1] - %29 = fmul double %real3, %imag9 ; <double> [#uses=1] - %30 = fsub double %28, %29 ; <double> [#uses=1] - %31 = fdiv double %30, %26 ; <double> [#uses=1] - %real10 = getelementptr %1* %0, i32 0, i32 0 ; <double*> [#uses=1] - store double %27, double* %real10, align 8 - %imag11 = getelementptr %1* %0, i32 0, i32 1 ; <double*> [#uses=1] - store double %31, double* %imag11, align 8 - %32 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %33 = getelementptr %1* %0, i32 0, i32 0 ; <double*> [#uses=1] - %34 = load double* %33, align 8 ; <double> [#uses=1] - store double %34, double* %32, align 4 - %35 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %36 = getelementptr %1* %0, i32 0, i32 1 ; <double*> [#uses=1] - %37 = load double* %36, align 8 ; <double> [#uses=1] - store double %37, double* %35, align 4 - call void @llvm.dbg.stoppoint(i32 1254, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram252 to %0*)) - ret %"struct.std::complex<double>"* %this -} - -define linkonce double* @_ZN10polynomialIdEixEj(%"struct.polynomial<double>"* %this, i32 %term) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram522 to %0*)) - call void @llvm.dbg.stoppoint(i32 308, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - %1 = load double** %0, align 4 ; <double*> [#uses=1] - %2 = getelementptr double* %1, i32 %term ; <double*> [#uses=1] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram522 to %0*)) - ret double* %2 -} - -define linkonce i32 @_ZNK10polynomialIdE6degreeEv(%"struct.polynomial<double>"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram527 to %0*)) - call void @llvm.dbg.stoppoint(i32 113, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %1 = load i32* %0, align 4 ; <i32> [#uses=1] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram527 to %0*)) - ret i32 %1 -} - -define linkonce %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %this, i32 %term) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram530 to %0*)) - call void @llvm.dbg.stoppoint(i32 308, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 1 ; <%"struct.std::complex<double>"**> [#uses=1] - %1 = load %"struct.std::complex<double>"** %0, align 4 ; <%"struct.std::complex<double>"*> [#uses=1] - %2 = getelementptr %"struct.std::complex<double>"* %1, i32 %term ; <%"struct.std::complex<double>"*> [#uses=1] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram530 to %0*)) - ret %"struct.std::complex<double>"* %2 -} - -define linkonce %"struct.std::complex<double>"* @_ZNSt7complexIdEmLIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %this, %"struct.std::complex<double>"* %__z) nounwind { -entry: - %__t = alloca %1, align 8 ; <%1*> [#uses=7] - %0 = alloca %1, align 8 ; <%1*> [#uses=4] - %1 = alloca %1, align 8 ; <%1*> [#uses=4] - %memtmp = alloca %1, align 8 ; <%1*> [#uses=4] - %memtmp3 = alloca %1, align 8 ; <%1*> [#uses=4] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram538 to %0*)) - %2 = bitcast %1* %__t to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %2, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable542 to %0*)) - call void @llvm.dbg.stoppoint(i32 1289, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %3 = call double* @_ZNKSt7complexIdE4realEv(%"struct.std::complex<double>"* %__z) nounwind ; <double*> [#uses=1] - %4 = load double* %3, align 8 ; <double> [#uses=1] - %5 = getelementptr %1* %__t, i32 0, i32 1 ; <double*> [#uses=1] - %6 = load double* %5, align 8 ; <double> [#uses=1] - %real = getelementptr %1* %__t, i32 0, i32 0 ; <double*> [#uses=1] - store double %4, double* %real, align 8 - %imag = getelementptr %1* %__t, i32 0, i32 1 ; <double*> [#uses=1] - store double %6, double* %imag, align 8 - call void @llvm.dbg.stoppoint(i32 1290, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %7 = call double* @_ZNKSt7complexIdE4imagEv(%"struct.std::complex<double>"* %__z) nounwind ; <double*> [#uses=1] - %8 = load double* %7, align 8 ; <double> [#uses=1] - %imag2 = getelementptr %1* %__t, i32 0, i32 1 ; <double*> [#uses=1] - store double %8, double* %imag2, align 8 - call void @llvm.dbg.stoppoint(i32 1291, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %9 = getelementptr %1* %1, i32 0, i32 0 ; <double*> [#uses=1] - %10 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %11 = load double* %10, align 4 ; <double> [#uses=1] - store double %11, double* %9, align 8 - %12 = getelementptr %1* %1, i32 0, i32 1 ; <double*> [#uses=1] - %13 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %14 = load double* %13, align 4 ; <double> [#uses=1] - store double %14, double* %12, align 8 - %15 = getelementptr %1* %memtmp, i32 0, i32 0 ; <double*> [#uses=1] - %16 = getelementptr %1* %1, i32 0, i32 0 ; <double*> [#uses=1] - %17 = load double* %16, align 8 ; <double> [#uses=1] - store double %17, double* %15, align 8 - %18 = getelementptr %1* %memtmp, i32 0, i32 1 ; <double*> [#uses=1] - %19 = getelementptr %1* %1, i32 0, i32 1 ; <double*> [#uses=1] - %20 = load double* %19, align 8 ; <double> [#uses=1] - store double %20, double* %18, align 8 - %21 = getelementptr %1* %memtmp3, i32 0, i32 0 ; <double*> [#uses=1] - %22 = getelementptr %1* %__t, i32 0, i32 0 ; <double*> [#uses=1] - %23 = load double* %22, align 8 ; <double> [#uses=1] - store double %23, double* %21, align 8 - %24 = getelementptr %1* %memtmp3, i32 0, i32 1 ; <double*> [#uses=1] - %25 = getelementptr %1* %__t, i32 0, i32 1 ; <double*> [#uses=1] - %26 = load double* %25, align 8 ; <double> [#uses=1] - store double %26, double* %24, align 8 - %real4 = getelementptr %1* %memtmp, i32 0, i32 0 ; <double*> [#uses=1] - %real5 = load double* %real4, align 8 ; <double> [#uses=2] - %imag6 = getelementptr %1* %memtmp, i32 0, i32 1 ; <double*> [#uses=1] - %imag7 = load double* %imag6, align 8 ; <double> [#uses=2] - %real8 = getelementptr %1* %memtmp3, i32 0, i32 0 ; <double*> [#uses=1] - %real9 = load double* %real8, align 8 ; <double> [#uses=2] - %imag10 = getelementptr %1* %memtmp3, i32 0, i32 1 ; <double*> [#uses=1] - %imag11 = load double* %imag10, align 8 ; <double> [#uses=2] - %27 = fmul double %real5, %real9 ; <double> [#uses=1] - %28 = fmul double %imag7, %imag11 ; <double> [#uses=1] - %29 = fsub double %27, %28 ; <double> [#uses=1] - %30 = fmul double %real5, %imag11 ; <double> [#uses=1] - %31 = fmul double %real9, %imag7 ; <double> [#uses=1] - %32 = fadd double %30, %31 ; <double> [#uses=1] - %real12 = getelementptr %1* %0, i32 0, i32 0 ; <double*> [#uses=1] - store double %29, double* %real12, align 8 - %imag13 = getelementptr %1* %0, i32 0, i32 1 ; <double*> [#uses=1] - store double %32, double* %imag13, align 8 - %33 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %34 = getelementptr %1* %0, i32 0, i32 0 ; <double*> [#uses=1] - %35 = load double* %34, align 8 ; <double> [#uses=1] - store double %35, double* %33, align 4 - %36 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %37 = getelementptr %1* %0, i32 0, i32 1 ; <double*> [#uses=1] - %38 = load double* %37, align 8 ; <double> [#uses=1] - store double %38, double* %36, align 4 - call void @llvm.dbg.stoppoint(i32 1292, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram538 to %0*)) - ret %"struct.std::complex<double>"* %this -} - -define linkonce void @_ZN10polynomialIdE9deep_copyEPKd(%"struct.polynomial<double>"* %this, double* %source) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram543 to %0*)) - call void @llvm.dbg.stoppoint(i32 197, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb1 - -bb: ; preds = %bb1 - call void @llvm.dbg.stoppoint(i32 198, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - %1 = load double** %0, align 4 ; <double*> [#uses=1] - %2 = getelementptr double* %source, i32 %n.0 ; <double*> [#uses=1] - %3 = load double* %2, align 1 ; <double> [#uses=1] - %4 = getelementptr double* %1, i32 %n.0 ; <double*> [#uses=1] - store double %3, double* %4, align 1 - call void @llvm.dbg.stoppoint(i32 197, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %5 = add i32 %n.0, 1 ; <i32> [#uses=1] - br label %bb1 - -bb1: ; preds = %bb, %entry - %n.0 = phi i32 [ 0, %entry ], [ %5, %bb ] ; <i32> [#uses=4] - call void @llvm.dbg.stoppoint(i32 197, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %6 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %7 = load i32* %6, align 4 ; <i32> [#uses=1] - %8 = icmp ugt i32 %7, %n.0 ; <i1> [#uses=1] - br i1 %8, label %bb, label %return - -return: ; preds = %bb1 - call void @llvm.dbg.stoppoint(i32 198, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram543 to %0*)) - ret void -} - -define linkonce i32 @_ZN10polynomialIdE4log2Ej(i32 %n) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram549 to %0*)) - call void @llvm.dbg.stoppoint(i32 407, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb1 - -bb: ; preds = %bb1 - call void @llvm.dbg.stoppoint(i32 411, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = add i32 %c.0, 1 ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 412, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %1 = shl i32 %x.0, 1 ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 414, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %2 = icmp eq i32 %1, 0 ; <i1> [#uses=1] - br i1 %2, label %return, label %bb1 - -bb1: ; preds = %bb, %entry - %c.0 = phi i32 [ 0, %entry ], [ %0, %bb ] ; <i32> [#uses=2] - %x.0 = phi i32 [ 1, %entry ], [ %1, %bb ] ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 409, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %3 = icmp ult i32 %x.0, %n ; <i1> [#uses=1] - br i1 %3, label %bb, label %return - -return: ; preds = %bb1, %bb - %c.1 = phi i32 [ %c.0, %bb1 ], [ %0, %bb ] ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 418, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram549 to %0*)) - ret i32 %c.1 -} - -define linkonce void @_ZStmlIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %agg.result, %"struct.std::complex<double>"* %__x, %"struct.std::complex<double>"* %__y) nounwind { -entry: - %__r = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=1] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram559 to %0*)) - %0 = bitcast %"struct.std::complex<double>"* %__r to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %0, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable564 to %0*)) - call void @llvm.dbg.stoppoint(i32 380, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %1 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %2 = getelementptr %"struct.std::complex<double>"* %__x, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %3 = load double* %2, align 4 ; <double> [#uses=1] - store double %3, double* %1, align 4 - %4 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %5 = getelementptr %"struct.std::complex<double>"* %__x, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %6 = load double* %5, align 4 ; <double> [#uses=1] - store double %6, double* %4, align 4 - call void @llvm.dbg.stoppoint(i32 381, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %7 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEmLIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %agg.result, %"struct.std::complex<double>"* %__y) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram559 to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialISt7complexIdEE9deep_copyEPKS1_(%"struct.polynomial<std::complex<double> >"* %this, %"struct.std::complex<double>"* %source) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram565 to %0*)) - call void @llvm.dbg.stoppoint(i32 197, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb1 - -bb: ; preds = %bb1 - call void @llvm.dbg.stoppoint(i32 198, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 1 ; <%"struct.std::complex<double>"**> [#uses=1] - %1 = load %"struct.std::complex<double>"** %0, align 4 ; <%"struct.std::complex<double>"*> [#uses=2] - %2 = getelementptr %"struct.std::complex<double>"* %1, i32 %n.0, i32 0, i32 0 ; <double*> [#uses=1] - %3 = getelementptr %"struct.std::complex<double>"* %source, i32 %n.0, i32 0, i32 0 ; <double*> [#uses=1] - %4 = load double* %3, align 1 ; <double> [#uses=1] - store double %4, double* %2, align 1 - %5 = getelementptr %"struct.std::complex<double>"* %1, i32 %n.0, i32 0, i32 1 ; <double*> [#uses=1] - %6 = getelementptr %"struct.std::complex<double>"* %source, i32 %n.0, i32 0, i32 1 ; <double*> [#uses=1] - %7 = load double* %6, align 1 ; <double> [#uses=1] - store double %7, double* %5, align 1 - call void @llvm.dbg.stoppoint(i32 197, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %8 = add i32 %n.0, 1 ; <i32> [#uses=1] - br label %bb1 - -bb1: ; preds = %bb, %entry - %n.0 = phi i32 [ 0, %entry ], [ %8, %bb ] ; <i32> [#uses=6] - call void @llvm.dbg.stoppoint(i32 197, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %9 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %10 = load i32* %9, align 4 ; <i32> [#uses=1] - %11 = icmp ugt i32 %10, %n.0 ; <i1> [#uses=1] - br i1 %11, label %bb, label %return - -return: ; preds = %bb1 - call void @llvm.dbg.stoppoint(i32 198, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram565 to %0*)) - ret void -} - -define linkonce i32 @_ZNK10polynomialISt7complexIdEE6degreeEv(%"struct.polynomial<std::complex<double> >"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram569 to %0*)) - call void @llvm.dbg.stoppoint(i32 113, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %1 = load i32* %0, align 4 ; <i32> [#uses=1] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram569 to %0*)) - ret i32 %1 -} - -define linkonce void @_ZStngIdESt7complexIT_ERKS2_(%"struct.std::complex<double>"* noalias sret %agg.result, %"struct.std::complex<double>"* %__x) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram576 to %0*)) - call void @llvm.dbg.stoppoint(i32 444, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %0 = call double* @_ZNKSt7complexIdE4imagEv(%"struct.std::complex<double>"* %__x) nounwind ; <double*> [#uses=1] - %1 = load double* %0, align 8 ; <double> [#uses=1] - %2 = fsub double -0.000000e+00, %1 ; <double> [#uses=1] - %3 = call double* @_ZNKSt7complexIdE4realEv(%"struct.std::complex<double>"* %__x) nounwind ; <double*> [#uses=1] - %4 = load double* %3, align 8 ; <double> [#uses=1] - %5 = fsub double -0.000000e+00, %4 ; <double> [#uses=1] - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %agg.result, double %5, double %2) nounwind - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram576 to %0*)) - ret void -} - -define linkonce double @_ZNK10polynomialIdE3getEj(%"struct.polynomial<double>"* %this, i32 %term) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram578 to %0*)) - call void @llvm.dbg.stoppoint(i32 302, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - %1 = load double** %0, align 4 ; <double*> [#uses=1] - %2 = getelementptr double* %1, i32 %term ; <double*> [#uses=1] - %3 = load double* %2, align 1 ; <double> [#uses=1] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram578 to %0*)) - ret double %3 -} - -define linkonce i32 @_ZN10polynomialIdE9flip_bitsEjj(i32 %k, i32 %bits) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram581 to %0*)) - call void @llvm.dbg.stoppoint(i32 425, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = add i32 %bits, -1 ; <i32> [#uses=1] - %1 = shl i32 1, %0 ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 427, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb3 - -bb: ; preds = %bb3 - call void @llvm.dbg.stoppoint(i32 431, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %2 = and i32 %rm.0, %k ; <i32> [#uses=1] - %3 = icmp ne i32 %2, 0 ; <i1> [#uses=1] - %4 = select i1 %3, i32 %lm.0, i32 0 ; <i32> [#uses=1] - %.r.1 = or i32 %r.1, %4 ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 434, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %5 = lshr i32 %lm.0, 1 ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 435, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %6 = shl i32 %rm.0, 1 ; <i32> [#uses=1] - br label %bb3 - -bb3: ; preds = %bb, %entry - %r.1 = phi i32 [ 0, %entry ], [ %.r.1, %bb ] ; <i32> [#uses=2] - %rm.0 = phi i32 [ 1, %entry ], [ %6, %bb ] ; <i32> [#uses=2] - %lm.0 = phi i32 [ %1, %entry ], [ %5, %bb ] ; <i32> [#uses=3] - call void @llvm.dbg.stoppoint(i32 429, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %7 = icmp eq i32 %lm.0, 0 ; <i1> [#uses=1] - br i1 %7, label %return, label %bb - -return: ; preds = %bb3 - call void @llvm.dbg.stoppoint(i32 438, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram581 to %0*)) - ret i32 %r.1 -} - -define linkonce %"struct.std::complex<double>"* @_ZNSt7complexIdEdVIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %this, %"struct.std::complex<double>"* %__z) nounwind { -entry: - %__t = alloca %1, align 8 ; <%1*> [#uses=7] - %0 = alloca %1, align 8 ; <%1*> [#uses=4] - %1 = alloca %1, align 8 ; <%1*> [#uses=4] - %memtmp = alloca %1, align 8 ; <%1*> [#uses=4] - %memtmp3 = alloca %1, align 8 ; <%1*> [#uses=4] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram593 to %0*)) - %2 = bitcast %1* %__t to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %2, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable596 to %0*)) - call void @llvm.dbg.stoppoint(i32 1300, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %3 = call double* @_ZNKSt7complexIdE4realEv(%"struct.std::complex<double>"* %__z) nounwind ; <double*> [#uses=1] - %4 = load double* %3, align 8 ; <double> [#uses=1] - %5 = getelementptr %1* %__t, i32 0, i32 1 ; <double*> [#uses=1] - %6 = load double* %5, align 8 ; <double> [#uses=1] - %real = getelementptr %1* %__t, i32 0, i32 0 ; <double*> [#uses=1] - store double %4, double* %real, align 8 - %imag = getelementptr %1* %__t, i32 0, i32 1 ; <double*> [#uses=1] - store double %6, double* %imag, align 8 - call void @llvm.dbg.stoppoint(i32 1301, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %7 = call double* @_ZNKSt7complexIdE4imagEv(%"struct.std::complex<double>"* %__z) nounwind ; <double*> [#uses=1] - %8 = load double* %7, align 8 ; <double> [#uses=1] - %imag2 = getelementptr %1* %__t, i32 0, i32 1 ; <double*> [#uses=1] - store double %8, double* %imag2, align 8 - call void @llvm.dbg.stoppoint(i32 1302, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %9 = getelementptr %1* %1, i32 0, i32 0 ; <double*> [#uses=1] - %10 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %11 = load double* %10, align 4 ; <double> [#uses=1] - store double %11, double* %9, align 8 - %12 = getelementptr %1* %1, i32 0, i32 1 ; <double*> [#uses=1] - %13 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %14 = load double* %13, align 4 ; <double> [#uses=1] - store double %14, double* %12, align 8 - %15 = getelementptr %1* %memtmp, i32 0, i32 0 ; <double*> [#uses=1] - %16 = getelementptr %1* %1, i32 0, i32 0 ; <double*> [#uses=1] - %17 = load double* %16, align 8 ; <double> [#uses=1] - store double %17, double* %15, align 8 - %18 = getelementptr %1* %memtmp, i32 0, i32 1 ; <double*> [#uses=1] - %19 = getelementptr %1* %1, i32 0, i32 1 ; <double*> [#uses=1] - %20 = load double* %19, align 8 ; <double> [#uses=1] - store double %20, double* %18, align 8 - %21 = getelementptr %1* %memtmp3, i32 0, i32 0 ; <double*> [#uses=1] - %22 = getelementptr %1* %__t, i32 0, i32 0 ; <double*> [#uses=1] - %23 = load double* %22, align 8 ; <double> [#uses=1] - store double %23, double* %21, align 8 - %24 = getelementptr %1* %memtmp3, i32 0, i32 1 ; <double*> [#uses=1] - %25 = getelementptr %1* %__t, i32 0, i32 1 ; <double*> [#uses=1] - %26 = load double* %25, align 8 ; <double> [#uses=1] - store double %26, double* %24, align 8 - %real4 = getelementptr %1* %memtmp, i32 0, i32 0 ; <double*> [#uses=1] - %real5 = load double* %real4, align 8 ; <double> [#uses=2] - %imag6 = getelementptr %1* %memtmp, i32 0, i32 1 ; <double*> [#uses=1] - %imag7 = load double* %imag6, align 8 ; <double> [#uses=2] - %real8 = getelementptr %1* %memtmp3, i32 0, i32 0 ; <double*> [#uses=1] - %real9 = load double* %real8, align 8 ; <double> [#uses=4] - %imag10 = getelementptr %1* %memtmp3, i32 0, i32 1 ; <double*> [#uses=1] - %imag11 = load double* %imag10, align 8 ; <double> [#uses=4] - %27 = fmul double %real5, %real9 ; <double> [#uses=1] - %28 = fmul double %imag7, %imag11 ; <double> [#uses=1] - %29 = fadd double %27, %28 ; <double> [#uses=1] - %30 = fmul double %real9, %real9 ; <double> [#uses=1] - %31 = fmul double %imag11, %imag11 ; <double> [#uses=1] - %32 = fadd double %30, %31 ; <double> [#uses=2] - %33 = fdiv double %29, %32 ; <double> [#uses=1] - %34 = fmul double %imag7, %real9 ; <double> [#uses=1] - %35 = fmul double %real5, %imag11 ; <double> [#uses=1] - %36 = fsub double %34, %35 ; <double> [#uses=1] - %37 = fdiv double %36, %32 ; <double> [#uses=1] - %real12 = getelementptr %1* %0, i32 0, i32 0 ; <double*> [#uses=1] - store double %33, double* %real12, align 8 - %imag13 = getelementptr %1* %0, i32 0, i32 1 ; <double*> [#uses=1] - store double %37, double* %imag13, align 8 - %38 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %39 = getelementptr %1* %0, i32 0, i32 0 ; <double*> [#uses=1] - %40 = load double* %39, align 8 ; <double> [#uses=1] - store double %40, double* %38, align 4 - %41 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %42 = getelementptr %1* %0, i32 0, i32 1 ; <double*> [#uses=1] - %43 = load double* %42, align 8 ; <double> [#uses=1] - store double %43, double* %41, align 4 - call void @llvm.dbg.stoppoint(i32 1303, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram593 to %0*)) - ret %"struct.std::complex<double>"* %this -} - -define linkonce void @_ZStdvIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %agg.result, %"struct.std::complex<double>"* %__x, %"struct.std::complex<double>"* %__y) { -entry: - %__r = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=1] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram599 to %0*)) - %0 = bitcast %"struct.std::complex<double>"* %__r to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %0, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable602 to %0*)) - call void @llvm.dbg.stoppoint(i32 410, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %1 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %2 = getelementptr %"struct.std::complex<double>"* %__x, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %3 = load double* %2, align 4 ; <double> [#uses=1] - store double %3, double* %1, align 4 - %4 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %5 = getelementptr %"struct.std::complex<double>"* %__x, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %6 = load double* %5, align 4 ; <double> [#uses=1] - store double %6, double* %4, align 4 - call void @llvm.dbg.stoppoint(i32 411, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %7 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEdVIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %agg.result, %"struct.std::complex<double>"* %__y) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram599 to %0*)) - ret void -} - -define linkonce %"struct.std::complex<double>"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %this, %"struct.std::complex<double>"* %__z) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram605 to %0*)) - call void @llvm.dbg.stoppoint(i32 1270, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %0 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %1 = load double* %0, align 4 ; <double> [#uses=1] - %2 = call double* @_ZNKSt7complexIdE4realEv(%"struct.std::complex<double>"* %__z) nounwind ; <double*> [#uses=1] - %3 = load double* %2, align 8 ; <double> [#uses=1] - %4 = fadd double %1, %3 ; <double> [#uses=1] - %5 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - store double %4, double* %5, align 4 - call void @llvm.dbg.stoppoint(i32 1271, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %6 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %7 = load double* %6, align 4 ; <double> [#uses=1] - %8 = call double* @_ZNKSt7complexIdE4imagEv(%"struct.std::complex<double>"* %__z) nounwind ; <double*> [#uses=1] - %9 = load double* %8, align 8 ; <double> [#uses=1] - %10 = fadd double %7, %9 ; <double> [#uses=1] - %11 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - store double %10, double* %11, align 4 - call void @llvm.dbg.stoppoint(i32 1272, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram605 to %0*)) - ret %"struct.std::complex<double>"* %this -} - -define linkonce void @_ZStplIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %agg.result, %"struct.std::complex<double>"* %__x, %"struct.std::complex<double>"* %__y) { -entry: - %__r = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=1] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram610 to %0*)) - %0 = bitcast %"struct.std::complex<double>"* %__r to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %0, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable613 to %0*)) - call void @llvm.dbg.stoppoint(i32 320, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %1 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %2 = getelementptr %"struct.std::complex<double>"* %__x, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %3 = load double* %2, align 4 ; <double> [#uses=1] - store double %3, double* %1, align 4 - %4 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %5 = getelementptr %"struct.std::complex<double>"* %__x, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %6 = load double* %5, align 4 ; <double> [#uses=1] - store double %6, double* %4, align 4 - call void @llvm.dbg.stoppoint(i32 321, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %7 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEpLIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %agg.result, %"struct.std::complex<double>"* %__y) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram610 to %0*)) - ret void -} - -define linkonce %"struct.std::complex<double>"* @_ZNSt7complexIdEmIIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %this, %"struct.std::complex<double>"* %__z) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram616 to %0*)) - call void @llvm.dbg.stoppoint(i32 1279, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %0 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %1 = load double* %0, align 4 ; <double> [#uses=1] - %2 = call double* @_ZNKSt7complexIdE4realEv(%"struct.std::complex<double>"* %__z) nounwind ; <double*> [#uses=1] - %3 = load double* %2, align 8 ; <double> [#uses=1] - %4 = fsub double %1, %3 ; <double> [#uses=1] - %5 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - store double %4, double* %5, align 4 - call void @llvm.dbg.stoppoint(i32 1280, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %6 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %7 = load double* %6, align 4 ; <double> [#uses=1] - %8 = call double* @_ZNKSt7complexIdE4imagEv(%"struct.std::complex<double>"* %__z) nounwind ; <double*> [#uses=1] - %9 = load double* %8, align 8 ; <double> [#uses=1] - %10 = fsub double %7, %9 ; <double> [#uses=1] - %11 = getelementptr %"struct.std::complex<double>"* %this, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - store double %10, double* %11, align 4 - call void @llvm.dbg.stoppoint(i32 1281, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram616 to %0*)) - ret %"struct.std::complex<double>"* %this -} - -define linkonce void @_ZStmiIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %agg.result, %"struct.std::complex<double>"* %__x, %"struct.std::complex<double>"* %__y) { -entry: - %__r = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=1] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram620 to %0*)) - %0 = bitcast %"struct.std::complex<double>"* %__r to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %0, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable623 to %0*)) - call void @llvm.dbg.stoppoint(i32 350, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %1 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %2 = getelementptr %"struct.std::complex<double>"* %__x, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %3 = load double* %2, align 4 ; <double> [#uses=1] - store double %3, double* %1, align 4 - %4 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %5 = getelementptr %"struct.std::complex<double>"* %__x, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %6 = load double* %5, align 4 ; <double> [#uses=1] - store double %6, double* %4, align 4 - call void @llvm.dbg.stoppoint(i32 351, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %7 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEmIIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %agg.result, %"struct.std::complex<double>"* %__y) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram620 to %0*)) - ret void -} - -define linkonce void @_ZNK10polynomialISt7complexIdEE3getEj(%"struct.std::complex<double>"* noalias sret %agg.result, %"struct.polynomial<std::complex<double> >"* %this, i32 %term) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram624 to %0*)) - call void @llvm.dbg.stoppoint(i32 302, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 1 ; <%"struct.std::complex<double>"**> [#uses=1] - %1 = load %"struct.std::complex<double>"** %0, align 4 ; <%"struct.std::complex<double>"*> [#uses=2] - %2 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %3 = getelementptr %"struct.std::complex<double>"* %1, i32 %term, i32 0, i32 0 ; <double*> [#uses=1] - %4 = load double* %3, align 1 ; <double> [#uses=1] - store double %4, double* %2, align 1 - %5 = getelementptr %"struct.std::complex<double>"* %agg.result, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %6 = getelementptr %"struct.std::complex<double>"* %1, i32 %term, i32 0, i32 1 ; <double*> [#uses=1] - %7 = load double* %6, align 1 ; <double> [#uses=1] - store double %7, double* %5, align 1 - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram624 to %0*)) - ret void -} - -declare void @_ZNSt8ios_base4InitC1Ev(%"struct.std::allocator<char>"*) - -declare i32 @__cxa_atexit(void (i8*)*, i8*, i8*) nounwind - -define internal void @__tcf_0(i8* nocapture %unnamed_arg) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram662 to %0*)) - call void @llvm.dbg.stoppoint(i32 77, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit636 to %0*)) - call void @_ZNSt8ios_base4InitD1Ev(%"struct.std::allocator<char>"* @_ZStL8__ioinit) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram662 to %0*)) - ret void -} - -declare void @_ZNSt8ios_base4InitD1Ev(%"struct.std::allocator<char>"*) - -define linkonce void @_ZN10polynomialIdE7releaseEv(%"struct.polynomial<double>"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram665 to %0*)) - call void @llvm.dbg.stoppoint(i32 190, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - %1 = load double** %0, align 4 ; <double*> [#uses=1] - %2 = icmp eq double* %1, null ; <i1> [#uses=1] - br i1 %2, label %return, label %bb - -bb: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 190, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %3 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - %4 = load double** %3, align 4 ; <double*> [#uses=1] - %5 = bitcast double* %4 to i8* ; <i8*> [#uses=1] - call void @_ZdaPv(i8* %5) nounwind - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram665 to %0*)) - ret void - -return: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 190, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - ret void -} - -declare void @_ZdaPv(i8*) nounwind - -define linkonce void @_ZN10polynomialIdED0Ev(%"struct.polynomial<double>"* %this) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram668 to %0*)) - call void @llvm.dbg.stoppoint(i32 255, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 0 ; <i32 (...)***> [#uses=1] - store i32 (...)** getelementptr ([4 x i32 (...)*]* @_ZTV10polynomialIdE, i32 0, i32 2), i32 (...)*** %0, align 4 - call void @llvm.dbg.stoppoint(i32 254, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdE7releaseEv(%"struct.polynomial<double>"* %this) nounwind - call void @llvm.dbg.stoppoint(i32 254, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %1 = bitcast %"struct.polynomial<double>"* %this to i8* ; <i8*> [#uses=1] - call void @_ZdlPv(i8* %1) nounwind - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram668 to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %this) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram691 to %0*)) - call void @llvm.dbg.stoppoint(i32 255, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 0 ; <i32 (...)***> [#uses=1] - store i32 (...)** getelementptr ([4 x i32 (...)*]* @_ZTV10polynomialIdE, i32 0, i32 2), i32 (...)*** %0, align 4 - call void @llvm.dbg.stoppoint(i32 254, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdE7releaseEv(%"struct.polynomial<double>"* %this) nounwind - call void @llvm.dbg.stoppoint(i32 254, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - ret void -} - -declare void @_ZdlPv(i8*) nounwind - -define linkonce void @_ZN10polynomialISt7complexIdEE7releaseEv(%"struct.polynomial<std::complex<double> >"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram693 to %0*)) - call void @llvm.dbg.stoppoint(i32 190, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 1 ; <%"struct.std::complex<double>"**> [#uses=1] - %1 = load %"struct.std::complex<double>"** %0, align 4 ; <%"struct.std::complex<double>"*> [#uses=1] - %2 = icmp eq %"struct.std::complex<double>"* %1, null ; <i1> [#uses=1] - br i1 %2, label %return, label %bb - -bb: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 190, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %3 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 1 ; <%"struct.std::complex<double>"**> [#uses=1] - %4 = load %"struct.std::complex<double>"** %3, align 4 ; <%"struct.std::complex<double>"*> [#uses=1] - %5 = bitcast %"struct.std::complex<double>"* %4 to i8* ; <i8*> [#uses=1] - call void @_ZdaPv(i8* %5) nounwind - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram693 to %0*)) - ret void - -return: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 190, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialISt7complexIdEED0Ev(%"struct.polynomial<std::complex<double> >"* %this) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram696 to %0*)) - call void @llvm.dbg.stoppoint(i32 255, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 0 ; <i32 (...)***> [#uses=1] - store i32 (...)** getelementptr ([4 x i32 (...)*]* @_ZTV10polynomialISt7complexIdEE, i32 0, i32 2), i32 (...)*** %0, align 4 - call void @llvm.dbg.stoppoint(i32 254, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialISt7complexIdEE7releaseEv(%"struct.polynomial<std::complex<double> >"* %this) nounwind - call void @llvm.dbg.stoppoint(i32 254, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %1 = bitcast %"struct.polynomial<std::complex<double> >"* %this to i8* ; <i8*> [#uses=1] - call void @_ZdlPv(i8* %1) nounwind - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram696 to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialISt7complexIdEED1Ev(%"struct.polynomial<std::complex<double> >"* %this) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram708 to %0*)) - call void @llvm.dbg.stoppoint(i32 255, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 0 ; <i32 (...)***> [#uses=1] - store i32 (...)** getelementptr ([4 x i32 (...)*]* @_ZTV10polynomialISt7complexIdEE, i32 0, i32 2), i32 (...)*** %0, align 4 - call void @llvm.dbg.stoppoint(i32 254, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialISt7complexIdEE7releaseEv(%"struct.polynomial<std::complex<double> >"* %this) nounwind - call void @llvm.dbg.stoppoint(i32 254, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialIdE7acquireEv(%"struct.polynomial<double>"* %this) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram710 to %0*)) - call void @llvm.dbg.stoppoint(i32 183, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %1 = load i32* %0, align 4 ; <i32> [#uses=1] - %2 = shl i32 %1, 3 ; <i32> [#uses=1] - %3 = call i8* @_Znaj(i32 %2) ; <i8*> [#uses=1] - %4 = bitcast i8* %3 to double* ; <double*> [#uses=1] - %5 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - store double* %4, double** %5, align 4 - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram710 to %0*)) - ret void -} - -declare i8* @_Znaj(i32) - -define linkonce void @_ZN10polynomialIdEC1Ej(%"struct.polynomial<double>"* %this, i32 %degree) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram713 to %0*)) - call void @llvm.dbg.stoppoint(i32 213, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 0 ; <i32 (...)***> [#uses=1] - store i32 (...)** getelementptr ([4 x i32 (...)*]* @_ZTV10polynomialIdE, i32 0, i32 2), i32 (...)*** %0, align 4 - %1 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - store double* null, double** %1, align 4 - %2 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - store i32 %degree, i32* %2, align 4 - call void @llvm.dbg.stoppoint(i32 215, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdE7acquireEv(%"struct.polynomial<double>"* %this) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram713 to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialIdEC1ERKS0_(%"struct.polynomial<double>"* %this, %"struct.polynomial<double>"* %source) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram717 to %0*)) - call void @llvm.dbg.stoppoint(i32 244, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 0 ; <i32 (...)***> [#uses=1] - store i32 (...)** getelementptr ([4 x i32 (...)*]* @_ZTV10polynomialIdE, i32 0, i32 2), i32 (...)*** %0, align 4 - %1 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - store double* null, double** %1, align 4 - %2 = getelementptr %"struct.polynomial<double>"* %source, i32 0, i32 2 ; <i32*> [#uses=1] - %3 = load i32* %2, align 4 ; <i32> [#uses=1] - %4 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - store i32 %3, i32* %4, align 4 - call void @llvm.dbg.stoppoint(i32 246, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdE7acquireEv(%"struct.polynomial<double>"* %this) - call void @llvm.dbg.stoppoint(i32 247, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %5 = getelementptr %"struct.polynomial<double>"* %source, i32 0, i32 1 ; <double**> [#uses=1] - %6 = load double** %5, align 4 ; <double*> [#uses=1] - call void @_ZN10polynomialIdE9deep_copyEPKd(%"struct.polynomial<double>"* %this, double* %6) nounwind - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram717 to %0*)) - ret void -} - -define linkonce %"struct.polynomial<double>"* @_ZN10polynomialIdE7stretchEj(%"struct.polynomial<double>"* %this, i32 %degrees) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram720 to %0*)) - call void @llvm.dbg.stoppoint(i32 278, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = icmp eq i32 %degrees, 0 ; <i1> [#uses=1] - br i1 %0, label %return, label %bb - -bb: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 280, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %1 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - %2 = load double** %1, align 4 ; <double*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 281, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %3 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %4 = load i32* %3, align 4 ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 283, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %5 = add i32 %4, %degrees ; <i32> [#uses=1] - %6 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - store i32 %5, i32* %6, align 4 - call void @llvm.dbg.stoppoint(i32 284, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdE7acquireEv(%"struct.polynomial<double>"* %this) - call void @llvm.dbg.stoppoint(i32 286, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb2 - -bb1: ; preds = %bb2 - call void @llvm.dbg.stoppoint(i32 289, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %7 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - %8 = load double** %7, align 4 ; <double*> [#uses=1] - %9 = getelementptr double* %2, i32 %n.0 ; <double*> [#uses=1] - %10 = load double* %9, align 1 ; <double> [#uses=1] - %11 = getelementptr double* %8, i32 %n.0 ; <double*> [#uses=1] - store double %10, double* %11, align 1 - call void @llvm.dbg.stoppoint(i32 288, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %12 = add i32 %n.0, 1 ; <i32> [#uses=1] - br label %bb2 - -bb2: ; preds = %bb1, %bb - %n.0 = phi i32 [ 0, %bb ], [ %12, %bb1 ] ; <i32> [#uses=5] - call void @llvm.dbg.stoppoint(i32 288, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %13 = icmp ult i32 %n.0, %4 ; <i1> [#uses=1] - br i1 %13, label %bb1, label %bb5 - -bb4: ; preds = %bb5 - call void @llvm.dbg.stoppoint(i32 292, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %14 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 1 ; <double**> [#uses=1] - %15 = load double** %14, align 4 ; <double*> [#uses=1] - %16 = getelementptr double* %15, i32 %n.1 ; <double*> [#uses=1] - store double 0.000000e+00, double* %16, align 1 - call void @llvm.dbg.stoppoint(i32 291, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %17 = add i32 %n.1, 1 ; <i32> [#uses=1] - br label %bb5 - -bb5: ; preds = %bb4, %bb2 - %n.1 = phi i32 [ %17, %bb4 ], [ %n.0, %bb2 ] ; <i32> [#uses=3] - call void @llvm.dbg.stoppoint(i32 291, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %18 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %19 = load i32* %18, align 4 ; <i32> [#uses=1] - %20 = icmp ugt i32 %19, %n.1 ; <i1> [#uses=1] - br i1 %20, label %bb4, label %return - -return: ; preds = %bb5, %entry - call void @llvm.dbg.stoppoint(i32 295, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram720 to %0*)) - ret %"struct.polynomial<double>"* %this -} - -define linkonce %"struct.polynomial<double>"* @_ZN10polynomialIdEaSERKS0_(%"struct.polynomial<double>"* %this, %"struct.polynomial<double>"* %source) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram729 to %0*)) - call void @llvm.dbg.stoppoint(i32 261, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %1 = load i32* %0, align 4 ; <i32> [#uses=1] - %2 = getelementptr %"struct.polynomial<double>"* %source, i32 0, i32 2 ; <i32*> [#uses=1] - %3 = load i32* %2, align 4 ; <i32> [#uses=1] - %4 = icmp eq i32 %1, %3 ; <i1> [#uses=1] - br i1 %4, label %bb1, label %bb - -bb: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 263, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdE7releaseEv(%"struct.polynomial<double>"* %this) nounwind - call void @llvm.dbg.stoppoint(i32 265, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %5 = getelementptr %"struct.polynomial<double>"* %source, i32 0, i32 2 ; <i32*> [#uses=1] - %6 = load i32* %5, align 4 ; <i32> [#uses=1] - %7 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - store i32 %6, i32* %7, align 4 - call void @llvm.dbg.stoppoint(i32 266, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdE7acquireEv(%"struct.polynomial<double>"* %this) - br label %bb1 - -bb1: ; preds = %bb, %entry - call void @llvm.dbg.stoppoint(i32 269, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %8 = getelementptr %"struct.polynomial<double>"* %source, i32 0, i32 1 ; <double**> [#uses=1] - %9 = load double** %8, align 4 ; <double*> [#uses=1] - call void @_ZN10polynomialIdE9deep_copyEPKd(%"struct.polynomial<double>"* %this, double* %9) nounwind - call void @llvm.dbg.stoppoint(i32 271, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram729 to %0*)) - ret %"struct.polynomial<double>"* %this -} - -define linkonce void @_ZN10polynomialISt7complexIdEE7acquireEv(%"struct.polynomial<std::complex<double> >"* %this) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram732 to %0*)) - call void @llvm.dbg.stoppoint(i32 183, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %1 = load i32* %0, align 4 ; <i32> [#uses=2] - %2 = shl i32 %1, 4 ; <i32> [#uses=1] - %3 = call i8* @_Znaj(i32 %2) ; <i8*> [#uses=1] - %4 = bitcast i8* %3 to %"struct.std::complex<double>"* ; <%"struct.std::complex<double>"*> [#uses=2] - br label %bb1 - -bb: ; preds = %bb1 - call void @llvm.dbg.stoppoint(i32 183, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %.0, double 0.000000e+00, double 0.000000e+00) nounwind - %5 = getelementptr %"struct.std::complex<double>"* %.0, i32 1 ; <%"struct.std::complex<double>"*> [#uses=1] - br label %bb1 - -bb1: ; preds = %bb, %entry - %.01.in = phi i32 [ %1, %entry ], [ %.01, %bb ] ; <i32> [#uses=1] - %.0 = phi %"struct.std::complex<double>"* [ %4, %entry ], [ %5, %bb ] ; <%"struct.std::complex<double>"*> [#uses=2] - %.01 = add i32 %.01.in, -1 ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 183, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %6 = icmp eq i32 %.01, -1 ; <i1> [#uses=1] - br i1 %6, label %bb2, label %bb - -bb2: ; preds = %bb1 - call void @llvm.dbg.stoppoint(i32 183, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %7 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 1 ; <%"struct.std::complex<double>"**> [#uses=1] - store %"struct.std::complex<double>"* %4, %"struct.std::complex<double>"** %7, align 4 - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram732 to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialISt7complexIdEEC1Ej(%"struct.polynomial<std::complex<double> >"* %this, i32 %degree) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram735 to %0*)) - call void @llvm.dbg.stoppoint(i32 213, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 0 ; <i32 (...)***> [#uses=1] - store i32 (...)** getelementptr ([4 x i32 (...)*]* @_ZTV10polynomialISt7complexIdEE, i32 0, i32 2), i32 (...)*** %0, align 4 - %1 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 1 ; <%"struct.std::complex<double>"**> [#uses=1] - store %"struct.std::complex<double>"* null, %"struct.std::complex<double>"** %1, align 4 - %2 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - store i32 %degree, i32* %2, align 4 - call void @llvm.dbg.stoppoint(i32 215, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialISt7complexIdEE7acquireEv(%"struct.polynomial<std::complex<double> >"* %this) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram735 to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialIdE11bit_reverseERKS0_(%"struct.polynomial<std::complex<double> >"* noalias sret %agg.result, %"struct.polynomial<double>"* %poly) { -entry: - %result = alloca %"struct.polynomial<std::complex<double> >", align 8 ; <%"struct.polynomial<std::complex<double> >"*> [#uses=1] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram738 to %0*)) - %0 = bitcast %"struct.polynomial<std::complex<double> >"* %result to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %0, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable744 to %0*)) - call void @llvm.dbg.stoppoint(i32 471, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %1 = call i32 @_ZNK10polynomialIdE6degreeEv(%"struct.polynomial<double>"* %poly) nounwind ; <i32> [#uses=1] - %2 = call i32 @_ZN10polynomialIdE4log2Ej(i32 %1) nounwind ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 473, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %3 = call i32 @_ZNK10polynomialIdE6degreeEv(%"struct.polynomial<double>"* %poly) nounwind ; <i32> [#uses=1] - call void @_ZN10polynomialISt7complexIdEEC1Ej(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %3) - call void @llvm.dbg.stoppoint(i32 475, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb6 - -bb: ; preds = %bb6 - call void @llvm.dbg.stoppoint(i32 476, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %4 = call double @_ZNK10polynomialIdE3getEj(%"struct.polynomial<double>"* %poly, i32 %n.0) nounwind ; <double> [#uses=1] - %5 = call i32 @_ZN10polynomialIdE9flip_bitsEjj(i32 %n.0, i32 %2) nounwind ; <i32> [#uses=1] - %6 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %5) nounwind ; <%"struct.std::complex<double>"*> [#uses=1] - %7 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEaSEd(%"struct.std::complex<double>"* %6, double %4) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.stoppoint(i32 475, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %8 = add i32 %n.0, 1 ; <i32> [#uses=1] - br label %bb6 - -bb6: ; preds = %bb, %entry - %n.0 = phi i32 [ 0, %entry ], [ %8, %bb ] ; <i32> [#uses=4] - call void @llvm.dbg.stoppoint(i32 475, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %9 = call i32 @_ZNK10polynomialIdE6degreeEv(%"struct.polynomial<double>"* %poly) nounwind ; <i32> [#uses=1] - %10 = icmp ugt i32 %9, %n.0 ; <i1> [#uses=1] - br i1 %10, label %bb, label %return - -return: ; preds = %bb6 - call void @llvm.dbg.stoppoint(i32 475, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram738 to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialIdE11bit_reverseERKS_ISt7complexIdEE(%"struct.polynomial<std::complex<double> >"* noalias sret %agg.result, %"struct.polynomial<std::complex<double> >"* %poly) { -entry: - %result = alloca %"struct.polynomial<std::complex<double> >", align 8 ; <%"struct.polynomial<std::complex<double> >"*> [#uses=1] - %memtmp7 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=3] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram747 to %0*)) - %0 = bitcast %"struct.polynomial<std::complex<double> >"* %result to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %0, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable751 to %0*)) - call void @llvm.dbg.stoppoint(i32 485, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %1 = call i32 @_ZNK10polynomialISt7complexIdEE6degreeEv(%"struct.polynomial<std::complex<double> >"* %poly) nounwind ; <i32> [#uses=1] - %2 = call i32 @_ZN10polynomialIdE4log2Ej(i32 %1) nounwind ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 487, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %3 = call i32 @_ZNK10polynomialISt7complexIdEE6degreeEv(%"struct.polynomial<std::complex<double> >"* %poly) nounwind ; <i32> [#uses=1] - call void @_ZN10polynomialISt7complexIdEEC1Ej(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %3) - call void @llvm.dbg.stoppoint(i32 489, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb8 - -bb: ; preds = %bb8 - call void @llvm.dbg.stoppoint(i32 490, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %4 = call i32 @_ZN10polynomialIdE9flip_bitsEjj(i32 %n.0, i32 %2) nounwind ; <i32> [#uses=1] - %5 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %4) nounwind ; <%"struct.std::complex<double>"*> [#uses=2] - call void @_ZNK10polynomialISt7complexIdEE3getEj(%"struct.std::complex<double>"* noalias sret %memtmp7, %"struct.polynomial<std::complex<double> >"* %poly, i32 %n.0) nounwind - %6 = getelementptr %"struct.std::complex<double>"* %5, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %7 = getelementptr %"struct.std::complex<double>"* %memtmp7, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %8 = load double* %7, align 8 ; <double> [#uses=1] - store double %8, double* %6, align 4 - %9 = getelementptr %"struct.std::complex<double>"* %5, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %10 = getelementptr %"struct.std::complex<double>"* %memtmp7, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %11 = load double* %10, align 8 ; <double> [#uses=1] - store double %11, double* %9, align 4 - call void @llvm.dbg.stoppoint(i32 489, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %12 = add i32 %n.0, 1 ; <i32> [#uses=1] - br label %bb8 - -bb8: ; preds = %bb, %entry - %n.0 = phi i32 [ 0, %entry ], [ %12, %bb ] ; <i32> [#uses=4] - call void @llvm.dbg.stoppoint(i32 489, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %13 = call i32 @_ZNK10polynomialISt7complexIdEE6degreeEv(%"struct.polynomial<std::complex<double> >"* %poly) nounwind ; <i32> [#uses=1] - %14 = icmp ugt i32 %13, %n.0 ; <i1> [#uses=1] - br i1 %14, label %bb, label %return - -return: ; preds = %bb8 - call void @llvm.dbg.stoppoint(i32 489, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram747 to %0*)) - ret void -} - -define linkonce %"struct.polynomial<std::complex<double> >"* @_ZN10polynomialISt7complexIdEEaSERKS2_(%"struct.polynomial<std::complex<double> >"* %this, %"struct.polynomial<std::complex<double> >"* %source) { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram753 to %0*)) - call void @llvm.dbg.stoppoint(i32 261, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %0 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %1 = load i32* %0, align 4 ; <i32> [#uses=1] - %2 = getelementptr %"struct.polynomial<std::complex<double> >"* %source, i32 0, i32 2 ; <i32*> [#uses=1] - %3 = load i32* %2, align 4 ; <i32> [#uses=1] - %4 = icmp eq i32 %1, %3 ; <i1> [#uses=1] - br i1 %4, label %bb1, label %bb - -bb: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 263, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialISt7complexIdEE7releaseEv(%"struct.polynomial<std::complex<double> >"* %this) nounwind - call void @llvm.dbg.stoppoint(i32 265, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %5 = getelementptr %"struct.polynomial<std::complex<double> >"* %source, i32 0, i32 2 ; <i32*> [#uses=1] - %6 = load i32* %5, align 4 ; <i32> [#uses=1] - %7 = getelementptr %"struct.polynomial<std::complex<double> >"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - store i32 %6, i32* %7, align 4 - call void @llvm.dbg.stoppoint(i32 266, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialISt7complexIdEE7acquireEv(%"struct.polynomial<std::complex<double> >"* %this) - br label %bb1 - -bb1: ; preds = %bb, %entry - call void @llvm.dbg.stoppoint(i32 269, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %8 = getelementptr %"struct.polynomial<std::complex<double> >"* %source, i32 0, i32 1 ; <%"struct.std::complex<double>"**> [#uses=1] - %9 = load %"struct.std::complex<double>"** %8, align 4 ; <%"struct.std::complex<double>"*> [#uses=1] - call void @_ZN10polynomialISt7complexIdEE9deep_copyEPKS1_(%"struct.polynomial<std::complex<double> >"* %this, %"struct.std::complex<double>"* %9) nounwind - call void @llvm.dbg.stoppoint(i32 271, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram753 to %0*)) - ret %"struct.polynomial<std::complex<double> >"* %this -} - -define linkonce i32 @_ZN10polynomialIdE11stretch_fftEv(%"struct.polynomial<double>"* %this) { -entry: - %0 = alloca %"struct.std::allocator<char>", align 8 ; <%"struct.std::allocator<char>"*> [#uses=4] - %1 = alloca %"struct.std::string", align 8 ; <%"struct.std::string"*> [#uses=4] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram756 to %0*)) - call void @llvm.dbg.stoppoint(i32 445, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb - -bb: ; preds = %bb1, %entry - %n.0 = phi i32 [ 1, %entry ], [ %5, %bb1 ] ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 449, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %2 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %3 = load i32* %2, align 4 ; <i32> [#uses=1] - %4 = icmp ugt i32 %3, %n.0 ; <i1> [#uses=1] - %5 = shl i32 %n.0, 1 ; <i32> [#uses=4] - br i1 %4, label %bb1, label %bb17 - -bb1: ; preds = %bb - call void @llvm.dbg.stoppoint(i32 454, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %6 = icmp eq i32 %5, 0 ; <i1> [#uses=1] - br i1 %6, label %bb2, label %bb - -bb2: ; preds = %bb1 - call void @llvm.dbg.stoppoint(i32 455, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZNSaIcEC1Ev(%"struct.std::allocator<char>"* %0) nounwind - invoke void @_ZNSsC1EPKcRKSaIcE(%"struct.std::string"* %1, i8* getelementptr ([35 x i8]* @.str759, i32 0, i32 0), %"struct.std::allocator<char>"* %0) - to label %invcont unwind label %lpad - -invcont: ; preds = %bb2 - call void @llvm.dbg.stoppoint(i32 455, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %7 = call i8* @__cxa_allocate_exception(i32 8) nounwind ; <i8*> [#uses=3] - %8 = bitcast i8* %7 to %"struct.std::overflow_error"* ; <%"struct.std::overflow_error"*> [#uses=1] - invoke void @_ZNSt14overflow_errorC1ERKSs(%"struct.std::overflow_error"* %8, %"struct.std::string"* %1) - to label %invcont3 unwind label %lpad23 - -invcont3: ; preds = %invcont - call void @llvm.dbg.stoppoint(i32 455, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZNSsD1Ev(%"struct.std::string"* %1) - to label %bb11 unwind label %lpad31 - -bb11: ; preds = %invcont3 - call void @llvm.dbg.stoppoint(i32 455, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZNSaIcED1Ev(%"struct.std::allocator<char>"* %0) nounwind - call void @__cxa_throw(i8* %7, i8* bitcast (%struct.__si_class_type_info_pseudo* @_ZTISt14overflow_error to i8*), void (i8*)* bitcast (void (%"struct.std::overflow_error"*)* @_ZNSt14overflow_errorD1Ev to void (i8*)*)) noreturn - unreachable - -bb17: ; preds = %bb - call void @llvm.dbg.stoppoint(i32 459, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %9 = getelementptr %"struct.polynomial<double>"* %this, i32 0, i32 2 ; <i32*> [#uses=1] - %10 = load i32* %9, align 4 ; <i32> [#uses=2] - %11 = sub i32 %5, %10 ; <i32> [#uses=3] - call void @llvm.dbg.stoppoint(i32 461, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %12 = icmp eq i32 %5, %10 ; <i1> [#uses=1] - br i1 %12, label %return, label %bb18 - -bb18: ; preds = %bb17 - call void @llvm.dbg.stoppoint(i32 462, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %13 = call %"struct.polynomial<double>"* @_ZN10polynomialIdE7stretchEj(%"struct.polynomial<double>"* %this, i32 %11) ; <%"struct.polynomial<double>"*> [#uses=0] - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram756 to %0*)) - ret i32 %11 - -return: ; preds = %bb17 - call void @llvm.dbg.stoppoint(i32 464, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - ret i32 %11 - -lpad: ; preds = %bb2 - %eh_ptr = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select22 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - br label %ppad - -lpad23: ; preds = %invcont - %eh_ptr24 = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select26 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr24, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 455, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @__cxa_free_exception(i8* %7) nounwind - call void @llvm.dbg.stoppoint(i32 455, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZNSsD1Ev(%"struct.std::string"* %1) - to label %ppad unwind label %lpad27 - -lpad27: ; preds = %lpad23 - %eh_ptr28 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select30 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr28, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 455, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -lpad31: ; preds = %invcont3 - %eh_ptr32 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select34 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr32, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 455, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -ppad: ; preds = %lpad23, %lpad - %eh_exception.0 = phi i8* [ %eh_ptr, %lpad ], [ %eh_ptr24, %lpad23 ] ; <i8*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 455, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZNSaIcED1Ev(%"struct.std::allocator<char>"* %0) nounwind - call void @_Unwind_Resume(i8* %eh_exception.0) - unreachable -} - -declare void @_ZNSaIcEC1Ev(%"struct.std::allocator<char>"*) nounwind - -declare void @_ZNSsC1EPKcRKSaIcE(%"struct.std::string"*, i8*, %"struct.std::allocator<char>"*) - -declare i8* @__cxa_allocate_exception(i32) nounwind - -declare void @_ZNSt14overflow_errorC1ERKSs(%"struct.std::overflow_error"*, %"struct.std::string"*) - -declare void @_ZNSsD1Ev(%"struct.std::string"*) - -declare i8* @llvm.eh.exception() nounwind - -declare i32 @llvm.eh.selector.i32(i8*, i8*, ...) nounwind - -declare void @__cxa_free_exception(i8*) nounwind - -declare void @_ZSt9terminatev() noreturn nounwind - -declare void @_ZNSaIcED1Ev(%"struct.std::allocator<char>"*) nounwind - -declare void @__cxa_throw(i8*, i8*, void (i8*)*) noreturn - -define linkonce void @_ZNSt14overflow_errorD1Ev(%"struct.std::overflow_error"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1727 to %0*)) - call void @llvm.dbg.stoppoint(i32 134, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*)) - %0 = getelementptr %"struct.std::overflow_error"* %this, i32 0, i32 0, i32 0, i32 0 ; <i32 (...)***> [#uses=1] - store i32 (...)** getelementptr ([5 x i32 (...)*]* @_ZTVSt14overflow_error, i32 0, i32 2), i32 (...)*** %0, align 4 - %1 = getelementptr %"struct.std::overflow_error"* %this, i32 0, i32 0 ; <%"struct.std::runtime_error"*> [#uses=1] - call void @_ZNSt13runtime_errorD2Ev(%"struct.std::runtime_error"* %1) nounwind - call void @llvm.dbg.stoppoint(i32 134, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*)) - ret void -} - -declare i32 @__gxx_personality_v0(...) - -declare void @_Unwind_Resume(i8*) - -define linkonce void @_ZNSt14overflow_errorD0Ev(%"struct.std::overflow_error"* %this) nounwind { -entry: - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1736 to %0*)) - call void @llvm.dbg.stoppoint(i32 134, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*)) - %0 = getelementptr %"struct.std::overflow_error"* %this, i32 0, i32 0, i32 0, i32 0 ; <i32 (...)***> [#uses=1] - store i32 (...)** getelementptr ([5 x i32 (...)*]* @_ZTVSt14overflow_error, i32 0, i32 2), i32 (...)*** %0, align 4 - %1 = getelementptr %"struct.std::overflow_error"* %this, i32 0, i32 0 ; <%"struct.std::runtime_error"*> [#uses=1] - call void @_ZNSt13runtime_errorD2Ev(%"struct.std::runtime_error"* %1) nounwind - call void @llvm.dbg.stoppoint(i32 134, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit770 to %0*)) - %2 = bitcast %"struct.std::overflow_error"* %this to i8* ; <i8*> [#uses=1] - call void @_ZdlPv(i8* %2) nounwind - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1736 to %0*)) - ret void -} - -declare i8* @_ZNKSt13runtime_error4whatEv(%"struct.std::runtime_error"*) nounwind - -declare void @_ZNSt13runtime_errorD2Ev(%"struct.std::runtime_error"*) nounwind - -define linkonce void @_ZSt13__complex_expCd(%1* noalias sret %agg.result, %1 %__z) nounwind { -entry: - %0 = alloca %1, align 8 ; <%1*> [#uses=4] - %memtmp = alloca %1, align 8 ; <%1*> [#uses=3] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1742 to %0*)) - call void @llvm.dbg.stoppoint(i32 730, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - call void @cexp(%1* noalias sret %memtmp, %1 %__z) nounwind - %1 = getelementptr %1* %0, i32 0, i32 0 ; <double*> [#uses=1] - %2 = getelementptr %1* %memtmp, i32 0, i32 0 ; <double*> [#uses=1] - %3 = load double* %2, align 8 ; <double> [#uses=1] - store double %3, double* %1, align 8 - %4 = getelementptr %1* %0, i32 0, i32 1 ; <double*> [#uses=1] - %5 = getelementptr %1* %memtmp, i32 0, i32 1 ; <double*> [#uses=1] - %6 = load double* %5, align 8 ; <double> [#uses=1] - store double %6, double* %4, align 8 - %7 = getelementptr %1* %agg.result, i32 0, i32 0 ; <double*> [#uses=1] - %8 = getelementptr %1* %0, i32 0, i32 0 ; <double*> [#uses=1] - %9 = load double* %8, align 8 ; <double> [#uses=1] - store double %9, double* %7, align 8 - %10 = getelementptr %1* %agg.result, i32 0, i32 1 ; <double*> [#uses=1] - %11 = getelementptr %1* %0, i32 0, i32 1 ; <double*> [#uses=1] - %12 = load double* %11, align 8 ; <double> [#uses=1] - store double %12, double* %10, align 8 - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1742 to %0*)) - ret void -} - -declare void @cexp(%1* noalias sret, %1) nounwind - -define linkonce void @_ZSt3expIdESt7complexIT_ERKS2_(%"struct.std::complex<double>"* noalias sret %agg.result, %"struct.std::complex<double>"* %__z) nounwind { -entry: - %0 = alloca %1, align 8 ; <%1*> [#uses=3] - %1 = alloca %1, align 8 ; <%1*> [#uses=3] - %memtmp = alloca %1, align 8 ; <%1*> [#uses=3] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1746 to %0*)) - call void @llvm.dbg.stoppoint(i32 738, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit5 to %0*)) - %2 = call %1* @_ZNKSt7complexIdE5__repEv(%"struct.std::complex<double>"* %__z) nounwind ; <%1*> [#uses=2] - %3 = getelementptr %1* %1, i32 0, i32 0 ; <double*> [#uses=1] - %4 = getelementptr %1* %2, i32 0, i32 0 ; <double*> [#uses=1] - %5 = load double* %4, align 8 ; <double> [#uses=1] - store double %5, double* %3, align 8 - %6 = getelementptr %1* %1, i32 0, i32 1 ; <double*> [#uses=1] - %7 = getelementptr %1* %2, i32 0, i32 1 ; <double*> [#uses=1] - %8 = load double* %7, align 8 ; <double> [#uses=1] - store double %8, double* %6, align 8 - %9 = load %1* %1, align 8 ; <%1> [#uses=1] - call void @_ZSt13__complex_expCd(%1* noalias sret %memtmp, %1 %9) nounwind - %10 = getelementptr %1* %0, i32 0, i32 0 ; <double*> [#uses=1] - %11 = getelementptr %1* %memtmp, i32 0, i32 0 ; <double*> [#uses=1] - %12 = load double* %11, align 8 ; <double> [#uses=1] - store double %12, double* %10, align 8 - %13 = getelementptr %1* %0, i32 0, i32 1 ; <double*> [#uses=1] - %14 = getelementptr %1* %memtmp, i32 0, i32 1 ; <double*> [#uses=1] - %15 = load double* %14, align 8 ; <double> [#uses=1] - store double %15, double* %13, align 8 - %16 = load %1* %0, align 8 ; <%1> [#uses=1] - call void @_ZNSt7complexIdEC1ECd(%"struct.std::complex<double>"* %agg.result, %1 %16) nounwind - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1746 to %0*)) - ret void -} - -define linkonce void @_ZN10polynomialIdE3fftERKS0_(%"struct.polynomial<std::complex<double> >"* noalias sret %agg.result, %"struct.polynomial<double>"* %poly) { -entry: - %result = alloca %"struct.polynomial<std::complex<double> >", align 8 ; <%"struct.polynomial<std::complex<double> >"*> [#uses=1] - %u = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=6] - %t = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=6] - %w = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=5] - %wm = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=5] - %0 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=2] - %1 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=2] - %memtmp20 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=3] - %memtmp23 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=3] - %memtmp24 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=3] - %memtmp26 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=3] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1748 to %0*)) - %2 = bitcast %"struct.polynomial<std::complex<double> >"* %result to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %2, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1751 to %0*)) - %3 = bitcast %"struct.std::complex<double>"* %u to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %3, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1753 to %0*)) - %4 = bitcast %"struct.std::complex<double>"* %t to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %4, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1755 to %0*)) - %5 = bitcast %"struct.std::complex<double>"* %w to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %5, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1757 to %0*)) - %6 = bitcast %"struct.std::complex<double>"* %wm to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %6, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1759 to %0*)) - call void @llvm.dbg.stoppoint(i32 499, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %7 = call i32 @_ZNK10polynomialIdE6degreeEv(%"struct.polynomial<double>"* %poly) nounwind ; <i32> [#uses=1] - %8 = call i32 @_ZN10polynomialIdE4log2Ej(i32 %7) nounwind ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 501, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %wm, double 0.000000e+00, double 0.000000e+00) nounwind - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %w, double 0.000000e+00, double 0.000000e+00) nounwind - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %t, double 0.000000e+00, double 0.000000e+00) nounwind - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %u, double 0.000000e+00, double 0.000000e+00) nounwind - call void @llvm.dbg.stoppoint(i32 503, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdE11bit_reverseERKS0_(%"struct.polynomial<std::complex<double> >"* noalias sret %agg.result, %"struct.polynomial<double>"* %poly) - call void @llvm.dbg.stoppoint(i32 508, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb32 - -bb: ; preds = %bb32 - call void @llvm.dbg.stoppoint(i32 510, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %9 = uitofp i32 %m.0 to double ; <double> [#uses=1] - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %0, double %9, double 0.000000e+00) nounwind - invoke void @_ZStdvIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %1, %"struct.std::complex<double>"* @_ZN10polynomialIdE4PI2IE, %"struct.std::complex<double>"* %0) - to label %invcont unwind label %lpad - -invcont: ; preds = %bb - call void @llvm.dbg.stoppoint(i32 510, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt3expIdESt7complexIT_ERKS2_(%"struct.std::complex<double>"* noalias sret %memtmp20, %"struct.std::complex<double>"* %1) nounwind - %10 = getelementptr %"struct.std::complex<double>"* %wm, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %11 = getelementptr %"struct.std::complex<double>"* %memtmp20, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %12 = load double* %11, align 8 ; <double> [#uses=1] - store double %12, double* %10, align 8 - %13 = getelementptr %"struct.std::complex<double>"* %wm, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %14 = getelementptr %"struct.std::complex<double>"* %memtmp20, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %15 = load double* %14, align 8 ; <double> [#uses=1] - store double %15, double* %13, align 8 - call void @llvm.dbg.stoppoint(i32 511, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %16 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEaSEd(%"struct.std::complex<double>"* %w, double 1.000000e+00) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.stoppoint(i32 513, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb30 - -bb22: ; preds = %bb28 - call void @llvm.dbg.stoppoint(i32 517, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %17 = add i32 %k.0, %m2.0 ; <i32> [#uses=1] - %18 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %17) nounwind ; <%"struct.std::complex<double>"*> [#uses=1] - call void @_ZStmlIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %memtmp23, %"struct.std::complex<double>"* %w, %"struct.std::complex<double>"* %18) nounwind - %19 = getelementptr %"struct.std::complex<double>"* %t, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %20 = getelementptr %"struct.std::complex<double>"* %memtmp23, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %21 = load double* %20, align 8 ; <double> [#uses=1] - store double %21, double* %19, align 8 - %22 = getelementptr %"struct.std::complex<double>"* %t, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %23 = getelementptr %"struct.std::complex<double>"* %memtmp23, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %24 = load double* %23, align 8 ; <double> [#uses=1] - store double %24, double* %22, align 8 - call void @llvm.dbg.stoppoint(i32 518, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %25 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %k.0) nounwind ; <%"struct.std::complex<double>"*> [#uses=2] - %26 = getelementptr %"struct.std::complex<double>"* %u, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %27 = getelementptr %"struct.std::complex<double>"* %25, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %28 = load double* %27, align 4 ; <double> [#uses=1] - store double %28, double* %26, align 8 - %29 = getelementptr %"struct.std::complex<double>"* %u, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %30 = getelementptr %"struct.std::complex<double>"* %25, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %31 = load double* %30, align 4 ; <double> [#uses=1] - store double %31, double* %29, align 8 - call void @llvm.dbg.stoppoint(i32 519, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %32 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %k.0) nounwind ; <%"struct.std::complex<double>"*> [#uses=2] - invoke void @_ZStplIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %memtmp24, %"struct.std::complex<double>"* %u, %"struct.std::complex<double>"* %t) - to label %invcont25 unwind label %lpad - -invcont25: ; preds = %bb22 - %33 = getelementptr %"struct.std::complex<double>"* %32, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %34 = getelementptr %"struct.std::complex<double>"* %memtmp24, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %35 = load double* %34, align 8 ; <double> [#uses=1] - store double %35, double* %33, align 4 - %36 = getelementptr %"struct.std::complex<double>"* %32, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %37 = getelementptr %"struct.std::complex<double>"* %memtmp24, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %38 = load double* %37, align 8 ; <double> [#uses=1] - store double %38, double* %36, align 4 - call void @llvm.dbg.stoppoint(i32 520, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %39 = add i32 %k.0, %m2.0 ; <i32> [#uses=1] - %40 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %39) nounwind ; <%"struct.std::complex<double>"*> [#uses=2] - invoke void @_ZStmiIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %memtmp26, %"struct.std::complex<double>"* %u, %"struct.std::complex<double>"* %t) - to label %invcont27 unwind label %lpad - -invcont27: ; preds = %invcont25 - %41 = getelementptr %"struct.std::complex<double>"* %40, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %42 = getelementptr %"struct.std::complex<double>"* %memtmp26, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %43 = load double* %42, align 8 ; <double> [#uses=1] - store double %43, double* %41, align 4 - %44 = getelementptr %"struct.std::complex<double>"* %40, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %45 = getelementptr %"struct.std::complex<double>"* %memtmp26, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %46 = load double* %45, align 8 ; <double> [#uses=1] - store double %46, double* %44, align 4 - call void @llvm.dbg.stoppoint(i32 515, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %47 = add i32 %k.0, %m.0 ; <i32> [#uses=1] - br label %bb28 - -bb28: ; preds = %bb30, %invcont27 - %k.0 = phi i32 [ %47, %invcont27 ], [ %j.0, %bb30 ] ; <i32> [#uses=6] - call void @llvm.dbg.stoppoint(i32 515, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %48 = call i32 @_ZNK10polynomialIdE6degreeEv(%"struct.polynomial<double>"* %poly) nounwind ; <i32> [#uses=1] - %49 = add i32 %48, -1 ; <i32> [#uses=1] - %50 = icmp ult i32 %49, %k.0 ; <i1> [#uses=1] - br i1 %50, label %bb29, label %bb22 - -bb29: ; preds = %bb28 - call void @llvm.dbg.stoppoint(i32 523, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %51 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEmLIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %w, %"struct.std::complex<double>"* %wm) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.stoppoint(i32 513, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %52 = add i32 %j.0, 1 ; <i32> [#uses=1] - br label %bb30 - -bb30: ; preds = %bb29, %invcont - %j.0 = phi i32 [ 0, %invcont ], [ %52, %bb29 ] ; <i32> [#uses=3] - call void @llvm.dbg.stoppoint(i32 513, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %53 = add i32 %m2.0, -1 ; <i32> [#uses=1] - %54 = icmp ult i32 %53, %j.0 ; <i1> [#uses=1] - br i1 %54, label %bb31, label %bb28 - -bb31: ; preds = %bb30 - call void @llvm.dbg.stoppoint(i32 526, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %55 = shl i32 %m.0, 1 ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 527, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %56 = shl i32 %m2.0, 1 ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 508, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %57 = add i32 %s.0, 1 ; <i32> [#uses=1] - br label %bb32 - -bb32: ; preds = %bb31, %entry - %m.0 = phi i32 [ 2, %entry ], [ %55, %bb31 ] ; <i32> [#uses=3] - %m2.0 = phi i32 [ 1, %entry ], [ %56, %bb31 ] ; <i32> [#uses=4] - %s.0 = phi i32 [ 0, %entry ], [ %57, %bb31 ] ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 508, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %58 = icmp ult i32 %s.0, %8 ; <i1> [#uses=1] - br i1 %58, label %bb, label %return - -return: ; preds = %bb32 - call void @llvm.dbg.stoppoint(i32 530, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1748 to %0*)) - ret void - -lpad: ; preds = %invcont25, %bb22, %bb - %eh_ptr = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select40 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 530, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialISt7complexIdEED1Ev(%"struct.polynomial<std::complex<double> >"* %agg.result) - to label %Unwind unwind label %lpad41 - -lpad41: ; preds = %lpad - %eh_ptr42 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select44 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr42, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 530, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -Unwind: ; preds = %lpad - call void @_Unwind_Resume(i8* %eh_ptr) - unreachable -} - -define linkonce void @_ZN10polynomialIdE11inverse_fftERKS_ISt7complexIdEE(%"struct.polynomial<std::complex<double> >"* noalias sret %agg.result, %"struct.polynomial<std::complex<double> >"* %poly) { -entry: - %result = alloca %"struct.polynomial<std::complex<double> >", align 8 ; <%"struct.polynomial<std::complex<double> >"*> [#uses=1] - %u = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=6] - %t = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=6] - %w = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=5] - %wm = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=5] - %0 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=2] - %1 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=2] - %2 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=2] - %memtmp22 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=3] - %memtmp25 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=3] - %memtmp26 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=3] - %memtmp28 = alloca %"struct.std::complex<double>", align 8 ; <%"struct.std::complex<double>"*> [#uses=3] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1771 to %0*)) - %3 = bitcast %"struct.polynomial<std::complex<double> >"* %result to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %3, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1774 to %0*)) - %4 = bitcast %"struct.std::complex<double>"* %u to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %4, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1775 to %0*)) - %5 = bitcast %"struct.std::complex<double>"* %t to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %5, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1776 to %0*)) - %6 = bitcast %"struct.std::complex<double>"* %w to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %6, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1777 to %0*)) - %7 = bitcast %"struct.std::complex<double>"* %wm to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %7, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1778 to %0*)) - call void @llvm.dbg.stoppoint(i32 537, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %8 = call i32 @_ZNK10polynomialISt7complexIdEE6degreeEv(%"struct.polynomial<std::complex<double> >"* %poly) nounwind ; <i32> [#uses=1] - %9 = call i32 @_ZN10polynomialIdE4log2Ej(i32 %8) nounwind ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 539, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %wm, double 0.000000e+00, double 0.000000e+00) nounwind - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %w, double 0.000000e+00, double 0.000000e+00) nounwind - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %t, double 0.000000e+00, double 0.000000e+00) nounwind - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %u, double 0.000000e+00, double 0.000000e+00) nounwind - call void @llvm.dbg.stoppoint(i32 541, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdE11bit_reverseERKS_ISt7complexIdEE(%"struct.polynomial<std::complex<double> >"* noalias sret %agg.result, %"struct.polynomial<std::complex<double> >"* %poly) - call void @llvm.dbg.stoppoint(i32 546, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb34 - -bb: ; preds = %bb34 - call void @llvm.dbg.stoppoint(i32 548, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %10 = uitofp i32 %m.0 to double ; <double> [#uses=1] - call void @_ZNSt7complexIdEC1Edd(%"struct.std::complex<double>"* %1, double %10, double 0.000000e+00) nounwind - call void @_ZStngIdESt7complexIT_ERKS2_(%"struct.std::complex<double>"* noalias sret %0, %"struct.std::complex<double>"* @_ZN10polynomialIdE4PI2IE) nounwind - invoke void @_ZStdvIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %2, %"struct.std::complex<double>"* %0, %"struct.std::complex<double>"* %1) - to label %invcont unwind label %lpad - -invcont: ; preds = %bb - call void @llvm.dbg.stoppoint(i32 548, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt3expIdESt7complexIT_ERKS2_(%"struct.std::complex<double>"* noalias sret %memtmp22, %"struct.std::complex<double>"* %2) nounwind - %11 = getelementptr %"struct.std::complex<double>"* %wm, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %12 = getelementptr %"struct.std::complex<double>"* %memtmp22, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %13 = load double* %12, align 8 ; <double> [#uses=1] - store double %13, double* %11, align 8 - %14 = getelementptr %"struct.std::complex<double>"* %wm, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %15 = getelementptr %"struct.std::complex<double>"* %memtmp22, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %16 = load double* %15, align 8 ; <double> [#uses=1] - store double %16, double* %14, align 8 - call void @llvm.dbg.stoppoint(i32 549, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %17 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEaSEd(%"struct.std::complex<double>"* %w, double 1.000000e+00) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.stoppoint(i32 551, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb32 - -bb24: ; preds = %bb30 - call void @llvm.dbg.stoppoint(i32 555, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %18 = add i32 %k.0, %m2.0 ; <i32> [#uses=1] - %19 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %18) nounwind ; <%"struct.std::complex<double>"*> [#uses=1] - call void @_ZStmlIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %memtmp25, %"struct.std::complex<double>"* %w, %"struct.std::complex<double>"* %19) nounwind - %20 = getelementptr %"struct.std::complex<double>"* %t, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %21 = getelementptr %"struct.std::complex<double>"* %memtmp25, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %22 = load double* %21, align 8 ; <double> [#uses=1] - store double %22, double* %20, align 8 - %23 = getelementptr %"struct.std::complex<double>"* %t, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %24 = getelementptr %"struct.std::complex<double>"* %memtmp25, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %25 = load double* %24, align 8 ; <double> [#uses=1] - store double %25, double* %23, align 8 - call void @llvm.dbg.stoppoint(i32 556, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %26 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %k.0) nounwind ; <%"struct.std::complex<double>"*> [#uses=2] - %27 = getelementptr %"struct.std::complex<double>"* %u, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %28 = getelementptr %"struct.std::complex<double>"* %26, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %29 = load double* %28, align 4 ; <double> [#uses=1] - store double %29, double* %27, align 8 - %30 = getelementptr %"struct.std::complex<double>"* %u, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %31 = getelementptr %"struct.std::complex<double>"* %26, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %32 = load double* %31, align 4 ; <double> [#uses=1] - store double %32, double* %30, align 8 - call void @llvm.dbg.stoppoint(i32 557, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %33 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %k.0) nounwind ; <%"struct.std::complex<double>"*> [#uses=2] - invoke void @_ZStplIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %memtmp26, %"struct.std::complex<double>"* %u, %"struct.std::complex<double>"* %t) - to label %invcont27 unwind label %lpad - -invcont27: ; preds = %bb24 - %34 = getelementptr %"struct.std::complex<double>"* %33, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %35 = getelementptr %"struct.std::complex<double>"* %memtmp26, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %36 = load double* %35, align 8 ; <double> [#uses=1] - store double %36, double* %34, align 4 - %37 = getelementptr %"struct.std::complex<double>"* %33, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %38 = getelementptr %"struct.std::complex<double>"* %memtmp26, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %39 = load double* %38, align 8 ; <double> [#uses=1] - store double %39, double* %37, align 4 - call void @llvm.dbg.stoppoint(i32 558, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %40 = add i32 %k.0, %m2.0 ; <i32> [#uses=1] - %41 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %40) nounwind ; <%"struct.std::complex<double>"*> [#uses=2] - invoke void @_ZStmiIdESt7complexIT_ERKS2_S4_(%"struct.std::complex<double>"* noalias sret %memtmp28, %"struct.std::complex<double>"* %u, %"struct.std::complex<double>"* %t) - to label %invcont29 unwind label %lpad - -invcont29: ; preds = %invcont27 - %42 = getelementptr %"struct.std::complex<double>"* %41, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %43 = getelementptr %"struct.std::complex<double>"* %memtmp28, i32 0, i32 0, i32 0 ; <double*> [#uses=1] - %44 = load double* %43, align 8 ; <double> [#uses=1] - store double %44, double* %42, align 4 - %45 = getelementptr %"struct.std::complex<double>"* %41, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %46 = getelementptr %"struct.std::complex<double>"* %memtmp28, i32 0, i32 0, i32 1 ; <double*> [#uses=1] - %47 = load double* %46, align 8 ; <double> [#uses=1] - store double %47, double* %45, align 4 - call void @llvm.dbg.stoppoint(i32 553, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %48 = add i32 %k.0, %m.0 ; <i32> [#uses=1] - br label %bb30 - -bb30: ; preds = %bb32, %invcont29 - %k.0 = phi i32 [ %48, %invcont29 ], [ %j.0, %bb32 ] ; <i32> [#uses=6] - call void @llvm.dbg.stoppoint(i32 553, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %49 = call i32 @_ZNK10polynomialISt7complexIdEE6degreeEv(%"struct.polynomial<std::complex<double> >"* %poly) nounwind ; <i32> [#uses=1] - %50 = add i32 %49, -1 ; <i32> [#uses=1] - %51 = icmp ult i32 %50, %k.0 ; <i1> [#uses=1] - br i1 %51, label %bb31, label %bb24 - -bb31: ; preds = %bb30 - call void @llvm.dbg.stoppoint(i32 561, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %52 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEmLIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %w, %"struct.std::complex<double>"* %wm) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.stoppoint(i32 551, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %53 = add i32 %j.0, 1 ; <i32> [#uses=1] - br label %bb32 - -bb32: ; preds = %bb31, %invcont - %j.0 = phi i32 [ 0, %invcont ], [ %53, %bb31 ] ; <i32> [#uses=3] - call void @llvm.dbg.stoppoint(i32 551, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %54 = add i32 %m2.0, -1 ; <i32> [#uses=1] - %55 = icmp ult i32 %54, %j.0 ; <i1> [#uses=1] - br i1 %55, label %bb33, label %bb30 - -bb33: ; preds = %bb32 - call void @llvm.dbg.stoppoint(i32 564, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %56 = shl i32 %m.0, 1 ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 565, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %57 = shl i32 %m2.0, 1 ; <i32> [#uses=1] - call void @llvm.dbg.stoppoint(i32 546, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %58 = add i32 %s.0, 1 ; <i32> [#uses=1] - br label %bb34 - -bb34: ; preds = %bb33, %entry - %m.0 = phi i32 [ 2, %entry ], [ %56, %bb33 ] ; <i32> [#uses=3] - %m2.0 = phi i32 [ 1, %entry ], [ %57, %bb33 ] ; <i32> [#uses=4] - %s.0 = phi i32 [ 0, %entry ], [ %58, %bb33 ] ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 546, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %59 = icmp ult i32 %s.0, %9 ; <i1> [#uses=1] - br i1 %59, label %bb, label %bb37 - -bb36: ; preds = %bb37 - call void @llvm.dbg.stoppoint(i32 569, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %60 = call i32 @_ZNK10polynomialISt7complexIdEE6degreeEv(%"struct.polynomial<std::complex<double> >"* %poly) nounwind ; <i32> [#uses=1] - %61 = uitofp i32 %60 to double ; <double> [#uses=1] - %62 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %agg.result, i32 %j.1) nounwind ; <%"struct.std::complex<double>"*> [#uses=1] - %63 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEdVEd(%"struct.std::complex<double>"* %62, double %61) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.stoppoint(i32 568, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %64 = add i32 %j.1, 1 ; <i32> [#uses=1] - br label %bb37 - -bb37: ; preds = %bb36, %bb34 - %j.1 = phi i32 [ %64, %bb36 ], [ 0, %bb34 ] ; <i32> [#uses=3] - call void @llvm.dbg.stoppoint(i32 568, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %65 = call i32 @_ZNK10polynomialISt7complexIdEE6degreeEv(%"struct.polynomial<std::complex<double> >"* %poly) nounwind ; <i32> [#uses=1] - %66 = icmp ugt i32 %65, %j.1 ; <i1> [#uses=1] - br i1 %66, label %bb36, label %return - -return: ; preds = %bb37 - call void @llvm.dbg.stoppoint(i32 571, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1771 to %0*)) - ret void - -lpad: ; preds = %invcont27, %bb24, %bb - %eh_ptr = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select46 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 571, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialISt7complexIdEED1Ev(%"struct.polynomial<std::complex<double> >"* %agg.result) - to label %Unwind unwind label %lpad47 - -lpad47: ; preds = %lpad - %eh_ptr48 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select50 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr48, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 571, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -Unwind: ; preds = %lpad - call void @_Unwind_Resume(i8* %eh_ptr) - unreachable -} - -define linkonce void @_ZNK10polynomialIdEmlERKS0_(%"struct.polynomial<double>"* noalias sret %agg.result, %"struct.polynomial<double>"* %this, %"struct.polynomial<double>"* %poly) { -entry: - %result = alloca %"struct.polynomial<double>", align 8 ; <%"struct.polynomial<double>"*> [#uses=1] - %dft2 = alloca %"struct.polynomial<std::complex<double> >", align 8 ; <%"struct.polynomial<std::complex<double> >"*> [#uses=7] - %dft1 = alloca %"struct.polynomial<std::complex<double> >", align 8 ; <%"struct.polynomial<std::complex<double> >"*> [#uses=6] - %a2 = alloca %"struct.polynomial<double>", align 8 ; <%"struct.polynomial<double>"*> [#uses=8] - %a1 = alloca %"struct.polynomial<double>", align 8 ; <%"struct.polynomial<double>"*> [#uses=9] - %0 = alloca %"struct.polynomial<std::complex<double> >", align 8 ; <%"struct.polynomial<std::complex<double> >"*> [#uses=4] - call void @llvm.dbg.func.start(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1785 to %0*)) - %1 = bitcast %"struct.polynomial<double>"* %result to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %1, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1791 to %0*)) - %2 = bitcast %"struct.polynomial<std::complex<double> >"* %dft2 to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %2, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1795 to %0*)) - %3 = bitcast %"struct.polynomial<std::complex<double> >"* %dft1 to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %3, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1797 to %0*)) - %4 = bitcast %"struct.polynomial<double>"* %a2 to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %4, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1799 to %0*)) - %5 = bitcast %"struct.polynomial<double>"* %a1 to %0* ; <%0*> [#uses=1] - call void @llvm.dbg.declare(%0* %5, %0* bitcast (%llvm.dbg.variable.type* @llvm.dbg.variable1801 to %0*)) - call void @llvm.dbg.stoppoint(i32 590, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdEC1ERKS0_(%"struct.polynomial<double>"* %a1, %"struct.polynomial<double>"* %this) - call void @llvm.dbg.stoppoint(i32 591, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdEC1ERKS0_(%"struct.polynomial<double>"* %a2, %"struct.polynomial<double>"* %poly) - to label %invcont unwind label %lpad - -invcont: ; preds = %entry - call void @llvm.dbg.stoppoint(i32 594, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %6 = call i32 @_ZNK10polynomialIdE6degreeEv(%"struct.polynomial<double>"* %a1) nounwind ; <i32> [#uses=1] - %7 = call i32 @_ZNK10polynomialIdE6degreeEv(%"struct.polynomial<double>"* %a2) nounwind ; <i32> [#uses=1] - %8 = icmp ugt i32 %6, %7 ; <i1> [#uses=1] - br i1 %8, label %bb, label %bb26 - -bb: ; preds = %invcont - call void @llvm.dbg.stoppoint(i32 595, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %9 = invoke i32 @_ZN10polynomialIdE11stretch_fftEv(%"struct.polynomial<double>"* %a1) - to label %invcont24 unwind label %lpad76 ; <i32> [#uses=1] - -invcont24: ; preds = %bb - call void @llvm.dbg.stoppoint(i32 595, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %10 = invoke %"struct.polynomial<double>"* @_ZN10polynomialIdE7stretchEj(%"struct.polynomial<double>"* %a2, i32 %9) - to label %bb29 unwind label %lpad76 ; <%"struct.polynomial<double>"*> [#uses=0] - -bb26: ; preds = %invcont - call void @llvm.dbg.stoppoint(i32 597, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %11 = invoke i32 @_ZN10polynomialIdE11stretch_fftEv(%"struct.polynomial<double>"* %a2) - to label %invcont27 unwind label %lpad76 ; <i32> [#uses=1] - -invcont27: ; preds = %bb26 - call void @llvm.dbg.stoppoint(i32 597, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %12 = invoke %"struct.polynomial<double>"* @_ZN10polynomialIdE7stretchEj(%"struct.polynomial<double>"* %a1, i32 %11) - to label %bb29 unwind label %lpad76 ; <%"struct.polynomial<double>"*> [#uses=0] - -bb29: ; preds = %invcont27, %invcont24 - call void @llvm.dbg.stoppoint(i32 600, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdE3fftERKS0_(%"struct.polynomial<std::complex<double> >"* noalias sret %dft1, %"struct.polynomial<double>"* %a1) - to label %invcont30 unwind label %lpad76 - -invcont30: ; preds = %bb29 - call void @llvm.dbg.stoppoint(i32 601, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdE3fftERKS0_(%"struct.polynomial<std::complex<double> >"* noalias sret %dft2, %"struct.polynomial<double>"* %a2) - to label %invcont31 unwind label %lpad80 - -invcont31: ; preds = %invcont30 - call void @llvm.dbg.stoppoint(i32 604, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %13 = call i32 @_ZNK10polynomialIdE6degreeEv(%"struct.polynomial<double>"* %a1) nounwind ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 606, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - br label %bb33 - -bb32: ; preds = %bb33 - call void @llvm.dbg.stoppoint(i32 607, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %14 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %dft2, i32 %k15.0) nounwind ; <%"struct.std::complex<double>"*> [#uses=1] - %15 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %dft1, i32 %k15.0) nounwind ; <%"struct.std::complex<double>"*> [#uses=1] - %16 = call %"struct.std::complex<double>"* @_ZNSt7complexIdEmLIdEERS0_RKS_IT_E(%"struct.std::complex<double>"* %15, %"struct.std::complex<double>"* %14) nounwind ; <%"struct.std::complex<double>"*> [#uses=0] - call void @llvm.dbg.stoppoint(i32 606, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %17 = add i32 %k15.0, 1 ; <i32> [#uses=1] - br label %bb33 - -bb33: ; preds = %bb32, %invcont31 - %k15.0 = phi i32 [ 0, %invcont31 ], [ %17, %bb32 ] ; <i32> [#uses=4] - call void @llvm.dbg.stoppoint(i32 606, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %18 = icmp ult i32 %k15.0, %13 ; <i1> [#uses=1] - br i1 %18, label %bb32, label %bb34 - -bb34: ; preds = %bb33 - call void @llvm.dbg.stoppoint(i32 610, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdE11inverse_fftERKS_ISt7complexIdEE(%"struct.polynomial<std::complex<double> >"* noalias sret %0, %"struct.polynomial<std::complex<double> >"* %dft1) - to label %invcont35 unwind label %lpad84 - -invcont35: ; preds = %bb34 - call void @llvm.dbg.stoppoint(i32 610, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %19 = invoke %"struct.polynomial<std::complex<double> >"* @_ZN10polynomialISt7complexIdEEaSERKS2_(%"struct.polynomial<std::complex<double> >"* %dft2, %"struct.polynomial<std::complex<double> >"* %0) - to label %invcont36 unwind label %lpad88 ; <%"struct.polynomial<std::complex<double> >"*> [#uses=0] - -invcont36: ; preds = %invcont35 - call void @llvm.dbg.stoppoint(i32 610, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialISt7complexIdEED1Ev(%"struct.polynomial<std::complex<double> >"* %0) - to label %bb43 unwind label %lpad84 - -bb43: ; preds = %invcont36 - call void @llvm.dbg.stoppoint(i32 613, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %20 = add i32 %13, -1 ; <i32> [#uses=2] - call void @llvm.dbg.stoppoint(i32 614, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdEC1Ej(%"struct.polynomial<double>"* %agg.result, i32 %20) - to label %bb46 unwind label %lpad84 - -bb45: ; preds = %bb46 - call void @llvm.dbg.stoppoint(i32 617, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %21 = call double* @_ZN10polynomialIdEixEj(%"struct.polynomial<double>"* %agg.result, i32 %k.0) nounwind ; <double*> [#uses=1] - %22 = call %"struct.std::complex<double>"* @_ZN10polynomialISt7complexIdEEixEj(%"struct.polynomial<std::complex<double> >"* %dft2, i32 %k.0) nounwind ; <%"struct.std::complex<double>"*> [#uses=1] - %23 = call double* @_ZNSt7complexIdE4realEv(%"struct.std::complex<double>"* %22) nounwind ; <double*> [#uses=1] - %24 = load double* %23, align 8 ; <double> [#uses=1] - store double %24, double* %21, align 8 - call void @llvm.dbg.stoppoint(i32 616, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %25 = add i32 %k.0, 1 ; <i32> [#uses=1] - br label %bb46 - -bb46: ; preds = %bb45, %bb43 - %k.0 = phi i32 [ %25, %bb45 ], [ 0, %bb43 ] ; <i32> [#uses=4] - call void @llvm.dbg.stoppoint(i32 616, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - %26 = icmp ult i32 %k.0, %20 ; <i1> [#uses=1] - br i1 %26, label %bb45, label %bb47 - -bb47: ; preds = %bb46 - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialISt7complexIdEED1Ev(%"struct.polynomial<std::complex<double> >"* %dft2) - to label %bb54 unwind label %lpad80 - -bb54: ; preds = %bb47 - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialISt7complexIdEED1Ev(%"struct.polynomial<std::complex<double> >"* %dft1) - to label %bb61 unwind label %lpad76 - -bb61: ; preds = %bb54 - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %a2) - to label %bb68 unwind label %lpad - -bb68: ; preds = %bb61 - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %a1) - call void @llvm.dbg.region.end(%0* bitcast (%llvm.dbg.subprogram.type* @llvm.dbg.subprogram1785 to %0*)) - ret void - -lpad: ; preds = %bb61, %entry - %eh_ptr = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select75 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - br label %ppad - -lpad76: ; preds = %bb54, %bb29, %invcont27, %bb26, %invcont24, %bb - %eh_ptr77 = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select79 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr77, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - br label %ppad112 - -lpad80: ; preds = %bb47, %invcont30 - %eh_ptr81 = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select83 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr81, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - br label %ppad113 - -lpad84: ; preds = %bb43, %invcont36, %bb34 - %eh_ptr85 = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select87 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr85, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - br label %ppad114 - -lpad88: ; preds = %invcont35 - %eh_ptr89 = call i8* @llvm.eh.exception() ; <i8*> [#uses=2] - %eh_select91 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr89, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i8* null) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 610, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialISt7complexIdEED1Ev(%"struct.polynomial<std::complex<double> >"* %0) - to label %ppad114 unwind label %lpad92 - -lpad92: ; preds = %lpad88 - %eh_ptr93 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select95 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr93, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 610, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -lpad96: ; preds = %ppad114 - %eh_ptr97 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select99 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr97, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -lpad100: ; preds = %ppad113 - %eh_ptr101 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select103 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr101, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -lpad104: ; preds = %ppad112 - %eh_ptr105 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select107 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr105, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -lpad108: ; preds = %ppad - %eh_ptr109 = call i8* @llvm.eh.exception() ; <i8*> [#uses=1] - %eh_select111 = call i32 (i8*, i8*, ...)* @llvm.eh.selector.i32(i8* %eh_ptr109, i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*), i32 1) ; <i32> [#uses=0] - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - call void @_ZSt9terminatev() noreturn nounwind - unreachable - -ppad: ; preds = %ppad112, %lpad - %eh_exception.3 = phi i8* [ %eh_ptr, %lpad ], [ %eh_exception.2, %ppad112 ] ; <i8*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %a1) - to label %Unwind unwind label %lpad108 - -ppad112: ; preds = %ppad113, %lpad76 - %eh_exception.2 = phi i8* [ %eh_ptr77, %lpad76 ], [ %eh_exception.1, %ppad113 ] ; <i8*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialIdED1Ev(%"struct.polynomial<double>"* %a2) - to label %ppad unwind label %lpad104 - -ppad113: ; preds = %ppad114, %lpad80 - %eh_exception.1 = phi i8* [ %eh_ptr81, %lpad80 ], [ %eh_exception.0, %ppad114 ] ; <i8*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialISt7complexIdEED1Ev(%"struct.polynomial<std::complex<double> >"* %dft1) - to label %ppad112 unwind label %lpad100 - -ppad114: ; preds = %lpad88, %lpad84 - %eh_exception.0 = phi i8* [ %eh_ptr85, %lpad84 ], [ %eh_ptr89, %lpad88 ] ; <i8*> [#uses=1] - call void @llvm.dbg.stoppoint(i32 620, i32 0, %0* bitcast (%llvm.dbg.compile_unit.type* @llvm.dbg.compile_unit to %0*)) - invoke void @_ZN10polynomialISt7complexIdEED1Ev(%"struct.polynomial<std::complex<double> >"* %dft2) - to label %ppad113 unwind label %lpad96 - -Unwind: ; preds = %ppad - call void @_Unwind_Resume(i8* %eh_exception.3) - unreachable -} - -declare i32 @strcmp(i8* nocapture, i8* nocapture) nounwind readonly - -declare %"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZNSolsEd(%"struct.std::basic_ostream<char,std::char_traits<char> >"*, double) - -declare %"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc(%"struct.std::basic_ostream<char,std::char_traits<char> >"*, i8*) - -declare %"struct.std::basic_ostream<char,std::char_traits<char> >"* @_ZNSo5flushEv(%"struct.std::basic_ostream<char,std::char_traits<char> >"*) - -declare extern_weak i32 @pthread_once(i32*, void ()*) - -declare extern_weak i8* @pthread_getspecific(i32) - -declare extern_weak i32 @pthread_setspecific(i32, i8*) - -declare extern_weak i32 @pthread_create(i32*, %struct.pthread_attr_t*, i8* (i8*)*, i8*) - -declare extern_weak i32 @pthread_cancel(i32) - -declare extern_weak i32 @pthread_mutex_lock(%struct.pthread_mutex_t*) - -declare extern_weak i32 @pthread_mutex_trylock(%struct.pthread_mutex_t*) - -declare extern_weak i32 @pthread_mutex_unlock(%struct.pthread_mutex_t*) - -declare extern_weak i32 @pthread_mutex_init(%struct.pthread_mutex_t*, %struct..0._50*) - -declare extern_weak i32 @pthread_key_create(i32*, void (i8*)*) - -declare extern_weak i32 @pthread_key_delete(i32) - -declare extern_weak i32 @pthread_mutexattr_init(%struct..0._50*) - -declare extern_weak i32 @pthread_mutexattr_settype(%struct..0._50*, i32) - -declare extern_weak i32 @pthread_mutexattr_destroy(%struct..0._50*) - -declare i32 @memcmp(i8* nocapture, i8* nocapture, i32) nounwind readonly diff --git a/test/Transforms/SimplifyCFG/2009-06-15-InvokeCrash.ll b/test/Transforms/SimplifyCFG/2009-06-15-InvokeCrash.ll index 72a15b1..f99cbe1 100644 --- a/test/Transforms/SimplifyCFG/2009-06-15-InvokeCrash.ll +++ b/test/Transforms/SimplifyCFG/2009-06-15-InvokeCrash.ll @@ -144,11 +144,11 @@ invcont57: ; preds = %invcont51 store double %tmp64, double* %tmp62 %tmp65 = call double* @_ZN6QSizeF6rwidthEv( %struct.QPointF* %scaledPageSize ) ; <double*> [#uses=2] %tmp67 = load double* %tmp65 ; <double> [#uses=1] - %tmp69 = mul double %tmp67, %tmp48 ; <double> [#uses=1] + %tmp69 = fmul double %tmp67, %tmp48 ; <double> [#uses=1] store double %tmp69, double* %tmp65 %tmp71 = call double* @_ZN6QSizeF7rheightEv( %struct.QPointF* %scaledPageSize ) ; <double*> [#uses=2] %tmp73 = load double* %tmp71 ; <double> [#uses=1] - %tmp75 = mul double %tmp73, %tmp54 ; <double> [#uses=1] + %tmp75 = fmul double %tmp73, %tmp54 ; <double> [#uses=1] store double %tmp75, double* %tmp71 %tmp78 = getelementptr %struct.QPrinter* %printer, i32 0, i32 0 ; <%struct.QPaintDevice*> [#uses=1] %tmp80 = invoke i32 @_ZNK12QPaintDevice6heightEv( %struct.QPaintDevice* %tmp78 ) @@ -190,7 +190,7 @@ invcont104: ; preds = %invcont103 to label %invcont106 unwind label %cleanup329 ; <i32> [#uses=1] invcont106: ; preds = %invcont104 %tmp108 = sitofp i32 %tmp107 to double ; <double> [#uses=1] - %tmp109 = mul double %tmp108, 0x3FE93264C993264C ; <double> [#uses=1] + %tmp109 = fmul double %tmp108, 0x3FE93264C993264C ; <double> [#uses=1] %tmp109.upgrd.17 = fptosi double %tmp109 to i32 ; <i32> [#uses=3] %tmp.upgrd.18 = call %struct.QTextBlockGroup* @_ZNK13QTextDocument9rootFrameEv( %struct.QAbstractTextDocumentLayout* %tmp95 ) ; <%struct.QTextBlockGroup*> [#uses=1] invoke void @_ZNK10QTextFrame11frameFormatEv( %struct.QTextBlockFormat* sret %fmt, %struct.QTextBlockGroup* %tmp.upgrd.18 ) @@ -237,7 +237,7 @@ invcont124: ; preds = %invcont122 store double %tmp137, double* %tmp135 %tmp138 = call double @_ZNK6QRectF6heightEv( %struct.QRectF* %body ) ; <double> [#uses=1] %tmp139 = sitofp i32 %tmp109.upgrd.17 to double ; <double> [#uses=1] - %tmp140 = sub double %tmp138, %tmp139 ; <double> [#uses=1] + %tmp140 = fsub double %tmp138, %tmp139 ; <double> [#uses=1] %tmp142 = invoke %struct.QPaintDevice* @_ZNK8QPainter6deviceEv( %struct.QPainter* %p ) to label %invcont141 unwind label %cleanup192 ; <%struct.QPaintDevice*> [#uses=1] invcont141: ; preds = %invcont124 @@ -251,7 +251,7 @@ invcont146: ; preds = %invcont144 to label %invcont148 unwind label %cleanup168 ; <i32> [#uses=1] invcont148: ; preds = %invcont146 %tmp149.upgrd.21 = sitofp i32 %tmp149 to double ; <double> [#uses=1] - %tmp150 = add double %tmp140, %tmp149.upgrd.21 ; <double> [#uses=1] + %tmp150 = fadd double %tmp140, %tmp149.upgrd.21 ; <double> [#uses=1] %tmp152 = invoke %struct.QPaintDevice* @_ZNK8QPainter6deviceEv( %struct.QPainter* %p ) to label %invcont151 unwind label %cleanup168 ; <%struct.QPaintDevice*> [#uses=1] invcont151: ; preds = %invcont148 @@ -261,10 +261,10 @@ invcont153: ; preds = %invcont151 %tmp155 = mul i32 %tmp154, 5 ; <i32> [#uses=1] %tmp156 = sdiv i32 %tmp155, 72 ; <i32> [#uses=1] %tmp156.upgrd.22 = sitofp i32 %tmp156 to double ; <double> [#uses=1] - %tmp157 = add double %tmp150, %tmp156.upgrd.22 ; <double> [#uses=1] + %tmp157 = fadd double %tmp150, %tmp156.upgrd.22 ; <double> [#uses=1] %tmp158 = call double @_ZNK6QRectF5widthEv( %struct.QRectF* %body ) ; <double> [#uses=1] %tmp159 = sitofp i32 %tmp109.upgrd.17 to double ; <double> [#uses=1] - %tmp160 = sub double %tmp158, %tmp159 ; <double> [#uses=1] + %tmp160 = fsub double %tmp158, %tmp159 ; <double> [#uses=1] call void @_ZN7QPointFC1Edd( %struct.QPointF* %tmp2, double %tmp160, double %tmp157 ) %tmp161 = getelementptr %struct.QPointF* %pageNumberPos, i32 0, i32 0 ; <double*> [#uses=1] %tmp162 = getelementptr %struct.QPointF* %tmp2, i32 0, i32 0 ; <double*> [#uses=1] diff --git a/test/Unit/lit.cfg b/test/Unit/lit.cfg index 34372bb..5fe0732 100644 --- a/test/Unit/lit.cfg +++ b/test/Unit/lit.cfg @@ -23,7 +23,14 @@ config.test_format = lit.formats.GoogleTest(llvm_build_mode, 'Tests') ### -import os +# If necessary, point the dynamic loader at libLLVM.so. +if config.enable_shared: + libdir = os.path.join(config.llvm_obj_root, config.llvm_build_mode, 'lib') + shlibpath = config.environment.get(config.shlibpath_var,'') + if shlibpath: + shlibpath = ':' + shlibpath + shlibpath = libdir + shlibpath + config.environment[config.shlibpath_var] = shlibpath # Check that the object root is known. if config.test_exec_root is None: diff --git a/test/Unit/lit.site.cfg.in b/test/Unit/lit.site.cfg.in index c190ffa..51b5bc4 100644 --- a/test/Unit/lit.site.cfg.in +++ b/test/Unit/lit.site.cfg.in @@ -5,6 +5,8 @@ config.llvm_obj_root = "@LLVM_BINARY_DIR@" config.llvm_tools_dir = "@LLVM_TOOLS_DIR@" config.llvmgcc_dir = "@LLVMGCCDIR@" config.llvm_build_mode = "@LLVM_BUILD_MODE@" +config.enable_shared = @ENABLE_SHARED@ +config.shlibpath_var = "@SHLIBPATH_VAR@" # Let the main config do the real work. lit.load_config(config, "@LLVM_SOURCE_DIR@/test/Unit/lit.cfg") diff --git a/test/lib/llvm.exp b/test/lib/llvm.exp index 319cc11..19a2729 100644 --- a/test/lib/llvm.exp +++ b/test/lib/llvm.exp @@ -47,7 +47,7 @@ proc execOneLine { test PRS outcome lineno line } { # cases. proc substitute { line test tmpFile } { global srcroot objroot srcdir objdir subdir target_triplet - global llvmgcc llvmgxx llvmgcc_version llvmgccmajvers ocamlopt + global llvmgcc llvmgxx ocamlopt global gccpath gxxpath compile_c compile_cxx link shlibext llvmlibsdir global llvmdsymutil valgrind grep gas bugpoint_topts set path [file join $srcdir $subdir] @@ -119,7 +119,7 @@ proc substitute { line test tmpFile } { # This procedure runs the set of tests for the test_source_files array. proc RunLLVMTests { test_source_files } { - global srcroot objroot srcdir objdir subdir target_triplet llvmgcc_version + global srcroot objroot srcdir objdir subdir target_triplet set timeout 60 set path [file join $objdir $subdir] @@ -194,12 +194,6 @@ proc RunLLVMTests { test_source_files } { if {$targetPASS != 1} { set outcome XFAIL } - } elseif { [regexp {llvmgcc(([0-9]+)|([0-9]+[.][0-9]+))} $target match submatch submatch2] } { - if { [regexp ^($submatch)$|^(($submatch)(\.)) $llvmgcc_version match] } { - if {$targetPASS != 1} { - set outcome XFAIL - } - } } } } elseif {[regexp {XTARGET:[ *](.+)} $line match targets]} { @@ -213,11 +207,6 @@ proc RunLLVMTests { test_source_files } { } elseif { [regexp $target $target_triplet match] } { set targetPASS 1 set outcome PASS - } elseif { [regexp {llvmgcc(([0-9]+)|([0-9]+[.][0-9]+))} $target match submatch submatch2] } { - if { [regexp ^($submatch)$|^(($submatch)(\.)) $llvmgcc_version match] } { - set targetPASS 1 - set outcome PASS - } } } } diff --git a/test/lit.cfg b/test/lit.cfg index 0894d9b..929871a 100644 --- a/test/lit.cfg +++ b/test/lit.cfg @@ -40,13 +40,14 @@ if llvm_obj_root is not None: config.environment['PATH'] = path llvmgcc_dir = getattr(config, 'llvmgcc_dir', None) - if not llvm_tools_dir: - lit.fatal('No llvm-gcc dir set!') if llvmgcc_dir: path = os.path.pathsep.join((os.path.join(llvmgcc_dir, 'bin'), config.environment['PATH'])) config.environment['PATH'] = path +# Propogate 'HOME' through the environment. +config.environment['HOME'] = os.environ['HOME'] + ### import os diff --git a/test/site.exp.in b/test/site.exp.in index f88d361..c760c2c 100644 --- a/test/site.exp.in +++ b/test/site.exp.in @@ -3,7 +3,6 @@ set target_triplet "@TARGET_TRIPLE@" set TARGETS_TO_BUILD "@TARGETS_TO_BUILD@" set llvmgcc_langs "@LLVMGCC_LANGS@" -set llvmgcc_version "@LLVMGCC_VERSION@" set llvmtoolsdir "@LLVM_TOOLS_DIR@" set llvmlibsdir "@LLVM_LIBS_DIR@" set llvm_bindings "@LLVM_BINDINGS@" @@ -18,7 +17,6 @@ set compile_cxx "@TEST_COMPILE_CXX_CMD@" set link "@TEST_LINK_CMD@" set llvmgcc "@LLVMGCC@" set llvmgxx "@LLVMGXX@" -set llvmgccmajvers "@LLVMGCCMAJVERS@" set bugpoint_topts "@BUGPOINT_TOPTS@" set shlibext "@SHLIBEXT@" set ocamlopt "@OCAMLOPT@" diff --git a/tools/Makefile b/tools/Makefile index c9b9ff2..b6637a9 100644 --- a/tools/Makefile +++ b/tools/Makefile @@ -44,4 +44,9 @@ ifeq ($(TARGET_OS), $(filter $(TARGET_OS), Cygwin MingW)) DIRS := $(filter-out lto gold, $(DIRS)) endif +# Only build edis if X86 target support is enabled. +ifeq ($(filter $(TARGETS_TO_BUILD), X86),) + PARALLEL_DIRS := $(filter-out edis, $(PARALLEL_DIRS)) +endif + include $(LEVEL)/Makefile.common diff --git a/tools/bugpoint/CrashDebugger.cpp b/tools/bugpoint/CrashDebugger.cpp index b348a08..b51bdb4 100644 --- a/tools/bugpoint/CrashDebugger.cpp +++ b/tools/bugpoint/CrashDebugger.cpp @@ -295,7 +295,7 @@ bool ReduceCrashingBlocks::TestBlocks(std::vector<const BasicBlock*> &BBs) { TerminatorInst *BBTerm = BB->getTerminator(); - if (isa<StructType>(BBTerm->getType())) + if (BBTerm->getType()->isStructTy()) BBTerm->replaceAllUsesWith(UndefValue::get(BBTerm->getType())); else if (BB->getTerminator()->getType() != Type::getVoidTy(BB->getContext())) diff --git a/tools/bugpoint/ExtractFunction.cpp b/tools/bugpoint/ExtractFunction.cpp index 70011a7..eaa2c53 100644 --- a/tools/bugpoint/ExtractFunction.cpp +++ b/tools/bugpoint/ExtractFunction.cpp @@ -73,7 +73,7 @@ Module *BugDriver::deleteInstructionFromProgram(const Instruction *I, Instruction *TheInst = RI; // Got the corresponding instruction! // If this instruction produces a value, replace any users with null values - if (isa<StructType>(TheInst->getType())) + if (TheInst->getType()->isStructTy()) TheInst->replaceAllUsesWith(UndefValue::get(TheInst->getType())); else if (TheInst->getType() != Type::getVoidTy(I->getContext())) TheInst->replaceAllUsesWith(Constant::getNullValue(TheInst->getType())); diff --git a/tools/llc/llc.cpp b/tools/llc/llc.cpp index fe34bd1..810ba42 100644 --- a/tools/llc/llc.cpp +++ b/tools/llc/llc.cpp @@ -305,6 +305,14 @@ int main(int argc, char **argv) { case '3': OLvl = CodeGenOpt::Aggressive; break; } + // Request that addPassesToEmitFile run the Verifier after running + // passes which modify the IR. +#ifndef NDEBUG + bool DisableVerify = false; +#else + bool DisableVerify = true; +#endif + // If this target requires addPassesToEmitWholeFile, do it now. This is // used by strange things like the C backend. if (Target.WantsWholeFile()) { @@ -320,7 +328,8 @@ int main(int argc, char **argv) { PM.add(createVerifierPass()); // Ask the target to add backend passes as necessary. - if (Target.addPassesToEmitWholeFile(PM, *Out, FileType, OLvl)) { + if (Target.addPassesToEmitWholeFile(PM, *Out, FileType, OLvl, + DisableVerify)) { errs() << argv[0] << ": target does not support generation of this" << " file type!\n"; if (Out != &fouts()) delete Out; @@ -347,7 +356,8 @@ int main(int argc, char **argv) { // Override default to generate verbose assembly. Target.setAsmVerbosityDefault(true); - if (Target.addPassesToEmitFile(Passes, *Out, FileType, OLvl)) { + if (Target.addPassesToEmitFile(Passes, *Out, FileType, OLvl, + DisableVerify)) { errs() << argv[0] << ": target does not support generation of this" << " file type!\n"; if (Out != &fouts()) delete Out; diff --git a/tools/llvm-config/CMakeLists.txt b/tools/llvm-config/CMakeLists.txt index 8a710ea..7a43dba 100644 --- a/tools/llvm-config/CMakeLists.txt +++ b/tools/llvm-config/CMakeLists.txt @@ -18,8 +18,6 @@ execute_process(COMMAND date #set(LLVMGCCDIR "") #set(LLVMGCC "") #set(LLVMGXX "") -#set(LLVMGCC_VERSION "") -#set(LLVMGCC_MAJVERS "") test_big_endian(IS_BIG_ENDIAN) if( IS_BIG_ENDIAN ) set(ENDIAN "big") diff --git a/tools/llvm-shlib/Makefile b/tools/llvm-shlib/Makefile new file mode 100644 index 0000000..ce29bf8 --- /dev/null +++ b/tools/llvm-shlib/Makefile @@ -0,0 +1,60 @@ +##===- tools/shlib/Makefile --------------------------------*- Makefile -*-===## +# +# The LLVM Compiler Infrastructure +# +# This file is distributed under the University of Illinois Open Source +# License. See LICENSE.TXT for details. +# +##===----------------------------------------------------------------------===## + +LEVEL = ../.. + +LIBRARYNAME = LLVM-$(LLVMVersion) + +NO_BUILD_ARCHIVE = 1 +LINK_LIBS_IN_SHARED = 1 +SHARED_LIBRARY = 1 + +include $(LEVEL)/Makefile.common + +# Include all archives in libLLVM.(so|dylib) except the ones that have +# their own dynamic libraries. +Archives := $(wildcard $(LibDir)/libLLVM*.a) +SharedLibraries := $(wildcard $(LibDir)/libLLVM*$(SHLIBEXT)) +IncludeInLibLlvm := $(filter-out $(basename $(SharedLibraries)).a, $(Archives)) +LLVMLibsOptions := $(IncludeInLibLlvm:$(LibDir)/lib%.a=-l%) +LLVMLibsPaths := $(IncludeInLibLlvm) + +$(LibName.SO): $(LLVMLibsPaths) + +ifeq ($(HOST_OS),Darwin) + # set dylib internal version number to llvmCore submission number + ifdef LLVM_SUBMIT_VERSION + LLVMLibsOptions := $(LLVMLibsOptions) -Wl,-current_version \ + -Wl,$(LLVM_SUBMIT_VERSION).$(LLVM_SUBMIT_SUBVERSION) \ + -Wl,-compatibility_version -Wl,1 + endif + # Include everything from the .a's into the shared library. + LLVMLibsOptions := $(LLVMLibsOptions) -all_load + # extra options to override libtool defaults + LLVMLibsOptions := $(LLVMLibsOptions) \ + -avoid-version \ + -Wl,-dead_strip \ + -Wl,-seg1addr -Wl,0xE0000000 + + # Mac OS X 10.4 and earlier tools do not allow a second -install_name on command line + DARWIN_VERS := $(shell echo $(TARGET_TRIPLE) | sed 's/.*darwin\([0-9]*\).*/\1/') + ifneq ($(DARWIN_VERS),8) + LLVMLibsOptions := $(LLVMLibsOptions) \ + -Wl,-install_name \ + -Wl,"@executable_path/../lib/lib$(LIBRARYNAME)$(SHLIBEXT)" + endif +endif + +ifeq ($(HOST_OS), Linux) + # Include everything from the .a's into the shared library. + LLVMLibsOptions := -Wl,--whole-archive $(LLVMLibsOptions) \ + -Wl,--no-whole-archive + # Don't allow unresolved symbols. + LLVMLibsOptions += -Wl,--no-undefined +endif diff --git a/tools/llvmc/example/mcc16/plugins/PIC16Base/PIC16Base.td b/tools/llvmc/example/mcc16/plugins/PIC16Base/PIC16Base.td index f13b9f8..25149ad 100644 --- a/tools/llvmc/example/mcc16/plugins/PIC16Base/PIC16Base.td +++ b/tools/llvmc/example/mcc16/plugins/PIC16Base/PIC16Base.td @@ -56,7 +56,7 @@ def OptionList : OptionList<[ // (parameter_option "pre-RA-sched", // (help "Example of an option that is passed to llc")), (parameter_option "regalloc", - (help "Register allocator to use.(possible values: simple, linearscan, pbqp, local. default = linearscan)")), + (help "Register allocator to use (possible values: simple, linearscan, pbqp, local; default=linearscan)")), (prefix_list_option "Wa,", (comma_separated), (help "Pass options to assembler (Run 'gpasm -help' for assembler options)")), (prefix_list_option "Wl,", (comma_separated), @@ -72,16 +72,14 @@ class clang_based<string language, string cmd, string ext_E> : Tool< [(in_language language), (out_language "llvm-bitcode"), (output_suffix "bc"), - (cmd_line (case - (switch_on "E"), - (case - (not_empty "o"), !strconcat(cmd, " -E $INFILE -o $OUTFILE"), - (default), !strconcat(cmd, " -E $INFILE")), - (default), !strconcat(cmd, " $INFILE -o $OUTFILE"))), - (actions (case - (and (multiple_input_files), (or (switch_on "S"), (switch_on "c"))), - (error "cannot specify -o with -c or -S with multiple files"), - (switch_on "E"), [(stop_compilation), (output_suffix ext_E)], + (command cmd), + (actions (case + (and (multiple_input_files), + (or (switch_on "S"), (switch_on "c"))), + (error "cannot specify -o with -c or -S with multiple files"), + (switch_on "E"), [(forward "E"), + (stop_compilation), (output_suffix ext_E)], + (and (switch_on "E"), (empty "o")), (no_out_file), (switch_on "bc"),[(stop_compilation), (output_suffix "bc")], (switch_on "g"), (append_cmd "-g"), (switch_on "w"), (append_cmd "-w"), @@ -116,12 +114,13 @@ def llvm_ld : Tool<[ (in_language "llvm-bitcode"), (out_language "llvm-bitcode"), (output_suffix "bc"), - (cmd_line "$CALL(GetBinDir)llvm-ld -L $CALL(GetStdLibsDir) -disable-gvn -disable-licm-promotion -disable-mem2reg $INFILE -b $OUTFILE -l std"), + (command "$CALL(GetBinDir)llvm-ld -L $CALL(GetStdLibsDir) -disable-licm-promotion -l std"), + (out_file_option "-b"), (actions (case (switch_on "O0"), (append_cmd "-disable-opt"), (switch_on "O1"), (append_cmd "-disable-opt"), // Whenever O3 is not specified on the command line, default i.e. disable-inlining will always be added. - (switch_on "O2"), (append_cmd ""), + (switch_on "O2"), (append_cmd ""), (switch_on "O3"), (append_cmd ""), (default), (append_cmd "-disable-inlining"))), (join) @@ -134,12 +133,13 @@ def llvm_ld_optimizer : Tool<[ (output_suffix "bc"), // FIXME: we are still not disabling licm-promotion. // -disable-licm-promotion and building stdn library causes c16-71 to fail. - (cmd_line "$CALL(GetBinDir)llvm-ld -disable-gvn -disable-mem2reg $INFILE -b $OUTFILE"), + (command "$CALL(GetBinDir)llvm-ld "), + (out_file_option "-b"), (actions (case (switch_on "O0"), (append_cmd "-disable-opt"), (switch_on "O1"), (append_cmd "-disable-opt"), // Whenever O3 is not specified on the command line, default i.e. disable-inlining will always be added. - (switch_on "O2"), (append_cmd ""), + (switch_on "O2"), (append_cmd ""), (switch_on "O3"), (append_cmd ""), (default), (append_cmd "-disable-inlining"))) ]>; @@ -149,7 +149,7 @@ def pic16passes : Tool<[ (in_language "llvm-bitcode"), (out_language "llvm-bitcode"), (output_suffix "obc"), - (cmd_line "$CALL(GetBinDir)opt -pic16cg -pic16overlay $INFILE -f -o $OUTFILE"), + (command "$CALL(GetBinDir)opt -pic16cloner -pic16overlay -f"), (actions (case (switch_on "O0"), (append_cmd "-disable-opt"))) ]>; @@ -158,7 +158,7 @@ def llc : Tool<[ (in_language "llvm-bitcode"), (out_language "assembler"), (output_suffix "s"), - (cmd_line "$CALL(GetBinDir)llc -march=pic16 -disable-jump-tables -pre-RA-sched=list-burr -f $INFILE -o $OUTFILE"), + (command "$CALL(GetBinDir)llc -march=pic16 -disable-jump-tables -pre-RA-sched=list-burr -f"), (actions (case (switch_on "S"), (stop_compilation), // (not_empty "Wllc,"), (unpack_values "Wllc,"), @@ -171,7 +171,7 @@ def gpasm : Tool<[ (in_language "assembler"), (out_language "object-code"), (output_suffix "o"), - (cmd_line "$CALL(GetBinDir)gpasm -z -r decimal -I $CALL(GetStdAsmHeadersDir) -C -c -w 2 $INFILE -o $OUTFILE"), + (command "$CALL(GetBinDir)gpasm -z -r decimal -I $CALL(GetStdAsmHeadersDir) -C -c -w 2"), (actions (case (switch_on "c"), (stop_compilation), (switch_on "g"), (append_cmd "-g"), @@ -184,7 +184,7 @@ def mplink : Tool<[ (in_language "object-code"), (out_language "executable"), (output_suffix "cof"), - (cmd_line "$CALL(GetBinDir)mplink -e -k $CALL(GetStdLinkerScriptsDir) -l $CALL(GetStdLibsDir) intrinsics.lib stdn.lib $INFILE -o $OUTFILE"), + (command "$CALL(GetBinDir)mplink -e -k $CALL(GetStdLinkerScriptsDir) -l $CALL(GetStdLibsDir) intrinsics.lib stdn.lib"), (actions (case (not_empty "Wl,"), (forward_value "Wl,"), (switch_on "X"), (append_cmd "-x"), @@ -217,13 +217,13 @@ def LanguageMap : LanguageMap<[ def CompilationGraph : CompilationGraph<[ Edge<"root", "clang_cc">, Edge<"root", "llvm_ld">, - OptionalEdge<"root", "llvm_ld_optimizer", (case + OptionalEdge<"root", "llvm_ld_optimizer", (case (switch_on "S"), (inc_weight), (switch_on "c"), (inc_weight))>, Edge<"root", "gpasm">, Edge<"root", "mplink">, Edge<"clang_cc", "llvm_ld">, - OptionalEdge<"clang_cc", "llvm_ld_optimizer", (case + OptionalEdge<"clang_cc", "llvm_ld_optimizer", (case (switch_on "S"), (inc_weight), (switch_on "c"), (inc_weight))>, Edge<"llvm_ld", "pic16passes">, diff --git a/tools/llvmc/plugins/Base/Base.td.in b/tools/llvmc/plugins/Base/Base.td.in index 1acd969..7b82313 100644 --- a/tools/llvmc/plugins/Base/Base.td.in +++ b/tools/llvmc/plugins/Base/Base.td.in @@ -49,7 +49,7 @@ def OptList : OptionList<[ (parameter_option "linker", (help "Choose linker (possible values: gcc, g++)")), (parameter_option "mtune", - (help "Target a specific CPU type"), (hidden)), + (help "Target a specific CPU type"), (hidden), (forward_not_split)), // TODO: Add a conditional compilation mechanism to make Darwin-only options // like '-arch' really Darwin-only. @@ -57,9 +57,9 @@ def OptList : OptionList<[ (parameter_option "arch", (help "Compile for the specified target architecture"), (hidden)), (parameter_option "march", - (help "A synonym for -mtune"), (hidden)), + (help "A synonym for -mtune"), (hidden), (forward_not_split)), (parameter_option "mcpu", - (help "A deprecated synonym for -mtune"), (hidden)), + (help "A deprecated synonym for -mtune"), (hidden), (forward_not_split)), (switch_option "mfix-and-continue", (help "Needed by gdb to load .o files dynamically"), (hidden)), (parameter_option "MF", @@ -76,14 +76,20 @@ def OptList : OptionList<[ (help "Specifies a framework to link against")), (parameter_list_option "weak_framework", (help "Specifies a framework to weakly link against"), (hidden)), + (parameter_option "filelist", (hidden), + (help "Link the files listed in file")), (prefix_list_option "F", (help "Add a directory to framework search path")), (prefix_list_option "I", (help "Add a directory to include path")), (prefix_list_option "D", (help "Define a macro")), + (parameter_list_option "Xpreprocessor", (hidden), + (help "Pass options to preprocessor")), (prefix_list_option "Wa,", (comma_separated), (help "Pass options to assembler")), + (parameter_list_option "Xassembler", (hidden), + (help "Pass options to assembler")), (prefix_list_option "Wllc,", (comma_separated), (help "Pass options to llc")), (prefix_list_option "L", @@ -92,6 +98,8 @@ def OptList : OptionList<[ (help "Search a library when linking")), (prefix_list_option "Wl,", (help "Pass options to linker")), + (parameter_list_option "Xlinker", (hidden), + (help "Pass options to linker")), (prefix_list_option "Wo,", (comma_separated), (help "Pass options to opt")), (prefix_list_option "m", @@ -105,6 +113,8 @@ def OptList : OptionList<[ (help "Remove unreachable blocks of code")), (switch_option "single_module", (hidden), (help "Build the library so it contains only one module")), + (parameter_option "install_name", (hidden), + (help "File name the library will be installed in")), (parameter_option "compatibility_version", (hidden), (help "Compatibility version number")), (parameter_option "current_version", (hidden), @@ -126,31 +136,31 @@ def Preprocess : OptionPreprocessor< // Tools -class llvm_gcc_based <string cmd_prefix, string in_lang, string E_ext> : Tool< +class llvm_gcc_based <string cmd_prefix, string in_lang, + string E_ext, string out_lang> : Tool< [(in_language in_lang), (out_language "llvm-bitcode"), - (output_suffix "bc"), - (cmd_line (case - (switch_on "E"), - (case (not_empty "o"), - !strconcat(cmd_prefix, " -E $INFILE -o $OUTFILE"), - (default), - !strconcat(cmd_prefix, " -E $INFILE")), - (switch_on "fsyntax-only"), - !strconcat(cmd_prefix, " -fsyntax-only $INFILE"), - (and (switch_on "S"), (switch_on "emit-llvm")), - !strconcat(cmd_prefix, " -S $INFILE -o $OUTFILE -emit-llvm"), - (default), - !strconcat(cmd_prefix, " -c $INFILE -o $OUTFILE -emit-llvm"))), + (output_suffix out_lang), + (command cmd_prefix), (actions (case - (and (multiple_input_files), (or (switch_on "S"), (switch_on "c"))), + (and (not_empty "o"), + (multiple_input_files), (or (switch_on "S"), (switch_on "c"))), (error "cannot specify -o with -c or -S with multiple files"), - (switch_on "E"), [(stop_compilation), (output_suffix E_ext)], + (switch_on "E"), + [(forward "E"), (stop_compilation), (output_suffix E_ext)], + (and (switch_on "E"), (empty "o")), (no_out_file), (switch_on ["emit-llvm", "S"]), [(output_suffix "ll"), (stop_compilation)], (switch_on ["emit-llvm", "c"]), (stop_compilation), - (switch_on "fsyntax-only"), (stop_compilation), + (switch_on "fsyntax-only"), [(forward "fsyntax-only"), + (no_out_file), (stop_compilation)], + (switch_on ["S", "emit-llvm"]), [(forward "S"), (forward "emit-llvm")], + (not (or (switch_on ["S", "emit-llvm"]), (switch_on "fsyntax-only"))), + [(append_cmd "-c"), (append_cmd "-emit-llvm")], + + // Forwards + (not_empty "Xpreprocessor"), (forward "Xpreprocessor"), (not_empty "include"), (forward "include"), (not_empty "iquote"), (forward "iquote"), (not_empty "save-temps"), (append_cmd "-save-temps"), @@ -176,12 +186,24 @@ class llvm_gcc_based <string cmd_prefix, string in_lang, string E_ext> : Tool< (sink) ]>; -def llvm_gcc_c : llvm_gcc_based<"@LLVMGCCCOMMAND@ -x c", "c", "i">; -def llvm_gcc_cpp : llvm_gcc_based<"@LLVMGXXCOMMAND@ -x c++", "c++", "i">; +def llvm_gcc_c : llvm_gcc_based<"@LLVMGCCCOMMAND@ -x c", "c", "i", "bc">; +def llvm_gcc_cpp : llvm_gcc_based<"@LLVMGXXCOMMAND@ -x c++", "c++", "i", "bc">; def llvm_gcc_m : llvm_gcc_based<"@LLVMGCCCOMMAND@ -x objective-c", - "objective-c", "mi">; + "objective-c", "mi", "bc">; def llvm_gcc_mxx : llvm_gcc_based<"@LLVMGCCCOMMAND@ -x objective-c++", - "objective-c++", "mi">; + "objective-c++", "mi", "bc">; + +def llvm_gcc_c_pch : llvm_gcc_based<"@LLVMGCCCOMMAND@ -x c-header", + "c-header", "i", "gch">; +def llvm_gcc_cpp_pch : llvm_gcc_based<"@LLVMGXXCOMMAND@ -x c++-header", + "c++-header", + "i", "gch">; +def llvm_gcc_m_pch : llvm_gcc_based<"@LLVMGCCCOMMAND@ -x objective-c-header", + "objective-c-header", + "mi", "gch">; +def llvm_gcc_mxx_pch + : llvm_gcc_based<"@LLVMGCCCOMMAND@ -x objective-c++-header", + "objective-c++-header", "mi", "gch">; def opt : Tool< [(in_language "llvm-bitcode"), @@ -191,14 +213,14 @@ def opt : Tool< (switch_on "O1"), (forward "O1"), (switch_on "O2"), (forward "O2"), (switch_on "O3"), (forward "O3"))), - (cmd_line "opt -f $INFILE -o $OUTFILE") + (command "opt -f") ]>; def llvm_as : Tool< [(in_language "llvm-assembler"), (out_language "llvm-bitcode"), (output_suffix "bc"), - (cmd_line "llvm-as $INFILE -o $OUTFILE"), + (command "llvm-as"), (actions (case (switch_on "emit-llvm"), (stop_compilation))) ]>; @@ -206,18 +228,19 @@ def llvm_gcc_assembler : Tool< [(in_language "assembler"), (out_language "object-code"), (output_suffix "o"), - (cmd_line "@LLVMGCCCOMMAND@ -c -x assembler $INFILE -o $OUTFILE"), + (command "@LLVMGCCCOMMAND@ -c -x assembler"), (actions (case (switch_on "c"), (stop_compilation), (not_empty "arch"), (forward "arch"), - (not_empty "Wa,"), (forward_value "Wa,"))) + (not_empty "Xassembler"), (forward "Xassembler"), + (not_empty "Wa,"), (forward "Wa,"))) ]>; def llc : Tool< [(in_language ["llvm-bitcode", "llvm-assembler"]), (out_language "assembler"), (output_suffix "s"), - (cmd_line "llc -f $INFILE -o $OUTFILE"), + (command "llc -f"), (actions (case (switch_on "S"), (stop_compilation), (switch_on "O0"), (forward "O0"), @@ -227,8 +250,8 @@ def llc : Tool< (switch_on "fPIC"), (append_cmd "-relocation-model=pic"), (switch_on "mdynamic-no-pic"), (append_cmd "-relocation-model=dynamic-no-pic"), - (not_empty "march"), (forward "mcpu"), - (not_empty "mtune"), (forward "mcpu"), + (not_empty "march"), (forward_as "mtune", "-mcpu"), + (not_empty "mtune"), (forward_as "mtune", "-mcpu"), (not_empty "mcpu"), (forward "mcpu"), (not_empty "m"), (forward_transformed_value "m", "ConvertToMAttr"), (not_empty "Wllc,"), (forward_value "Wllc,"))) @@ -239,7 +262,9 @@ class llvm_gcc_based_linker <string cmd_prefix> : Tool< [(in_language "object-code"), (out_language "executable"), (output_suffix "out"), - (cmd_line !strconcat(cmd_prefix, " $INFILE -o $OUTFILE")), + (command cmd_prefix), + (works_on_empty (case (not_empty "filelist"), true, + (default), false)), (join), (actions (case (switch_on "pthread"), (append_cmd "-lpthread"), @@ -248,9 +273,11 @@ class llvm_gcc_based_linker <string cmd_prefix> : Tool< (not_empty "arch"), (forward "arch"), (not_empty "framework"), (forward "framework"), (not_empty "weak_framework"), (forward "weak_framework"), + (not_empty "filelist"), (forward "filelist"), (switch_on "m32"), (forward "m32"), (switch_on "m64"), (forward "m64"), (not_empty "l"), (forward "l"), + (not_empty "Xlinker"), (forward "Xlinker"), (not_empty "Wl,"), (forward "Wl,"), (switch_on "dynamiclib"), (forward "dynamiclib"), (switch_on "prebind"), (forward "prebind"), @@ -258,8 +285,8 @@ class llvm_gcc_based_linker <string cmd_prefix> : Tool< (switch_on "single_module"), (forward "single_module"), (not_empty "compatibility_version"), (forward "compatibility_version"), - (not_empty "current_version"), - (forward "current_version"))) + (not_empty "current_version"), (forward "current_version"), + (not_empty "install_name"), (forward "install_name"))) ]>; // Default linker @@ -271,16 +298,20 @@ def llvm_gcc_cpp_linker : llvm_gcc_based_linker<"@LLVMGXXCOMMAND@">; def LanguageMap : LanguageMap< [LangToSuffixes<"c++", ["cc", "cp", "cxx", "cpp", "CPP", "c++", "C"]>, + LangToSuffixes<"c++-header", ["hpp"]>, LangToSuffixes<"c", ["c"]>, + LangToSuffixes<"c-header", ["h"]>, LangToSuffixes<"c-cpp-output", ["i"]>, LangToSuffixes<"objective-c-cpp-output", ["mi"]>, LangToSuffixes<"objective-c++", ["mm"]>, + LangToSuffixes<"objective-c++-header", ["hmm"]>, LangToSuffixes<"objective-c", ["m"]>, + LangToSuffixes<"objective-c-header", ["hm"]>, LangToSuffixes<"assembler", ["s"]>, LangToSuffixes<"assembler-with-cpp", ["S"]>, LangToSuffixes<"llvm-assembler", ["ll"]>, LangToSuffixes<"llvm-bitcode", ["bc"]>, - LangToSuffixes<"object-code", ["o"]>, + LangToSuffixes<"object-code", ["o", "*empty*"]>, LangToSuffixes<"executable", ["out"]> ]>; @@ -294,6 +325,11 @@ def CompilationGraph : CompilationGraph<[ Edge<"root", "llvm_gcc_mxx">, Edge<"root", "llc">, + Edge<"root", "llvm_gcc_c_pch">, + Edge<"root", "llvm_gcc_cpp_pch">, + Edge<"root", "llvm_gcc_m_pch">, + Edge<"root", "llvm_gcc_mxx_pch">, + Edge<"llvm_gcc_c", "llc">, Edge<"llvm_gcc_cpp", "llc">, Edge<"llvm_gcc_m", "llc">, diff --git a/tools/llvmc/plugins/Clang/Clang.td b/tools/llvmc/plugins/Clang/Clang.td index ac8ac15..988d9b1 100644 --- a/tools/llvmc/plugins/Clang/Clang.td +++ b/tools/llvmc/plugins/Clang/Clang.td @@ -1,12 +1,3 @@ -// A replacement for the Clang's ccc script. -// Depends on the Base plugin. -// To compile, use this command: -// -// cd $LLVMC2_DIR -// make DRIVER_NAME=ccc2 BUILTIN_PLUGINS=Clang -// -// Or just use the default llvmc, which now has this plugin enabled. - include "llvm/CompilerDriver/Common.td" def Priority : PluginPriority<1>; @@ -33,23 +24,17 @@ class clang_based<string language, string cmd, string ext_E> : Tool< [(in_language language), (out_language "llvm-bitcode"), (output_suffix "bc"), - (cmd_line (case - (switch_on "E"), - (case - (not_empty "o"), - !strconcat(cmd, " -E $INFILE -o $OUTFILE"), - (default), - !strconcat(cmd, " -E $INFILE")), - (and (switch_on "S"), (switch_on "emit-llvm")), - !strconcat(cmd, " -emit-llvm $INFILE -o $OUTFILE"), - (default), - !strconcat(cmd, " -emit-llvm-bc $INFILE -o $OUTFILE"))), + (command cmd), (actions (case (switch_on "E"), - [(stop_compilation), (output_suffix ext_E)], + [(forward "E"), (stop_compilation), (output_suffix ext_E)], + (and (switch_on "E"), (empty "o")), (no_out_file), (switch_on "fsyntax-only"), (stop_compilation), - (and (switch_on "S"), (switch_on "emit-llvm")), - [(stop_compilation), (output_suffix "ll")], - (and (switch_on "c"), (switch_on "emit-llvm")), + (switch_on ["S", "emit-llvm"]), + [(append_cmd "-emit-llvm"), + (stop_compilation), (output_suffix "ll")], + (not (switch_on ["S", "emit-llvm"])), + (append_cmd "-emit-llvm-bc"), + (switch_on ["c", "emit-llvm"]), (stop_compilation), (not_empty "include"), (forward "include"), (not_empty "I"), (forward "I"))), @@ -67,7 +52,7 @@ def as : Tool< [(in_language "assembler"), (out_language "object-code"), (output_suffix "o"), - (cmd_line "as $INFILE -o $OUTFILE"), + (command "as"), (actions (case (not_empty "Wa,"), (forward_value "Wa,"), (switch_on "c"), (stop_compilation))) ]>; @@ -77,7 +62,7 @@ def llvm_ld : Tool< [(in_language "object-code"), (out_language "executable"), (output_suffix "out"), - (cmd_line "llvm-ld -native -disable-internalize $INFILE -o $OUTFILE"), + (command "llvm-ld -native -disable-internalize"), (actions (case (switch_on "pthread"), (append_cmd "-lpthread"), (not_empty "L"), (forward "L"), diff --git a/tools/opt/opt.cpp b/tools/opt/opt.cpp index 5200180..b123d54 100644 --- a/tools/opt/opt.cpp +++ b/tools/opt/opt.cpp @@ -475,10 +475,11 @@ int main(int argc, char **argv) { errs() << argv[0] << ": cannot create pass: " << PassInf->getPassName() << "\n"; if (P) { + PassKind Kind = P->getPassKind(); addPass(Passes, P); if (AnalyzeOnly) { - switch (P->getPassKind()) { + switch (Kind) { case PT_BasicBlock: Passes.add(new BasicBlockPassPrinter(PassInf)); break; diff --git a/unittests/ADT/APFloatTest.cpp b/unittests/ADT/APFloatTest.cpp index b02cc3e..5719960 100644 --- a/unittests/ADT/APFloatTest.cpp +++ b/unittests/ADT/APFloatTest.cpp @@ -236,6 +236,13 @@ TEST(APFloatTest, fromDecimalString) { EXPECT_EQ(2.05e12, APFloat(APFloat::IEEEdouble, "002.05000e12").convertToDouble()); EXPECT_EQ(2.05e+12, APFloat(APFloat::IEEEdouble, "002.05000e+12").convertToDouble()); EXPECT_EQ(2.05e-12, APFloat(APFloat::IEEEdouble, "002.05000e-12").convertToDouble()); + + // These are "carefully selected" to overflow the fast log-base + // calculations in APFloat.cpp + EXPECT_TRUE(APFloat(APFloat::IEEEdouble, "99e99999").isInfinity()); + EXPECT_TRUE(APFloat(APFloat::IEEEdouble, "-99e99999").isInfinity()); + EXPECT_TRUE(APFloat(APFloat::IEEEdouble, "1e-99999").isPosZero()); + EXPECT_TRUE(APFloat(APFloat::IEEEdouble, "-1e-99999").isNegZero()); } TEST(APFloatTest, fromHexadecimalString) { @@ -337,6 +344,35 @@ TEST(APFloatTest, toString) { ASSERT_EQ("8.731834E+2", convertToString(873.1834, 0, 0)); } +static APInt nanbits(const fltSemantics &Sem, + bool SNaN, bool Negative, uint64_t fill) { + APInt apfill(64, fill); + if (SNaN) + return APFloat::getSNaN(Sem, Negative, &apfill).bitcastToAPInt(); + else + return APFloat::getQNaN(Sem, Negative, &apfill).bitcastToAPInt(); +} + +TEST(APFloatTest, makeNaN) { + ASSERT_EQ(0x7fc00000, nanbits(APFloat::IEEEsingle, false, false, 0)); + ASSERT_EQ(0xffc00000, nanbits(APFloat::IEEEsingle, false, true, 0)); + ASSERT_EQ(0x7fc0ae72, nanbits(APFloat::IEEEsingle, false, false, 0xae72)); + ASSERT_EQ(0x7fffae72, nanbits(APFloat::IEEEsingle, false, false, 0xffffae72)); + ASSERT_EQ(0x7fa00000, nanbits(APFloat::IEEEsingle, true, false, 0)); + ASSERT_EQ(0xffa00000, nanbits(APFloat::IEEEsingle, true, true, 0)); + ASSERT_EQ(0x7f80ae72, nanbits(APFloat::IEEEsingle, true, false, 0xae72)); + ASSERT_EQ(0x7fbfae72, nanbits(APFloat::IEEEsingle, true, false, 0xffffae72)); + + ASSERT_EQ(0x7ff8000000000000ULL, nanbits(APFloat::IEEEdouble, false, false, 0)); + ASSERT_EQ(0xfff8000000000000ULL, nanbits(APFloat::IEEEdouble, false, true, 0)); + ASSERT_EQ(0x7ff800000000ae72ULL, nanbits(APFloat::IEEEdouble, false, false, 0xae72)); + ASSERT_EQ(0x7fffffffffffae72ULL, nanbits(APFloat::IEEEdouble, false, false, 0xffffffffffffae72ULL)); + ASSERT_EQ(0x7ff4000000000000ULL, nanbits(APFloat::IEEEdouble, true, false, 0)); + ASSERT_EQ(0xfff4000000000000ULL, nanbits(APFloat::IEEEdouble, true, true, 0)); + ASSERT_EQ(0x7ff000000000ae72ULL, nanbits(APFloat::IEEEdouble, true, false, 0xae72)); + ASSERT_EQ(0x7ff7ffffffffae72ULL, nanbits(APFloat::IEEEdouble, true, false, 0xffffffffffffae72ULL)); +} + #ifdef GTEST_HAS_DEATH_TEST TEST(APFloatTest, SemanticsDeath) { EXPECT_DEATH(APFloat(APFloat::IEEEsingle, 0.0f).convertToDouble(), "Float semantics are not IEEEdouble"); diff --git a/unittests/Makefile.unittest b/unittests/Makefile.unittest index 6fbef54..b1c1d2c 100644 --- a/unittests/Makefile.unittest +++ b/unittests/Makefile.unittest @@ -28,6 +28,15 @@ CPP.Flags += -I$(LLVM_SRC_ROOT)/utils/unittest/googletest/include/ CPP.Flags += $(NO_VARIADIC_MACROS) TESTLIBS = -lGoogleTest -lUnitTestMain +ifeq ($(ENABLE_SHARED), 1) + # Add the absolute path to the dynamic library. This is ok because + # we'll never install unittests. + LD.Flags += $(RPATH) -Wl,$(LibDir) + # Also set {DYLD,LD}_LIBRARY_PATH because OSX ignores the rpath most + # of the time. + Run.Shared := $(SHLIBPATH_VAR)="$(LibDir)$${$(SHLIBPATH_VAR):+:}$$$(SHLIBPATH_VAR)" +endif + $(LLVMUnitTestExe): $(ObjectsO) $(ProjLibsPaths) $(LLVMLibsPaths) $(Echo) Linking $(BuildMode) unit test $(TESTNAME) $(StripWarnMsg) $(Verb) $(Link) -o $@ $(TOOLLINKOPTS) $(ObjectsO) $(ProjLibsOptions) \ @@ -38,6 +47,6 @@ $(LLVMUnitTestExe): $(ObjectsO) $(ProjLibsPaths) $(LLVMLibsPaths) all:: $(LLVMUnitTestExe) unitcheck:: $(LLVMUnitTestExe) - $(LLVMUnitTestExe) + $(Run.Shared) $(LLVMUnitTestExe) endif diff --git a/unittests/Support/AllocatorTest.cpp b/unittests/Support/AllocatorTest.cpp index 2a01f3a..6c0fca9 100644 --- a/unittests/Support/AllocatorTest.cpp +++ b/unittests/Support/AllocatorTest.cpp @@ -88,7 +88,7 @@ TEST(AllocatorTest, TestOverflow) { Alloc.Allocate(4096 - sizeof(MemSlab), 0); EXPECT_EQ(1U, Alloc.GetNumSlabs()); - // If we dont't allocate a new slab, then we will have overflowed. + // If we don't allocate a new slab, then we will have overflowed. Alloc.Allocate(1, 0); EXPECT_EQ(2U, Alloc.GetNumSlabs()); } diff --git a/unittests/Support/RegexTest.cpp b/unittests/Support/RegexTest.cpp index 44c7e55..65b66c3 100644 --- a/unittests/Support/RegexTest.cpp +++ b/unittests/Support/RegexTest.cpp @@ -62,4 +62,33 @@ TEST_F(RegexTest, Basics) { EXPECT_TRUE(r5.match(String)); } +TEST_F(RegexTest, Substitution) { + std::string Error; + + EXPECT_EQ("aNUMber", Regex("[0-9]+").sub("NUM", "a1234ber")); + + // Standard Escapes + EXPECT_EQ("a\\ber", Regex("[0-9]+").sub("\\\\", "a1234ber", &Error)); + EXPECT_EQ(Error, ""); + EXPECT_EQ("a\nber", Regex("[0-9]+").sub("\\n", "a1234ber", &Error)); + EXPECT_EQ(Error, ""); + EXPECT_EQ("a\tber", Regex("[0-9]+").sub("\\t", "a1234ber", &Error)); + EXPECT_EQ(Error, ""); + EXPECT_EQ("ajber", Regex("[0-9]+").sub("\\j", "a1234ber", &Error)); + EXPECT_EQ(Error, ""); + + EXPECT_EQ("aber", Regex("[0-9]+").sub("\\", "a1234ber", &Error)); + EXPECT_EQ(Error, "replacement string contained trailing backslash"); + + // Backreferences + EXPECT_EQ("aa1234bber", Regex("a[0-9]+b").sub("a\\0b", "a1234ber", &Error)); + EXPECT_EQ(Error, ""); + + EXPECT_EQ("a1234ber", Regex("a([0-9]+)b").sub("a\\1b", "a1234ber", &Error)); + EXPECT_EQ(Error, ""); + + EXPECT_EQ("aber", Regex("a[0-9]+b").sub("a\\100b", "a1234ber", &Error)); + EXPECT_EQ(Error, "invalid backreference string '100'"); +} + } diff --git a/utils/GenLibDeps.pl b/utils/GenLibDeps.pl index b320a91..f1f7e72 100755 --- a/utils/GenLibDeps.pl +++ b/utils/GenLibDeps.pl @@ -70,6 +70,8 @@ opendir DIR,$Directory; my @files = readdir DIR; closedir DIR; my @libs = grep(/libLLVM.*\.(dylib|so|a)$/,sort(@files)); +# Omit the all-of-llvm shared library. +@libs = grep(!/libLLVM-\d\.\d(svn)?\.(dylib|so)/, @libs); my @objs = grep(/LLVM.*\.o$/,sort(@files)); # Declare the hashes we will use to keep track of the library and object file diff --git a/utils/TableGen/CMakeLists.txt b/utils/TableGen/CMakeLists.txt index a2678a2..881b50a 100644 --- a/utils/TableGen/CMakeLists.txt +++ b/utils/TableGen/CMakeLists.txt @@ -11,6 +11,7 @@ add_executable(tblgen DAGISelEmitter.cpp DAGISelMatcherEmitter.cpp DAGISelMatcherGen.cpp + DAGISelMatcherOpt.cpp DAGISelMatcher.cpp DisassemblerEmitter.cpp EDEmitter.cpp diff --git a/utils/TableGen/CodeGenDAGPatterns.cpp b/utils/TableGen/CodeGenDAGPatterns.cpp index 94d3534..ce737bf 100644 --- a/utils/TableGen/CodeGenDAGPatterns.cpp +++ b/utils/TableGen/CodeGenDAGPatterns.cpp @@ -447,6 +447,30 @@ SDNodeInfo::SDNodeInfo(Record *R) : Def(R) { TypeConstraints.assign(ConstraintList.begin(), ConstraintList.end()); } +/// getKnownType - If the type constraints on this node imply a fixed type +/// (e.g. all stores return void, etc), then return it as an +/// MVT::SimpleValueType. Otherwise, return EEVT::isUnknown. +unsigned SDNodeInfo::getKnownType() const { + unsigned NumResults = getNumResults(); + assert(NumResults <= 1 && + "We only work with nodes with zero or one result so far!"); + + for (unsigned i = 0, e = TypeConstraints.size(); i != e; ++i) { + // Make sure that this applies to the correct node result. + if (TypeConstraints[i].OperandNo >= NumResults) // FIXME: need value # + continue; + + switch (TypeConstraints[i].ConstraintType) { + default: break; + case SDTypeConstraint::SDTCisVT: + return TypeConstraints[i].x.SDTCisVT_Info.VT; + case SDTypeConstraint::SDTCisPtrTy: + return MVT::iPTR; + } + } + return EEVT::isUnknown; +} + //===----------------------------------------------------------------------===// // TreePatternNode implementation // @@ -568,33 +592,36 @@ bool TreePatternNode::UpdateNodeType(const std::vector<unsigned char> &ExtVTs, return true; // unreachable } +static std::string GetTypeName(unsigned char TypeID) { + switch (TypeID) { + case MVT::Other: return "Other"; + case MVT::iAny: return "iAny"; + case MVT::fAny: return "fAny"; + case MVT::vAny: return "vAny"; + case EEVT::isUnknown: return "isUnknown"; + case MVT::iPTR: return "iPTR"; + case MVT::iPTRAny: return "iPTRAny"; + default: + std::string VTName = llvm::getName((MVT::SimpleValueType)TypeID); + // Strip off EVT:: prefix if present. + if (VTName.substr(0,5) == "MVT::") + VTName = VTName.substr(5); + return VTName; + } +} + void TreePatternNode::print(raw_ostream &OS) const { if (isLeaf()) { OS << *getLeafValue(); } else { - OS << "(" << getOperator()->getName(); + OS << '(' << getOperator()->getName(); } // FIXME: At some point we should handle printing all the value types for // nodes that are multiply typed. - switch (getExtTypeNum(0)) { - case MVT::Other: OS << ":Other"; break; - case MVT::iAny: OS << ":iAny"; break; - case MVT::fAny : OS << ":fAny"; break; - case MVT::vAny: OS << ":vAny"; break; - case EEVT::isUnknown: ; /*OS << ":?";*/ break; - case MVT::iPTR: OS << ":iPTR"; break; - case MVT::iPTRAny: OS << ":iPTRAny"; break; - default: { - std::string VTName = llvm::getName(getTypeNum(0)); - // Strip off EVT:: prefix if present. - if (VTName.substr(0,5) == "MVT::") - VTName = VTName.substr(5); - OS << ":" << VTName; - break; - } - } + if (getExtTypeNum(0) != EEVT::isUnknown) + OS << ':' << GetTypeName(getExtTypeNum(0)); if (!isLeaf()) { if (getNumChildren() != 0) { @@ -956,19 +983,25 @@ bool TreePatternNode::ApplyTypeConstraints(TreePattern &TP, bool NotRegisters) { MadeChange |= UpdateNodeType(MVT::isVoid, TP); } return MadeChange; - } else if (getOperator()->getName() == "implicit" || - getOperator()->getName() == "parallel") { + } + + if (getOperator()->getName() == "implicit" || + getOperator()->getName() == "parallel") { bool MadeChange = false; for (unsigned i = 0; i < getNumChildren(); ++i) MadeChange = getChild(i)->ApplyTypeConstraints(TP, NotRegisters); MadeChange |= UpdateNodeType(MVT::isVoid, TP); return MadeChange; - } else if (getOperator()->getName() == "COPY_TO_REGCLASS") { + } + + if (getOperator()->getName() == "COPY_TO_REGCLASS") { bool MadeChange = false; MadeChange |= getChild(0)->ApplyTypeConstraints(TP, NotRegisters); MadeChange |= getChild(1)->ApplyTypeConstraints(TP, NotRegisters); return MadeChange; - } else if (const CodeGenIntrinsic *Int = getIntrinsicInfo(CDP)) { + } + + if (const CodeGenIntrinsic *Int = getIntrinsicInfo(CDP)) { bool MadeChange = false; // Apply the result type to the node. @@ -992,7 +1025,9 @@ bool TreePatternNode::ApplyTypeConstraints(TreePattern &TP, bool NotRegisters) { MadeChange |= getChild(i)->ApplyTypeConstraints(TP, NotRegisters); } return MadeChange; - } else if (getOperator()->isSubClassOf("SDNode")) { + } + + if (getOperator()->isSubClassOf("SDNode")) { const SDNodeInfo &NI = CDP.getSDNodeInfo(getOperator()); bool MadeChange = NI.ApplyTypeConstraints(this, TP); @@ -1004,7 +1039,9 @@ bool TreePatternNode::ApplyTypeConstraints(TreePattern &TP, bool NotRegisters) { MadeChange |= UpdateNodeType(MVT::isVoid, TP); return MadeChange; - } else if (getOperator()->isSubClassOf("Instruction")) { + } + + if (getOperator()->isSubClassOf("Instruction")) { const DAGInstruction &Inst = CDP.getInstruction(getOperator()); bool MadeChange = false; unsigned NumResults = Inst.getNumResults(); @@ -1080,24 +1117,24 @@ bool TreePatternNode::ApplyTypeConstraints(TreePattern &TP, bool NotRegisters) { "' was provided too many operands!"); return MadeChange; - } else { - assert(getOperator()->isSubClassOf("SDNodeXForm") && "Unknown node type!"); - - // Node transforms always take one operand. - if (getNumChildren() != 1) - TP.error("Node transform '" + getOperator()->getName() + - "' requires one operand!"); - - // If either the output or input of the xform does not have exact - // type info. We assume they must be the same. Otherwise, it is perfectly - // legal to transform from one type to a completely different type. - if (!hasTypeSet() || !getChild(0)->hasTypeSet()) { - bool MadeChange = UpdateNodeType(getChild(0)->getExtTypes(), TP); - MadeChange |= getChild(0)->UpdateNodeType(getExtTypes(), TP); - return MadeChange; - } - return false; } + + assert(getOperator()->isSubClassOf("SDNodeXForm") && "Unknown node type!"); + + // Node transforms always take one operand. + if (getNumChildren() != 1) + TP.error("Node transform '" + getOperator()->getName() + + "' requires one operand!"); + + // If either the output or input of the xform does not have exact + // type info. We assume they must be the same. Otherwise, it is perfectly + // legal to transform from one type to a completely different type. + if (!hasTypeSet() || !getChild(0)->hasTypeSet()) { + bool MadeChange = UpdateNodeType(getChild(0)->getExtTypes(), TP); + MadeChange |= getChild(0)->UpdateNodeType(getExtTypes(), TP); + return MadeChange; + } + return false; } /// OnlyOnRHSOfCommutative - Return true if this value is only allowed on the @@ -1370,7 +1407,6 @@ void TreePattern::dump() const { print(errs()); } // CodeGenDAGPatterns implementation // -// FIXME: REMOVE OSTREAM ARGUMENT CodeGenDAGPatterns::CodeGenDAGPatterns(RecordKeeper &R) : Records(R) { Intrinsics = LoadIntrinsics(Records, false); TgtIntrinsics = LoadIntrinsics(Records, true); @@ -1569,10 +1605,10 @@ void CodeGenDAGPatterns::ParseDefaultOperands() { if (TPN->ContainsUnresolvedType()) { if (iter == 0) throw "Value #" + utostr(i) + " of PredicateOperand '" + - DefaultOps[iter][i]->getName() + "' doesn't have a concrete type!"; + DefaultOps[iter][i]->getName() +"' doesn't have a concrete type!"; else throw "Value #" + utostr(i) + " of OptionalDefOperand '" + - DefaultOps[iter][i]->getName() + "' doesn't have a concrete type!"; + DefaultOps[iter][i]->getName() +"' doesn't have a concrete type!"; } DefaultOpInfo.DefaultOps.push_back(TPN); } @@ -1616,21 +1652,21 @@ static bool HandleUse(TreePattern *I, TreePatternNode *Pat, TreePatternNode *&Slot = InstInputs[Pat->getName()]; if (!Slot) { Slot = Pat; + return true; + } + Record *SlotRec; + if (Slot->isLeaf()) { + SlotRec = dynamic_cast<DefInit*>(Slot->getLeafValue())->getDef(); } else { - Record *SlotRec; - if (Slot->isLeaf()) { - SlotRec = dynamic_cast<DefInit*>(Slot->getLeafValue())->getDef(); - } else { - assert(Slot->getNumChildren() == 0 && "can't be a use with children!"); - SlotRec = Slot->getOperator(); - } - - // Ensure that the inputs agree if we've already seen this input. - if (Rec != SlotRec) - I->error("All $" + Pat->getName() + " inputs must agree with each other"); - if (Slot->getExtTypes() != Pat->getExtTypes()) - I->error("All $" + Pat->getName() + " inputs must agree with each other"); + assert(Slot->getNumChildren() == 0 && "can't be a use with children!"); + SlotRec = Slot->getOperator(); } + + // Ensure that the inputs agree if we've already seen this input. + if (Rec != SlotRec) + I->error("All $" + Pat->getName() + " inputs must agree with each other"); + if (Slot->getExtTypes() != Pat->getExtTypes()) + I->error("All $" + Pat->getName() + " inputs must agree with each other"); return true; } @@ -1648,7 +1684,9 @@ FindPatternInputsAndOutputs(TreePattern *I, TreePatternNode *Pat, if (!isUse && Pat->getTransformFn()) I->error("Cannot specify a transform function for a non-input value!"); return; - } else if (Pat->getOperator()->getName() == "implicit") { + } + + if (Pat->getOperator()->getName() == "implicit") { for (unsigned i = 0, e = Pat->getNumChildren(); i != e; ++i) { TreePatternNode *Dest = Pat->getChild(i); if (!Dest->isLeaf()) @@ -1660,7 +1698,9 @@ FindPatternInputsAndOutputs(TreePattern *I, TreePatternNode *Pat, InstImpResults.push_back(Val->getDef()); } return; - } else if (Pat->getOperator()->getName() != "set") { + } + + if (Pat->getOperator()->getName() != "set") { // If this is not a set, verify that the children nodes are not void typed, // and recurse. for (unsigned i = 0, e = Pat->getNumChildren(); i != e; ++i) { @@ -1677,7 +1717,7 @@ FindPatternInputsAndOutputs(TreePattern *I, TreePatternNode *Pat, if (!isUse && Pat->getTransformFn()) I->error("Cannot specify a transform function for a non-input value!"); return; - } + } // Otherwise, this is a set, validate and collect instruction results. if (Pat->getNumChildren() == 0) @@ -2063,20 +2103,100 @@ void CodeGenDAGPatterns::ParseInstructions() { SrcPattern = Pattern; } - std::string Reason; - if (!SrcPattern->canPatternMatch(Reason, *this)) - I->error("Instruction can never match: " + Reason); - Record *Instr = II->first; - TreePatternNode *DstPattern = TheInst.getResultPattern(); - PatternsToMatch. - push_back(PatternToMatch(Instr->getValueAsListInit("Predicates"), - SrcPattern, DstPattern, TheInst.getImpResults(), - Instr->getValueAsInt("AddedComplexity"))); + AddPatternToMatch(I, + PatternToMatch(Instr->getValueAsListInit("Predicates"), + SrcPattern, + TheInst.getResultPattern(), + TheInst.getImpResults(), + Instr->getValueAsInt("AddedComplexity"), + Instr->getID())); } } +typedef std::pair<const TreePatternNode*, unsigned> NameRecord; + +static void FindNames(const TreePatternNode *P, + std::map<std::string, NameRecord> &Names, + const TreePattern *PatternTop) { + if (!P->getName().empty()) { + NameRecord &Rec = Names[P->getName()]; + // If this is the first instance of the name, remember the node. + if (Rec.second++ == 0) + Rec.first = P; + else if (Rec.first->getExtTypes() != P->getExtTypes()) + PatternTop->error("repetition of value: $" + P->getName() + + " where different uses have different types!"); + } + + if (!P->isLeaf()) { + for (unsigned i = 0, e = P->getNumChildren(); i != e; ++i) + FindNames(P->getChild(i), Names, PatternTop); + } +} + +void CodeGenDAGPatterns::AddPatternToMatch(const TreePattern *Pattern, + const PatternToMatch &PTM) { + // Do some sanity checking on the pattern we're about to match. + std::string Reason; + if (!PTM.getSrcPattern()->canPatternMatch(Reason, *this)) + Pattern->error("Pattern can never match: " + Reason); + + // If the source pattern's root is a complex pattern, that complex pattern + // must specify the nodes it can potentially match. + if (const ComplexPattern *CP = + PTM.getSrcPattern()->getComplexPatternInfo(*this)) + if (CP->getRootNodes().empty()) + Pattern->error("ComplexPattern at root must specify list of opcodes it" + " could match"); + + + // Find all of the named values in the input and output, ensure they have the + // same type. + std::map<std::string, NameRecord> SrcNames, DstNames; + FindNames(PTM.getSrcPattern(), SrcNames, Pattern); + FindNames(PTM.getDstPattern(), DstNames, Pattern); + + // Scan all of the named values in the destination pattern, rejecting them if + // they don't exist in the input pattern. + for (std::map<std::string, NameRecord>::iterator + I = DstNames.begin(), E = DstNames.end(); I != E; ++I) { + if (SrcNames[I->first].first == 0) + Pattern->error("Pattern has input without matching name in output: $" + + I->first); + +#if 0 + const std::vector<unsigned char> &SrcTypeVec = + SrcNames[I->first].first->getExtTypes(); + const std::vector<unsigned char> &DstTypeVec = + I->second.first->getExtTypes(); + if (SrcTypeVec == DstTypeVec) continue; + + std::string SrcType, DstType; + for (unsigned i = 0, e = SrcTypeVec.size(); i != e; ++i) + SrcType += ":" + GetTypeName(SrcTypeVec[i]); + for (unsigned i = 0, e = DstTypeVec.size(); i != e; ++i) + DstType += ":" + GetTypeName(DstTypeVec[i]); + + Pattern->error("Variable $" + I->first + + " has different types in source (" + SrcType + + ") and dest (" + DstType + ") pattern!"); +#endif + } + + // Scan all of the named values in the source pattern, rejecting them if the + // name isn't used in the dest, and isn't used to tie two values together. + for (std::map<std::string, NameRecord>::iterator + I = SrcNames.begin(), E = SrcNames.end(); I != E; ++I) + if (DstNames[I->first].first == 0 && SrcNames[I->first].second == 1) + Pattern->error("Pattern has dead named input: $" + I->first); + + PatternsToMatch.push_back(PTM); +} + + + void CodeGenDAGPatterns::InferInstructionFlags() { std::map<std::string, CodeGenInstruction> &InstrDescs = Target.getInstructions(); @@ -2204,15 +2324,13 @@ void CodeGenDAGPatterns::ParsePatterns() { TreePattern Temp(Result->getRecord(), DstPattern, false, *this); Temp.InferAllTypes(); - std::string Reason; - if (!Pattern->getTree(0)->canPatternMatch(Reason, *this)) - Pattern->error("Pattern can never match: " + Reason); - PatternsToMatch. - push_back(PatternToMatch(Patterns[i]->getValueAsListInit("Predicates"), - Pattern->getTree(0), - Temp.getOnlyTree(), InstImpResults, - Patterns[i]->getValueAsInt("AddedComplexity"))); + AddPatternToMatch(Pattern, + PatternToMatch(Patterns[i]->getValueAsListInit("Predicates"), + Pattern->getTree(0), + Temp.getOnlyTree(), InstImpResults, + Patterns[i]->getValueAsInt("AddedComplexity"), + Patterns[i]->getID())); } } @@ -2234,13 +2352,13 @@ static void CombineChildVariants(TreePatternNode *Orig, bool NotDone; do { #ifndef NDEBUG - if (DebugFlag && !Idxs.empty()) { - errs() << Orig->getOperator()->getName() << ": Idxs = [ "; - for (unsigned i = 0; i < Idxs.size(); ++i) { - errs() << Idxs[i] << " "; - } - errs() << "]\n"; - } + DEBUG(if (!Idxs.empty()) { + errs() << Orig->getOperator()->getName() << ": Idxs = [ "; + for (unsigned i = 0; i < Idxs.size(); ++i) { + errs() << Idxs[i] << " "; + } + errs() << "]\n"; + }); #endif // Create the variant and add it to the output list. std::vector<TreePatternNode*> NewChildren; @@ -2506,7 +2624,8 @@ void CodeGenDAGPatterns::GenerateVariants() { push_back(PatternToMatch(PatternsToMatch[i].getPredicates(), Variant, PatternsToMatch[i].getDstPattern(), PatternsToMatch[i].getDstRegs(), - PatternsToMatch[i].getAddedComplexity())); + PatternsToMatch[i].getAddedComplexity(), + Record::getNewUID())); } DEBUG(errs() << "\n"); diff --git a/utils/TableGen/CodeGenDAGPatterns.h b/utils/TableGen/CodeGenDAGPatterns.h index f1f7bca..37d633e 100644 --- a/utils/TableGen/CodeGenDAGPatterns.h +++ b/utils/TableGen/CodeGenDAGPatterns.h @@ -125,6 +125,11 @@ public: return TypeConstraints; } + /// getKnownType - If the type constraints on this node imply a fixed type + /// (e.g. all stores return void, etc), then return it as an + /// MVT::SimpleValueType. Otherwise, return EEVT::isUnknown. + unsigned getKnownType() const; + /// hasProperty - Return true if this node has the specified property. /// bool hasProperty(enum SDNP Prop) const { return Properties & (1 << Prop); } @@ -466,19 +471,21 @@ public: /// PatternToMatch - Used by CodeGenDAGPatterns to keep tab of patterns /// processed to produce isel. -struct PatternToMatch { +class PatternToMatch { +public: PatternToMatch(ListInit *preds, TreePatternNode *src, TreePatternNode *dst, const std::vector<Record*> &dstregs, - unsigned complexity): - Predicates(preds), SrcPattern(src), DstPattern(dst), Dstregs(dstregs), - AddedComplexity(complexity) {} + unsigned complexity, unsigned uid) + : Predicates(preds), SrcPattern(src), DstPattern(dst), + Dstregs(dstregs), AddedComplexity(complexity), ID(uid) {} ListInit *Predicates; // Top level predicate conditions to match. TreePatternNode *SrcPattern; // Source pattern to match. TreePatternNode *DstPattern; // Resulting pattern. std::vector<Record*> Dstregs; // Physical register defs being matched. unsigned AddedComplexity; // Add to matching pattern complexity. + unsigned ID; // Unique ID for the record. ListInit *getPredicates() const { return Predicates; } TreePatternNode *getSrcPattern() const { return SrcPattern; } @@ -574,7 +581,7 @@ public: abort(); } - const DAGDefaultOperand &getDefaultOperand(Record *R) { + const DAGDefaultOperand &getDefaultOperand(Record *R) const { assert(DefaultOperands.count(R) &&"Isn't an analyzed default operand!"); return DefaultOperands.find(R)->second; } @@ -624,6 +631,7 @@ private: void InferInstructionFlags(); void GenerateVariants(); + void AddPatternToMatch(const TreePattern *Pattern, const PatternToMatch &PTM); void FindPatternInputsAndOutputs(TreePattern *I, TreePatternNode *Pat, std::map<std::string, TreePatternNode*> &InstInputs, diff --git a/utils/TableGen/CodeGenInstruction.cpp b/utils/TableGen/CodeGenInstruction.cpp index d31502b..f5b52ec 100644 --- a/utils/TableGen/CodeGenInstruction.cpp +++ b/utils/TableGen/CodeGenInstruction.cpp @@ -117,7 +117,6 @@ CodeGenInstruction::CodeGenInstruction(Record *R, const std::string &AsmStr) hasCtrlDep = R->getValueAsBit("hasCtrlDep"); isNotDuplicable = R->getValueAsBit("isNotDuplicable"); hasSideEffects = R->getValueAsBit("hasSideEffects"); - mayHaveSideEffects = R->getValueAsBit("mayHaveSideEffects"); neverHasSideEffects = R->getValueAsBit("neverHasSideEffects"); isAsCheapAsAMove = R->getValueAsBit("isAsCheapAsAMove"); hasExtraSrcRegAllocReq = R->getValueAsBit("hasExtraSrcRegAllocReq"); @@ -125,7 +124,7 @@ CodeGenInstruction::CodeGenInstruction(Record *R, const std::string &AsmStr) hasOptionalDef = false; isVariadic = false; - if (mayHaveSideEffects + neverHasSideEffects + hasSideEffects > 1) + if (neverHasSideEffects + hasSideEffects > 1) throw R->getName() + ": multiple conflicting side-effect flags set!"; DagInit *DI; diff --git a/utils/TableGen/CodeGenInstruction.h b/utils/TableGen/CodeGenInstruction.h index 285da14..aae2cac 100644 --- a/utils/TableGen/CodeGenInstruction.h +++ b/utils/TableGen/CodeGenInstruction.h @@ -133,7 +133,6 @@ namespace llvm { bool isNotDuplicable; bool hasOptionalDef; bool hasSideEffects; - bool mayHaveSideEffects; bool neverHasSideEffects; bool isAsCheapAsAMove; bool hasExtraSrcRegAllocReq; diff --git a/utils/TableGen/DAGISelEmitter.cpp b/utils/TableGen/DAGISelEmitter.cpp index f0cebed..e0fa7c8 100644 --- a/utils/TableGen/DAGISelEmitter.cpp +++ b/utils/TableGen/DAGISelEmitter.cpp @@ -14,49 +14,13 @@ #include "DAGISelEmitter.h" #include "DAGISelMatcher.h" #include "Record.h" -#include "llvm/ADT/StringExtras.h" -#include "llvm/Support/CommandLine.h" #include "llvm/Support/Debug.h" -#include "llvm/Support/MathExtras.h" -#include "llvm/Support/Debug.h" -#include <algorithm> -#include <deque> -#include <iostream> using namespace llvm; -static cl::opt<bool> -GenDebug("gen-debug", cl::desc("Generate debug code"), cl::init(false)); - //===----------------------------------------------------------------------===// // DAGISelEmitter Helper methods // -/// getNodeName - The top level Select_* functions have an "SDNode* N" -/// argument. When expanding the pattern-matching code, the intermediate -/// variables have type SDValue. This function provides a uniform way to -/// reference the underlying "SDNode *" for both cases. -static std::string getNodeName(const std::string &S) { - if (S == "N") return S; - return S + ".getNode()"; -} - -/// getNodeValue - Similar to getNodeName, except it provides a uniform -/// way to access the SDValue for both cases. -static std::string getValueName(const std::string &S) { - if (S == "N") return "SDValue(N, 0)"; - return S; -} - -/// NodeIsComplexPattern - return true if N is a leaf node and a subclass of -/// ComplexPattern. -static bool NodeIsComplexPattern(TreePatternNode *N) { - return (N->isLeaf() && - dynamic_cast<DefInit*>(N->getLeafValue()) && - static_cast<DefInit*>(N->getLeafValue())->getDef()-> - isSubClassOf("ComplexPattern")); -} - - /// getPatternSize - Return the 'size' of this pattern. We want to match large /// patterns before small ones. This is used to determine the size of a /// pattern. @@ -96,7 +60,7 @@ static unsigned getPatternSize(TreePatternNode *P, CodeGenDAGPatterns &CGP) { else if (Child->isLeaf()) { if (dynamic_cast<IntInit*>(Child->getLeafValue())) Size += 5; // Matches a ConstantSDNode (+3) and a specific value (+2). - else if (NodeIsComplexPattern(Child)) + else if (Child->getComplexPatternInfo(CGP)) Size += getPatternSize(Child, CGP); else if (!Child->getPredicateFns().empty()) ++Size; @@ -142,106 +106,6 @@ static unsigned getResultPatternSize(TreePatternNode *P, return Cost; } -// PatternSortingPredicate - return true if we prefer to match LHS before RHS. -// In particular, we want to match maximal patterns first and lowest cost within -// a particular complexity first. -struct PatternSortingPredicate { - PatternSortingPredicate(CodeGenDAGPatterns &cgp) : CGP(cgp) {} - CodeGenDAGPatterns &CGP; - - typedef std::pair<unsigned, std::string> CodeLine; - typedef std::vector<CodeLine> CodeList; - typedef std::vector<std::pair<const PatternToMatch*, CodeList> > PatternList; - - bool operator()(const std::pair<const PatternToMatch*, CodeList> &LHSPair, - const std::pair<const PatternToMatch*, CodeList> &RHSPair) { - const PatternToMatch *LHS = LHSPair.first; - const PatternToMatch *RHS = RHSPair.first; - - unsigned LHSSize = getPatternSize(LHS->getSrcPattern(), CGP); - unsigned RHSSize = getPatternSize(RHS->getSrcPattern(), CGP); - LHSSize += LHS->getAddedComplexity(); - RHSSize += RHS->getAddedComplexity(); - if (LHSSize > RHSSize) return true; // LHS -> bigger -> less cost - if (LHSSize < RHSSize) return false; - - // If the patterns have equal complexity, compare generated instruction cost - unsigned LHSCost = getResultPatternCost(LHS->getDstPattern(), CGP); - unsigned RHSCost = getResultPatternCost(RHS->getDstPattern(), CGP); - if (LHSCost < RHSCost) return true; - if (LHSCost > RHSCost) return false; - - return getResultPatternSize(LHS->getDstPattern(), CGP) < - getResultPatternSize(RHS->getDstPattern(), CGP); - } -}; - -/// getRegisterValueType - Look up and return the ValueType of the specified -/// register. If the register is a member of multiple register classes which -/// have different associated types, return MVT::Other. -static MVT::SimpleValueType getRegisterValueType(Record *R, const CodeGenTarget &T) { - bool FoundRC = false; - MVT::SimpleValueType VT = MVT::Other; - const std::vector<CodeGenRegisterClass> &RCs = T.getRegisterClasses(); - std::vector<CodeGenRegisterClass>::const_iterator RC; - std::vector<Record*>::const_iterator Element; - - for (RC = RCs.begin() ; RC != RCs.end() ; RC++) { - Element = find((*RC).Elements.begin(), (*RC).Elements.end(), R); - if (Element != (*RC).Elements.end()) { - if (!FoundRC) { - FoundRC = true; - VT = (*RC).getValueTypeNum(0); - } else { - // In multiple RC's - if (VT != (*RC).getValueTypeNum(0)) { - // Types of the RC's do not agree. Return MVT::Other. The - // target is responsible for handling this. - return MVT::Other; - } - } - } - } - return VT; -} - -static std::string getOpcodeName(Record *Op, CodeGenDAGPatterns &CGP) { - return CGP.getSDNodeInfo(Op).getEnumName(); -} - -//===----------------------------------------------------------------------===// -// Node Transformation emitter implementation. -// -void DAGISelEmitter::EmitNodeTransforms(raw_ostream &OS) { - // Walk the pattern fragments, adding them to a map, which sorts them by - // name. - typedef std::map<std::string, CodeGenDAGPatterns::NodeXForm> NXsByNameTy; - NXsByNameTy NXsByName; - - for (CodeGenDAGPatterns::nx_iterator I = CGP.nx_begin(), E = CGP.nx_end(); - I != E; ++I) - NXsByName.insert(std::make_pair(I->first->getName(), I->second)); - - OS << "\n// Node transformations.\n"; - - for (NXsByNameTy::iterator I = NXsByName.begin(), E = NXsByName.end(); - I != E; ++I) { - Record *SDNode = I->second.first; - std::string Code = I->second.second; - - if (Code.empty()) continue; // Empty code? Skip it. - - std::string ClassName = CGP.getSDNodeInfo(SDNode).getSDClassName(); - const char *C2 = ClassName == "SDNode" ? "N" : "inN"; - - OS << "inline SDValue Transform_" << I->first << "(SDNode *" << C2 - << ") {\n"; - if (ClassName != "SDNode") - OS << " " << ClassName << " *N = cast<" << ClassName << ">(inN);\n"; - OS << Code << "\n}\n"; - } -} - //===----------------------------------------------------------------------===// // Predicate emitter implementation. // @@ -287,1688 +151,41 @@ void DAGISelEmitter::EmitPredicateFunctions(raw_ostream &OS) { OS << "\n\n"; } - -//===----------------------------------------------------------------------===// -// PatternCodeEmitter implementation. -// -class PatternCodeEmitter { -private: +namespace { +// PatternSortingPredicate - return true if we prefer to match LHS before RHS. +// In particular, we want to match maximal patterns first and lowest cost within +// a particular complexity first. +struct PatternSortingPredicate { + PatternSortingPredicate(CodeGenDAGPatterns &cgp) : CGP(cgp) {} CodeGenDAGPatterns &CGP; - - // Predicates. - std::string PredicateCheck; - // Pattern cost. - unsigned Cost; - // Instruction selector pattern. - TreePatternNode *Pattern; - // Matched instruction. - TreePatternNode *Instruction; - - // Node to name mapping - std::map<std::string, std::string> VariableMap; - // Node to operator mapping - std::map<std::string, Record*> OperatorMap; - // Name of the folded node which produces a flag. - std::pair<std::string, unsigned> FoldedFlag; - // Names of all the folded nodes which produce chains. - std::vector<std::pair<std::string, unsigned> > FoldedChains; - // Original input chain(s). - std::vector<std::pair<std::string, std::string> > OrigChains; - std::set<std::string> Duplicates; - - /// LSI - Load/Store information. - /// Save loads/stores matched by a pattern, and generate a MemOperandSDNode - /// for each memory access. This facilitates the use of AliasAnalysis in - /// the backend. - std::vector<std::string> LSI; - - /// GeneratedCode - This is the buffer that we emit code to. The first int - /// indicates whether this is an exit predicate (something that should be - /// tested, and if true, the match fails) [when 1], or normal code to emit - /// [when 0], or initialization code to emit [when 2]. - std::vector<std::pair<unsigned, std::string> > &GeneratedCode; - /// GeneratedDecl - This is the set of all SDValue declarations needed for - /// the set of patterns for each top-level opcode. - std::set<std::string> &GeneratedDecl; - /// TargetOpcodes - The target specific opcodes used by the resulting - /// instructions. - std::vector<std::string> &TargetOpcodes; - std::vector<std::string> &TargetVTs; - /// OutputIsVariadic - Records whether the instruction output pattern uses - /// variable_ops. This requires that the Emit function be passed an - /// additional argument to indicate where the input varargs operands - /// begin. - bool &OutputIsVariadic; - /// NumInputRootOps - Records the number of operands the root node of the - /// input pattern has. This information is used in the generated code to - /// pass to Emit functions when variable_ops processing is needed. - unsigned &NumInputRootOps; - - std::string ChainName; - unsigned TmpNo; - unsigned OpcNo; - unsigned VTNo; - void emitCheck(const std::string &S) { - if (!S.empty()) - GeneratedCode.push_back(std::make_pair(1, S)); - } - void emitCode(const std::string &S) { - if (!S.empty()) - GeneratedCode.push_back(std::make_pair(0, S)); - } - void emitInit(const std::string &S) { - if (!S.empty()) - GeneratedCode.push_back(std::make_pair(2, S)); - } - void emitDecl(const std::string &S) { - assert(!S.empty() && "Invalid declaration"); - GeneratedDecl.insert(S); - } - void emitOpcode(const std::string &Opc) { - TargetOpcodes.push_back(Opc); - OpcNo++; - } - void emitVT(const std::string &VT) { - TargetVTs.push_back(VT); - VTNo++; - } -public: - PatternCodeEmitter(CodeGenDAGPatterns &cgp, std::string predcheck, - TreePatternNode *pattern, TreePatternNode *instr, - std::vector<std::pair<unsigned, std::string> > &gc, - std::set<std::string> &gd, - std::vector<std::string> &to, - std::vector<std::string> &tv, - bool &oiv, - unsigned &niro) - : CGP(cgp), PredicateCheck(predcheck), Pattern(pattern), Instruction(instr), - GeneratedCode(gc), GeneratedDecl(gd), - TargetOpcodes(to), TargetVTs(tv), - OutputIsVariadic(oiv), NumInputRootOps(niro), - TmpNo(0), OpcNo(0), VTNo(0) {} - - /// EmitMatchCode - Emit a matcher for N, going to the label for PatternNo - /// if the match fails. At this point, we already know that the opcode for N - /// matches, and the SDNode for the result has the RootName specified name. - void EmitMatchCode(TreePatternNode *N, TreePatternNode *P, - const std::string &RootName, const std::string &ChainSuffix, - bool &FoundChain); - - void EmitChildMatchCode(TreePatternNode *Child, TreePatternNode *Parent, - const std::string &RootName, - const std::string &ChainSuffix, bool &FoundChain); - - /// EmitResultCode - Emit the action for a pattern. Now that it has matched - /// we actually have to build a DAG! - std::vector<std::string> - EmitResultCode(TreePatternNode *N, std::vector<Record*> DstRegs, - bool InFlagDecled, bool ResNodeDecled, - bool LikeLeaf = false, bool isRoot = false); - - /// InsertOneTypeCheck - Insert a type-check for an unresolved type in 'Pat' - /// and add it to the tree. 'Pat' and 'Other' are isomorphic trees except that - /// 'Pat' may be missing types. If we find an unresolved type to add a check - /// for, this returns true otherwise false if Pat has all types. - bool InsertOneTypeCheck(TreePatternNode *Pat, TreePatternNode *Other, - const std::string &Prefix, bool isRoot = false) { - // Did we find one? - if (Pat->getExtTypes() != Other->getExtTypes()) { - // Move a type over from 'other' to 'pat'. - Pat->setTypes(Other->getExtTypes()); - // The top level node type is checked outside of the select function. - if (!isRoot) - emitCheck(Prefix + ".getValueType() == " + - getName(Pat->getTypeNum(0))); - return true; - } - - unsigned OpNo = (unsigned)Pat->NodeHasProperty(SDNPHasChain, CGP); - for (unsigned i = 0, e = Pat->getNumChildren(); i != e; ++i, ++OpNo) - if (InsertOneTypeCheck(Pat->getChild(i), Other->getChild(i), - Prefix + utostr(OpNo))) - return true; - return false; - } - -private: - /// EmitInFlagSelectCode - Emit the flag operands for the DAG that is - /// being built. - void EmitInFlagSelectCode(TreePatternNode *N, const std::string &RootName, - bool &ChainEmitted, bool &InFlagDecled, - bool &ResNodeDecled, bool isRoot = false) { - const CodeGenTarget &T = CGP.getTargetInfo(); - unsigned OpNo = (unsigned)N->NodeHasProperty(SDNPHasChain, CGP); - bool HasInFlag = N->NodeHasProperty(SDNPInFlag, CGP); - for (unsigned i = 0, e = N->getNumChildren(); i != e; ++i, ++OpNo) { - TreePatternNode *Child = N->getChild(i); - if (!Child->isLeaf()) { - EmitInFlagSelectCode(Child, RootName + utostr(OpNo), ChainEmitted, - InFlagDecled, ResNodeDecled); - } else { - if (DefInit *DI = dynamic_cast<DefInit*>(Child->getLeafValue())) { - if (!Child->getName().empty()) { - std::string Name = RootName + utostr(OpNo); - if (Duplicates.find(Name) != Duplicates.end()) - // A duplicate! Do not emit a copy for this node. - continue; - } - - Record *RR = DI->getDef(); - if (RR->isSubClassOf("Register")) { - MVT::SimpleValueType RVT = getRegisterValueType(RR, T); - if (RVT == MVT::Flag) { - if (!InFlagDecled) { - emitCode("SDValue InFlag = " + - getValueName(RootName + utostr(OpNo)) + ";"); - InFlagDecled = true; - } else - emitCode("InFlag = " + - getValueName(RootName + utostr(OpNo)) + ";"); - } else { - if (!ChainEmitted) { - emitCode("SDValue Chain = CurDAG->getEntryNode();"); - ChainName = "Chain"; - ChainEmitted = true; - } - if (!InFlagDecled) { - emitCode("SDValue InFlag(0, 0);"); - InFlagDecled = true; - } - std::string Decl = (!ResNodeDecled) ? "SDNode *" : ""; - emitCode(Decl + "ResNode = CurDAG->getCopyToReg(" + ChainName + - ", " + getNodeName(RootName) + "->getDebugLoc()" + - ", " + getQualifiedName(RR) + - ", " + getValueName(RootName + utostr(OpNo)) + - ", InFlag).getNode();"); - ResNodeDecled = true; - emitCode(ChainName + " = SDValue(ResNode, 0);"); - emitCode("InFlag = SDValue(ResNode, 1);"); - } - } - } - } - } - - if (HasInFlag) { - if (!InFlagDecled) { - emitCode("SDValue InFlag = " + getNodeName(RootName) + - "->getOperand(" + utostr(OpNo) + ");"); - InFlagDecled = true; - } else - emitCode("InFlag = " + getNodeName(RootName) + - "->getOperand(" + utostr(OpNo) + ");"); - } - } -}; - - -/// EmitMatchCode - Emit a matcher for N, going to the label for PatternNo -/// if the match fails. At this point, we already know that the opcode for N -/// matches, and the SDNode for the result has the RootName specified name. -void PatternCodeEmitter::EmitMatchCode(TreePatternNode *N, TreePatternNode *P, - const std::string &RootName, - const std::string &ChainSuffix, - bool &FoundChain) { - - // Save loads/stores matched by a pattern. - if (!N->isLeaf() && N->getName().empty()) { - if (N->NodeHasProperty(SDNPMemOperand, CGP)) - LSI.push_back(getNodeName(RootName)); - } - - bool isRoot = (P == NULL); - // Emit instruction predicates. Each predicate is just a string for now. - if (isRoot) { - // Record input varargs info. - NumInputRootOps = N->getNumChildren(); - emitCheck(PredicateCheck); - } - - if (N->isLeaf()) { - if (IntInit *II = dynamic_cast<IntInit*>(N->getLeafValue())) { - emitCheck("cast<ConstantSDNode>(" + getNodeName(RootName) + - ")->getSExtValue() == INT64_C(" + - itostr(II->getValue()) + ")"); - return; - } else if (!NodeIsComplexPattern(N)) { - assert(0 && "Cannot match this as a leaf value!"); - abort(); - } - } - - // If this node has a name associated with it, capture it in VariableMap. If - // we already saw this in the pattern, emit code to verify dagness. - if (!N->getName().empty()) { - std::string &VarMapEntry = VariableMap[N->getName()]; - if (VarMapEntry.empty()) { - VarMapEntry = RootName; - } else { - // If we get here, this is a second reference to a specific name. Since - // we already have checked that the first reference is valid, we don't - // have to recursively match it, just check that it's the same as the - // previously named thing. - emitCheck(VarMapEntry + " == " + RootName); - return; - } - - if (!N->isLeaf()) - OperatorMap[N->getName()] = N->getOperator(); - } - - - // Emit code to load the child nodes and match their contents recursively. - unsigned OpNo = 0; - bool NodeHasChain = N->NodeHasProperty(SDNPHasChain, CGP); - bool HasChain = N->TreeHasProperty(SDNPHasChain, CGP); - if (HasChain) { - if (NodeHasChain) - OpNo = 1; - if (!isRoot) { - // Check if it's profitable to fold the node. e.g. Check for multiple uses - // of actual result? - std::string ParentName(RootName.begin(), RootName.end()-1); - emitCheck("IsProfitableToFold(" + getValueName(RootName) + - ", " + getNodeName(ParentName) + ", N)"); - if (NodeHasChain) { - // If the immediate use can somehow reach this node through another - // path, then can't fold it either or it will create a cycle. - // e.g. In the following diagram, XX can reach ld through YY. If - // ld is folded into XX, then YY is both a predecessor and a successor - // of XX. - // - // [ld] - // ^ ^ - // | | - // / \--- - // / [YY] - // | ^ - // [XX]-------| - - // We know we need the check if N's parent is not the root. - bool NeedCheck = P != Pattern; - if (!NeedCheck) { - const SDNodeInfo &PInfo = CGP.getSDNodeInfo(P->getOperator()); - NeedCheck = - P->getOperator() == CGP.get_intrinsic_void_sdnode() || - P->getOperator() == CGP.get_intrinsic_w_chain_sdnode() || - P->getOperator() == CGP.get_intrinsic_wo_chain_sdnode() || - PInfo.getNumOperands() > 1 || - PInfo.hasProperty(SDNPHasChain) || - PInfo.hasProperty(SDNPInFlag) || - PInfo.hasProperty(SDNPOptInFlag); - } - - if (NeedCheck) { - emitCheck("IsLegalToFold(" + getValueName(RootName) + - ", " + getNodeName(ParentName) + ", N)"); - } - } - } - - if (NodeHasChain) { - if (FoundChain) { - emitCheck("(" + ChainName + ".getNode() == " + - getNodeName(RootName) + " || " - "IsChainCompatible(" + ChainName + ".getNode(), " + - getNodeName(RootName) + "))"); - OrigChains.push_back(std::make_pair(ChainName, - getValueName(RootName))); - } else - FoundChain = true; - ChainName = "Chain" + ChainSuffix; - emitInit("SDValue " + ChainName + " = " + getNodeName(RootName) + - "->getOperand(0);"); - } - } - - // If there are node predicates for this, emit the calls. - for (unsigned i = 0, e = N->getPredicateFns().size(); i != e; ++i) - emitCheck(N->getPredicateFns()[i] + "(" + getNodeName(RootName) + ")"); - - // If this is an 'and R, 1234' where the operation is AND/OR and the RHS is - // a constant without a predicate fn that has more that one bit set, handle - // this as a special case. This is usually for targets that have special - // handling of certain large constants (e.g. alpha with it's 8/16/32-bit - // handling stuff). Using these instructions is often far more efficient - // than materializing the constant. Unfortunately, both the instcombiner - // and the dag combiner can often infer that bits are dead, and thus drop - // them from the mask in the dag. For example, it might turn 'AND X, 255' - // into 'AND X, 254' if it knows the low bit is set. Emit code that checks - // to handle this. - if (!N->isLeaf() && - (N->getOperator()->getName() == "and" || - N->getOperator()->getName() == "or") && - N->getChild(1)->isLeaf() && - N->getChild(1)->getPredicateFns().empty()) { - if (IntInit *II = dynamic_cast<IntInit*>(N->getChild(1)->getLeafValue())) { - if (!isPowerOf2_32(II->getValue())) { // Don't bother with single bits. - emitInit("SDValue " + RootName + "0" + " = " + - getNodeName(RootName) + "->getOperand(" + utostr(0) + ");"); - emitInit("SDValue " + RootName + "1" + " = " + - getNodeName(RootName) + "->getOperand(" + utostr(1) + ");"); - - unsigned NTmp = TmpNo++; - emitCode("ConstantSDNode *Tmp" + utostr(NTmp) + - " = dyn_cast<ConstantSDNode>(" + - getNodeName(RootName + "1") + ");"); - emitCheck("Tmp" + utostr(NTmp)); - const char *MaskPredicate = N->getOperator()->getName() == "or" - ? "CheckOrMask(" : "CheckAndMask("; - emitCheck(MaskPredicate + getValueName(RootName + "0") + - ", Tmp" + utostr(NTmp) + - ", INT64_C(" + itostr(II->getValue()) + "))"); - - EmitChildMatchCode(N->getChild(0), N, RootName + utostr(0), - ChainSuffix + utostr(0), FoundChain); - return; - } - } - } - - for (unsigned i = 0, e = N->getNumChildren(); i != e; ++i, ++OpNo) { - emitInit("SDValue " + getValueName(RootName + utostr(OpNo)) + " = " + - getNodeName(RootName) + "->getOperand(" + utostr(OpNo) + ");"); - - EmitChildMatchCode(N->getChild(i), N, RootName + utostr(OpNo), - ChainSuffix + utostr(OpNo), FoundChain); - } - - // Handle cases when root is a complex pattern. - const ComplexPattern *CP; - if (isRoot && N->isLeaf() && (CP = N->getComplexPatternInfo(CGP))) { - std::string Fn = CP->getSelectFunc(); - unsigned NumOps = CP->getNumOperands(); - for (unsigned i = 0; i < NumOps; ++i) { - emitDecl("CPTmp" + RootName + "_" + utostr(i)); - emitCode("SDValue CPTmp" + RootName + "_" + utostr(i) + ";"); - } - if (CP->hasProperty(SDNPHasChain)) { - emitDecl("CPInChain"); - emitDecl("Chain" + ChainSuffix); - emitCode("SDValue CPInChain;"); - emitCode("SDValue Chain" + ChainSuffix + ";"); - } - - std::string Code = Fn + "(" + - getNodeName(RootName) + ", " + - getValueName(RootName); - for (unsigned i = 0; i < NumOps; i++) - Code += ", CPTmp" + RootName + "_" + utostr(i); - if (CP->hasProperty(SDNPHasChain)) { - ChainName = "Chain" + ChainSuffix; - Code += ", CPInChain, Chain" + ChainSuffix; - } - emitCheck(Code + ")"); - } -} - -void PatternCodeEmitter::EmitChildMatchCode(TreePatternNode *Child, - TreePatternNode *Parent, - const std::string &RootName, - const std::string &ChainSuffix, - bool &FoundChain) { - if (!Child->isLeaf()) { - // If it's not a leaf, recursively match. - const SDNodeInfo &CInfo = CGP.getSDNodeInfo(Child->getOperator()); - emitCheck(getNodeName(RootName) + "->getOpcode() == " + - CInfo.getEnumName()); - EmitMatchCode(Child, Parent, RootName, ChainSuffix, FoundChain); - bool HasChain = false; - if (Child->NodeHasProperty(SDNPHasChain, CGP)) { - HasChain = true; - FoldedChains.push_back(std::make_pair(getValueName(RootName), - CInfo.getNumResults())); - } - if (Child->NodeHasProperty(SDNPOutFlag, CGP)) { - assert(FoldedFlag.first == "" && FoldedFlag.second == 0 && - "Pattern folded multiple nodes which produce flags?"); - FoldedFlag = std::make_pair(getValueName(RootName), - CInfo.getNumResults() + (unsigned)HasChain); - } - } else { - // If this child has a name associated with it, capture it in VarMap. If - // we already saw this in the pattern, emit code to verify dagness. - if (!Child->getName().empty()) { - std::string &VarMapEntry = VariableMap[Child->getName()]; - if (VarMapEntry.empty()) { - VarMapEntry = getValueName(RootName); - } else { - // If we get here, this is a second reference to a specific name. - // Since we already have checked that the first reference is valid, - // we don't have to recursively match it, just check that it's the - // same as the previously named thing. - emitCheck(VarMapEntry + " == " + getValueName(RootName)); - Duplicates.insert(getValueName(RootName)); - return; - } - } - - // Handle leaves of various types. - if (DefInit *DI = dynamic_cast<DefInit*>(Child->getLeafValue())) { - Record *LeafRec = DI->getDef(); - if (LeafRec->isSubClassOf("RegisterClass") || - LeafRec->isSubClassOf("PointerLikeRegClass")) { - // Handle register references. Nothing to do here. - } else if (LeafRec->isSubClassOf("Register")) { - // Handle register references. - } else if (LeafRec->isSubClassOf("ComplexPattern")) { - // Handle complex pattern. - const ComplexPattern *CP = Child->getComplexPatternInfo(CGP); - std::string Fn = CP->getSelectFunc(); - unsigned NumOps = CP->getNumOperands(); - for (unsigned i = 0; i < NumOps; ++i) { - emitDecl("CPTmp" + RootName + "_" + utostr(i)); - emitCode("SDValue CPTmp" + RootName + "_" + utostr(i) + ";"); - } - if (CP->hasProperty(SDNPHasChain)) { - const SDNodeInfo &PInfo = CGP.getSDNodeInfo(Parent->getOperator()); - FoldedChains.push_back(std::make_pair("CPInChain", - PInfo.getNumResults())); - ChainName = "Chain" + ChainSuffix; - emitDecl("CPInChain"); - emitDecl(ChainName); - emitCode("SDValue CPInChain;"); - emitCode("SDValue " + ChainName + ";"); - } - - std::string Code = Fn + "(N, "; - if (CP->hasProperty(SDNPHasChain)) { - std::string ParentName(RootName.begin(), RootName.end()-1); - Code += getValueName(ParentName) + ", "; - } - Code += getValueName(RootName); - for (unsigned i = 0; i < NumOps; i++) - Code += ", CPTmp" + RootName + "_" + utostr(i); - if (CP->hasProperty(SDNPHasChain)) - Code += ", CPInChain, Chain" + ChainSuffix; - emitCheck(Code + ")"); - } else if (LeafRec->getName() == "srcvalue") { - // Place holder for SRCVALUE nodes. Nothing to do here. - } else if (LeafRec->isSubClassOf("ValueType")) { - // Make sure this is the specified value type. - emitCheck("cast<VTSDNode>(" + getNodeName(RootName) + - ")->getVT() == MVT::" + LeafRec->getName()); - } else if (LeafRec->isSubClassOf("CondCode")) { - // Make sure this is the specified cond code. - emitCheck("cast<CondCodeSDNode>(" + getNodeName(RootName) + - ")->get() == ISD::" + LeafRec->getName()); - } else { -#ifndef NDEBUG - Child->dump(); - errs() << " "; -#endif - assert(0 && "Unknown leaf type!"); - } - - // If there are node predicates for this, emit the calls. - for (unsigned i = 0, e = Child->getPredicateFns().size(); i != e; ++i) - emitCheck(Child->getPredicateFns()[i] + "(" + getNodeName(RootName) + - ")"); - } else if (IntInit *II = - dynamic_cast<IntInit*>(Child->getLeafValue())) { - unsigned NTmp = TmpNo++; - emitCode("ConstantSDNode *Tmp"+ utostr(NTmp) + - " = dyn_cast<ConstantSDNode>("+ - getNodeName(RootName) + ");"); - emitCheck("Tmp" + utostr(NTmp)); - unsigned CTmp = TmpNo++; - emitCode("int64_t CN"+ utostr(CTmp) + - " = Tmp" + utostr(NTmp) + "->getSExtValue();"); - emitCheck("CN" + utostr(CTmp) + " == " - "INT64_C(" +itostr(II->getValue()) + ")"); - } else { -#ifndef NDEBUG - Child->dump(); -#endif - assert(0 && "Unknown leaf type!"); - } - } -} - -/// EmitResultCode - Emit the action for a pattern. Now that it has matched -/// we actually have to build a DAG! -std::vector<std::string> -PatternCodeEmitter::EmitResultCode(TreePatternNode *N, - std::vector<Record*> DstRegs, - bool InFlagDecled, bool ResNodeDecled, - bool LikeLeaf, bool isRoot) { - // List of arguments of getMachineNode() or SelectNodeTo(). - std::vector<std::string> NodeOps; - // This is something selected from the pattern we matched. - if (!N->getName().empty()) { - const std::string &VarName = N->getName(); - std::string Val = VariableMap[VarName]; - bool ModifiedVal = false; - if (Val.empty()) { - errs() << "Variable '" << VarName << " referenced but not defined " - << "and not caught earlier!\n"; - abort(); - } - if (Val[0] == 'T' && Val[1] == 'm' && Val[2] == 'p') { - // Already selected this operand, just return the tmpval. - NodeOps.push_back(getValueName(Val)); - return NodeOps; - } - - const ComplexPattern *CP; - unsigned ResNo = TmpNo++; - if (!N->isLeaf() && N->getOperator()->getName() == "imm") { - assert(N->getExtTypes().size() == 1 && "Multiple types not handled!"); - std::string CastType; - std::string TmpVar = "Tmp" + utostr(ResNo); - switch (N->getTypeNum(0)) { - default: - errs() << "Cannot handle " << getEnumName(N->getTypeNum(0)) - << " type as an immediate constant. Aborting\n"; - abort(); - case MVT::i1: CastType = "bool"; break; - case MVT::i8: CastType = "unsigned char"; break; - case MVT::i16: CastType = "unsigned short"; break; - case MVT::i32: CastType = "unsigned"; break; - case MVT::i64: CastType = "uint64_t"; break; - } - emitCode("SDValue " + TmpVar + - " = CurDAG->getTargetConstant(((" + CastType + - ") cast<ConstantSDNode>(" + Val + ")->getZExtValue()), " + - getEnumName(N->getTypeNum(0)) + ");"); - // Add Tmp<ResNo> to VariableMap, so that we don't multiply select this - // value if used multiple times by this pattern result. - Val = TmpVar; - ModifiedVal = true; - NodeOps.push_back(getValueName(Val)); - } else if (!N->isLeaf() && N->getOperator()->getName() == "fpimm") { - assert(N->getExtTypes().size() == 1 && "Multiple types not handled!"); - std::string TmpVar = "Tmp" + utostr(ResNo); - emitCode("SDValue " + TmpVar + - " = CurDAG->getTargetConstantFP(*cast<ConstantFPSDNode>(" + - Val + ")->getConstantFPValue(), cast<ConstantFPSDNode>(" + - Val + ")->getValueType(0));"); - // Add Tmp<ResNo> to VariableMap, so that we don't multiply select this - // value if used multiple times by this pattern result. - Val = TmpVar; - ModifiedVal = true; - NodeOps.push_back(getValueName(Val)); - } else if (!N->isLeaf() && N->getOperator()->getName() == "texternalsym"){ - Record *Op = OperatorMap[N->getName()]; - // Transform ExternalSymbol to TargetExternalSymbol - if (Op && Op->getName() == "externalsym") { - std::string TmpVar = "Tmp"+utostr(ResNo); - emitCode("SDValue " + TmpVar + " = CurDAG->getTarget" - "ExternalSymbol(cast<ExternalSymbolSDNode>(" + - Val + ")->getSymbol(), " + - getEnumName(N->getTypeNum(0)) + ");"); - // Add Tmp<ResNo> to VariableMap, so that we don't multiply select - // this value if used multiple times by this pattern result. - Val = TmpVar; - ModifiedVal = true; - } - NodeOps.push_back(getValueName(Val)); - } else if (!N->isLeaf() && (N->getOperator()->getName() == "tglobaladdr" - || N->getOperator()->getName() == "tglobaltlsaddr")) { - Record *Op = OperatorMap[N->getName()]; - // Transform GlobalAddress to TargetGlobalAddress - if (Op && (Op->getName() == "globaladdr" || - Op->getName() == "globaltlsaddr")) { - std::string TmpVar = "Tmp" + utostr(ResNo); - emitCode("SDValue " + TmpVar + " = CurDAG->getTarget" - "GlobalAddress(cast<GlobalAddressSDNode>(" + Val + - ")->getGlobal(), " + getEnumName(N->getTypeNum(0)) + - ");"); - // Add Tmp<ResNo> to VariableMap, so that we don't multiply select - // this value if used multiple times by this pattern result. - Val = TmpVar; - ModifiedVal = true; - } - NodeOps.push_back(getValueName(Val)); - } else if (!N->isLeaf() - && (N->getOperator()->getName() == "texternalsym" || - N->getOperator()->getName() == "tconstpool")) { - // Do not rewrite the variable name, since we don't generate a new - // temporary. - NodeOps.push_back(getValueName(Val)); - } else if (N->isLeaf() && (CP = N->getComplexPatternInfo(CGP))) { - for (unsigned i = 0; i < CP->getNumOperands(); ++i) { - NodeOps.push_back(getValueName("CPTmp" + Val + "_" + utostr(i))); - } - } else { - // This node, probably wrapped in a SDNodeXForm, behaves like a leaf - // node even if it isn't one. Don't select it. - if (!LikeLeaf) { - if (isRoot && N->isLeaf()) { - emitCode("ReplaceUses(SDValue(N, 0), " + Val + ");"); - emitCode("return NULL;"); - } - } - NodeOps.push_back(getValueName(Val)); - } - - if (ModifiedVal) - VariableMap[VarName] = Val; - return NodeOps; - } - if (N->isLeaf()) { - // If this is an explicit register reference, handle it. - if (DefInit *DI = dynamic_cast<DefInit*>(N->getLeafValue())) { - unsigned ResNo = TmpNo++; - if (DI->getDef()->isSubClassOf("Register")) { - emitCode("SDValue Tmp" + utostr(ResNo) + " = CurDAG->getRegister(" + - getQualifiedName(DI->getDef()) + ", " + - getEnumName(N->getTypeNum(0)) + ");"); - NodeOps.push_back(getValueName("Tmp" + utostr(ResNo))); - return NodeOps; - } else if (DI->getDef()->getName() == "zero_reg") { - emitCode("SDValue Tmp" + utostr(ResNo) + - " = CurDAG->getRegister(0, " + - getEnumName(N->getTypeNum(0)) + ");"); - NodeOps.push_back(getValueName("Tmp" + utostr(ResNo))); - return NodeOps; - } else if (DI->getDef()->isSubClassOf("RegisterClass")) { - // Handle a reference to a register class. This is used - // in COPY_TO_SUBREG instructions. - emitCode("SDValue Tmp" + utostr(ResNo) + - " = CurDAG->getTargetConstant(" + - getQualifiedName(DI->getDef()) + "RegClassID, " + - "MVT::i32);"); - NodeOps.push_back(getValueName("Tmp" + utostr(ResNo))); - return NodeOps; - } - } else if (IntInit *II = dynamic_cast<IntInit*>(N->getLeafValue())) { - unsigned ResNo = TmpNo++; - assert(N->getExtTypes().size() == 1 && "Multiple types not handled!"); - emitCode("SDValue Tmp" + utostr(ResNo) + - " = CurDAG->getTargetConstant(0x" + - utohexstr((uint64_t) II->getValue()) + - "ULL, " + getEnumName(N->getTypeNum(0)) + ");"); - NodeOps.push_back(getValueName("Tmp" + utostr(ResNo))); - return NodeOps; - } - -#ifndef NDEBUG - N->dump(); -#endif - assert(0 && "Unknown leaf type!"); - return NodeOps; - } - - Record *Op = N->getOperator(); - if (Op->isSubClassOf("Instruction")) { - const CodeGenTarget &CGT = CGP.getTargetInfo(); - CodeGenInstruction &II = CGT.getInstruction(Op->getName()); - const DAGInstruction &Inst = CGP.getInstruction(Op); - const TreePattern *InstPat = Inst.getPattern(); - // FIXME: Assume actual pattern comes before "implicit". - TreePatternNode *InstPatNode = - isRoot ? (InstPat ? InstPat->getTree(0) : Pattern) - : (InstPat ? InstPat->getTree(0) : NULL); - if (InstPatNode && !InstPatNode->isLeaf() && - InstPatNode->getOperator()->getName() == "set") { - InstPatNode = InstPatNode->getChild(InstPatNode->getNumChildren()-1); - } - bool IsVariadic = isRoot && II.isVariadic; - // FIXME: fix how we deal with physical register operands. - bool HasImpInputs = isRoot && Inst.getNumImpOperands() > 0; - bool HasImpResults = isRoot && DstRegs.size() > 0; - bool NodeHasOptInFlag = isRoot && - Pattern->TreeHasProperty(SDNPOptInFlag, CGP); - bool NodeHasInFlag = isRoot && - Pattern->TreeHasProperty(SDNPInFlag, CGP); - bool NodeHasOutFlag = isRoot && - Pattern->TreeHasProperty(SDNPOutFlag, CGP); - bool NodeHasChain = InstPatNode && - InstPatNode->TreeHasProperty(SDNPHasChain, CGP); - bool InputHasChain = isRoot && Pattern->NodeHasProperty(SDNPHasChain, CGP); - unsigned NumResults = Inst.getNumResults(); - unsigned NumDstRegs = HasImpResults ? DstRegs.size() : 0; - - // Record output varargs info. - OutputIsVariadic = IsVariadic; - - if (NodeHasOptInFlag) { - emitCode("bool HasInFlag = " - "(N->getOperand(N->getNumOperands()-1).getValueType() == " - "MVT::Flag);"); - } - if (IsVariadic) - emitCode("SmallVector<SDValue, 8> Ops" + utostr(OpcNo) + ";"); - - // How many results is this pattern expected to produce? - unsigned NumPatResults = 0; - for (unsigned i = 0, e = Pattern->getExtTypes().size(); i != e; i++) { - MVT::SimpleValueType VT = Pattern->getTypeNum(i); - if (VT != MVT::isVoid && VT != MVT::Flag) - NumPatResults++; - } - - if (OrigChains.size() > 0) { - // The original input chain is being ignored. If it is not just - // pointing to the op that's being folded, we should create a - // TokenFactor with it and the chain of the folded op as the new chain. - // We could potentially be doing multiple levels of folding, in that - // case, the TokenFactor can have more operands. - emitCode("SmallVector<SDValue, 8> InChains;"); - for (unsigned i = 0, e = OrigChains.size(); i < e; ++i) { - emitCode("if (" + OrigChains[i].first + ".getNode() != " + - OrigChains[i].second + ".getNode()) {"); - emitCode(" InChains.push_back(" + OrigChains[i].first + ");"); - emitCode("}"); - } - emitCode("InChains.push_back(" + ChainName + ");"); - emitCode(ChainName + " = CurDAG->getNode(ISD::TokenFactor, " - "N->getDebugLoc(), MVT::Other, " - "&InChains[0], InChains.size());"); - if (GenDebug) { - emitCode("CurDAG->setSubgraphColor(" + ChainName +".getNode(), \"yellow\");"); - emitCode("CurDAG->setSubgraphColor(" + ChainName +".getNode(), \"black\");"); - } - } - - // Loop over all of the operands of the instruction pattern, emitting code - // to fill them all in. The node 'N' usually has number children equal to - // the number of input operands of the instruction. However, in cases - // where there are predicate operands for an instruction, we need to fill - // in the 'execute always' values. Match up the node operands to the - // instruction operands to do this. - std::vector<std::string> AllOps; - for (unsigned ChildNo = 0, InstOpNo = NumResults; - InstOpNo != II.OperandList.size(); ++InstOpNo) { - std::vector<std::string> Ops; - - // Determine what to emit for this operand. - Record *OperandNode = II.OperandList[InstOpNo].Rec; - if ((OperandNode->isSubClassOf("PredicateOperand") || - OperandNode->isSubClassOf("OptionalDefOperand")) && - !CGP.getDefaultOperand(OperandNode).DefaultOps.empty()) { - // This is a predicate or optional def operand; emit the - // 'default ops' operands. - const DAGDefaultOperand &DefaultOp = - CGP.getDefaultOperand(II.OperandList[InstOpNo].Rec); - for (unsigned i = 0, e = DefaultOp.DefaultOps.size(); i != e; ++i) { - Ops = EmitResultCode(DefaultOp.DefaultOps[i], DstRegs, - InFlagDecled, ResNodeDecled); - AllOps.insert(AllOps.end(), Ops.begin(), Ops.end()); - } - } else { - // Otherwise this is a normal operand or a predicate operand without - // 'execute always'; emit it. - Ops = EmitResultCode(N->getChild(ChildNo), DstRegs, - InFlagDecled, ResNodeDecled); - AllOps.insert(AllOps.end(), Ops.begin(), Ops.end()); - ++ChildNo; - } - } - - // Emit all the chain and CopyToReg stuff. - bool ChainEmitted = NodeHasChain; - if (NodeHasInFlag || HasImpInputs) - EmitInFlagSelectCode(Pattern, "N", ChainEmitted, - InFlagDecled, ResNodeDecled, true); - if (NodeHasOptInFlag || NodeHasInFlag || HasImpInputs) { - if (!InFlagDecled) { - emitCode("SDValue InFlag(0, 0);"); - InFlagDecled = true; - } - if (NodeHasOptInFlag) { - emitCode("if (HasInFlag) {"); - emitCode(" InFlag = N->getOperand(N->getNumOperands()-1);"); - emitCode("}"); - } - } - - unsigned ResNo = TmpNo++; - - unsigned OpsNo = OpcNo; - std::string CodePrefix; - bool ChainAssignmentNeeded = NodeHasChain && !isRoot; - std::deque<std::string> After; - std::string NodeName; - if (!isRoot) { - NodeName = "Tmp" + utostr(ResNo); - CodePrefix = "SDValue " + NodeName + "("; - } else { - NodeName = "ResNode"; - if (!ResNodeDecled) { - CodePrefix = "SDNode *" + NodeName + " = "; - ResNodeDecled = true; - } else - CodePrefix = NodeName + " = "; - } - - std::string Code = "Opc" + utostr(OpcNo); - - if (!isRoot || (InputHasChain && !NodeHasChain)) - // For call to "getMachineNode()". - Code += ", N->getDebugLoc()"; - - emitOpcode(II.Namespace + "::" + II.TheDef->getName()); - - // Output order: results, chain, flags - // Result types. - if (NumResults > 0 && N->getTypeNum(0) != MVT::isVoid) { - Code += ", VT" + utostr(VTNo); - emitVT(getEnumName(N->getTypeNum(0))); - } - // Add types for implicit results in physical registers, scheduler will - // care of adding copyfromreg nodes. - for (unsigned i = 0; i < NumDstRegs; i++) { - Record *RR = DstRegs[i]; - if (RR->isSubClassOf("Register")) { - MVT::SimpleValueType RVT = getRegisterValueType(RR, CGT); - Code += ", " + getEnumName(RVT); - } - } - if (NodeHasChain) - Code += ", MVT::Other"; - if (NodeHasOutFlag) - Code += ", MVT::Flag"; - - // Inputs. - if (IsVariadic) { - for (unsigned i = 0, e = AllOps.size(); i != e; ++i) - emitCode("Ops" + utostr(OpsNo) + ".push_back(" + AllOps[i] + ");"); - AllOps.clear(); - - // Figure out whether any operands at the end of the op list are not - // part of the variable section. - std::string EndAdjust; - if (NodeHasInFlag || HasImpInputs) - EndAdjust = "-1"; // Always has one flag. - else if (NodeHasOptInFlag) - EndAdjust = "-(HasInFlag?1:0)"; // May have a flag. - - emitCode("for (unsigned i = NumInputRootOps + " + utostr(NodeHasChain) + - ", e = N->getNumOperands()" + EndAdjust + "; i != e; ++i) {"); - - emitCode(" Ops" + utostr(OpsNo) + ".push_back(N->getOperand(i));"); - emitCode("}"); - } - - // Populate MemRefs with entries for each memory accesses covered by - // this pattern. - if (isRoot && !LSI.empty()) { - std::string MemRefs = "MemRefs" + utostr(OpsNo); - emitCode("MachineSDNode::mmo_iterator " + MemRefs + " = " - "MF->allocateMemRefsArray(" + utostr(LSI.size()) + ");"); - for (unsigned i = 0, e = LSI.size(); i != e; ++i) - emitCode(MemRefs + "[" + utostr(i) + "] = " - "cast<MemSDNode>(" + LSI[i] + ")->getMemOperand();"); - After.push_back("cast<MachineSDNode>(ResNode)->setMemRefs(" + - MemRefs + ", " + MemRefs + " + " + utostr(LSI.size()) + - ");"); - } - - if (NodeHasChain) { - if (IsVariadic) - emitCode("Ops" + utostr(OpsNo) + ".push_back(" + ChainName + ");"); - else - AllOps.push_back(ChainName); - } - - if (IsVariadic) { - if (NodeHasInFlag || HasImpInputs) - emitCode("Ops" + utostr(OpsNo) + ".push_back(InFlag);"); - else if (NodeHasOptInFlag) { - emitCode("if (HasInFlag)"); - emitCode(" Ops" + utostr(OpsNo) + ".push_back(InFlag);"); - } - Code += ", &Ops" + utostr(OpsNo) + "[0], Ops" + utostr(OpsNo) + - ".size()"; - } else if (NodeHasInFlag || NodeHasOptInFlag || HasImpInputs) - AllOps.push_back("InFlag"); - - unsigned NumOps = AllOps.size(); - if (NumOps) { - if (!NodeHasOptInFlag && NumOps < 4) { - for (unsigned i = 0; i != NumOps; ++i) - Code += ", " + AllOps[i]; - } else { - std::string OpsCode = "SDValue Ops" + utostr(OpsNo) + "[] = { "; - for (unsigned i = 0; i != NumOps; ++i) { - OpsCode += AllOps[i]; - if (i != NumOps-1) - OpsCode += ", "; - } - emitCode(OpsCode + " };"); - Code += ", Ops" + utostr(OpsNo) + ", "; - if (NodeHasOptInFlag) { - Code += "HasInFlag ? "; - Code += utostr(NumOps) + " : " + utostr(NumOps-1); - } else - Code += utostr(NumOps); - } - } - - if (!isRoot) - Code += "), 0"; - - std::vector<std::string> ReplaceFroms; - std::vector<std::string> ReplaceTos; - if (!isRoot) { - NodeOps.push_back("Tmp" + utostr(ResNo)); - } else { - - if (NodeHasOutFlag) { - if (!InFlagDecled) { - After.push_back("SDValue InFlag(ResNode, " + - utostr(NumResults+NumDstRegs+(unsigned)NodeHasChain) + - ");"); - InFlagDecled = true; - } else - After.push_back("InFlag = SDValue(ResNode, " + - utostr(NumResults+NumDstRegs+(unsigned)NodeHasChain) + - ");"); - } - - for (unsigned j = 0, e = FoldedChains.size(); j < e; j++) { - ReplaceFroms.push_back("SDValue(" + - FoldedChains[j].first + ".getNode(), " + - utostr(FoldedChains[j].second) + - ")"); - ReplaceTos.push_back("SDValue(ResNode, " + - utostr(NumResults+NumDstRegs) + ")"); - } - - if (NodeHasOutFlag) { - if (FoldedFlag.first != "") { - ReplaceFroms.push_back("SDValue(" + FoldedFlag.first + ".getNode(), " + - utostr(FoldedFlag.second) + ")"); - ReplaceTos.push_back("InFlag"); - } else { - assert(Pattern->NodeHasProperty(SDNPOutFlag, CGP)); - ReplaceFroms.push_back("SDValue(N, " + - utostr(NumPatResults + (unsigned)InputHasChain) - + ")"); - ReplaceTos.push_back("InFlag"); - } - } - - if (!ReplaceFroms.empty() && InputHasChain) { - ReplaceFroms.push_back("SDValue(N, " + - utostr(NumPatResults) + ")"); - ReplaceTos.push_back("SDValue(" + ChainName + ".getNode(), " + - ChainName + ".getResNo()" + ")"); - ChainAssignmentNeeded |= NodeHasChain; - } - - // User does not expect the instruction would produce a chain! - if ((!InputHasChain && NodeHasChain) && NodeHasOutFlag) { - ; - } else if (InputHasChain && !NodeHasChain) { - // One of the inner node produces a chain. - assert(!NodeHasOutFlag && "Node has flag but not chain!"); - ReplaceFroms.push_back("SDValue(N, " + - utostr(NumPatResults) + ")"); - ReplaceTos.push_back(ChainName); - } - } - - if (ChainAssignmentNeeded) { - // Remember which op produces the chain. - std::string ChainAssign; - if (!isRoot) - ChainAssign = ChainName + " = SDValue(" + NodeName + - ".getNode(), " + utostr(NumResults+NumDstRegs) + ");"; - else - ChainAssign = ChainName + " = SDValue(" + NodeName + - ", " + utostr(NumResults+NumDstRegs) + ");"; - - After.push_front(ChainAssign); - } - - if (ReplaceFroms.size() == 1) { - After.push_back("ReplaceUses(" + ReplaceFroms[0] + ", " + - ReplaceTos[0] + ");"); - } else if (!ReplaceFroms.empty()) { - After.push_back("const SDValue Froms[] = {"); - for (unsigned i = 0, e = ReplaceFroms.size(); i != e; ++i) - After.push_back(" " + ReplaceFroms[i] + (i + 1 != e ? "," : "")); - After.push_back("};"); - After.push_back("const SDValue Tos[] = {"); - for (unsigned i = 0, e = ReplaceFroms.size(); i != e; ++i) - After.push_back(" " + ReplaceTos[i] + (i + 1 != e ? "," : "")); - After.push_back("};"); - After.push_back("ReplaceUses(Froms, Tos, " + - itostr(ReplaceFroms.size()) + ");"); - } - - // We prefer to use SelectNodeTo since it avoids allocation when - // possible and it avoids CSE map recalculation for the node's - // users, however it's tricky to use in a non-root context. - // - // We also don't use SelectNodeTo if the pattern replacement is being - // used to jettison a chain result, since morphing the node in place - // would leave users of the chain dangling. - // - if (!isRoot || (InputHasChain && !NodeHasChain)) { - Code = "CurDAG->getMachineNode(" + Code; - } else { - Code = "CurDAG->SelectNodeTo(N, " + Code; - } - if (isRoot) { - if (After.empty()) - CodePrefix = "return "; - else - After.push_back("return ResNode;"); - } - - emitCode(CodePrefix + Code + ");"); - - if (GenDebug) { - if (!isRoot) { - emitCode("CurDAG->setSubgraphColor(" + - NodeName +".getNode(), \"yellow\");"); - emitCode("CurDAG->setSubgraphColor(" + - NodeName +".getNode(), \"black\");"); - } else { - emitCode("CurDAG->setSubgraphColor(" + NodeName +", \"yellow\");"); - emitCode("CurDAG->setSubgraphColor(" + NodeName +", \"black\");"); - } - } - - for (unsigned i = 0, e = After.size(); i != e; ++i) - emitCode(After[i]); - - return NodeOps; - } - if (Op->isSubClassOf("SDNodeXForm")) { - assert(N->getNumChildren() == 1 && "node xform should have one child!"); - // PatLeaf node - the operand may or may not be a leaf node. But it should - // behave like one. - std::vector<std::string> Ops = - EmitResultCode(N->getChild(0), DstRegs, InFlagDecled, - ResNodeDecled, true); - unsigned ResNo = TmpNo++; - emitCode("SDValue Tmp" + utostr(ResNo) + " = Transform_" + Op->getName() - + "(" + Ops.back() + ".getNode());"); - NodeOps.push_back("Tmp" + utostr(ResNo)); - if (isRoot) - emitCode("return Tmp" + utostr(ResNo) + ".getNode();"); - return NodeOps; - } - - N->dump(); - errs() << "\n"; - throw std::string("Unknown node in result pattern!"); -} - - -/// EmitCodeForPattern - Given a pattern to match, emit code to the specified -/// stream to match the pattern, and generate the code for the match if it -/// succeeds. Returns true if the pattern is not guaranteed to match. -void DAGISelEmitter::GenerateCodeForPattern(const PatternToMatch &Pattern, - std::vector<std::pair<unsigned, std::string> > &GeneratedCode, - std::set<std::string> &GeneratedDecl, - std::vector<std::string> &TargetOpcodes, - std::vector<std::string> &TargetVTs, - bool &OutputIsVariadic, - unsigned &NumInputRootOps) { - OutputIsVariadic = false; - NumInputRootOps = 0; - - PatternCodeEmitter Emitter(CGP, Pattern.getPredicateCheck(), - Pattern.getSrcPattern(), Pattern.getDstPattern(), - GeneratedCode, GeneratedDecl, - TargetOpcodes, TargetVTs, - OutputIsVariadic, NumInputRootOps); - - // Emit the matcher, capturing named arguments in VariableMap. - bool FoundChain = false; - Emitter.EmitMatchCode(Pattern.getSrcPattern(), NULL, "N", "", FoundChain); - - // TP - Get *SOME* tree pattern, we don't care which. It is only used for - // diagnostics, which we know are impossible at this point. - TreePattern &TP = *CGP.pf_begin()->second; - - // At this point, we know that we structurally match the pattern, but the - // types of the nodes may not match. Figure out the fewest number of type - // comparisons we need to emit. For example, if there is only one integer - // type supported by a target, there should be no type comparisons at all for - // integer patterns! - // - // To figure out the fewest number of type checks needed, clone the pattern, - // remove the types, then perform type inference on the pattern as a whole. - // If there are unresolved types, emit an explicit check for those types, - // apply the type to the tree, then rerun type inference. Iterate until all - // types are resolved. - // - TreePatternNode *Pat = Pattern.getSrcPattern()->clone(); - Pat->RemoveAllTypes(); - - do { - // Resolve/propagate as many types as possible. - try { - bool MadeChange = true; - while (MadeChange) - MadeChange = Pat->ApplyTypeConstraints(TP, - true/*Ignore reg constraints*/); - } catch (...) { - assert(0 && "Error: could not find consistent types for something we" - " already decided was ok!"); - abort(); - } - - // Insert a check for an unresolved type and add it to the tree. If we find - // an unresolved type to add a check for, this returns true and we iterate, - // otherwise we are done. - } while (Emitter.InsertOneTypeCheck(Pat, Pattern.getSrcPattern(), "N", true)); - - Emitter.EmitResultCode(Pattern.getDstPattern(), Pattern.getDstRegs(), - false, false, false, true); - delete Pat; -} - -/// EraseCodeLine - Erase one code line from all of the patterns. If removing -/// a line causes any of them to be empty, remove them and return true when -/// done. -static bool EraseCodeLine(std::vector<std::pair<const PatternToMatch*, - std::vector<std::pair<unsigned, std::string> > > > - &Patterns) { - bool ErasedPatterns = false; - for (unsigned i = 0, e = Patterns.size(); i != e; ++i) { - Patterns[i].second.pop_back(); - if (Patterns[i].second.empty()) { - Patterns.erase(Patterns.begin()+i); - --i; --e; - ErasedPatterns = true; - } - } - return ErasedPatterns; -} - -/// EmitPatterns - Emit code for at least one pattern, but try to group common -/// code together between the patterns. -void DAGISelEmitter::EmitPatterns(std::vector<std::pair<const PatternToMatch*, - std::vector<std::pair<unsigned, std::string> > > > - &Patterns, unsigned Indent, - raw_ostream &OS) { - typedef std::pair<unsigned, std::string> CodeLine; - typedef std::vector<CodeLine> CodeList; - typedef std::vector<std::pair<const PatternToMatch*, CodeList> > PatternList; - - if (Patterns.empty()) return; - - // Figure out how many patterns share the next code line. Explicitly copy - // FirstCodeLine so that we don't invalidate a reference when changing - // Patterns. - const CodeLine FirstCodeLine = Patterns.back().second.back(); - unsigned LastMatch = Patterns.size()-1; - while (LastMatch != 0 && Patterns[LastMatch-1].second.back() == FirstCodeLine) - --LastMatch; - - // If not all patterns share this line, split the list into two pieces. The - // first chunk will use this line, the second chunk won't. - if (LastMatch != 0) { - PatternList Shared(Patterns.begin()+LastMatch, Patterns.end()); - PatternList Other(Patterns.begin(), Patterns.begin()+LastMatch); - - // FIXME: Emit braces? - if (Shared.size() == 1) { - const PatternToMatch &Pattern = *Shared.back().first; - OS << "\n" << std::string(Indent, ' ') << "// Pattern: "; - Pattern.getSrcPattern()->print(OS); - OS << "\n" << std::string(Indent, ' ') << "// Emits: "; - Pattern.getDstPattern()->print(OS); - OS << "\n"; - unsigned AddedComplexity = Pattern.getAddedComplexity(); - OS << std::string(Indent, ' ') << "// Pattern complexity = " - << getPatternSize(Pattern.getSrcPattern(), CGP) + AddedComplexity - << " cost = " - << getResultPatternCost(Pattern.getDstPattern(), CGP) - << " size = " - << getResultPatternSize(Pattern.getDstPattern(), CGP) << "\n"; - } - if (FirstCodeLine.first != 1) { - OS << std::string(Indent, ' ') << "{\n"; - Indent += 2; - } - EmitPatterns(Shared, Indent, OS); - if (FirstCodeLine.first != 1) { - Indent -= 2; - OS << std::string(Indent, ' ') << "}\n"; - } + bool operator()(const PatternToMatch *LHS, + const PatternToMatch *RHS) { + unsigned LHSSize = getPatternSize(LHS->getSrcPattern(), CGP); + unsigned RHSSize = getPatternSize(RHS->getSrcPattern(), CGP); + LHSSize += LHS->getAddedComplexity(); + RHSSize += RHS->getAddedComplexity(); + if (LHSSize > RHSSize) return true; // LHS -> bigger -> less cost + if (LHSSize < RHSSize) return false; - if (Other.size() == 1) { - const PatternToMatch &Pattern = *Other.back().first; - OS << "\n" << std::string(Indent, ' ') << "// Pattern: "; - Pattern.getSrcPattern()->print(OS); - OS << "\n" << std::string(Indent, ' ') << "// Emits: "; - Pattern.getDstPattern()->print(OS); - OS << "\n"; - unsigned AddedComplexity = Pattern.getAddedComplexity(); - OS << std::string(Indent, ' ') << "// Pattern complexity = " - << getPatternSize(Pattern.getSrcPattern(), CGP) + AddedComplexity - << " cost = " - << getResultPatternCost(Pattern.getDstPattern(), CGP) - << " size = " - << getResultPatternSize(Pattern.getDstPattern(), CGP) << "\n"; - } - EmitPatterns(Other, Indent, OS); - return; - } - - // Remove this code from all of the patterns that share it. - bool ErasedPatterns = EraseCodeLine(Patterns); - - bool isPredicate = FirstCodeLine.first == 1; - - // Otherwise, every pattern in the list has this line. Emit it. - if (!isPredicate) { - // Normal code. - OS << std::string(Indent, ' ') << FirstCodeLine.second << "\n"; - } else { - OS << std::string(Indent, ' ') << "if (" << FirstCodeLine.second; + // If the patterns have equal complexity, compare generated instruction cost + unsigned LHSCost = getResultPatternCost(LHS->getDstPattern(), CGP); + unsigned RHSCost = getResultPatternCost(RHS->getDstPattern(), CGP); + if (LHSCost < RHSCost) return true; + if (LHSCost > RHSCost) return false; - // If the next code line is another predicate, and if all of the pattern - // in this group share the same next line, emit it inline now. Do this - // until we run out of common predicates. - while (!ErasedPatterns && Patterns.back().second.back().first == 1) { - // Check that all of the patterns in Patterns end with the same predicate. - bool AllEndWithSamePredicate = true; - for (unsigned i = 0, e = Patterns.size(); i != e; ++i) - if (Patterns[i].second.back() != Patterns.back().second.back()) { - AllEndWithSamePredicate = false; - break; - } - // If all of the predicates aren't the same, we can't share them. - if (!AllEndWithSamePredicate) break; - - // Otherwise we can. Emit it shared now. - OS << " &&\n" << std::string(Indent+4, ' ') - << Patterns.back().second.back().second; - ErasedPatterns = EraseCodeLine(Patterns); - } + unsigned LHSPatSize = getResultPatternSize(LHS->getDstPattern(), CGP); + unsigned RHSPatSize = getResultPatternSize(RHS->getDstPattern(), CGP); + if (LHSPatSize < RHSPatSize) return true; + if (LHSPatSize > RHSPatSize) return false; - OS << ") {\n"; - Indent += 2; + // Sort based on the UID of the pattern, giving us a deterministic ordering. + assert(LHS == RHS || LHS->ID != RHS->ID); + return LHS->ID < RHS->ID; } - - EmitPatterns(Patterns, Indent, OS); - - if (isPredicate) - OS << std::string(Indent-2, ' ') << "}\n"; -} - -static std::string getLegalCName(std::string OpName) { - std::string::size_type pos = OpName.find("::"); - if (pos != std::string::npos) - OpName.replace(pos, 2, "_"); - return OpName; +}; } -void DAGISelEmitter::EmitInstructionSelector(raw_ostream &OS) { - const CodeGenTarget &Target = CGP.getTargetInfo(); - - // Get the namespace to insert instructions into. - std::string InstNS = Target.getInstNamespace(); - if (!InstNS.empty()) InstNS += "::"; - - // Group the patterns by their top-level opcodes. - std::map<std::string, std::vector<const PatternToMatch*> > PatternsByOpcode; - // All unique target node emission functions. - std::map<std::string, unsigned> EmitFunctions; - for (CodeGenDAGPatterns::ptm_iterator I = CGP.ptm_begin(), - E = CGP.ptm_end(); I != E; ++I) { - const PatternToMatch &Pattern = *I; - TreePatternNode *Node = Pattern.getSrcPattern(); - if (!Node->isLeaf()) { - PatternsByOpcode[getOpcodeName(Node->getOperator(), CGP)]. - push_back(&Pattern); - } else { - const ComplexPattern *CP; - if (dynamic_cast<IntInit*>(Node->getLeafValue())) { - PatternsByOpcode[getOpcodeName(CGP.getSDNodeNamed("imm"), CGP)]. - push_back(&Pattern); - } else if ((CP = Node->getComplexPatternInfo(CGP))) { - std::vector<Record*> OpNodes = CP->getRootNodes(); - for (unsigned j = 0, e = OpNodes.size(); j != e; j++) { - PatternsByOpcode[getOpcodeName(OpNodes[j], CGP)] - .insert(PatternsByOpcode[getOpcodeName(OpNodes[j], CGP)].begin(), - &Pattern); - } - } else { - errs() << "Unrecognized opcode '"; - Node->dump(); - errs() << "' on tree pattern '"; - errs() << Pattern.getDstPattern()->getOperator()->getName() << "'!\n"; - exit(1); - } - } - } - - // For each opcode, there might be multiple select functions, one per - // ValueType of the node (or its first operand if it doesn't produce a - // non-chain result. - std::map<std::string, std::vector<std::string> > OpcodeVTMap; - - // Emit one Select_* method for each top-level opcode. We do this instead of - // emitting one giant switch statement to support compilers where this will - // result in the recursive functions taking less stack space. - for (std::map<std::string, std::vector<const PatternToMatch*> >::iterator - PBOI = PatternsByOpcode.begin(), E = PatternsByOpcode.end(); - PBOI != E; ++PBOI) { - const std::string &OpName = PBOI->first; - std::vector<const PatternToMatch*> &PatternsOfOp = PBOI->second; - assert(!PatternsOfOp.empty() && "No patterns but map has entry?"); - - // Split them into groups by type. - std::map<MVT::SimpleValueType, - std::vector<const PatternToMatch*> > PatternsByType; - for (unsigned i = 0, e = PatternsOfOp.size(); i != e; ++i) { - const PatternToMatch *Pat = PatternsOfOp[i]; - TreePatternNode *SrcPat = Pat->getSrcPattern(); - PatternsByType[SrcPat->getTypeNum(0)].push_back(Pat); - } - - for (std::map<MVT::SimpleValueType, - std::vector<const PatternToMatch*> >::iterator - II = PatternsByType.begin(), EE = PatternsByType.end(); II != EE; - ++II) { - MVT::SimpleValueType OpVT = II->first; - std::vector<const PatternToMatch*> &Patterns = II->second; - typedef std::pair<unsigned, std::string> CodeLine; - typedef std::vector<CodeLine> CodeList; - typedef CodeList::iterator CodeListI; - - std::vector<std::pair<const PatternToMatch*, CodeList> > CodeForPatterns; - std::vector<std::vector<std::string> > PatternOpcodes; - std::vector<std::vector<std::string> > PatternVTs; - std::vector<std::set<std::string> > PatternDecls; - std::vector<bool> OutputIsVariadicFlags; - std::vector<unsigned> NumInputRootOpsCounts; - for (unsigned i = 0, e = Patterns.size(); i != e; ++i) { - CodeList GeneratedCode; - std::set<std::string> GeneratedDecl; - std::vector<std::string> TargetOpcodes; - std::vector<std::string> TargetVTs; - bool OutputIsVariadic; - unsigned NumInputRootOps; - GenerateCodeForPattern(*Patterns[i], GeneratedCode, GeneratedDecl, - TargetOpcodes, TargetVTs, - OutputIsVariadic, NumInputRootOps); - CodeForPatterns.push_back(std::make_pair(Patterns[i], GeneratedCode)); - PatternDecls.push_back(GeneratedDecl); - PatternOpcodes.push_back(TargetOpcodes); - PatternVTs.push_back(TargetVTs); - OutputIsVariadicFlags.push_back(OutputIsVariadic); - NumInputRootOpsCounts.push_back(NumInputRootOps); - } - - // Factor target node emission code (emitted by EmitResultCode) into - // separate functions. Uniquing and share them among all instruction - // selection routines. - for (unsigned i = 0, e = CodeForPatterns.size(); i != e; ++i) { - CodeList &GeneratedCode = CodeForPatterns[i].second; - std::vector<std::string> &TargetOpcodes = PatternOpcodes[i]; - std::vector<std::string> &TargetVTs = PatternVTs[i]; - std::set<std::string> Decls = PatternDecls[i]; - bool OutputIsVariadic = OutputIsVariadicFlags[i]; - unsigned NumInputRootOps = NumInputRootOpsCounts[i]; - std::vector<std::string> AddedInits; - int CodeSize = (int)GeneratedCode.size(); - int LastPred = -1; - for (int j = CodeSize-1; j >= 0; --j) { - if (LastPred == -1 && GeneratedCode[j].first == 1) - LastPred = j; - else if (LastPred != -1 && GeneratedCode[j].first == 2) - AddedInits.push_back(GeneratedCode[j].second); - } - - std::string CalleeCode = "(SDNode *N"; - std::string CallerCode = "(N"; - for (unsigned j = 0, e = TargetOpcodes.size(); j != e; ++j) { - CalleeCode += ", unsigned Opc" + utostr(j); - CallerCode += ", " + TargetOpcodes[j]; - } - for (unsigned j = 0, e = TargetVTs.size(); j != e; ++j) { - CalleeCode += ", MVT::SimpleValueType VT" + utostr(j); - CallerCode += ", " + TargetVTs[j]; - } - for (std::set<std::string>::iterator - I = Decls.begin(), E = Decls.end(); I != E; ++I) { - std::string Name = *I; - CalleeCode += ", SDValue &" + Name; - CallerCode += ", " + Name; - } - - if (OutputIsVariadic) { - CalleeCode += ", unsigned NumInputRootOps"; - CallerCode += ", " + utostr(NumInputRootOps); - } - - CallerCode += ");"; - CalleeCode += ") {\n"; - - for (std::vector<std::string>::const_reverse_iterator - I = AddedInits.rbegin(), E = AddedInits.rend(); I != E; ++I) - CalleeCode += " " + *I + "\n"; - - for (int j = LastPred+1; j < CodeSize; ++j) - CalleeCode += " " + GeneratedCode[j].second + "\n"; - for (int j = LastPred+1; j < CodeSize; ++j) - GeneratedCode.pop_back(); - CalleeCode += "}\n"; - - // Uniquing the emission routines. - unsigned EmitFuncNum; - std::map<std::string, unsigned>::iterator EFI = - EmitFunctions.find(CalleeCode); - if (EFI != EmitFunctions.end()) { - EmitFuncNum = EFI->second; - } else { - EmitFuncNum = EmitFunctions.size(); - EmitFunctions.insert(std::make_pair(CalleeCode, EmitFuncNum)); - // Prevent emission routines from being inlined to reduce selection - // routines stack frame sizes. - OS << "DISABLE_INLINE "; - OS << "SDNode *Emit_" << utostr(EmitFuncNum) << CalleeCode; - } - - // Replace the emission code within selection routines with calls to the - // emission functions. - if (GenDebug) - GeneratedCode.push_back(std::make_pair(0, "CurDAG->setSubgraphColor(N, \"red\");")); - CallerCode = "SDNode *Result = Emit_" + utostr(EmitFuncNum) + CallerCode; - GeneratedCode.push_back(std::make_pair(3, CallerCode)); - if (GenDebug) { - GeneratedCode.push_back(std::make_pair(0, "if(Result) {")); - GeneratedCode.push_back(std::make_pair(0, " CurDAG->setSubgraphColor(Result, \"yellow\");")); - GeneratedCode.push_back(std::make_pair(0, " CurDAG->setSubgraphColor(Result, \"black\");")); - GeneratedCode.push_back(std::make_pair(0, "}")); - //GeneratedCode.push_back(std::make_pair(0, "CurDAG->setSubgraphColor(N, \"black\");")); - } - GeneratedCode.push_back(std::make_pair(0, "return Result;")); - } - - // Print function. - std::string OpVTStr; - if (OpVT == MVT::iPTR) { - OpVTStr = "_iPTR"; - } else if (OpVT == MVT::iPTRAny) { - OpVTStr = "_iPTRAny"; - } else if (OpVT == MVT::isVoid) { - // Nodes with a void result actually have a first result type of either - // Other (a chain) or Flag. Since there is no one-to-one mapping from - // void to this case, we handle it specially here. - } else { - OpVTStr = "_" + getEnumName(OpVT).substr(5); // Skip 'MVT::' - } - std::map<std::string, std::vector<std::string> >::iterator OpVTI = - OpcodeVTMap.find(OpName); - if (OpVTI == OpcodeVTMap.end()) { - std::vector<std::string> VTSet; - VTSet.push_back(OpVTStr); - OpcodeVTMap.insert(std::make_pair(OpName, VTSet)); - } else - OpVTI->second.push_back(OpVTStr); - - // We want to emit all of the matching code now. However, we want to emit - // the matches in order of minimal cost. Sort the patterns so the least - // cost one is at the start. - std::stable_sort(CodeForPatterns.begin(), CodeForPatterns.end(), - PatternSortingPredicate(CGP)); - - // Scan the code to see if all of the patterns are reachable and if it is - // possible that the last one might not match. - bool mightNotMatch = true; - for (unsigned i = 0, e = CodeForPatterns.size(); i != e; ++i) { - CodeList &GeneratedCode = CodeForPatterns[i].second; - mightNotMatch = false; - - for (unsigned j = 0, e = GeneratedCode.size(); j != e; ++j) { - if (GeneratedCode[j].first == 1) { // predicate. - mightNotMatch = true; - break; - } - } - - // If this pattern definitely matches, and if it isn't the last one, the - // patterns after it CANNOT ever match. Error out. - if (mightNotMatch == false && i != CodeForPatterns.size()-1) { - errs() << "Pattern '"; - CodeForPatterns[i].first->getSrcPattern()->print(errs()); - errs() << "' is impossible to select!\n"; - exit(1); - } - } - - // Loop through and reverse all of the CodeList vectors, as we will be - // accessing them from their logical front, but accessing the end of a - // vector is more efficient. - for (unsigned i = 0, e = CodeForPatterns.size(); i != e; ++i) { - CodeList &GeneratedCode = CodeForPatterns[i].second; - std::reverse(GeneratedCode.begin(), GeneratedCode.end()); - } - - // Next, reverse the list of patterns itself for the same reason. - std::reverse(CodeForPatterns.begin(), CodeForPatterns.end()); - - OS << "SDNode *Select_" << getLegalCName(OpName) - << OpVTStr << "(SDNode *N) {\n"; - - // Emit all of the patterns now, grouped together to share code. - EmitPatterns(CodeForPatterns, 2, OS); - - // If the last pattern has predicates (which could fail) emit code to - // catch the case where nothing handles a pattern. - if (mightNotMatch) { - OS << "\n"; - if (OpName != "ISD::INTRINSIC_W_CHAIN" && - OpName != "ISD::INTRINSIC_WO_CHAIN" && - OpName != "ISD::INTRINSIC_VOID") - OS << " CannotYetSelect(N);\n"; - else - OS << " CannotYetSelectIntrinsic(N);\n"; - - OS << " return NULL;\n"; - } - OS << "}\n\n"; - } - } - - OS << "// The main instruction selector code.\n" - << "SDNode *SelectCode(SDNode *N) {\n" - << " MVT::SimpleValueType NVT = N->getValueType(0).getSimpleVT().SimpleTy;\n" - << " switch (N->getOpcode()) {\n" - << " default:\n" - << " assert(!N->isMachineOpcode() && \"Node already selected!\");\n" - << " break;\n" - << " case ISD::EntryToken: // These nodes remain the same.\n" - << " case ISD::BasicBlock:\n" - << " case ISD::Register:\n" - << " case ISD::HANDLENODE:\n" - << " case ISD::TargetConstant:\n" - << " case ISD::TargetConstantFP:\n" - << " case ISD::TargetConstantPool:\n" - << " case ISD::TargetFrameIndex:\n" - << " case ISD::TargetExternalSymbol:\n" - << " case ISD::TargetBlockAddress:\n" - << " case ISD::TargetJumpTable:\n" - << " case ISD::TargetGlobalTLSAddress:\n" - << " case ISD::TargetGlobalAddress:\n" - << " case ISD::TokenFactor:\n" - << " case ISD::CopyFromReg:\n" - << " case ISD::CopyToReg: {\n" - << " return NULL;\n" - << " }\n" - << " case ISD::AssertSext:\n" - << " case ISD::AssertZext: {\n" - << " ReplaceUses(SDValue(N, 0), N->getOperand(0));\n" - << " return NULL;\n" - << " }\n" - << " case ISD::INLINEASM: return Select_INLINEASM(N);\n" - << " case ISD::EH_LABEL: return Select_EH_LABEL(N);\n" - << " case ISD::UNDEF: return Select_UNDEF(N);\n"; - - // Loop over all of the case statements, emiting a call to each method we - // emitted above. - for (std::map<std::string, std::vector<const PatternToMatch*> >::iterator - PBOI = PatternsByOpcode.begin(), E = PatternsByOpcode.end(); - PBOI != E; ++PBOI) { - const std::string &OpName = PBOI->first; - // Potentially multiple versions of select for this opcode. One for each - // ValueType of the node (or its first true operand if it doesn't produce a - // result. - std::map<std::string, std::vector<std::string> >::iterator OpVTI = - OpcodeVTMap.find(OpName); - std::vector<std::string> &OpVTs = OpVTI->second; - OS << " case " << OpName << ": {\n"; - // If we have only one variant and it's the default, elide the - // switch. Marginally faster, and makes MSVC happier. - if (OpVTs.size()==1 && OpVTs[0].empty()) { - OS << " return Select_" << getLegalCName(OpName) << "(N);\n"; - OS << " break;\n"; - OS << " }\n"; - continue; - } - // Keep track of whether we see a pattern that has an iPtr result. - bool HasPtrPattern = false; - bool HasDefaultPattern = false; - - OS << " switch (NVT) {\n"; - for (unsigned i = 0, e = OpVTs.size(); i < e; ++i) { - std::string &VTStr = OpVTs[i]; - if (VTStr.empty()) { - HasDefaultPattern = true; - continue; - } - - // If this is a match on iPTR: don't emit it directly, we need special - // code. - if (VTStr == "_iPTR") { - HasPtrPattern = true; - continue; - } - OS << " case MVT::" << VTStr.substr(1) << ":\n" - << " return Select_" << getLegalCName(OpName) - << VTStr << "(N);\n"; - } - OS << " default:\n"; - - // If there is an iPTR result version of this pattern, emit it here. - if (HasPtrPattern) { - OS << " if (TLI.getPointerTy() == NVT)\n"; - OS << " return Select_" << getLegalCName(OpName) <<"_iPTR(N);\n"; - } - if (HasDefaultPattern) { - OS << " return Select_" << getLegalCName(OpName) << "(N);\n"; - } - OS << " break;\n"; - OS << " }\n"; - OS << " break;\n"; - OS << " }\n"; - } - - OS << " } // end of big switch.\n\n" - << " if (N->getOpcode() != ISD::INTRINSIC_W_CHAIN &&\n" - << " N->getOpcode() != ISD::INTRINSIC_WO_CHAIN &&\n" - << " N->getOpcode() != ISD::INTRINSIC_VOID) {\n" - << " CannotYetSelect(N);\n" - << " } else {\n" - << " CannotYetSelectIntrinsic(N);\n" - << " }\n" - << " return NULL;\n" - << "}\n\n"; -} void DAGISelEmitter::run(raw_ostream &OS) { EmitSourceFileHeader("DAG Instruction Selector for the " + @@ -1978,46 +195,45 @@ void DAGISelEmitter::run(raw_ostream &OS) { << "// *** instruction selector class. These functions are really " << "methods.\n\n"; - OS << "// Include standard, target-independent definitions and methods used\n" - << "// by the instruction selector.\n"; - OS << "#include \"llvm/CodeGen/DAGISelHeader.h\"\n\n"; - - EmitNodeTransforms(OS); + DEBUG(errs() << "\n\nALL PATTERNS TO MATCH:\n\n"; + for (CodeGenDAGPatterns::ptm_iterator I = CGP.ptm_begin(), + E = CGP.ptm_end(); I != E; ++I) { + errs() << "PATTERN: "; I->getSrcPattern()->dump(); + errs() << "\nRESULT: "; I->getDstPattern()->dump(); + errs() << "\n"; + }); + + // FIXME: These are being used by hand written code, gross. EmitPredicateFunctions(OS); - - DEBUG(errs() << "\n\nALL PATTERNS TO MATCH:\n\n"); + + // Add all the patterns to a temporary list so we can sort them. + std::vector<const PatternToMatch*> Patterns; for (CodeGenDAGPatterns::ptm_iterator I = CGP.ptm_begin(), E = CGP.ptm_end(); - I != E; ++I) { - DEBUG(errs() << "PATTERN: "; I->getSrcPattern()->dump()); - DEBUG(errs() << "\nRESULT: "; I->getDstPattern()->dump()); - DEBUG(errs() << "\n"); - } + I != E; ++I) + Patterns.push_back(&*I); + + // We want to process the matches in order of minimal cost. Sort the patterns + // so the least cost one is at the start. + std::stable_sort(Patterns.begin(), Patterns.end(), + PatternSortingPredicate(CGP)); - // At this point, we have full information about the 'Patterns' we need to - // parse, both implicitly from instructions as well as from explicit pattern - // definitions. Emit the resultant instruction selector. - EmitInstructionSelector(OS); -#if 0 - MatcherNode *Matcher = 0; - // Walk the patterns backwards, building a matcher for each and adding it to - // the matcher for the whole target. - for (CodeGenDAGPatterns::ptm_iterator I = CGP.ptm_begin(), - E = CGP.ptm_end(); I != E;) { - const PatternToMatch &Pattern = *--E; - MatcherNode *N = ConvertPatternToMatcher(Pattern, CGP); - - if (Matcher == 0) - Matcher = N; - else - Matcher = new PushMatcherNode(N, Matcher); + // Convert each variant of each pattern into a Matcher. + std::vector<Matcher*> PatternMatchers; + for (unsigned i = 0, e = Patterns.size(); i != e; ++i) { + for (unsigned Variant = 0; ; ++Variant) { + if (Matcher *M = ConvertPatternToMatcher(*Patterns[i], Variant, CGP)) + PatternMatchers.push_back(M); + else + break; + } } - - - EmitMatcherTable(Matcher, OS); - - + + Matcher *TheMatcher = new ScopeMatcher(&PatternMatchers[0], + PatternMatchers.size()); + + TheMatcher = OptimizeMatcher(TheMatcher, CGP); //Matcher->dump(); - delete Matcher; -#endif + EmitMatcherTable(TheMatcher, CGP, OS); + delete TheMatcher; } diff --git a/utils/TableGen/DAGISelEmitter.h b/utils/TableGen/DAGISelEmitter.h index d5b889b..5ffdde8 100644 --- a/utils/TableGen/DAGISelEmitter.h +++ b/utils/TableGen/DAGISelEmitter.h @@ -31,24 +31,8 @@ public: // run - Output the isel, returning true on failure. void run(raw_ostream &OS); - - private: - void EmitNodeTransforms(raw_ostream &OS); void EmitPredicateFunctions(raw_ostream &OS); - - void GenerateCodeForPattern(const PatternToMatch &Pattern, - std::vector<std::pair<unsigned, std::string> > &GeneratedCode, - std::set<std::string> &GeneratedDecl, - std::vector<std::string> &TargetOpcodes, - std::vector<std::string> &TargetVTs, - bool &OutputIsVariadic, - unsigned &NumInputRootOps); - void EmitPatterns(std::vector<std::pair<const PatternToMatch*, - std::vector<std::pair<unsigned, std::string> > > > &Patterns, - unsigned Indent, raw_ostream &OS); - - void EmitInstructionSelector(raw_ostream &OS); }; } // End llvm namespace diff --git a/utils/TableGen/DAGISelMatcher.cpp b/utils/TableGen/DAGISelMatcher.cpp index 3f75558..c4f1cbf 100644 --- a/utils/TableGen/DAGISelMatcher.cpp +++ b/utils/TableGen/DAGISelMatcher.cpp @@ -10,109 +10,338 @@ #include "DAGISelMatcher.h" #include "CodeGenDAGPatterns.h" #include "CodeGenTarget.h" +#include "Record.h" #include "llvm/Support/raw_ostream.h" +#include "llvm/ADT/StringExtras.h" using namespace llvm; -void MatcherNode::dump() const { - print(errs()); +void Matcher::dump() const { + print(errs(), 0); } -void EmitNodeMatcherNode::print(raw_ostream &OS, unsigned indent) const { - OS.indent(indent) << "EmitNode: Src = " << *Pattern.getSrcPattern() << "\n"; - OS.indent(indent) << "EmitNode: Dst = " << *Pattern.getDstPattern() << "\n"; +void Matcher::print(raw_ostream &OS, unsigned indent) const { + printImpl(OS, indent); + if (Next) + return Next->print(OS, indent); } -void MatcherNodeWithChild::printChild(raw_ostream &OS, unsigned indent) const { - if (Child) - return Child->print(OS, indent); - OS.indent(indent) << "<null child>\n"; +void Matcher::printOne(raw_ostream &OS) const { + printImpl(OS, 0); } +ScopeMatcher::~ScopeMatcher() { + for (unsigned i = 0, e = Children.size(); i != e; ++i) + delete Children[i]; +} + + +// printImpl methods. -void PushMatcherNode::print(raw_ostream &OS, unsigned indent) const { - OS.indent(indent) << "Push\n"; - printChild(OS, indent+2); - Failure->print(OS, indent); +void ScopeMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "Scope\n"; + for (unsigned i = 0, e = getNumChildren(); i != e; ++i) { + if (getChild(i) == 0) + OS.indent(indent+1) << "NULL POINTER\n"; + else + getChild(i)->print(OS, indent+2); + } } -void RecordMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void RecordMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "Record\n"; - printChild(OS, indent); } -void MoveChildMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void RecordChildMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "RecordChild: " << ChildNo << '\n'; +} + +void RecordMemRefMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "RecordMemRef\n"; +} + +void CaptureFlagInputMatcher::printImpl(raw_ostream &OS, unsigned indent) const{ + OS.indent(indent) << "CaptureFlagInput\n"; +} + +void MoveChildMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "MoveChild " << ChildNo << '\n'; - printChild(OS, indent); } -void MoveParentMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void MoveParentMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "MoveParent\n"; - printChild(OS, indent); } -void CheckSameMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void CheckSameMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckSame " << MatchNumber << '\n'; - printChild(OS, indent); } -void CheckPatternPredicateMatcherNode:: -print(raw_ostream &OS, unsigned indent) const { +void CheckPatternPredicateMatcher:: +printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckPatternPredicate " << Predicate << '\n'; - printChild(OS, indent); } -void CheckPredicateMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void CheckPredicateMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckPredicate " << PredName << '\n'; - printChild(OS, indent); } -void CheckOpcodeMatcherNode::print(raw_ostream &OS, unsigned indent) const { - OS.indent(indent) << "CheckOpcode " << OpcodeName << '\n'; - printChild(OS, indent); +void CheckOpcodeMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "CheckOpcode " << Opcode.getEnumName() << '\n'; } -void CheckTypeMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void SwitchOpcodeMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "SwitchOpcode: {\n"; + for (unsigned i = 0, e = Cases.size(); i != e; ++i) { + OS.indent(indent) << "case " << Cases[i].first->getEnumName() << ":\n"; + Cases[i].second->print(OS, indent+2); + } + OS.indent(indent) << "}\n"; +} + + +void CheckTypeMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckType " << getEnumName(Type) << '\n'; - printChild(OS, indent); } -void CheckIntegerMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void SwitchTypeMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "SwitchType: {\n"; + for (unsigned i = 0, e = Cases.size(); i != e; ++i) { + OS.indent(indent) << "case " << getEnumName(Cases[i].first) << ":\n"; + Cases[i].second->print(OS, indent+2); + } + OS.indent(indent) << "}\n"; +} + +void CheckChildTypeMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "CheckChildType " << ChildNo << " " + << getEnumName(Type) << '\n'; +} + + +void CheckIntegerMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckInteger " << Value << '\n'; - printChild(OS, indent); } -void CheckCondCodeMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void CheckCondCodeMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckCondCode ISD::" << CondCodeName << '\n'; - printChild(OS, indent); } -void CheckValueTypeMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void CheckValueTypeMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckValueType MVT::" << TypeName << '\n'; - printChild(OS, indent); } -void CheckComplexPatMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void CheckComplexPatMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckComplexPat " << Pattern.getSelectFunc() << '\n'; - printChild(OS, indent); } -void CheckAndImmMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void CheckAndImmMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckAndImm " << Value << '\n'; - printChild(OS, indent); } -void CheckOrImmMatcherNode::print(raw_ostream &OS, unsigned indent) const { +void CheckOrImmMatcher::printImpl(raw_ostream &OS, unsigned indent) const { OS.indent(indent) << "CheckOrImm " << Value << '\n'; - printChild(OS, indent); } -void CheckProfitableToFoldMatcherNode::print(raw_ostream &OS, - unsigned indent) const { - OS.indent(indent) << "CheckProfitableToFold\n"; - printChild(OS, indent); +void CheckFoldableChainNodeMatcher::printImpl(raw_ostream &OS, + unsigned indent) const { + OS.indent(indent) << "CheckFoldableChainNode\n"; +} + +void EmitIntegerMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "EmitInteger " << Val << " VT=" << VT << '\n'; +} + +void EmitStringIntegerMatcher:: +printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "EmitStringInteger " << Val << " VT=" << VT << '\n'; +} + +void EmitRegisterMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "EmitRegister "; + if (Reg) + OS << Reg->getName(); + else + OS << "zero_reg"; + OS << " VT=" << VT << '\n'; +} + +void EmitConvertToTargetMatcher:: +printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "EmitConvertToTarget " << Slot << '\n'; +} + +void EmitMergeInputChainsMatcher:: +printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "EmitMergeInputChains <todo: args>\n"; +} + +void EmitCopyToRegMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "EmitCopyToReg <todo: args>\n"; +} + +void EmitNodeXFormMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "EmitNodeXForm " << NodeXForm->getName() + << " Slot=" << Slot << '\n'; +} + + +void EmitNodeMatcherCommon::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent); + OS << (isa<MorphNodeToMatcher>(this) ? "MorphNodeTo: " : "EmitNode: ") + << OpcodeName << ": <todo flags> "; + + for (unsigned i = 0, e = VTs.size(); i != e; ++i) + OS << ' ' << getEnumName(VTs[i]); + OS << '('; + for (unsigned i = 0, e = Operands.size(); i != e; ++i) + OS << Operands[i] << ' '; + OS << ")\n"; +} + +void MarkFlagResultsMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "MarkFlagResults <todo: args>\n"; +} + +void CompleteMatchMatcher::printImpl(raw_ostream &OS, unsigned indent) const { + OS.indent(indent) << "CompleteMatch <todo args>\n"; + OS.indent(indent) << "Src = " << *Pattern.getSrcPattern() << "\n"; + OS.indent(indent) << "Dst = " << *Pattern.getDstPattern() << "\n"; +} + +// getHashImpl Implementation. + +unsigned CheckPatternPredicateMatcher::getHashImpl() const { + return HashString(Predicate); +} + +unsigned CheckPredicateMatcher::getHashImpl() const { + return HashString(PredName); +} + +unsigned CheckOpcodeMatcher::getHashImpl() const { + return HashString(Opcode.getEnumName()); +} + +unsigned CheckCondCodeMatcher::getHashImpl() const { + return HashString(CondCodeName); +} + +unsigned CheckValueTypeMatcher::getHashImpl() const { + return HashString(TypeName); +} + +unsigned EmitStringIntegerMatcher::getHashImpl() const { + return HashString(Val) ^ VT; +} + +template<typename It> +static unsigned HashUnsigneds(It I, It E) { + unsigned Result = 0; + for (; I != E; ++I) + Result = (Result<<3) ^ *I; + return Result; +} + +unsigned EmitMergeInputChainsMatcher::getHashImpl() const { + return HashUnsigneds(ChainNodes.begin(), ChainNodes.end()); +} + +bool CheckOpcodeMatcher::isEqualImpl(const Matcher *M) const { + // Note: pointer equality isn't enough here, we have to check the enum names + // to ensure that the nodes are for the same opcode. + return cast<CheckOpcodeMatcher>(M)->Opcode.getEnumName() == + Opcode.getEnumName(); +} + + +bool EmitNodeMatcherCommon::isEqualImpl(const Matcher *m) const { + const EmitNodeMatcherCommon *M = cast<EmitNodeMatcherCommon>(m); + return M->OpcodeName == OpcodeName && M->VTs == VTs && + M->Operands == Operands && M->HasChain == HasChain && + M->HasInFlag == HasInFlag && M->HasOutFlag == HasOutFlag && + M->HasMemRefs == HasMemRefs && + M->NumFixedArityOperands == NumFixedArityOperands; +} + +unsigned EmitNodeMatcherCommon::getHashImpl() const { + return (HashString(OpcodeName) << 4) | Operands.size(); } -void CheckLegalToFoldMatcherNode::print(raw_ostream &OS, unsigned indent) const{ - OS.indent(indent) << "CheckLegalToFold\n"; - printChild(OS, indent); + +unsigned MarkFlagResultsMatcher::getHashImpl() const { + return HashUnsigneds(FlagResultNodes.begin(), FlagResultNodes.end()); +} + +unsigned CompleteMatchMatcher::getHashImpl() const { + return HashUnsigneds(Results.begin(), Results.end()) ^ + ((unsigned)(intptr_t)&Pattern << 8); +} + +// isContradictoryImpl Implementations. + +static bool TypesAreContradictory(MVT::SimpleValueType T1, + MVT::SimpleValueType T2) { + // If the two types are the same, then they are the same, so they don't + // contradict. + if (T1 == T2) return false; + + // If either type is about iPtr, then they don't conflict unless the other + // one is not a scalar integer type. + if (T1 == MVT::iPTR) + return !MVT(T2).isInteger() || MVT(T2).isVector(); + + if (T2 == MVT::iPTR) + return !MVT(T1).isInteger() || MVT(T1).isVector(); + + // Otherwise, they are two different non-iPTR types, they conflict. + return true; +} + +bool CheckOpcodeMatcher::isContradictoryImpl(const Matcher *M) const { + if (const CheckOpcodeMatcher *COM = dyn_cast<CheckOpcodeMatcher>(M)) { + // One node can't have two different opcodes! + // Note: pointer equality isn't enough here, we have to check the enum names + // to ensure that the nodes are for the same opcode. + return COM->getOpcode().getEnumName() != getOpcode().getEnumName(); + } + + // If the node has a known type, and if the type we're checking for is + // different, then we know they contradict. For example, a check for + // ISD::STORE will never be true at the same time a check for Type i32 is. + if (const CheckTypeMatcher *CT = dyn_cast<CheckTypeMatcher>(M)) { + // FIXME: What result is this referring to? + unsigned NodeType; + if (getOpcode().getNumResults() == 0) + NodeType = MVT::isVoid; + else + NodeType = getOpcode().getKnownType(); + if (NodeType != EEVT::isUnknown) + return TypesAreContradictory((MVT::SimpleValueType)NodeType, + CT->getType()); + } + + return false; +} + +bool CheckTypeMatcher::isContradictoryImpl(const Matcher *M) const { + if (const CheckTypeMatcher *CT = dyn_cast<CheckTypeMatcher>(M)) + return TypesAreContradictory(getType(), CT->getType()); + return false; +} + +bool CheckChildTypeMatcher::isContradictoryImpl(const Matcher *M) const { + if (const CheckChildTypeMatcher *CC = dyn_cast<CheckChildTypeMatcher>(M)) { + // If the two checks are about different nodes, we don't know if they + // conflict! + if (CC->getChildNo() != getChildNo()) + return false; + + return TypesAreContradictory(getType(), CC->getType()); + } + return false; +} + +bool CheckIntegerMatcher::isContradictoryImpl(const Matcher *M) const { + if (const CheckIntegerMatcher *CIM = dyn_cast<CheckIntegerMatcher>(M)) + return CIM->getValue() != getValue(); + return false; } diff --git a/utils/TableGen/DAGISelMatcher.h b/utils/TableGen/DAGISelMatcher.h index b40fbf9..c2e8171 100644 --- a/utils/TableGen/DAGISelMatcher.h +++ b/utils/TableGen/DAGISelMatcher.h @@ -13,379 +13,1038 @@ #include "llvm/CodeGen/ValueTypes.h" #include "llvm/ADT/OwningPtr.h" #include "llvm/ADT/StringRef.h" +#include "llvm/ADT/SmallVector.h" #include "llvm/Support/Casting.h" namespace llvm { class CodeGenDAGPatterns; - class MatcherNode; + class Matcher; class PatternToMatch; class raw_ostream; class ComplexPattern; + class Record; + class SDNodeInfo; -MatcherNode *ConvertPatternToMatcher(const PatternToMatch &Pattern, - const CodeGenDAGPatterns &CGP); - -void EmitMatcherTable(const MatcherNode *Matcher, raw_ostream &OS); +Matcher *ConvertPatternToMatcher(const PatternToMatch &Pattern,unsigned Variant, + const CodeGenDAGPatterns &CGP); +Matcher *OptimizeMatcher(Matcher *Matcher, const CodeGenDAGPatterns &CGP); +void EmitMatcherTable(const Matcher *Matcher, const CodeGenDAGPatterns &CGP, + raw_ostream &OS); -/// MatcherNode - Base class for all the the DAG ISel Matcher representation +/// Matcher - Base class for all the the DAG ISel Matcher representation /// nodes. -class MatcherNode { +class Matcher { + // The next matcher node that is executed after this one. Null if this is the + // last stage of a match. + OwningPtr<Matcher> Next; public: enum KindTy { - EmitNode, - Push, // [Push, Dest0, Dest1, Dest2, Dest3] - Record, // [Record] - MoveChild, // [MoveChild, Child#] - MoveParent, // [MoveParent] + // Matcher state manipulation. + Scope, // Push a checking scope. + RecordNode, // Record the current node. + RecordChild, // Record a child of the current node. + RecordMemRef, // Record the memref in the current node. + CaptureFlagInput, // If the current node has an input flag, save it. + MoveChild, // Move current node to specified child. + MoveParent, // Move current node to parent. - CheckSame, // [CheckSame, N] Fail if not same as prev match. + // Predicate checking. + CheckSame, // Fail if not same as prev match. CheckPatternPredicate, - CheckPredicate, // [CheckPredicate, P] Fail if predicate fails. - CheckOpcode, // [CheckOpcode, Opcode] Fail if not opcode. - CheckType, // [CheckType, MVT] Fail if not correct type. - CheckInteger, // [CheckInteger, int0,int1,int2,...int7] Fail if wrong val. - CheckCondCode, // [CheckCondCode, CondCode] Fail if not condcode. + CheckPredicate, // Fail if node predicate fails. + CheckOpcode, // Fail if not opcode. + SwitchOpcode, // Dispatch based on opcode. + CheckType, // Fail if not correct type. + SwitchType, // Dispatch based on type. + CheckChildType, // Fail if child has wrong type. + CheckInteger, // Fail if wrong val. + CheckCondCode, // Fail if not condcode. CheckValueType, CheckComplexPat, CheckAndImm, CheckOrImm, - CheckProfitableToFold, - CheckLegalToFold + CheckFoldableChainNode, + + // Node creation/emisssion. + EmitInteger, // Create a TargetConstant + EmitStringInteger, // Create a TargetConstant from a string. + EmitRegister, // Create a register. + EmitConvertToTarget, // Convert a imm/fpimm to target imm/fpimm + EmitMergeInputChains, // Merge together a chains for an input. + EmitCopyToReg, // Emit a copytoreg into a physreg. + EmitNode, // Create a DAG node + EmitNodeXForm, // Run a SDNodeXForm + MarkFlagResults, // Indicate which interior nodes have flag results. + CompleteMatch, // Finish a match and update the results. + MorphNodeTo // Build a node, finish a match and update results. }; const KindTy Kind; - + protected: - MatcherNode(KindTy K) : Kind(K) {} + Matcher(KindTy K) : Kind(K) {} public: - virtual ~MatcherNode() {} + virtual ~Matcher() {} KindTy getKind() const { return Kind; } + + Matcher *getNext() { return Next.get(); } + const Matcher *getNext() const { return Next.get(); } + void setNext(Matcher *C) { Next.reset(C); } + Matcher *takeNext() { return Next.take(); } + + OwningPtr<Matcher> &getNextPtr() { return Next; } + static inline bool classof(const Matcher *) { return true; } - static inline bool classof(const MatcherNode *) { return true; } + bool isEqual(const Matcher *M) const { + if (getKind() != M->getKind()) return false; + return isEqualImpl(M); + } + + unsigned getHash() const { + // Clear the high bit so we don't conflict with tombstones etc. + return ((getHashImpl() << 4) ^ getKind()) & (~0U>>1); + } + + /// isSafeToReorderWithPatternPredicate - Return true if it is safe to sink a + /// PatternPredicate node past this one. + virtual bool isSafeToReorderWithPatternPredicate() const { + return false; + } + + /// isContradictory - Return true of these two matchers could never match on + /// the same node. + bool isContradictory(const Matcher *Other) const { + // Since this predicate is reflexive, we canonicalize the ordering so that + // we always match a node against nodes with kinds that are greater or equal + // to them. For example, we'll pass in a CheckType node as an argument to + // the CheckOpcode method, not the other way around. + if (getKind() < Other->getKind()) + return isContradictoryImpl(Other); + return Other->isContradictoryImpl(this); + } - virtual void print(raw_ostream &OS, unsigned indent = 0) const = 0; + void print(raw_ostream &OS, unsigned indent = 0) const; + void printOne(raw_ostream &OS) const; void dump() const; +protected: + virtual void printImpl(raw_ostream &OS, unsigned indent) const = 0; + virtual bool isEqualImpl(const Matcher *M) const = 0; + virtual unsigned getHashImpl() const = 0; + virtual bool isContradictoryImpl(const Matcher *M) const { return false; } }; -/// EmitNodeMatcherNode - This signals a successful match and generates a node. -class EmitNodeMatcherNode : public MatcherNode { - const PatternToMatch &Pattern; +/// ScopeMatcher - This attempts to match each of its children to find the first +/// one that successfully matches. If one child fails, it tries the next child. +/// If none of the children match then this check fails. It never has a 'next'. +class ScopeMatcher : public Matcher { + SmallVector<Matcher*, 4> Children; public: - EmitNodeMatcherNode(const PatternToMatch &pattern) - : MatcherNode(EmitNode), Pattern(pattern) {} - - const PatternToMatch &getPattern() const { return Pattern; } + ScopeMatcher(Matcher *const *children, unsigned numchildren) + : Matcher(Scope), Children(children, children+numchildren) { + } + virtual ~ScopeMatcher(); + + unsigned getNumChildren() const { return Children.size(); } + + Matcher *getChild(unsigned i) { return Children[i]; } + const Matcher *getChild(unsigned i) const { return Children[i]; } + + void resetChild(unsigned i, Matcher *N) { + delete Children[i]; + Children[i] = N; + } - static inline bool classof(const MatcherNode *N) { - return N->getKind() == EmitNode; + Matcher *takeChild(unsigned i) { + Matcher *Res = Children[i]; + Children[i] = 0; + return Res; + } + + void setNumChildren(unsigned NC) { + if (NC < Children.size()) { + // delete any children we're about to lose pointers to. + for (unsigned i = NC, e = Children.size(); i != e; ++i) + delete Children[i]; + } + Children.resize(NC); } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + static inline bool classof(const Matcher *N) { + return N->getKind() == Scope; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { return false; } + virtual unsigned getHashImpl() const { return 12312; } }; -/// MatcherNodeWithChild - Every node accept the final accept state has a child -/// that is executed after the node runs. This class captures this commonality. -class MatcherNodeWithChild : public MatcherNode { - OwningPtr<MatcherNode> Child; +/// RecordMatcher - Save the current node in the operand list. +class RecordMatcher : public Matcher { + /// WhatFor - This is a string indicating why we're recording this. This + /// should only be used for comment generation not anything semantic. + std::string WhatFor; + + /// ResultNo - The slot number in the RecordedNodes vector that this will be, + /// just printed as a comment. + unsigned ResultNo; public: - MatcherNodeWithChild(KindTy K) : MatcherNode(K) {} + RecordMatcher(const std::string &whatfor, unsigned resultNo) + : Matcher(RecordNode), WhatFor(whatfor), ResultNo(resultNo) {} - MatcherNode *getChild() { return Child.get(); } - const MatcherNode *getChild() const { return Child.get(); } - void setChild(MatcherNode *C) { Child.reset(C); } + const std::string &getWhatFor() const { return WhatFor; } + unsigned getResultNo() const { return ResultNo; } - static inline bool classof(const MatcherNode *N) { - return N->getKind() != EmitNode; + static inline bool classof(const Matcher *N) { + return N->getKind() == RecordNode; } -protected: - void printChild(raw_ostream &OS, unsigned indent) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { return true; } + virtual unsigned getHashImpl() const { return 0; } }; - -/// PushMatcherNode - This pushes a failure scope on the stack and evaluates -/// 'child'. If 'child' fails to match, it pops its scope and attempts to -/// match 'Failure'. -class PushMatcherNode : public MatcherNodeWithChild { - OwningPtr<MatcherNode> Failure; + +/// RecordChildMatcher - Save a numbered child of the current node, or fail +/// the match if it doesn't exist. This is logically equivalent to: +/// MoveChild N + RecordNode + MoveParent. +class RecordChildMatcher : public Matcher { + unsigned ChildNo; + + /// WhatFor - This is a string indicating why we're recording this. This + /// should only be used for comment generation not anything semantic. + std::string WhatFor; + + /// ResultNo - The slot number in the RecordedNodes vector that this will be, + /// just printed as a comment. + unsigned ResultNo; public: - PushMatcherNode(MatcherNode *child = 0, MatcherNode *failure = 0) - : MatcherNodeWithChild(Push), Failure(failure) { - setChild(child); + RecordChildMatcher(unsigned childno, const std::string &whatfor, + unsigned resultNo) + : Matcher(RecordChild), ChildNo(childno), WhatFor(whatfor), + ResultNo(resultNo) {} + + unsigned getChildNo() const { return ChildNo; } + const std::string &getWhatFor() const { return WhatFor; } + unsigned getResultNo() const { return ResultNo; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == RecordChild; } - MatcherNode *getFailure() { return Failure.get(); } - const MatcherNode *getFailure() const { return Failure.get(); } - void setFailure(MatcherNode *N) { Failure.reset(N); } + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } - static inline bool classof(const MatcherNode *N) { - return N->getKind() == Push; +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<RecordChildMatcher>(M)->getChildNo() == getChildNo(); + } + virtual unsigned getHashImpl() const { return getChildNo(); } +}; + +/// RecordMemRefMatcher - Save the current node's memref. +class RecordMemRefMatcher : public Matcher { +public: + RecordMemRefMatcher() : Matcher(RecordMemRef) {} + + static inline bool classof(const Matcher *N) { + return N->getKind() == RecordMemRef; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { return true; } + virtual unsigned getHashImpl() const { return 0; } }; -/// RecordMatcherNode - Save the current node in the operand list. -class RecordMatcherNode : public MatcherNodeWithChild { + +/// CaptureFlagInputMatcher - If the current record has a flag input, record +/// it so that it is used as an input to the generated code. +class CaptureFlagInputMatcher : public Matcher { public: - RecordMatcherNode() : MatcherNodeWithChild(Record) {} + CaptureFlagInputMatcher() : Matcher(CaptureFlagInput) {} - static inline bool classof(const MatcherNode *N) { - return N->getKind() == Record; + static inline bool classof(const Matcher *N) { + return N->getKind() == CaptureFlagInput; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { return true; } + virtual unsigned getHashImpl() const { return 0; } }; -/// MoveChildMatcherNode - This tells the interpreter to move into the +/// MoveChildMatcher - This tells the interpreter to move into the /// specified child node. -class MoveChildMatcherNode : public MatcherNodeWithChild { +class MoveChildMatcher : public Matcher { unsigned ChildNo; public: - MoveChildMatcherNode(unsigned childNo) - : MatcherNodeWithChild(MoveChild), ChildNo(childNo) {} + MoveChildMatcher(unsigned childNo) : Matcher(MoveChild), ChildNo(childNo) {} unsigned getChildNo() const { return ChildNo; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == MoveChild; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<MoveChildMatcher>(M)->getChildNo() == getChildNo(); + } + virtual unsigned getHashImpl() const { return getChildNo(); } }; -/// MoveParentMatcherNode - This tells the interpreter to move to the parent +/// MoveParentMatcher - This tells the interpreter to move to the parent /// of the current node. -class MoveParentMatcherNode : public MatcherNodeWithChild { +class MoveParentMatcher : public Matcher { public: - MoveParentMatcherNode() - : MatcherNodeWithChild(MoveParent) {} + MoveParentMatcher() : Matcher(MoveParent) {} - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == MoveParent; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { return true; } + virtual unsigned getHashImpl() const { return 0; } }; -/// CheckSameMatcherNode - This checks to see if this node is exactly the same +/// CheckSameMatcher - This checks to see if this node is exactly the same /// node as the specified match that was recorded with 'Record'. This is used /// when patterns have the same name in them, like '(mul GPR:$in, GPR:$in)'. -class CheckSameMatcherNode : public MatcherNodeWithChild { +class CheckSameMatcher : public Matcher { unsigned MatchNumber; public: - CheckSameMatcherNode(unsigned matchnumber) - : MatcherNodeWithChild(CheckSame), MatchNumber(matchnumber) {} + CheckSameMatcher(unsigned matchnumber) + : Matcher(CheckSame), MatchNumber(matchnumber) {} unsigned getMatchNumber() const { return MatchNumber; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckSame; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckSameMatcher>(M)->getMatchNumber() == getMatchNumber(); + } + virtual unsigned getHashImpl() const { return getMatchNumber(); } }; -/// CheckPatternPredicateMatcherNode - This checks the target-specific predicate +/// CheckPatternPredicateMatcher - This checks the target-specific predicate /// to see if the entire pattern is capable of matching. This predicate does /// not take a node as input. This is used for subtarget feature checks etc. -class CheckPatternPredicateMatcherNode : public MatcherNodeWithChild { +class CheckPatternPredicateMatcher : public Matcher { std::string Predicate; public: - CheckPatternPredicateMatcherNode(StringRef predicate) - : MatcherNodeWithChild(CheckPatternPredicate), Predicate(predicate) {} + CheckPatternPredicateMatcher(StringRef predicate) + : Matcher(CheckPatternPredicate), Predicate(predicate) {} StringRef getPredicate() const { return Predicate; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckPatternPredicate; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckPatternPredicateMatcher>(M)->getPredicate() == Predicate; + } + virtual unsigned getHashImpl() const; }; -/// CheckPredicateMatcherNode - This checks the target-specific predicate to +/// CheckPredicateMatcher - This checks the target-specific predicate to /// see if the node is acceptable. -class CheckPredicateMatcherNode : public MatcherNodeWithChild { +class CheckPredicateMatcher : public Matcher { StringRef PredName; public: - CheckPredicateMatcherNode(StringRef predname) - : MatcherNodeWithChild(CheckPredicate), PredName(predname) {} + CheckPredicateMatcher(StringRef predname) + : Matcher(CheckPredicate), PredName(predname) {} StringRef getPredicateName() const { return PredName; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckPredicate; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + // TODO: Ok? + //virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckPredicateMatcher>(M)->PredName == PredName; + } + virtual unsigned getHashImpl() const; }; -/// CheckOpcodeMatcherNode - This checks to see if the current node has the +/// CheckOpcodeMatcher - This checks to see if the current node has the /// specified opcode, if not it fails to match. -class CheckOpcodeMatcherNode : public MatcherNodeWithChild { - StringRef OpcodeName; +class CheckOpcodeMatcher : public Matcher { + const SDNodeInfo &Opcode; public: - CheckOpcodeMatcherNode(StringRef opcodename) - : MatcherNodeWithChild(CheckOpcode), OpcodeName(opcodename) {} + CheckOpcodeMatcher(const SDNodeInfo &opcode) + : Matcher(CheckOpcode), Opcode(opcode) {} - StringRef getOpcodeName() const { return OpcodeName; } + const SDNodeInfo &getOpcode() const { return Opcode; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckOpcode; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const; + virtual unsigned getHashImpl() const; + virtual bool isContradictoryImpl(const Matcher *M) const; }; + +/// SwitchOpcodeMatcher - Switch based on the current node's opcode, dispatching +/// to one matcher per opcode. If the opcode doesn't match any of the cases, +/// then the match fails. This is semantically equivalent to a Scope node where +/// every child does a CheckOpcode, but is much faster. +class SwitchOpcodeMatcher : public Matcher { + SmallVector<std::pair<const SDNodeInfo*, Matcher*>, 8> Cases; +public: + SwitchOpcodeMatcher(const std::pair<const SDNodeInfo*, Matcher*> *cases, + unsigned numcases) + : Matcher(SwitchOpcode), Cases(cases, cases+numcases) {} + + static inline bool classof(const Matcher *N) { + return N->getKind() == SwitchOpcode; + } + + unsigned getNumCases() const { return Cases.size(); } -/// CheckTypeMatcherNode - This checks to see if the current node has the + const SDNodeInfo &getCaseOpcode(unsigned i) const { return *Cases[i].first; } + Matcher *getCaseMatcher(unsigned i) { return Cases[i].second; } + const Matcher *getCaseMatcher(unsigned i) const { return Cases[i].second; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { return false; } + virtual unsigned getHashImpl() const { return 4123; } +}; + +/// CheckTypeMatcher - This checks to see if the current node has the /// specified type, if not it fails to match. -class CheckTypeMatcherNode : public MatcherNodeWithChild { +class CheckTypeMatcher : public Matcher { MVT::SimpleValueType Type; public: - CheckTypeMatcherNode(MVT::SimpleValueType type) - : MatcherNodeWithChild(CheckType), Type(type) {} + CheckTypeMatcher(MVT::SimpleValueType type) + : Matcher(CheckType), Type(type) {} MVT::SimpleValueType getType() const { return Type; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckType; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckTypeMatcher>(M)->Type == Type; + } + virtual unsigned getHashImpl() const { return Type; } + virtual bool isContradictoryImpl(const Matcher *M) const; +}; + +/// SwitchTypeMatcher - Switch based on the current node's type, dispatching +/// to one matcher per case. If the type doesn't match any of the cases, +/// then the match fails. This is semantically equivalent to a Scope node where +/// every child does a CheckType, but is much faster. +class SwitchTypeMatcher : public Matcher { + SmallVector<std::pair<MVT::SimpleValueType, Matcher*>, 8> Cases; +public: + SwitchTypeMatcher(const std::pair<MVT::SimpleValueType, Matcher*> *cases, + unsigned numcases) + : Matcher(SwitchType), Cases(cases, cases+numcases) {} + + static inline bool classof(const Matcher *N) { + return N->getKind() == SwitchType; + } + + unsigned getNumCases() const { return Cases.size(); } + + MVT::SimpleValueType getCaseType(unsigned i) const { return Cases[i].first; } + Matcher *getCaseMatcher(unsigned i) { return Cases[i].second; } + const Matcher *getCaseMatcher(unsigned i) const { return Cases[i].second; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { return false; } + virtual unsigned getHashImpl() const { return 4123; } }; + + +/// CheckChildTypeMatcher - This checks to see if a child node has the +/// specified type, if not it fails to match. +class CheckChildTypeMatcher : public Matcher { + unsigned ChildNo; + MVT::SimpleValueType Type; +public: + CheckChildTypeMatcher(unsigned childno, MVT::SimpleValueType type) + : Matcher(CheckChildType), ChildNo(childno), Type(type) {} + + unsigned getChildNo() const { return ChildNo; } + MVT::SimpleValueType getType() const { return Type; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == CheckChildType; + } + + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } -/// CheckIntegerMatcherNode - This checks to see if the current node is a +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckChildTypeMatcher>(M)->ChildNo == ChildNo && + cast<CheckChildTypeMatcher>(M)->Type == Type; + } + virtual unsigned getHashImpl() const { return (Type << 3) | ChildNo; } + virtual bool isContradictoryImpl(const Matcher *M) const; +}; + + +/// CheckIntegerMatcher - This checks to see if the current node is a /// ConstantSDNode with the specified integer value, if not it fails to match. -class CheckIntegerMatcherNode : public MatcherNodeWithChild { +class CheckIntegerMatcher : public Matcher { int64_t Value; public: - CheckIntegerMatcherNode(int64_t value) - : MatcherNodeWithChild(CheckInteger), Value(value) {} + CheckIntegerMatcher(int64_t value) + : Matcher(CheckInteger), Value(value) {} int64_t getValue() const { return Value; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckInteger; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckIntegerMatcher>(M)->Value == Value; + } + virtual unsigned getHashImpl() const { return Value; } + virtual bool isContradictoryImpl(const Matcher *M) const; }; -/// CheckCondCodeMatcherNode - This checks to see if the current node is a +/// CheckCondCodeMatcher - This checks to see if the current node is a /// CondCodeSDNode with the specified condition, if not it fails to match. -class CheckCondCodeMatcherNode : public MatcherNodeWithChild { +class CheckCondCodeMatcher : public Matcher { StringRef CondCodeName; public: - CheckCondCodeMatcherNode(StringRef condcodename) - : MatcherNodeWithChild(CheckCondCode), CondCodeName(condcodename) {} + CheckCondCodeMatcher(StringRef condcodename) + : Matcher(CheckCondCode), CondCodeName(condcodename) {} StringRef getCondCodeName() const { return CondCodeName; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckCondCode; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckCondCodeMatcher>(M)->CondCodeName == CondCodeName; + } + virtual unsigned getHashImpl() const; }; -/// CheckValueTypeMatcherNode - This checks to see if the current node is a +/// CheckValueTypeMatcher - This checks to see if the current node is a /// VTSDNode with the specified type, if not it fails to match. -class CheckValueTypeMatcherNode : public MatcherNodeWithChild { +class CheckValueTypeMatcher : public Matcher { StringRef TypeName; public: - CheckValueTypeMatcherNode(StringRef type_name) - : MatcherNodeWithChild(CheckValueType), TypeName(type_name) {} + CheckValueTypeMatcher(StringRef type_name) + : Matcher(CheckValueType), TypeName(type_name) {} StringRef getTypeName() const { return TypeName; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckValueType; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckValueTypeMatcher>(M)->TypeName == TypeName; + } + virtual unsigned getHashImpl() const; }; -/// CheckComplexPatMatcherNode - This node runs the specified ComplexPattern on +/// CheckComplexPatMatcher - This node runs the specified ComplexPattern on /// the current node. -class CheckComplexPatMatcherNode : public MatcherNodeWithChild { +class CheckComplexPatMatcher : public Matcher { const ComplexPattern &Pattern; public: - CheckComplexPatMatcherNode(const ComplexPattern &pattern) - : MatcherNodeWithChild(CheckComplexPat), Pattern(pattern) {} + CheckComplexPatMatcher(const ComplexPattern &pattern) + : Matcher(CheckComplexPat), Pattern(pattern) {} - static inline bool classof(const MatcherNode *N) { + const ComplexPattern &getPattern() const { return Pattern; } + + static inline bool classof(const Matcher *N) { return N->getKind() == CheckComplexPat; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + // Not safe to move a pattern predicate past a complex pattern. + virtual bool isSafeToReorderWithPatternPredicate() const { return false; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return &cast<CheckComplexPatMatcher>(M)->Pattern == &Pattern; + } + virtual unsigned getHashImpl() const { + return (unsigned)(intptr_t)&Pattern; + } }; -/// CheckAndImmMatcherNode - This checks to see if the current node is an 'and' +/// CheckAndImmMatcher - This checks to see if the current node is an 'and' /// with something equivalent to the specified immediate. -class CheckAndImmMatcherNode : public MatcherNodeWithChild { +class CheckAndImmMatcher : public Matcher { int64_t Value; public: - CheckAndImmMatcherNode(int64_t value) - : MatcherNodeWithChild(CheckAndImm), Value(value) {} + CheckAndImmMatcher(int64_t value) + : Matcher(CheckAndImm), Value(value) {} int64_t getValue() const { return Value; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckAndImm; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckAndImmMatcher>(M)->Value == Value; + } + virtual unsigned getHashImpl() const { return Value; } }; -/// CheckOrImmMatcherNode - This checks to see if the current node is an 'and' +/// CheckOrImmMatcher - This checks to see if the current node is an 'and' /// with something equivalent to the specified immediate. -class CheckOrImmMatcherNode : public MatcherNodeWithChild { +class CheckOrImmMatcher : public Matcher { int64_t Value; public: - CheckOrImmMatcherNode(int64_t value) - : MatcherNodeWithChild(CheckOrImm), Value(value) {} + CheckOrImmMatcher(int64_t value) + : Matcher(CheckOrImm), Value(value) {} int64_t getValue() const { return Value; } - static inline bool classof(const MatcherNode *N) { + static inline bool classof(const Matcher *N) { return N->getKind() == CheckOrImm; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CheckOrImmMatcher>(M)->Value == Value; + } + virtual unsigned getHashImpl() const { return Value; } }; -/// CheckProfitableToFoldMatcherNode - This checks to see if the current node is -/// worthwhile to try to fold into a large pattern. -class CheckProfitableToFoldMatcherNode : public MatcherNodeWithChild { +/// CheckFoldableChainNodeMatcher - This checks to see if the current node +/// (which defines a chain operand) is safe to fold into a larger pattern. +class CheckFoldableChainNodeMatcher : public Matcher { public: - CheckProfitableToFoldMatcherNode() - : MatcherNodeWithChild(CheckProfitableToFold) {} + CheckFoldableChainNodeMatcher() + : Matcher(CheckFoldableChainNode) {} - static inline bool classof(const MatcherNode *N) { - return N->getKind() == CheckProfitableToFold; + static inline bool classof(const Matcher *N) { + return N->getKind() == CheckFoldableChainNode; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; + virtual bool isSafeToReorderWithPatternPredicate() const { return true; } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { return true; } + virtual unsigned getHashImpl() const { return 0; } }; -/// CheckLegalToFoldMatcherNode - This checks to see if the current node is -/// legal to try to fold into a large pattern. -class CheckLegalToFoldMatcherNode : public MatcherNodeWithChild { +/// EmitIntegerMatcher - This creates a new TargetConstant. +class EmitIntegerMatcher : public Matcher { + int64_t Val; + MVT::SimpleValueType VT; public: - CheckLegalToFoldMatcherNode() - : MatcherNodeWithChild(CheckLegalToFold) {} + EmitIntegerMatcher(int64_t val, MVT::SimpleValueType vt) + : Matcher(EmitInteger), Val(val), VT(vt) {} - static inline bool classof(const MatcherNode *N) { - return N->getKind() == CheckLegalToFold; + int64_t getValue() const { return Val; } + MVT::SimpleValueType getVT() const { return VT; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == EmitInteger; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<EmitIntegerMatcher>(M)->Val == Val && + cast<EmitIntegerMatcher>(M)->VT == VT; + } + virtual unsigned getHashImpl() const { return (Val << 4) | VT; } +}; + +/// EmitStringIntegerMatcher - A target constant whose value is represented +/// by a string. +class EmitStringIntegerMatcher : public Matcher { + std::string Val; + MVT::SimpleValueType VT; +public: + EmitStringIntegerMatcher(const std::string &val, MVT::SimpleValueType vt) + : Matcher(EmitStringInteger), Val(val), VT(vt) {} + + const std::string &getValue() const { return Val; } + MVT::SimpleValueType getVT() const { return VT; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == EmitStringInteger; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<EmitStringIntegerMatcher>(M)->Val == Val && + cast<EmitStringIntegerMatcher>(M)->VT == VT; + } + virtual unsigned getHashImpl() const; +}; + +/// EmitRegisterMatcher - This creates a new TargetConstant. +class EmitRegisterMatcher : public Matcher { + /// Reg - The def for the register that we're emitting. If this is null, then + /// this is a reference to zero_reg. + Record *Reg; + MVT::SimpleValueType VT; +public: + EmitRegisterMatcher(Record *reg, MVT::SimpleValueType vt) + : Matcher(EmitRegister), Reg(reg), VT(vt) {} + + Record *getReg() const { return Reg; } + MVT::SimpleValueType getVT() const { return VT; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == EmitRegister; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<EmitRegisterMatcher>(M)->Reg == Reg && + cast<EmitRegisterMatcher>(M)->VT == VT; + } + virtual unsigned getHashImpl() const { + return ((unsigned)(intptr_t)Reg) << 4 | VT; + } +}; + +/// EmitConvertToTargetMatcher - Emit an operation that reads a specified +/// recorded node and converts it from being a ISD::Constant to +/// ISD::TargetConstant, likewise for ConstantFP. +class EmitConvertToTargetMatcher : public Matcher { + unsigned Slot; +public: + EmitConvertToTargetMatcher(unsigned slot) + : Matcher(EmitConvertToTarget), Slot(slot) {} + + unsigned getSlot() const { return Slot; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == EmitConvertToTarget; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<EmitConvertToTargetMatcher>(M)->Slot == Slot; + } + virtual unsigned getHashImpl() const { return Slot; } +}; + +/// EmitMergeInputChainsMatcher - Emit a node that merges a list of input +/// chains together with a token factor. The list of nodes are the nodes in the +/// matched pattern that have chain input/outputs. This node adds all input +/// chains of these nodes if they are not themselves a node in the pattern. +class EmitMergeInputChainsMatcher : public Matcher { + SmallVector<unsigned, 3> ChainNodes; +public: + EmitMergeInputChainsMatcher(const unsigned *nodes, unsigned NumNodes) + : Matcher(EmitMergeInputChains), ChainNodes(nodes, nodes+NumNodes) {} + + unsigned getNumNodes() const { return ChainNodes.size(); } + + unsigned getNode(unsigned i) const { + assert(i < ChainNodes.size()); + return ChainNodes[i]; + } + + static inline bool classof(const Matcher *N) { + return N->getKind() == EmitMergeInputChains; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<EmitMergeInputChainsMatcher>(M)->ChainNodes == ChainNodes; + } + virtual unsigned getHashImpl() const; +}; + +/// EmitCopyToRegMatcher - Emit a CopyToReg node from a value to a physreg, +/// pushing the chain and flag results. +/// +class EmitCopyToRegMatcher : public Matcher { + unsigned SrcSlot; // Value to copy into the physreg. + Record *DestPhysReg; +public: + EmitCopyToRegMatcher(unsigned srcSlot, Record *destPhysReg) + : Matcher(EmitCopyToReg), SrcSlot(srcSlot), DestPhysReg(destPhysReg) {} + + unsigned getSrcSlot() const { return SrcSlot; } + Record *getDestPhysReg() const { return DestPhysReg; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == EmitCopyToReg; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<EmitCopyToRegMatcher>(M)->SrcSlot == SrcSlot && + cast<EmitCopyToRegMatcher>(M)->DestPhysReg == DestPhysReg; + } + virtual unsigned getHashImpl() const { + return SrcSlot ^ ((unsigned)(intptr_t)DestPhysReg << 4); + } +}; + + + +/// EmitNodeXFormMatcher - Emit an operation that runs an SDNodeXForm on a +/// recorded node and records the result. +class EmitNodeXFormMatcher : public Matcher { + unsigned Slot; + Record *NodeXForm; +public: + EmitNodeXFormMatcher(unsigned slot, Record *nodeXForm) + : Matcher(EmitNodeXForm), Slot(slot), NodeXForm(nodeXForm) {} + + unsigned getSlot() const { return Slot; } + Record *getNodeXForm() const { return NodeXForm; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == EmitNodeXForm; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<EmitNodeXFormMatcher>(M)->Slot == Slot && + cast<EmitNodeXFormMatcher>(M)->NodeXForm == NodeXForm; + } + virtual unsigned getHashImpl() const { + return Slot ^ ((unsigned)(intptr_t)NodeXForm << 4); + } +}; + +/// EmitNodeMatcherCommon - Common class shared between EmitNode and +/// MorphNodeTo. +class EmitNodeMatcherCommon : public Matcher { + std::string OpcodeName; + const SmallVector<MVT::SimpleValueType, 3> VTs; + const SmallVector<unsigned, 6> Operands; + bool HasChain, HasInFlag, HasOutFlag, HasMemRefs; + + /// NumFixedArityOperands - If this is a fixed arity node, this is set to -1. + /// If this is a varidic node, this is set to the number of fixed arity + /// operands in the root of the pattern. The rest are appended to this node. + int NumFixedArityOperands; +public: + EmitNodeMatcherCommon(const std::string &opcodeName, + const MVT::SimpleValueType *vts, unsigned numvts, + const unsigned *operands, unsigned numops, + bool hasChain, bool hasInFlag, bool hasOutFlag, + bool hasmemrefs, + int numfixedarityoperands, bool isMorphNodeTo) + : Matcher(isMorphNodeTo ? MorphNodeTo : EmitNode), OpcodeName(opcodeName), + VTs(vts, vts+numvts), Operands(operands, operands+numops), + HasChain(hasChain), HasInFlag(hasInFlag), HasOutFlag(hasOutFlag), + HasMemRefs(hasmemrefs), NumFixedArityOperands(numfixedarityoperands) {} + + const std::string &getOpcodeName() const { return OpcodeName; } + + unsigned getNumVTs() const { return VTs.size(); } + MVT::SimpleValueType getVT(unsigned i) const { + assert(i < VTs.size()); + return VTs[i]; + } + + unsigned getNumOperands() const { return Operands.size(); } + unsigned getOperand(unsigned i) const { + assert(i < Operands.size()); + return Operands[i]; + } + + const SmallVectorImpl<MVT::SimpleValueType> &getVTList() const { return VTs; } + const SmallVectorImpl<unsigned> &getOperandList() const { return Operands; } + + + bool hasChain() const { return HasChain; } + bool hasInFlag() const { return HasInFlag; } + bool hasOutFlag() const { return HasOutFlag; } + bool hasMemRefs() const { return HasMemRefs; } + int getNumFixedArityOperands() const { return NumFixedArityOperands; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == EmitNode || N->getKind() == MorphNodeTo; } - virtual void print(raw_ostream &OS, unsigned indent = 0) const; +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const; + virtual unsigned getHashImpl() const; +}; + +/// EmitNodeMatcher - This signals a successful match and generates a node. +class EmitNodeMatcher : public EmitNodeMatcherCommon { + unsigned FirstResultSlot; +public: + EmitNodeMatcher(const std::string &opcodeName, + const MVT::SimpleValueType *vts, unsigned numvts, + const unsigned *operands, unsigned numops, + bool hasChain, bool hasInFlag, bool hasOutFlag, + bool hasmemrefs, + int numfixedarityoperands, unsigned firstresultslot) + : EmitNodeMatcherCommon(opcodeName, vts, numvts, operands, numops, hasChain, + hasInFlag, hasOutFlag, hasmemrefs, + numfixedarityoperands, false), + FirstResultSlot(firstresultslot) {} + + unsigned getFirstResultSlot() const { return FirstResultSlot; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == EmitNode; + } + +}; + +class MorphNodeToMatcher : public EmitNodeMatcherCommon { + const PatternToMatch &Pattern; +public: + MorphNodeToMatcher(const std::string &opcodeName, + const MVT::SimpleValueType *vts, unsigned numvts, + const unsigned *operands, unsigned numops, + bool hasChain, bool hasInFlag, bool hasOutFlag, + bool hasmemrefs, + int numfixedarityoperands, const PatternToMatch &pattern) + : EmitNodeMatcherCommon(opcodeName, vts, numvts, operands, numops, hasChain, + hasInFlag, hasOutFlag, hasmemrefs, + numfixedarityoperands, true), + Pattern(pattern) { + } + + const PatternToMatch &getPattern() const { return Pattern; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == MorphNodeTo; + } +}; + +/// MarkFlagResultsMatcher - This node indicates which non-root nodes in the +/// pattern produce flags. This allows CompleteMatchMatcher to update them +/// with the output flag of the resultant code. +class MarkFlagResultsMatcher : public Matcher { + SmallVector<unsigned, 3> FlagResultNodes; +public: + MarkFlagResultsMatcher(const unsigned *nodes, unsigned NumNodes) + : Matcher(MarkFlagResults), FlagResultNodes(nodes, nodes+NumNodes) {} + + unsigned getNumNodes() const { return FlagResultNodes.size(); } + + unsigned getNode(unsigned i) const { + assert(i < FlagResultNodes.size()); + return FlagResultNodes[i]; + } + + static inline bool classof(const Matcher *N) { + return N->getKind() == MarkFlagResults; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<MarkFlagResultsMatcher>(M)->FlagResultNodes == FlagResultNodes; + } + virtual unsigned getHashImpl() const; +}; + +/// CompleteMatchMatcher - Complete a match by replacing the results of the +/// pattern with the newly generated nodes. This also prints a comment +/// indicating the source and dest patterns. +class CompleteMatchMatcher : public Matcher { + SmallVector<unsigned, 2> Results; + const PatternToMatch &Pattern; +public: + CompleteMatchMatcher(const unsigned *results, unsigned numresults, + const PatternToMatch &pattern) + : Matcher(CompleteMatch), Results(results, results+numresults), + Pattern(pattern) {} + + unsigned getNumResults() const { return Results.size(); } + unsigned getResult(unsigned R) const { return Results[R]; } + const PatternToMatch &getPattern() const { return Pattern; } + + static inline bool classof(const Matcher *N) { + return N->getKind() == CompleteMatch; + } + +private: + virtual void printImpl(raw_ostream &OS, unsigned indent) const; + virtual bool isEqualImpl(const Matcher *M) const { + return cast<CompleteMatchMatcher>(M)->Results == Results && + &cast<CompleteMatchMatcher>(M)->Pattern == &Pattern; + } + virtual unsigned getHashImpl() const; }; + } // end namespace llvm #endif diff --git a/utils/TableGen/DAGISelMatcherEmitter.cpp b/utils/TableGen/DAGISelMatcherEmitter.cpp index c0ad169..1f04050 100644 --- a/utils/TableGen/DAGISelMatcherEmitter.cpp +++ b/utils/TableGen/DAGISelMatcherEmitter.cpp @@ -13,71 +13,50 @@ #include "DAGISelMatcher.h" #include "CodeGenDAGPatterns.h" +#include "Record.h" +#include "llvm/ADT/DenseMap.h" #include "llvm/ADT/SmallString.h" #include "llvm/ADT/StringMap.h" +#include "llvm/Support/CommandLine.h" #include "llvm/Support/FormattedStream.h" using namespace llvm; -namespace { enum { - CommentIndent = 25 + CommentIndent = 30 }; -} -/// ClassifyInt - Classify an integer by size, return '1','2','4','8' if this -/// fits in 1, 2, 4, or 8 sign extended bytes. -static char ClassifyInt(int64_t Val) { - if (Val == int8_t(Val)) return '1'; - if (Val == int16_t(Val)) return '2'; - if (Val == int32_t(Val)) return '4'; - return '8'; -} - -/// EmitInt - Emit the specified integer, returning the number of bytes emitted. -static unsigned EmitInt(int64_t Val, formatted_raw_ostream &OS) { - unsigned BytesEmitted = 1; - OS << (int)(unsigned char)Val << ", "; - if (Val == int8_t(Val)) { - OS << "\n"; - return BytesEmitted; - } - - OS << (int)(unsigned char)(Val >> 8) << ", "; - ++BytesEmitted; - - if (Val != int16_t(Val)) { - OS << (int)(unsigned char)(Val >> 16) << ',' - << (int)(unsigned char)(Val >> 24) << ','; - BytesEmitted += 2; - - if (Val != int32_t(Val)) { - OS << (int)(unsigned char)(Val >> 32) << ',' - << (int)(unsigned char)(Val >> 40) << ',' - << (int)(unsigned char)(Val >> 48) << ',' - << (int)(unsigned char)(Val >> 56) << ','; - BytesEmitted += 4; - } - } - - OS.PadToColumn(CommentIndent) << "// " << Val << '\n'; - return BytesEmitted; -} +// To reduce generated source code size. +static cl::opt<bool> +OmitComments("omit-comments", cl::desc("Do not generate comments"), + cl::init(false)); namespace { class MatcherTableEmitter { - formatted_raw_ostream &OS; - StringMap<unsigned> NodePredicateMap, PatternPredicateMap; std::vector<std::string> NodePredicates, PatternPredicates; - + + DenseMap<const ComplexPattern*, unsigned> ComplexPatternMap; + std::vector<const ComplexPattern*> ComplexPatterns; + + + DenseMap<Record*, unsigned> NodeXFormMap; + std::vector<Record*> NodeXForms; + + // Per opcode frequence count. + std::vector<unsigned> Histogram; public: - MatcherTableEmitter(formatted_raw_ostream &os) : OS(os) {} + MatcherTableEmitter() {} - unsigned EmitMatcherAndChildren(const MatcherNode *N, unsigned Indent); + unsigned EmitMatcherList(const Matcher *N, unsigned Indent, + unsigned StartIdx, formatted_raw_ostream &OS); - void EmitPredicateFunctions(); + void EmitPredicateFunctions(const CodeGenDAGPatterns &CGP, + formatted_raw_ostream &OS); + + void EmitHistogram(formatted_raw_ostream &OS); private: - unsigned EmitMatcher(const MatcherNode *N, unsigned Indent); + unsigned EmitMatcher(const Matcher *N, unsigned Indent, unsigned CurrentIdx, + formatted_raw_ostream &OS); unsigned getNodePredicate(StringRef PredName) { unsigned &Entry = NodePredicateMap[PredName]; @@ -95,184 +74,679 @@ private: } return Entry-1; } + + unsigned getComplexPat(const ComplexPattern &P) { + unsigned &Entry = ComplexPatternMap[&P]; + if (Entry == 0) { + ComplexPatterns.push_back(&P); + Entry = ComplexPatterns.size(); + } + return Entry-1; + } + + unsigned getNodeXFormID(Record *Rec) { + unsigned &Entry = NodeXFormMap[Rec]; + if (Entry == 0) { + NodeXForms.push_back(Rec); + Entry = NodeXForms.size(); + } + return Entry-1; + } + }; } // end anonymous namespace. +static unsigned GetVBRSize(unsigned Val) { + if (Val <= 127) return 1; + + unsigned NumBytes = 0; + while (Val >= 128) { + Val >>= 7; + ++NumBytes; + } + return NumBytes+1; +} + +/// EmitVBRValue - Emit the specified value as a VBR, returning the number of +/// bytes emitted. +static uint64_t EmitVBRValue(uint64_t Val, raw_ostream &OS) { + if (Val <= 127) { + OS << Val << ", "; + return 1; + } + + uint64_t InVal = Val; + unsigned NumBytes = 0; + while (Val >= 128) { + OS << (Val&127) << "|128,"; + Val >>= 7; + ++NumBytes; + } + OS << Val; + if (!OmitComments) + OS << "/*" << InVal << "*/"; + OS << ", "; + return NumBytes+1; +} + /// EmitMatcherOpcodes - Emit bytes for the specified matcher and return /// the number of bytes emitted. unsigned MatcherTableEmitter:: -EmitMatcher(const MatcherNode *N, unsigned Indent) { +EmitMatcher(const Matcher *N, unsigned Indent, unsigned CurrentIdx, + formatted_raw_ostream &OS) { OS.PadToColumn(Indent*2); switch (N->getKind()) { - case MatcherNode::Push: assert(0 && "Should be handled by caller"); - case MatcherNode::EmitNode: - OS << "OPC_Emit, /*XXX*/"; - OS.PadToColumn(CommentIndent) << "// Src: " - << *cast<EmitNodeMatcherNode>(N)->getPattern().getSrcPattern() << '\n'; - OS.PadToColumn(CommentIndent) << "// Dst: " - << *cast<EmitNodeMatcherNode>(N)->getPattern().getDstPattern() << '\n'; + case Matcher::Scope: { + const ScopeMatcher *SM = cast<ScopeMatcher>(N); + assert(SM->getNext() == 0 && "Shouldn't have next after scope"); + + unsigned StartIdx = CurrentIdx; + + // Emit all of the children. + for (unsigned i = 0, e = SM->getNumChildren(); i != e; ++i) { + if (i == 0) { + OS << "OPC_Scope, "; + ++CurrentIdx; + } else { + if (!OmitComments) { + OS << "/*" << CurrentIdx << "*/"; + OS.PadToColumn(Indent*2) << "/*Scope*/ "; + } else + OS.PadToColumn(Indent*2); + } + + // We need to encode the child and the offset of the failure code before + // emitting either of them. Handle this by buffering the output into a + // string while we get the size. Unfortunately, the offset of the + // children depends on the VBR size of the child, so for large children we + // have to iterate a bit. + SmallString<128> TmpBuf; + unsigned ChildSize = 0; + unsigned VBRSize = 0; + do { + VBRSize = GetVBRSize(ChildSize); + + TmpBuf.clear(); + raw_svector_ostream OS(TmpBuf); + formatted_raw_ostream FOS(OS); + ChildSize = EmitMatcherList(SM->getChild(i), Indent+1, + CurrentIdx+VBRSize, FOS); + } while (GetVBRSize(ChildSize) != VBRSize); + + assert(ChildSize != 0 && "Should not have a zero-sized child!"); + + CurrentIdx += EmitVBRValue(ChildSize, OS); + if (!OmitComments) { + OS << "/*->" << CurrentIdx+ChildSize << "*/"; + + if (i == 0) + OS.PadToColumn(CommentIndent) << "// " << SM->getNumChildren() + << " children in Scope"; + } + + OS << '\n' << TmpBuf.str(); + CurrentIdx += ChildSize; + } + + // Emit a zero as a sentinel indicating end of 'Scope'. + if (!OmitComments) + OS << "/*" << CurrentIdx << "*/"; + OS.PadToColumn(Indent*2) << "0, "; + if (!OmitComments) + OS << "/*End of Scope*/"; + OS << '\n'; + return CurrentIdx - StartIdx + 1; + } + + case Matcher::RecordNode: + OS << "OPC_RecordNode,"; + if (!OmitComments) + OS.PadToColumn(CommentIndent) << "// #" + << cast<RecordMatcher>(N)->getResultNo() << " = " + << cast<RecordMatcher>(N)->getWhatFor(); + OS << '\n'; return 1; - case MatcherNode::Record: - OS << "OPC_Record,\n"; + + case Matcher::RecordChild: + OS << "OPC_RecordChild" << cast<RecordChildMatcher>(N)->getChildNo() + << ','; + if (!OmitComments) + OS.PadToColumn(CommentIndent) << "// #" + << cast<RecordChildMatcher>(N)->getResultNo() << " = " + << cast<RecordChildMatcher>(N)->getWhatFor(); + OS << '\n'; + return 1; + + case Matcher::RecordMemRef: + OS << "OPC_RecordMemRef,\n"; + return 1; + + case Matcher::CaptureFlagInput: + OS << "OPC_CaptureFlagInput,\n"; return 1; - case MatcherNode::MoveChild: - OS << "OPC_MoveChild, " - << cast<MoveChildMatcherNode>(N)->getChildNo() << ",\n"; + + case Matcher::MoveChild: + OS << "OPC_MoveChild, " << cast<MoveChildMatcher>(N)->getChildNo() << ",\n"; return 2; - case MatcherNode::MoveParent: + case Matcher::MoveParent: OS << "OPC_MoveParent,\n"; return 1; - case MatcherNode::CheckSame: + case Matcher::CheckSame: OS << "OPC_CheckSame, " - << cast<CheckSameMatcherNode>(N)->getMatchNumber() << ",\n"; + << cast<CheckSameMatcher>(N)->getMatchNumber() << ",\n"; return 2; - case MatcherNode::CheckPatternPredicate: { - StringRef Pred = cast<CheckPatternPredicateMatcherNode>(N)->getPredicate(); + case Matcher::CheckPatternPredicate: { + StringRef Pred = cast<CheckPatternPredicateMatcher>(N)->getPredicate(); OS << "OPC_CheckPatternPredicate, " << getPatternPredicate(Pred) << ','; - OS.PadToColumn(CommentIndent) << "// " << Pred << '\n'; + if (!OmitComments) + OS.PadToColumn(CommentIndent) << "// " << Pred; + OS << '\n'; return 2; } - case MatcherNode::CheckPredicate: { - StringRef Pred = cast<CheckPredicateMatcherNode>(N)->getPredicateName(); + case Matcher::CheckPredicate: { + StringRef Pred = cast<CheckPredicateMatcher>(N)->getPredicateName(); OS << "OPC_CheckPredicate, " << getNodePredicate(Pred) << ','; - OS.PadToColumn(CommentIndent) << "// " << Pred << '\n'; + if (!OmitComments) + OS.PadToColumn(CommentIndent) << "// " << Pred; + OS << '\n'; return 2; } - case MatcherNode::CheckOpcode: + case Matcher::CheckOpcode: OS << "OPC_CheckOpcode, " - << cast<CheckOpcodeMatcherNode>(N)->getOpcodeName() << ",\n"; + << cast<CheckOpcodeMatcher>(N)->getOpcode().getEnumName() << ",\n"; return 2; - case MatcherNode::CheckType: + case Matcher::SwitchOpcode: + case Matcher::SwitchType: { + unsigned StartIdx = CurrentIdx; + + unsigned NumCases; + if (const SwitchOpcodeMatcher *SOM = dyn_cast<SwitchOpcodeMatcher>(N)) { + OS << "OPC_SwitchOpcode "; + NumCases = SOM->getNumCases(); + } else { + OS << "OPC_SwitchType "; + NumCases = cast<SwitchTypeMatcher>(N)->getNumCases(); + } + + if (!OmitComments) + OS << "/*" << NumCases << " cases */"; + OS << ", "; + ++CurrentIdx; + + // For each case we emit the size, then the opcode, then the matcher. + for (unsigned i = 0, e = NumCases; i != e; ++i) { + const Matcher *Child; + if (const SwitchOpcodeMatcher *SOM = dyn_cast<SwitchOpcodeMatcher>(N)) + Child = SOM->getCaseMatcher(i); + else + Child = cast<SwitchTypeMatcher>(N)->getCaseMatcher(i); + + // We need to encode the opcode and the offset of the case code before + // emitting the case code. Handle this by buffering the output into a + // string while we get the size. Unfortunately, the offset of the + // children depends on the VBR size of the child, so for large children we + // have to iterate a bit. + SmallString<128> TmpBuf; + unsigned ChildSize = 0; + unsigned VBRSize = 0; + do { + VBRSize = GetVBRSize(ChildSize); + + TmpBuf.clear(); + raw_svector_ostream OS(TmpBuf); + formatted_raw_ostream FOS(OS); + ChildSize = EmitMatcherList(Child, Indent+1, CurrentIdx+VBRSize+1, FOS); + } while (GetVBRSize(ChildSize) != VBRSize); + + assert(ChildSize != 0 && "Should not have a zero-sized child!"); + + if (i != 0) { + OS.PadToColumn(Indent*2); + if (!OmitComments) + OS << (isa<SwitchOpcodeMatcher>(N) ? + "/*SwitchOpcode*/ " : "/*SwitchType*/ "); + } + + // Emit the VBR. + CurrentIdx += EmitVBRValue(ChildSize, OS); + + OS << ' '; + if (const SwitchOpcodeMatcher *SOM = dyn_cast<SwitchOpcodeMatcher>(N)) + OS << SOM->getCaseOpcode(i).getEnumName(); + else + OS << getEnumName(cast<SwitchTypeMatcher>(N)->getCaseType(i)); + OS << ','; + + if (!OmitComments) + OS << "// ->" << CurrentIdx+ChildSize+1; + OS << '\n'; + ++CurrentIdx; + OS << TmpBuf.str(); + CurrentIdx += ChildSize; + } + + // Emit the final zero to terminate the switch. + OS.PadToColumn(Indent*2) << "0, "; + if (!OmitComments) + OS << (isa<SwitchOpcodeMatcher>(N) ? + "// EndSwitchOpcode" : "// EndSwitchType"); + + OS << '\n'; + ++CurrentIdx; + return CurrentIdx-StartIdx; + } + + case Matcher::CheckType: OS << "OPC_CheckType, " - << getEnumName(cast<CheckTypeMatcherNode>(N)->getType()) << ",\n"; + << getEnumName(cast<CheckTypeMatcher>(N)->getType()) << ",\n"; return 2; - - case MatcherNode::CheckInteger: { - int64_t Val = cast<CheckIntegerMatcherNode>(N)->getValue(); - OS << "OPC_CheckInteger" << ClassifyInt(Val) << ", "; - return EmitInt(Val, OS)+1; - } - case MatcherNode::CheckCondCode: + + case Matcher::CheckChildType: + OS << "OPC_CheckChild" + << cast<CheckChildTypeMatcher>(N)->getChildNo() << "Type, " + << getEnumName(cast<CheckChildTypeMatcher>(N)->getType()) << ",\n"; + return 2; + + case Matcher::CheckInteger: { + OS << "OPC_CheckInteger, "; + unsigned Bytes=1+EmitVBRValue(cast<CheckIntegerMatcher>(N)->getValue(), OS); + OS << '\n'; + return Bytes; + } + case Matcher::CheckCondCode: OS << "OPC_CheckCondCode, ISD::" - << cast<CheckCondCodeMatcherNode>(N)->getCondCodeName() << ",\n"; + << cast<CheckCondCodeMatcher>(N)->getCondCodeName() << ",\n"; return 2; - case MatcherNode::CheckValueType: + case Matcher::CheckValueType: OS << "OPC_CheckValueType, MVT::" - << cast<CheckValueTypeMatcherNode>(N)->getTypeName() << ",\n"; + << cast<CheckValueTypeMatcher>(N)->getTypeName() << ",\n"; return 2; - case MatcherNode::CheckComplexPat: - OS << "OPC_CheckComplexPat, 0/*XXX*/,\n"; + case Matcher::CheckComplexPat: { + const ComplexPattern &Pattern = + cast<CheckComplexPatMatcher>(N)->getPattern(); + OS << "OPC_CheckComplexPat, " << getComplexPat(Pattern) << ','; + if (!OmitComments) { + OS.PadToColumn(CommentIndent) << "// " << Pattern.getSelectFunc(); + OS << ": " << Pattern.getNumOperands() << " operands"; + if (Pattern.hasProperty(SDNPHasChain)) + OS << " + chain result and input"; + } + OS << '\n'; return 2; + } - case MatcherNode::CheckAndImm: { - int64_t Val = cast<CheckAndImmMatcherNode>(N)->getValue(); - OS << "OPC_CheckAndImm" << ClassifyInt(Val) << ", "; - return EmitInt(Val, OS)+1; + case Matcher::CheckAndImm: { + OS << "OPC_CheckAndImm, "; + unsigned Bytes=1+EmitVBRValue(cast<CheckAndImmMatcher>(N)->getValue(), OS); + OS << '\n'; + return Bytes; } - case MatcherNode::CheckOrImm: { - int64_t Val = cast<CheckOrImmMatcherNode>(N)->getValue(); - OS << "OPC_CheckOrImm" << ClassifyInt(Val) << ", "; - return EmitInt(Val, OS)+1; + case Matcher::CheckOrImm: { + OS << "OPC_CheckOrImm, "; + unsigned Bytes = 1+EmitVBRValue(cast<CheckOrImmMatcher>(N)->getValue(), OS); + OS << '\n'; + return Bytes; } - case MatcherNode::CheckProfitableToFold: - OS << "OPC_IsProfitableToFold,\n"; - return 1; - case MatcherNode::CheckLegalToFold: - OS << "OPC_IsLegalToFold,\n"; + + case Matcher::CheckFoldableChainNode: + OS << "OPC_CheckFoldableChainNode,\n"; return 1; + + case Matcher::EmitInteger: { + int64_t Val = cast<EmitIntegerMatcher>(N)->getValue(); + OS << "OPC_EmitInteger, " + << getEnumName(cast<EmitIntegerMatcher>(N)->getVT()) << ", "; + unsigned Bytes = 2+EmitVBRValue(Val, OS); + OS << '\n'; + return Bytes; + } + case Matcher::EmitStringInteger: { + const std::string &Val = cast<EmitStringIntegerMatcher>(N)->getValue(); + // These should always fit into one byte. + OS << "OPC_EmitInteger, " + << getEnumName(cast<EmitStringIntegerMatcher>(N)->getVT()) << ", " + << Val << ",\n"; + return 3; + } + + case Matcher::EmitRegister: + OS << "OPC_EmitRegister, " + << getEnumName(cast<EmitRegisterMatcher>(N)->getVT()) << ", "; + if (Record *R = cast<EmitRegisterMatcher>(N)->getReg()) + OS << getQualifiedName(R) << ",\n"; + else { + OS << "0 "; + if (!OmitComments) + OS << "/*zero_reg*/"; + OS << ",\n"; + } + return 3; + + case Matcher::EmitConvertToTarget: + OS << "OPC_EmitConvertToTarget, " + << cast<EmitConvertToTargetMatcher>(N)->getSlot() << ",\n"; + return 2; + + case Matcher::EmitMergeInputChains: { + const EmitMergeInputChainsMatcher *MN = + cast<EmitMergeInputChainsMatcher>(N); + OS << "OPC_EmitMergeInputChains, " << MN->getNumNodes() << ", "; + for (unsigned i = 0, e = MN->getNumNodes(); i != e; ++i) + OS << MN->getNode(i) << ", "; + OS << '\n'; + return 2+MN->getNumNodes(); + } + case Matcher::EmitCopyToReg: + OS << "OPC_EmitCopyToReg, " + << cast<EmitCopyToRegMatcher>(N)->getSrcSlot() << ", " + << getQualifiedName(cast<EmitCopyToRegMatcher>(N)->getDestPhysReg()) + << ",\n"; + return 3; + case Matcher::EmitNodeXForm: { + const EmitNodeXFormMatcher *XF = cast<EmitNodeXFormMatcher>(N); + OS << "OPC_EmitNodeXForm, " << getNodeXFormID(XF->getNodeXForm()) << ", " + << XF->getSlot() << ','; + if (!OmitComments) + OS.PadToColumn(CommentIndent) << "// "<<XF->getNodeXForm()->getName(); + OS <<'\n'; + return 3; + } + + case Matcher::EmitNode: + case Matcher::MorphNodeTo: { + const EmitNodeMatcherCommon *EN = cast<EmitNodeMatcherCommon>(N); + OS << (isa<EmitNodeMatcher>(EN) ? "OPC_EmitNode" : "OPC_MorphNodeTo"); + OS << ", TARGET_OPCODE(" << EN->getOpcodeName() << "), 0"; + + if (EN->hasChain()) OS << "|OPFL_Chain"; + if (EN->hasInFlag()) OS << "|OPFL_FlagInput"; + if (EN->hasOutFlag()) OS << "|OPFL_FlagOutput"; + if (EN->hasMemRefs()) OS << "|OPFL_MemRefs"; + if (EN->getNumFixedArityOperands() != -1) + OS << "|OPFL_Variadic" << EN->getNumFixedArityOperands(); + OS << ",\n"; + + OS.PadToColumn(Indent*2+4) << EN->getNumVTs(); + if (!OmitComments) + OS << "/*#VTs*/"; + OS << ", "; + for (unsigned i = 0, e = EN->getNumVTs(); i != e; ++i) + OS << getEnumName(EN->getVT(i)) << ", "; + + OS << EN->getNumOperands(); + if (!OmitComments) + OS << "/*#Ops*/"; + OS << ", "; + unsigned NumOperandBytes = 0; + for (unsigned i = 0, e = EN->getNumOperands(); i != e; ++i) + NumOperandBytes += EmitVBRValue(EN->getOperand(i), OS); + + if (!OmitComments) { + // Print the result #'s for EmitNode. + if (const EmitNodeMatcher *E = dyn_cast<EmitNodeMatcher>(EN)) { + if (unsigned NumResults = EN->getNumVTs()) { + OS.PadToColumn(CommentIndent) << "// Results = "; + unsigned First = E->getFirstResultSlot(); + for (unsigned i = 0; i != NumResults; ++i) + OS << "#" << First+i << " "; + } + } + OS << '\n'; + + if (const MorphNodeToMatcher *SNT = dyn_cast<MorphNodeToMatcher>(N)) { + OS.PadToColumn(Indent*2) << "// Src: " + << *SNT->getPattern().getSrcPattern() << '\n'; + OS.PadToColumn(Indent*2) << "// Dst: " + << *SNT->getPattern().getDstPattern() << '\n'; + } + } else + OS << '\n'; + + return 6+EN->getNumVTs()+NumOperandBytes; + } + case Matcher::MarkFlagResults: { + const MarkFlagResultsMatcher *CFR = cast<MarkFlagResultsMatcher>(N); + OS << "OPC_MarkFlagResults, " << CFR->getNumNodes() << ", "; + unsigned NumOperandBytes = 0; + for (unsigned i = 0, e = CFR->getNumNodes(); i != e; ++i) + NumOperandBytes += EmitVBRValue(CFR->getNode(i), OS); + OS << '\n'; + return 2+NumOperandBytes; + } + case Matcher::CompleteMatch: { + const CompleteMatchMatcher *CM = cast<CompleteMatchMatcher>(N); + OS << "OPC_CompleteMatch, " << CM->getNumResults() << ", "; + unsigned NumResultBytes = 0; + for (unsigned i = 0, e = CM->getNumResults(); i != e; ++i) + NumResultBytes += EmitVBRValue(CM->getResult(i), OS); + OS << '\n'; + if (!OmitComments) { + OS.PadToColumn(Indent*2) << "// Src: " + << *CM->getPattern().getSrcPattern() << '\n'; + OS.PadToColumn(Indent*2) << "// Dst: " + << *CM->getPattern().getDstPattern(); + } + OS << '\n'; + return 2 + NumResultBytes; + } } assert(0 && "Unreachable"); return 0; } -/// EmitMatcherAndChildren - Emit the bytes for the specified matcher subtree. +/// EmitMatcherList - Emit the bytes for the specified matcher subtree. unsigned MatcherTableEmitter:: -EmitMatcherAndChildren(const MatcherNode *N, unsigned Indent) { +EmitMatcherList(const Matcher *N, unsigned Indent, unsigned CurrentIdx, + formatted_raw_ostream &OS) { unsigned Size = 0; - while (1) { - // Push is a special case since it is binary. - if (const PushMatcherNode *PMN = dyn_cast<PushMatcherNode>(N)) { - // We need to encode the child and the offset of the failure code before - // emitting either of them. Handle this by buffering the output into a - // string while we get the size. - SmallString<128> TmpBuf; - unsigned ChildSize; - { - raw_svector_ostream OS(TmpBuf); - formatted_raw_ostream FOS(OS); - ChildSize = - EmitMatcherAndChildren(cast<PushMatcherNode>(N)->getChild(),Indent+1); - } - - if (ChildSize > 255) { - errs() << - "Tblgen internal error: can't handle predicate this complex yet\n"; - exit(1); - } - - OS.PadToColumn(Indent*2); - OS << "OPC_Push, " << ChildSize << ",\n"; - OS << TmpBuf.str(); + while (N) { + if (unsigned(N->getKind()) >= Histogram.size()) + Histogram.resize(N->getKind()+1); + Histogram[N->getKind()]++; + if (!OmitComments) + OS << "/*" << CurrentIdx << "*/"; + unsigned MatcherSize = EmitMatcher(N, Indent, CurrentIdx, OS); + Size += MatcherSize; + CurrentIdx += MatcherSize; + + // If there are other nodes in this list, iterate to them, otherwise we're + // done. + N = N->getNext(); + } + return Size; +} + +void MatcherTableEmitter::EmitPredicateFunctions(const CodeGenDAGPatterns &CGP, + formatted_raw_ostream &OS) { + // Emit pattern predicates. + if (!PatternPredicates.empty()) { + OS << "bool CheckPatternPredicate(unsigned PredNo) const {\n"; + OS << " switch (PredNo) {\n"; + OS << " default: assert(0 && \"Invalid predicate in table?\");\n"; + for (unsigned i = 0, e = PatternPredicates.size(); i != e; ++i) + OS << " case " << i << ": return " << PatternPredicates[i] << ";\n"; + OS << " }\n"; + OS << "}\n\n"; + } + + // Emit Node predicates. + // FIXME: Annoyingly, these are stored by name, which we never even emit. Yay? + StringMap<TreePattern*> PFsByName; + + for (CodeGenDAGPatterns::pf_iterator I = CGP.pf_begin(), E = CGP.pf_end(); + I != E; ++I) + PFsByName[I->first->getName()] = I->second; + + if (!NodePredicates.empty()) { + OS << "bool CheckNodePredicate(SDNode *Node, unsigned PredNo) const {\n"; + OS << " switch (PredNo) {\n"; + OS << " default: assert(0 && \"Invalid predicate in table?\");\n"; + for (unsigned i = 0, e = NodePredicates.size(); i != e; ++i) { + // FIXME: Storing this by name is horrible. + TreePattern *P =PFsByName[NodePredicates[i].substr(strlen("Predicate_"))]; + assert(P && "Unknown name?"); - Size += 2 + ChildSize; + // Emit the predicate code corresponding to this pattern. + std::string Code = P->getRecord()->getValueAsCode("Predicate"); + assert(!Code.empty() && "No code in this predicate"); + OS << " case " << i << ": { // " << NodePredicates[i] << '\n'; + std::string ClassName; + if (P->getOnlyTree()->isLeaf()) + ClassName = "SDNode"; + else + ClassName = + CGP.getSDNodeInfo(P->getOnlyTree()->getOperator()).getSDClassName(); + if (ClassName == "SDNode") + OS << " SDNode *N = Node;\n"; + else + OS << " " << ClassName << "*N = cast<" << ClassName << ">(Node);\n"; + OS << Code << "\n }\n"; + } + OS << " }\n"; + OS << "}\n\n"; + } + + // Emit CompletePattern matchers. + // FIXME: This should be const. + if (!ComplexPatterns.empty()) { + OS << "bool CheckComplexPattern(SDNode *Root, SDValue N,\n"; + OS << " unsigned PatternNo, SmallVectorImpl<SDValue> &Result) {\n"; + OS << " switch (PatternNo) {\n"; + OS << " default: assert(0 && \"Invalid pattern # in table?\");\n"; + for (unsigned i = 0, e = ComplexPatterns.size(); i != e; ++i) { + const ComplexPattern &P = *ComplexPatterns[i]; + unsigned NumOps = P.getNumOperands(); + + if (P.hasProperty(SDNPHasChain)) + ++NumOps; // Get the chained node too. - N = PMN->getFailure(); - continue; + OS << " case " << i << ":\n"; + OS << " Result.resize(Result.size()+" << NumOps << ");\n"; + OS << " return " << P.getSelectFunc(); + + OS << "(Root, N"; + for (unsigned i = 0; i != NumOps; ++i) + OS << ", Result[Result.size()-" << (NumOps-i) << ']'; + OS << ");\n"; } + OS << " }\n"; + OS << "}\n\n"; + } - Size += EmitMatcher(N, Indent); + + // Emit SDNodeXForm handlers. + // FIXME: This should be const. + if (!NodeXForms.empty()) { + OS << "SDValue RunSDNodeXForm(SDValue V, unsigned XFormNo) {\n"; + OS << " switch (XFormNo) {\n"; + OS << " default: assert(0 && \"Invalid xform # in table?\");\n"; - // If there are children of this node, iterate to them, otherwise we're - // done. - if (const MatcherNodeWithChild *MNWC = dyn_cast<MatcherNodeWithChild>(N)) - N = MNWC->getChild(); - else - return Size; + // FIXME: The node xform could take SDValue's instead of SDNode*'s. + for (unsigned i = 0, e = NodeXForms.size(); i != e; ++i) { + const CodeGenDAGPatterns::NodeXForm &Entry = + CGP.getSDNodeTransform(NodeXForms[i]); + + Record *SDNode = Entry.first; + const std::string &Code = Entry.second; + + OS << " case " << i << ": { "; + if (!OmitComments) + OS << "// " << NodeXForms[i]->getName(); + OS << '\n'; + + std::string ClassName = CGP.getSDNodeInfo(SDNode).getSDClassName(); + if (ClassName == "SDNode") + OS << " SDNode *N = V.getNode();\n"; + else + OS << " " << ClassName << " *N = cast<" << ClassName + << ">(V.getNode());\n"; + OS << Code << "\n }\n"; + } + OS << " }\n"; + OS << "}\n\n"; } } -void MatcherTableEmitter::EmitPredicateFunctions() { - OS << "bool CheckPatternPredicate(unsigned PredNo) const {\n"; - OS << " switch (PredNo) {\n"; - OS << " default: assert(0 && \"Invalid predicate in table?\");\n"; - for (unsigned i = 0, e = PatternPredicates.size(); i != e; ++i) - OS << " case " << i << ": return " << PatternPredicates[i] << ";\n"; - OS << " }\n"; - OS << "}\n\n"; - - OS << "bool CheckNodePredicate(SDNode *N, unsigned PredNo) const {\n"; - OS << " switch (PredNo) {\n"; - OS << " default: assert(0 && \"Invalid predicate in table?\");\n"; - for (unsigned i = 0, e = NodePredicates.size(); i != e; ++i) - OS << " case " << i << ": return " << NodePredicates[i] << "(N);\n"; - OS << " }\n"; - OS << "}\n\n"; +void MatcherTableEmitter::EmitHistogram(formatted_raw_ostream &OS) { + if (OmitComments) + return; + OS << " // Opcode Histogram:\n"; + for (unsigned i = 0, e = Histogram.size(); i != e; ++i) { + OS << " // #"; + switch ((Matcher::KindTy)i) { + case Matcher::Scope: OS << "OPC_Scope"; break; + case Matcher::RecordNode: OS << "OPC_RecordNode"; break; + case Matcher::RecordChild: OS << "OPC_RecordChild"; break; + case Matcher::RecordMemRef: OS << "OPC_RecordMemRef"; break; + case Matcher::CaptureFlagInput: OS << "OPC_CaptureFlagInput"; break; + case Matcher::MoveChild: OS << "OPC_MoveChild"; break; + case Matcher::MoveParent: OS << "OPC_MoveParent"; break; + case Matcher::CheckSame: OS << "OPC_CheckSame"; break; + case Matcher::CheckPatternPredicate: + OS << "OPC_CheckPatternPredicate"; break; + case Matcher::CheckPredicate: OS << "OPC_CheckPredicate"; break; + case Matcher::CheckOpcode: OS << "OPC_CheckOpcode"; break; + case Matcher::SwitchOpcode: OS << "OPC_SwitchOpcode"; break; + case Matcher::CheckType: OS << "OPC_CheckType"; break; + case Matcher::SwitchType: OS << "OPC_SwitchType"; break; + case Matcher::CheckChildType: OS << "OPC_CheckChildType"; break; + case Matcher::CheckInteger: OS << "OPC_CheckInteger"; break; + case Matcher::CheckCondCode: OS << "OPC_CheckCondCode"; break; + case Matcher::CheckValueType: OS << "OPC_CheckValueType"; break; + case Matcher::CheckComplexPat: OS << "OPC_CheckComplexPat"; break; + case Matcher::CheckAndImm: OS << "OPC_CheckAndImm"; break; + case Matcher::CheckOrImm: OS << "OPC_CheckOrImm"; break; + case Matcher::CheckFoldableChainNode: + OS << "OPC_CheckFoldableChainNode"; break; + case Matcher::EmitInteger: OS << "OPC_EmitInteger"; break; + case Matcher::EmitStringInteger: OS << "OPC_EmitStringInteger"; break; + case Matcher::EmitRegister: OS << "OPC_EmitRegister"; break; + case Matcher::EmitConvertToTarget: OS << "OPC_EmitConvertToTarget"; break; + case Matcher::EmitMergeInputChains: OS << "OPC_EmitMergeInputChains"; break; + case Matcher::EmitCopyToReg: OS << "OPC_EmitCopyToReg"; break; + case Matcher::EmitNode: OS << "OPC_EmitNode"; break; + case Matcher::MorphNodeTo: OS << "OPC_MorphNodeTo"; break; + case Matcher::EmitNodeXForm: OS << "OPC_EmitNodeXForm"; break; + case Matcher::MarkFlagResults: OS << "OPC_MarkFlagResults"; break; + case Matcher::CompleteMatch: OS << "OPC_CompleteMatch"; break; + } + + OS.PadToColumn(40) << " = " << Histogram[i] << '\n'; + } + OS << '\n'; } -void llvm::EmitMatcherTable(const MatcherNode *Matcher, raw_ostream &O) { +void llvm::EmitMatcherTable(const Matcher *TheMatcher, + const CodeGenDAGPatterns &CGP, raw_ostream &O) { formatted_raw_ostream OS(O); OS << "// The main instruction selector code.\n"; - OS << "SDNode *SelectCode2(SDNode *N) {\n"; + OS << "SDNode *SelectCode(SDNode *N) {\n"; - MatcherTableEmitter MatcherEmitter(OS); + MatcherTableEmitter MatcherEmitter; + OS << " // Opcodes are emitted as 2 bytes, TARGET_OPCODE handles this.\n"; + OS << " #define TARGET_OPCODE(X) X & 255, unsigned(X) >> 8\n"; OS << " static const unsigned char MatcherTable[] = {\n"; - unsigned TotalSize = MatcherEmitter.EmitMatcherAndChildren(Matcher, 2); + unsigned TotalSize = MatcherEmitter.EmitMatcherList(TheMatcher, 5, 0, OS); OS << " 0\n }; // Total Array size is " << (TotalSize+1) << " bytes\n\n"; + + MatcherEmitter.EmitHistogram(OS); + + OS << " #undef TARGET_OPCODE\n"; OS << " return SelectCodeCommon(N, MatcherTable,sizeof(MatcherTable));\n}\n"; - OS << "\n"; + OS << '\n'; // Next up, emit the function for node and pattern predicates: - MatcherEmitter.EmitPredicateFunctions(); + MatcherEmitter.EmitPredicateFunctions(CGP, OS); } diff --git a/utils/TableGen/DAGISelMatcherGen.cpp b/utils/TableGen/DAGISelMatcherGen.cpp index 07ab472..4482803 100644 --- a/utils/TableGen/DAGISelMatcherGen.cpp +++ b/utils/TableGen/DAGISelMatcherGen.cpp @@ -10,9 +10,40 @@ #include "DAGISelMatcher.h" #include "CodeGenDAGPatterns.h" #include "Record.h" +#include "llvm/ADT/SmallVector.h" #include "llvm/ADT/StringMap.h" +#include <utility> using namespace llvm; + +/// getRegisterValueType - Look up and return the ValueType of the specified +/// register. If the register is a member of multiple register classes which +/// have different associated types, return MVT::Other. +static MVT::SimpleValueType getRegisterValueType(Record *R, + const CodeGenTarget &T) { + bool FoundRC = false; + MVT::SimpleValueType VT = MVT::Other; + const std::vector<CodeGenRegisterClass> &RCs = T.getRegisterClasses(); + std::vector<Record*>::const_iterator Element; + + for (unsigned rc = 0, e = RCs.size(); rc != e; ++rc) { + const CodeGenRegisterClass &RC = RCs[rc]; + if (!std::count(RC.Elements.begin(), RC.Elements.end(), R)) + continue; + + if (!FoundRC) { + FoundRC = true; + VT = RC.getValueTypeNum(0); + continue; + } + + // If this occurs in multiple register classes, they all have to agree. + assert(VT == RC.getValueTypeNum(0)); + } + return VT; +} + + namespace { class MatcherGen { const PatternToMatch &Pattern; @@ -27,10 +58,31 @@ namespace { /// number that they were captured as. These are biased by 1 to make /// insertion easier. StringMap<unsigned> VariableMap; + + /// NextRecordedOperandNo - As we emit opcodes to record matched values in + /// the RecordedNodes array, this keeps track of which slot will be next to + /// record into. unsigned NextRecordedOperandNo; - MatcherNodeWithChild *Matcher; - MatcherNodeWithChild *CurPredicate; + /// MatchedChainNodes - This maintains the position in the recorded nodes + /// array of all of the recorded input nodes that have chains. + SmallVector<unsigned, 2> MatchedChainNodes; + + /// MatchedFlagResultNodes - This maintains the position in the recorded + /// nodes array of all of the recorded input nodes that have flag results. + SmallVector<unsigned, 2> MatchedFlagResultNodes; + + /// PhysRegInputs - List list has an entry for each explicitly specified + /// physreg input to the pattern. The first elt is the Register node, the + /// second is the recorded slot number the input pattern match saved it in. + SmallVector<std::pair<Record*, unsigned>, 2> PhysRegInputs; + + /// Matcher - This is the top level of the generated matcher, the result. + Matcher *TheMatcher; + + /// CurPredicate - As we emit matcher nodes, this points to the latest check + /// which should have future checks stuck into its Next position. + Matcher *CurPredicate; public: MatcherGen(const PatternToMatch &pattern, const CodeGenDAGPatterns &cgp); @@ -38,25 +90,51 @@ namespace { delete PatWithNoTypes; } - void EmitMatcherCode(); + bool EmitMatcherCode(unsigned Variant); + void EmitResultCode(); - MatcherNodeWithChild *GetMatcher() const { return Matcher; } - MatcherNodeWithChild *GetCurPredicate() const { return CurPredicate; } + Matcher *GetMatcher() const { return TheMatcher; } + Matcher *GetCurPredicate() const { return CurPredicate; } private: - void AddMatcherNode(MatcherNodeWithChild *NewNode); + void AddMatcher(Matcher *NewNode); void InferPossibleTypes(); + + // Matcher Generation. void EmitMatchCode(const TreePatternNode *N, TreePatternNode *NodeNoTypes); void EmitLeafMatchCode(const TreePatternNode *N); void EmitOperatorMatchCode(const TreePatternNode *N, TreePatternNode *NodeNoTypes); - }; + + // Result Code Generation. + unsigned getNamedArgumentSlot(StringRef Name) { + unsigned VarMapEntry = VariableMap[Name]; + assert(VarMapEntry != 0 && + "Variable referenced but not defined and not caught earlier!"); + return VarMapEntry-1; + } + + /// GetInstPatternNode - Get the pattern for an instruction. + const TreePatternNode *GetInstPatternNode(const DAGInstruction &Ins, + const TreePatternNode *N); + + void EmitResultOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &ResultOps); + void EmitResultOfNamedOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &ResultOps); + void EmitResultLeafAsOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &ResultOps); + void EmitResultInstructionAsOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &ResultOps); + void EmitResultSDNodeXFormAsOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &ResultOps); + }; } // end anon namespace. MatcherGen::MatcherGen(const PatternToMatch &pattern, const CodeGenDAGPatterns &cgp) : Pattern(pattern), CGP(cgp), NextRecordedOperandNo(0), - Matcher(0), CurPredicate(0) { + TheMatcher(0), CurPredicate(0) { // We need to produce the matcher tree for the patterns source pattern. To do // this we need to match the structure as well as the types. To do the type // matching, we want to figure out the fewest number of type checks we need to @@ -97,23 +175,40 @@ void MatcherGen::InferPossibleTypes() { } -/// AddMatcherNode - Add a matcher node to the current graph we're building. -void MatcherGen::AddMatcherNode(MatcherNodeWithChild *NewNode) { +/// AddMatcher - Add a matcher node to the current graph we're building. +void MatcherGen::AddMatcher(Matcher *NewNode) { if (CurPredicate != 0) - CurPredicate->setChild(NewNode); + CurPredicate->setNext(NewNode); else - Matcher = NewNode; + TheMatcher = NewNode; CurPredicate = NewNode; } +//===----------------------------------------------------------------------===// +// Pattern Match Generation +//===----------------------------------------------------------------------===// /// EmitLeafMatchCode - Generate matching code for leaf nodes. void MatcherGen::EmitLeafMatchCode(const TreePatternNode *N) { assert(N->isLeaf() && "Not a leaf?"); + + // If there are node predicates for this node, generate their checks. + for (unsigned i = 0, e = N->getPredicateFns().size(); i != e; ++i) + AddMatcher(new CheckPredicateMatcher(N->getPredicateFns()[i])); + // Direct match against an integer constant. - if (IntInit *II = dynamic_cast<IntInit*>(N->getLeafValue())) - return AddMatcherNode(new CheckIntegerMatcherNode(II->getValue())); + if (IntInit *II = dynamic_cast<IntInit*>(N->getLeafValue())) { + // If this is the root of the dag we're matching, we emit a redundant opcode + // check to ensure that this gets folded into the normal top-level + // OpcodeSwitch. + if (N == Pattern.getSrcPattern()) { + const SDNodeInfo &NI = CGP.getSDNodeInfo(CGP.getSDNodeNamed("imm")); + AddMatcher(new CheckOpcodeMatcher(NI)); + } + + return AddMatcher(new CheckIntegerMatcher(II->getValue())); + } DefInit *DI = dynamic_cast<DefInit*>(N->getLeafValue()); if (DI == 0) { @@ -125,21 +220,58 @@ void MatcherGen::EmitLeafMatchCode(const TreePatternNode *N) { if (// Handle register references. Nothing to do here, they always match. LeafRec->isSubClassOf("RegisterClass") || LeafRec->isSubClassOf("PointerLikeRegClass") || - LeafRec->isSubClassOf("Register") || // Place holder for SRCVALUE nodes. Nothing to do here. LeafRec->getName() == "srcvalue") return; + + // If we have a physreg reference like (mul gpr:$src, EAX) then we need to + // record the register + if (LeafRec->isSubClassOf("Register")) { + AddMatcher(new RecordMatcher("physreg input "+LeafRec->getName(), + NextRecordedOperandNo)); + PhysRegInputs.push_back(std::make_pair(LeafRec, NextRecordedOperandNo++)); + return; + } if (LeafRec->isSubClassOf("ValueType")) - return AddMatcherNode(new CheckValueTypeMatcherNode(LeafRec->getName())); + return AddMatcher(new CheckValueTypeMatcher(LeafRec->getName())); if (LeafRec->isSubClassOf("CondCode")) - return AddMatcherNode(new CheckCondCodeMatcherNode(LeafRec->getName())); + return AddMatcher(new CheckCondCodeMatcher(LeafRec->getName())); if (LeafRec->isSubClassOf("ComplexPattern")) { + // We can't model ComplexPattern uses that don't have their name taken yet. + // The OPC_CheckComplexPattern operation implicitly records the results. + if (N->getName().empty()) { + errs() << "We expect complex pattern uses to have names: " << *N << "\n"; + exit(1); + } + // Handle complex pattern. const ComplexPattern &CP = CGP.getComplexPattern(LeafRec); - return AddMatcherNode(new CheckComplexPatMatcherNode(CP)); + + // Emit a CheckComplexPat operation, which does the match (aborting if it + // fails) and pushes the matched operands onto the recorded nodes list. + AddMatcher(new CheckComplexPatMatcher(CP)); + + // Record the right number of operands. + NextRecordedOperandNo += CP.getNumOperands(); + if (CP.hasProperty(SDNPHasChain)) + ++NextRecordedOperandNo; // Chained node operand. + + // If the complex pattern has a chain, then we need to keep track of the + // fact that we just recorded a chain input. The chain input will be + // matched as the last operand of the predicate if it was successful. + if (CP.hasProperty(SDNPHasChain)) { + // It is the last operand recorded. + assert(NextRecordedOperandNo > 1 && + "Should have recorded input/result chains at least!"); + MatchedChainNodes.push_back(NextRecordedOperandNo-1); + } + + // TODO: Complex patterns can't have output flags, if they did, we'd want + // to record them. + return; } errs() << "Unknown leaf kind: " << *N << "\n"; @@ -163,61 +295,82 @@ void MatcherGen::EmitOperatorMatchCode(const TreePatternNode *N, // to handle this. if ((N->getOperator()->getName() == "and" || N->getOperator()->getName() == "or") && - N->getChild(1)->isLeaf() && N->getChild(1)->getPredicateFns().empty()) { + N->getChild(1)->isLeaf() && N->getChild(1)->getPredicateFns().empty() && + N->getPredicateFns().empty()) { if (IntInit *II = dynamic_cast<IntInit*>(N->getChild(1)->getLeafValue())) { if (!isPowerOf2_32(II->getValue())) { // Don't bother with single bits. + // If this is at the root of the pattern, we emit a redundant + // CheckOpcode so that the following checks get factored properly under + // a single opcode check. + if (N == Pattern.getSrcPattern()) + AddMatcher(new CheckOpcodeMatcher(CInfo)); + + // Emit the CheckAndImm/CheckOrImm node. if (N->getOperator()->getName() == "and") - AddMatcherNode(new CheckAndImmMatcherNode(II->getValue())); + AddMatcher(new CheckAndImmMatcher(II->getValue())); else - AddMatcherNode(new CheckOrImmMatcherNode(II->getValue())); + AddMatcher(new CheckOrImmMatcher(II->getValue())); // Match the LHS of the AND as appropriate. - AddMatcherNode(new MoveChildMatcherNode(0)); + AddMatcher(new MoveChildMatcher(0)); EmitMatchCode(N->getChild(0), NodeNoTypes->getChild(0)); - AddMatcherNode(new MoveParentMatcherNode()); + AddMatcher(new MoveParentMatcher()); return; } } } // Check that the current opcode lines up. - AddMatcherNode(new CheckOpcodeMatcherNode(CInfo.getEnumName())); + AddMatcher(new CheckOpcodeMatcher(CInfo)); + + // If there are node predicates for this node, generate their checks. + for (unsigned i = 0, e = N->getPredicateFns().size(); i != e; ++i) + AddMatcher(new CheckPredicateMatcher(N->getPredicateFns()[i])); + + + // If this node has memory references (i.e. is a load or store), tell the + // interpreter to capture them in the memref array. + if (N->NodeHasProperty(SDNPMemOperand, CGP)) + AddMatcher(new RecordMemRefMatcher()); // If this node has a chain, then the chain is operand #0 is the SDNode, and // the child numbers of the node are all offset by one. unsigned OpNo = 0; - if (N->NodeHasProperty(SDNPHasChain, CGP)) + if (N->NodeHasProperty(SDNPHasChain, CGP)) { + // Record the node and remember it in our chained nodes list. + AddMatcher(new RecordMatcher("'" + N->getOperator()->getName() + + "' chained node", + NextRecordedOperandNo)); + // Remember all of the input chains our pattern will match. + MatchedChainNodes.push_back(NextRecordedOperandNo++); + + // Don't look at the input chain when matching the tree pattern to the + // SDNode. OpNo = 1; - // If this node is not the root and the subtree underneath it produces a - // chain, then the result of matching the node is also produce a chain. - // Beyond that, this means that we're also folding (at least) the root node - // into the node that produce the chain (for example, matching - // "(add reg, (load ptr))" as a add_with_memory on X86). This is problematic, - // if the 'reg' node also uses the load (say, its chain). Graphically: - // - // [LD] - // ^ ^ - // | \ DAG's like cheese. - // / | - // / [YY] - // | ^ - // [XX]--/ - // - // It would be invalid to fold XX and LD. In this case, folding the two - // nodes together would induce a cycle in the DAG, making it a cyclic DAG (!). - // To prevent this, we emit a dynamic check for legality before allowing this - // to be folded. - // - const TreePatternNode *Root = Pattern.getSrcPattern(); - if (N != Root && // Not the root of the pattern. - N->TreeHasProperty(SDNPHasChain, CGP)) { // Has a chain somewhere in tree. - - AddMatcherNode(new CheckProfitableToFoldMatcherNode()); - - // If this non-root node produces a chain, we may need to emit a validity - // check. - if (OpNo != 0) { + // If this node is not the root and the subtree underneath it produces a + // chain, then the result of matching the node is also produce a chain. + // Beyond that, this means that we're also folding (at least) the root node + // into the node that produce the chain (for example, matching + // "(add reg, (load ptr))" as a add_with_memory on X86). This is + // problematic, if the 'reg' node also uses the load (say, its chain). + // Graphically: + // + // [LD] + // ^ ^ + // | \ DAG's like cheese. + // / | + // / [YY] + // | ^ + // [XX]--/ + // + // It would be invalid to fold XX and LD. In this case, folding the two + // nodes together would induce a cycle in the DAG, making it a 'cyclic DAG' + // To prevent this, we emit a dynamic check for legality before allowing + // this to be folded. + // + const TreePatternNode *Root = Pattern.getSrcPattern(); + if (N != Root) { // Not the root of the pattern. // If there is a node between the root and this node, then we definitely // need to emit the check. bool NeedCheck = !Root->hasChild(N); @@ -239,16 +392,35 @@ void MatcherGen::EmitOperatorMatchCode(const TreePatternNode *N, } if (NeedCheck) - AddMatcherNode(new CheckLegalToFoldMatcherNode()); + AddMatcher(new CheckFoldableChainNodeMatcher()); } } + + // If this node has an output flag and isn't the root, remember it. + if (N->NodeHasProperty(SDNPOutFlag, CGP) && + N != Pattern.getSrcPattern()) { + // TODO: This redundantly records nodes with both flags and chains. + + // Record the node and remember it in our chained nodes list. + AddMatcher(new RecordMatcher("'" + N->getOperator()->getName() + + "' flag output node", + NextRecordedOperandNo)); + // Remember all of the nodes with output flags our pattern will match. + MatchedFlagResultNodes.push_back(NextRecordedOperandNo++); + } + + // If this node is known to have an input flag or if it *might* have an input + // flag, capture it as the flag input of the pattern. + if (N->NodeHasProperty(SDNPOptInFlag, CGP) || + N->NodeHasProperty(SDNPInFlag, CGP)) + AddMatcher(new CaptureFlagInputMatcher()); for (unsigned i = 0, e = N->getNumChildren(); i != e; ++i, ++OpNo) { // Get the code suitable for matching this child. Move to the child, check // it then move back to the parent. - AddMatcherNode(new MoveChildMatcherNode(i)); + AddMatcher(new MoveChildMatcher(OpNo)); EmitMatchCode(N->getChild(i), NodeNoTypes->getChild(i)); - AddMatcherNode(new MoveParentMatcherNode()); + AddMatcher(new MoveParentMatcher()); } } @@ -258,70 +430,424 @@ void MatcherGen::EmitMatchCode(const TreePatternNode *N, // If N and NodeNoTypes don't agree on a type, then this is a case where we // need to do a type check. Emit the check, apply the tyep to NodeNoTypes and // reinfer any correlated types. + unsigned NodeType = EEVT::isUnknown; if (NodeNoTypes->getExtTypes() != N->getExtTypes()) { - AddMatcherNode(new CheckTypeMatcherNode(N->getTypeNum(0))); + NodeType = N->getTypeNum(0); NodeNoTypes->setTypes(N->getExtTypes()); InferPossibleTypes(); } - // If this node has a name associated with it, capture it in VariableMap. If // we already saw this in the pattern, emit code to verify dagness. if (!N->getName().empty()) { unsigned &VarMapEntry = VariableMap[N->getName()]; if (VarMapEntry == 0) { + // If it is a named node, we must emit a 'Record' opcode. + AddMatcher(new RecordMatcher("$" + N->getName(), NextRecordedOperandNo)); VarMapEntry = ++NextRecordedOperandNo; - AddMatcherNode(new RecordMatcherNode()); } else { // If we get here, this is a second reference to a specific name. Since // we already have checked that the first reference is valid, we don't // have to recursively match it, just check that it's the same as the // previously named thing. - AddMatcherNode(new CheckSameMatcherNode(VarMapEntry-1)); + AddMatcher(new CheckSameMatcher(VarMapEntry-1)); return; } } - // If there are node predicates for this node, generate their checks. - for (unsigned i = 0, e = N->getPredicateFns().size(); i != e; ++i) - AddMatcherNode(new CheckPredicateMatcherNode(N->getPredicateFns()[i])); - if (N->isLeaf()) EmitLeafMatchCode(N); else EmitOperatorMatchCode(N, NodeNoTypes); + + if (NodeType != EEVT::isUnknown) + AddMatcher(new CheckTypeMatcher((MVT::SimpleValueType)NodeType)); + } -void MatcherGen::EmitMatcherCode() { +/// EmitMatcherCode - Generate the code that matches the predicate of this +/// pattern for the specified Variant. If the variant is invalid this returns +/// true and does not generate code, if it is valid, it returns false. +bool MatcherGen::EmitMatcherCode(unsigned Variant) { + // If the root of the pattern is a ComplexPattern and if it is specified to + // match some number of root opcodes, these are considered to be our variants. + // Depending on which variant we're generating code for, emit the root opcode + // check. + if (const ComplexPattern *CP = + Pattern.getSrcPattern()->getComplexPatternInfo(CGP)) { + const std::vector<Record*> &OpNodes = CP->getRootNodes(); + assert(!OpNodes.empty() &&"Complex Pattern must specify what it can match"); + if (Variant >= OpNodes.size()) return true; + + AddMatcher(new CheckOpcodeMatcher(CGP.getSDNodeInfo(OpNodes[Variant]))); + } else { + if (Variant != 0) return true; + } + // If the pattern has a predicate on it (e.g. only enabled when a subtarget // feature is around, do the check). + // FIXME: This should get emitted after the match code below to encourage + // sharing. This can't happen until we get an X86ISD::AddrMode node made by + // dag combine, eliminating the horrible side-effect-full stuff from + // X86's MatchAddress. if (!Pattern.getPredicateCheck().empty()) - AddMatcherNode(new - CheckPatternPredicateMatcherNode(Pattern.getPredicateCheck())); - + AddMatcher(new CheckPatternPredicateMatcher(Pattern.getPredicateCheck())); + // Emit the matcher for the pattern structure and types. EmitMatchCode(Pattern.getSrcPattern(), PatWithNoTypes); + return false; +} + + +//===----------------------------------------------------------------------===// +// Node Result Generation +//===----------------------------------------------------------------------===// + +void MatcherGen::EmitResultOfNamedOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &ResultOps){ + assert(!N->getName().empty() && "Operand not named!"); + + unsigned SlotNo = getNamedArgumentSlot(N->getName()); + + // A reference to a complex pattern gets all of the results of the complex + // pattern's match. + if (const ComplexPattern *CP = N->getComplexPatternInfo(CGP)) { + // The first slot entry is the node itself, the subsequent entries are the + // matched values. + for (unsigned i = 0, e = CP->getNumOperands(); i != e; ++i) + ResultOps.push_back(SlotNo+i+1); + return; + } + + // If this is an 'imm' or 'fpimm' node, make sure to convert it to the target + // version of the immediate so that it doesn't get selected due to some other + // node use. + if (!N->isLeaf()) { + StringRef OperatorName = N->getOperator()->getName(); + if (OperatorName == "imm" || OperatorName == "fpimm") { + AddMatcher(new EmitConvertToTargetMatcher(SlotNo)); + ResultOps.push_back(NextRecordedOperandNo++); + return; + } + } + + ResultOps.push_back(SlotNo); +} + +void MatcherGen::EmitResultLeafAsOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &ResultOps) { + assert(N->isLeaf() && "Must be a leaf"); + + if (IntInit *II = dynamic_cast<IntInit*>(N->getLeafValue())) { + AddMatcher(new EmitIntegerMatcher(II->getValue(),N->getTypeNum(0))); + ResultOps.push_back(NextRecordedOperandNo++); + return; + } + + // If this is an explicit register reference, handle it. + if (DefInit *DI = dynamic_cast<DefInit*>(N->getLeafValue())) { + if (DI->getDef()->isSubClassOf("Register")) { + AddMatcher(new EmitRegisterMatcher(DI->getDef(), + N->getTypeNum(0))); + ResultOps.push_back(NextRecordedOperandNo++); + return; + } + + if (DI->getDef()->getName() == "zero_reg") { + AddMatcher(new EmitRegisterMatcher(0, N->getTypeNum(0))); + ResultOps.push_back(NextRecordedOperandNo++); + return; + } + + // Handle a reference to a register class. This is used + // in COPY_TO_SUBREG instructions. + if (DI->getDef()->isSubClassOf("RegisterClass")) { + std::string Value = getQualifiedName(DI->getDef()) + "RegClassID"; + AddMatcher(new EmitStringIntegerMatcher(Value, MVT::i32)); + ResultOps.push_back(NextRecordedOperandNo++); + return; + } + } + + errs() << "unhandled leaf node: \n"; + N->dump(); +} + +/// GetInstPatternNode - Get the pattern for an instruction. +/// +const TreePatternNode *MatcherGen:: +GetInstPatternNode(const DAGInstruction &Inst, const TreePatternNode *N) { + const TreePattern *InstPat = Inst.getPattern(); + + // FIXME2?: Assume actual pattern comes before "implicit". + TreePatternNode *InstPatNode; + if (InstPat) + InstPatNode = InstPat->getTree(0); + else if (/*isRoot*/ N == Pattern.getDstPattern()) + InstPatNode = Pattern.getSrcPattern(); + else + return 0; + + if (InstPatNode && !InstPatNode->isLeaf() && + InstPatNode->getOperator()->getName() == "set") + InstPatNode = InstPatNode->getChild(InstPatNode->getNumChildren()-1); + + return InstPatNode; } +void MatcherGen:: +EmitResultInstructionAsOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &OutputOps) { + Record *Op = N->getOperator(); + const CodeGenTarget &CGT = CGP.getTargetInfo(); + CodeGenInstruction &II = CGT.getInstruction(Op->getName()); + const DAGInstruction &Inst = CGP.getInstruction(Op); + + // If we can, get the pattern for the instruction we're generating. We derive + // a variety of information from this pattern, such as whether it has a chain. + // + // FIXME2: This is extremely dubious for several reasons, not the least of + // which it gives special status to instructions with patterns that Pat<> + // nodes can't duplicate. + const TreePatternNode *InstPatNode = GetInstPatternNode(Inst, N); + + // NodeHasChain - Whether the instruction node we're creating takes chains. + bool NodeHasChain = InstPatNode && + InstPatNode->TreeHasProperty(SDNPHasChain, CGP); + + bool isRoot = N == Pattern.getDstPattern(); + + // TreeHasOutFlag - True if this tree has a flag. + bool TreeHasInFlag = false, TreeHasOutFlag = false; + if (isRoot) { + const TreePatternNode *SrcPat = Pattern.getSrcPattern(); + TreeHasInFlag = SrcPat->TreeHasProperty(SDNPOptInFlag, CGP) || + SrcPat->TreeHasProperty(SDNPInFlag, CGP); + + // FIXME2: this is checking the entire pattern, not just the node in + // question, doing this just for the root seems like a total hack. + TreeHasOutFlag = SrcPat->TreeHasProperty(SDNPOutFlag, CGP); + } + + // NumResults - This is the number of results produced by the instruction in + // the "outs" list. + unsigned NumResults = Inst.getNumResults(); -MatcherNode *llvm::ConvertPatternToMatcher(const PatternToMatch &Pattern, - const CodeGenDAGPatterns &CGP) { + // Loop over all of the operands of the instruction pattern, emitting code + // to fill them all in. The node 'N' usually has number children equal to + // the number of input operands of the instruction. However, in cases + // where there are predicate operands for an instruction, we need to fill + // in the 'execute always' values. Match up the node operands to the + // instruction operands to do this. + SmallVector<unsigned, 8> InstOps; + for (unsigned ChildNo = 0, InstOpNo = NumResults, e = II.OperandList.size(); + InstOpNo != e; ++InstOpNo) { + + // Determine what to emit for this operand. + Record *OperandNode = II.OperandList[InstOpNo].Rec; + if ((OperandNode->isSubClassOf("PredicateOperand") || + OperandNode->isSubClassOf("OptionalDefOperand")) && + !CGP.getDefaultOperand(OperandNode).DefaultOps.empty()) { + // This is a predicate or optional def operand; emit the + // 'default ops' operands. + const DAGDefaultOperand &DefaultOp = + CGP.getDefaultOperand(II.OperandList[InstOpNo].Rec); + for (unsigned i = 0, e = DefaultOp.DefaultOps.size(); i != e; ++i) + EmitResultOperand(DefaultOp.DefaultOps[i], InstOps); + continue; + } + + // Otherwise this is a normal operand or a predicate operand without + // 'execute always'; emit it. + EmitResultOperand(N->getChild(ChildNo), InstOps); + ++ChildNo; + } + + // If this node has an input flag or explicitly specified input physregs, we + // need to add chained and flagged copyfromreg nodes and materialize the flag + // input. + if (isRoot && !PhysRegInputs.empty()) { + // Emit all of the CopyToReg nodes for the input physical registers. These + // occur in patterns like (mul:i8 AL:i8, GR8:i8:$src). + for (unsigned i = 0, e = PhysRegInputs.size(); i != e; ++i) + AddMatcher(new EmitCopyToRegMatcher(PhysRegInputs[i].second, + PhysRegInputs[i].first)); + // Even if the node has no other flag inputs, the resultant node must be + // flagged to the CopyFromReg nodes we just generated. + TreeHasInFlag = true; + } + + // Result order: node results, chain, flags + + // Determine the result types. + SmallVector<MVT::SimpleValueType, 4> ResultVTs; + if (NumResults != 0 && N->getTypeNum(0) != MVT::isVoid) { + // FIXME2: If the node has multiple results, we should add them. For now, + // preserve existing behavior?! + ResultVTs.push_back(N->getTypeNum(0)); + } + + + // If this is the root instruction of a pattern that has physical registers in + // its result pattern, add output VTs for them. For example, X86 has: + // (set AL, (mul ...)) + // This also handles implicit results like: + // (implicit EFLAGS) + if (isRoot && Pattern.getDstRegs().size() != 0) { + for (unsigned i = 0; i != Pattern.getDstRegs().size(); ++i) + if (Pattern.getDstRegs()[i]->isSubClassOf("Register")) + ResultVTs.push_back(getRegisterValueType(Pattern.getDstRegs()[i], CGT)); + } + + // FIXME2: Instead of using the isVariadic flag on the instruction, we should + // have an SDNP that indicates variadicism. The TargetInstrInfo isVariadic + // property should be inferred from this when an instruction has a pattern. + int NumFixedArityOperands = -1; + if (isRoot && II.isVariadic) + NumFixedArityOperands = Pattern.getSrcPattern()->getNumChildren(); + + // If this is the root node and any of the nodes matched nodes in the input + // pattern have MemRefs in them, have the interpreter collect them and plop + // them onto this node. + // + // FIXME3: This is actively incorrect for result patterns where the root of + // the pattern is not the memory reference and is also incorrect when the + // result pattern has multiple memory-referencing instructions. For example, + // in the X86 backend, this pattern causes the memrefs to get attached to the + // CVTSS2SDrr instead of the MOVSSrm: + // + // def : Pat<(extloadf32 addr:$src), + // (CVTSS2SDrr (MOVSSrm addr:$src))>; + // + bool NodeHasMemRefs = + isRoot && Pattern.getSrcPattern()->TreeHasProperty(SDNPMemOperand, CGP); + + AddMatcher(new EmitNodeMatcher(II.Namespace+"::"+II.TheDef->getName(), + ResultVTs.data(), ResultVTs.size(), + InstOps.data(), InstOps.size(), + NodeHasChain, TreeHasInFlag, TreeHasOutFlag, + NodeHasMemRefs, NumFixedArityOperands, + NextRecordedOperandNo)); + + // The non-chain and non-flag results of the newly emitted node get recorded. + for (unsigned i = 0, e = ResultVTs.size(); i != e; ++i) { + if (ResultVTs[i] == MVT::Other || ResultVTs[i] == MVT::Flag) break; + OutputOps.push_back(NextRecordedOperandNo++); + } +} + +void MatcherGen:: +EmitResultSDNodeXFormAsOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &ResultOps) { + assert(N->getOperator()->isSubClassOf("SDNodeXForm") && "Not SDNodeXForm?"); + + // Emit the operand. + SmallVector<unsigned, 8> InputOps; + + // FIXME2: Could easily generalize this to support multiple inputs and outputs + // to the SDNodeXForm. For now we just support one input and one output like + // the old instruction selector. + assert(N->getNumChildren() == 1); + EmitResultOperand(N->getChild(0), InputOps); + + // The input currently must have produced exactly one result. + assert(InputOps.size() == 1 && "Unexpected input to SDNodeXForm"); + + AddMatcher(new EmitNodeXFormMatcher(InputOps[0], N->getOperator())); + ResultOps.push_back(NextRecordedOperandNo++); +} + +void MatcherGen::EmitResultOperand(const TreePatternNode *N, + SmallVectorImpl<unsigned> &ResultOps) { + // This is something selected from the pattern we matched. + if (!N->getName().empty()) + return EmitResultOfNamedOperand(N, ResultOps); + + if (N->isLeaf()) + return EmitResultLeafAsOperand(N, ResultOps); + + Record *OpRec = N->getOperator(); + if (OpRec->isSubClassOf("Instruction")) + return EmitResultInstructionAsOperand(N, ResultOps); + if (OpRec->isSubClassOf("SDNodeXForm")) + return EmitResultSDNodeXFormAsOperand(N, ResultOps); + errs() << "Unknown result node to emit code for: " << *N << '\n'; + throw std::string("Unknown node in result pattern!"); +} + +void MatcherGen::EmitResultCode() { + // Patterns that match nodes with (potentially multiple) chain inputs have to + // merge them together into a token factor. This informs the generated code + // what all the chained nodes are. + if (!MatchedChainNodes.empty()) + AddMatcher(new EmitMergeInputChainsMatcher + (MatchedChainNodes.data(), MatchedChainNodes.size())); + + // Codegen the root of the result pattern, capturing the resulting values. + SmallVector<unsigned, 8> Ops; + EmitResultOperand(Pattern.getDstPattern(), Ops); + + // At this point, we have however many values the result pattern produces. + // However, the input pattern might not need all of these. If there are + // excess values at the end (such as condition codes etc) just lop them off. + // This doesn't need to worry about flags or chains, just explicit results. + // + // FIXME2: This doesn't work because there is currently no way to get an + // accurate count of the # results the source pattern sets. This is because + // of the "parallel" construct in X86 land, which looks like this: + // + //def : Pat<(parallel (X86and_flag GR8:$src1, GR8:$src2), + // (implicit EFLAGS)), + // (AND8rr GR8:$src1, GR8:$src2)>; + // + // This idiom means to match the two-result node X86and_flag (which is + // declared as returning a single result, because we can't match multi-result + // nodes yet). In this case, we would have to know that the input has two + // results. However, mul8r is modelled exactly the same way, but without + // implicit defs included. The fix is to support multiple results directly + // and eliminate 'parallel'. + // + // FIXME2: When this is fixed, we should revert the terrible hack in the + // OPC_EmitNode code in the interpreter. +#if 0 + const TreePatternNode *Src = Pattern.getSrcPattern(); + unsigned NumSrcResults = Src->getTypeNum(0) != MVT::isVoid ? 1 : 0; + NumSrcResults += Pattern.getDstRegs().size(); + assert(Ops.size() >= NumSrcResults && "Didn't provide enough results"); + Ops.resize(NumSrcResults); +#endif + + // If the matched pattern covers nodes which define a flag result, emit a node + // that tells the matcher about them so that it can update their results. + if (!MatchedFlagResultNodes.empty()) + AddMatcher(new MarkFlagResultsMatcher(MatchedFlagResultNodes.data(), + MatchedFlagResultNodes.size())); + + AddMatcher(new CompleteMatchMatcher(Ops.data(), Ops.size(), Pattern)); +} + + +/// ConvertPatternToMatcher - Create the matcher for the specified pattern with +/// the specified variant. If the variant number is invalid, this returns null. +Matcher *llvm::ConvertPatternToMatcher(const PatternToMatch &Pattern, + unsigned Variant, + const CodeGenDAGPatterns &CGP) { MatcherGen Gen(Pattern, CGP); // Generate the code for the matcher. - Gen.EmitMatcherCode(); + if (Gen.EmitMatcherCode(Variant)) + return 0; - // If the match succeeds, then we generate Pattern. - EmitNodeMatcherNode *Result = new EmitNodeMatcherNode(Pattern); + // FIXME2: Kill extra MoveParent commands at the end of the matcher sequence. + // FIXME2: Split result code out to another table, and make the matcher end + // with an "Emit <index>" command. This allows result generation stuff to be + // shared and factored? - // Link it into the pattern. - if (MatcherNodeWithChild *Pred = Gen.GetCurPredicate()) { - Pred->setChild(Result); - return Gen.GetMatcher(); - } + // If the match succeeds, then we generate Pattern. + Gen.EmitResultCode(); // Unconditional match. - return Result; + return Gen.GetMatcher(); } diff --git a/utils/TableGen/DAGISelMatcherOpt.cpp b/utils/TableGen/DAGISelMatcherOpt.cpp new file mode 100644 index 0000000..dc077a9 --- /dev/null +++ b/utils/TableGen/DAGISelMatcherOpt.cpp @@ -0,0 +1,435 @@ +//===- DAGISelMatcherOpt.cpp - Optimize a DAG Matcher ---------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the DAG Matcher optimizer. +// +//===----------------------------------------------------------------------===// + +#define DEBUG_TYPE "isel-opt" +#include "DAGISelMatcher.h" +#include "CodeGenDAGPatterns.h" +#include "llvm/ADT/DenseSet.h" +#include "llvm/ADT/StringSet.h" +#include "llvm/Support/Debug.h" +#include "llvm/Support/raw_ostream.h" +#include <vector> +using namespace llvm; + +/// ContractNodes - Turn multiple matcher node patterns like 'MoveChild+Record' +/// into single compound nodes like RecordChild. +static void ContractNodes(OwningPtr<Matcher> &MatcherPtr, + const CodeGenDAGPatterns &CGP) { + // If we reached the end of the chain, we're done. + Matcher *N = MatcherPtr.get(); + if (N == 0) return; + + // If we have a scope node, walk down all of the children. + if (ScopeMatcher *Scope = dyn_cast<ScopeMatcher>(N)) { + for (unsigned i = 0, e = Scope->getNumChildren(); i != e; ++i) { + OwningPtr<Matcher> Child(Scope->takeChild(i)); + ContractNodes(Child, CGP); + Scope->resetChild(i, Child.take()); + } + return; + } + + // If we found a movechild node with a node that comes in a 'foochild' form, + // transform it. + if (MoveChildMatcher *MC = dyn_cast<MoveChildMatcher>(N)) { + Matcher *New = 0; + if (RecordMatcher *RM = dyn_cast<RecordMatcher>(MC->getNext())) + New = new RecordChildMatcher(MC->getChildNo(), RM->getWhatFor(), + RM->getResultNo()); + + if (CheckTypeMatcher *CT= dyn_cast<CheckTypeMatcher>(MC->getNext())) + New = new CheckChildTypeMatcher(MC->getChildNo(), CT->getType()); + + if (New) { + // Insert the new node. + New->setNext(MatcherPtr.take()); + MatcherPtr.reset(New); + // Remove the old one. + MC->setNext(MC->getNext()->takeNext()); + return ContractNodes(MatcherPtr, CGP); + } + } + + // Zap movechild -> moveparent. + if (MoveChildMatcher *MC = dyn_cast<MoveChildMatcher>(N)) + if (MoveParentMatcher *MP = + dyn_cast<MoveParentMatcher>(MC->getNext())) { + MatcherPtr.reset(MP->takeNext()); + return ContractNodes(MatcherPtr, CGP); + } + + // Turn EmitNode->MarkFlagResults->CompleteMatch into + // MarkFlagResults->EmitNode->CompleteMatch when we can to encourage + // MorphNodeTo formation. This is safe because MarkFlagResults never refers + // to the root of the pattern. + if (isa<EmitNodeMatcher>(N) && isa<MarkFlagResultsMatcher>(N->getNext()) && + isa<CompleteMatchMatcher>(N->getNext()->getNext())) { + // Unlink the two nodes from the list. + Matcher *EmitNode = MatcherPtr.take(); + Matcher *MFR = EmitNode->takeNext(); + Matcher *Tail = MFR->takeNext(); + + // Relink them. + MatcherPtr.reset(MFR); + MFR->setNext(EmitNode); + EmitNode->setNext(Tail); + return ContractNodes(MatcherPtr, CGP); + } + + // Turn EmitNode->CompleteMatch into MorphNodeTo if we can. + if (EmitNodeMatcher *EN = dyn_cast<EmitNodeMatcher>(N)) + if (CompleteMatchMatcher *CM = + dyn_cast<CompleteMatchMatcher>(EN->getNext())) { + // We can only use MorphNodeTo if the result values match up. + unsigned RootResultFirst = EN->getFirstResultSlot(); + bool ResultsMatch = true; + for (unsigned i = 0, e = CM->getNumResults(); i != e; ++i) + if (CM->getResult(i) != RootResultFirst+i) + ResultsMatch = false; + + // If the selected node defines a subset of the flag/chain results, we + // can't use MorphNodeTo. For example, we can't use MorphNodeTo if the + // matched pattern has a chain but the root node doesn't. + const PatternToMatch &Pattern = CM->getPattern(); + + if (!EN->hasChain() && + Pattern.getSrcPattern()->NodeHasProperty(SDNPHasChain, CGP)) + ResultsMatch = false; + + // If the matched node has a flag and the output root doesn't, we can't + // use MorphNodeTo. + // + // NOTE: Strictly speaking, we don't have to check for the flag here + // because the code in the pattern generator doesn't handle it right. We + // do it anyway for thoroughness. + if (!EN->hasOutFlag() && + Pattern.getSrcPattern()->NodeHasProperty(SDNPOutFlag, CGP)) + ResultsMatch = false; + + + // If the root result node defines more results than the source root node + // *and* has a chain or flag input, then we can't match it because it + // would end up replacing the extra result with the chain/flag. +#if 0 + if ((EN->hasFlag() || EN->hasChain()) && + EN->getNumNonChainFlagVTs() > ... need to get no results reliably ...) + ResultMatch = false; +#endif + + if (ResultsMatch) { + const SmallVectorImpl<MVT::SimpleValueType> &VTs = EN->getVTList(); + const SmallVectorImpl<unsigned> &Operands = EN->getOperandList(); + MatcherPtr.reset(new MorphNodeToMatcher(EN->getOpcodeName(), + VTs.data(), VTs.size(), + Operands.data(),Operands.size(), + EN->hasChain(), EN->hasInFlag(), + EN->hasOutFlag(), + EN->hasMemRefs(), + EN->getNumFixedArityOperands(), + Pattern)); + return; + } + + // FIXME2: Kill off all the SelectionDAG::SelectNodeTo and getMachineNode + // variants. + } + + ContractNodes(N->getNextPtr(), CGP); + + + // If we have a CheckType/CheckChildType/Record node followed by a + // CheckOpcode, invert the two nodes. We prefer to do structural checks + // before type checks, as this opens opportunities for factoring on targets + // like X86 where many operations are valid on multiple types. + if ((isa<CheckTypeMatcher>(N) || isa<CheckChildTypeMatcher>(N) || + isa<RecordMatcher>(N)) && + isa<CheckOpcodeMatcher>(N->getNext())) { + // Unlink the two nodes from the list. + Matcher *CheckType = MatcherPtr.take(); + Matcher *CheckOpcode = CheckType->takeNext(); + Matcher *Tail = CheckOpcode->takeNext(); + + // Relink them. + MatcherPtr.reset(CheckOpcode); + CheckOpcode->setNext(CheckType); + CheckType->setNext(Tail); + return ContractNodes(MatcherPtr, CGP); + } +} + +/// SinkPatternPredicates - Pattern predicates can be checked at any level of +/// the matching tree. The generator dumps them at the top level of the pattern +/// though, which prevents factoring from being able to see past them. This +/// optimization sinks them as far down into the pattern as possible. +/// +/// Conceptually, we'd like to sink these predicates all the way to the last +/// matcher predicate in the series. However, it turns out that some +/// ComplexPatterns have side effects on the graph, so we really don't want to +/// run a the complex pattern if the pattern predicate will fail. For this +/// reason, we refuse to sink the pattern predicate past a ComplexPattern. +/// +static void SinkPatternPredicates(OwningPtr<Matcher> &MatcherPtr) { + // Recursively scan for a PatternPredicate. + // If we reached the end of the chain, we're done. + Matcher *N = MatcherPtr.get(); + if (N == 0) return; + + // Walk down all members of a scope node. + if (ScopeMatcher *Scope = dyn_cast<ScopeMatcher>(N)) { + for (unsigned i = 0, e = Scope->getNumChildren(); i != e; ++i) { + OwningPtr<Matcher> Child(Scope->takeChild(i)); + SinkPatternPredicates(Child); + Scope->resetChild(i, Child.take()); + } + return; + } + + // If this node isn't a CheckPatternPredicateMatcher we keep scanning until + // we find one. + CheckPatternPredicateMatcher *CPPM =dyn_cast<CheckPatternPredicateMatcher>(N); + if (CPPM == 0) + return SinkPatternPredicates(N->getNextPtr()); + + // Ok, we found one, lets try to sink it. Check if we can sink it past the + // next node in the chain. If not, we won't be able to change anything and + // might as well bail. + if (!CPPM->getNext()->isSafeToReorderWithPatternPredicate()) + return; + + // Okay, we know we can sink it past at least one node. Unlink it from the + // chain and scan for the new insertion point. + MatcherPtr.take(); // Don't delete CPPM. + MatcherPtr.reset(CPPM->takeNext()); + + N = MatcherPtr.get(); + while (N->getNext()->isSafeToReorderWithPatternPredicate()) + N = N->getNext(); + + // At this point, we want to insert CPPM after N. + CPPM->setNext(N->takeNext()); + N->setNext(CPPM); +} + +/// FactorNodes - Turn matches like this: +/// Scope +/// OPC_CheckType i32 +/// ABC +/// OPC_CheckType i32 +/// XYZ +/// into: +/// OPC_CheckType i32 +/// Scope +/// ABC +/// XYZ +/// +static void FactorNodes(OwningPtr<Matcher> &MatcherPtr) { + // If we reached the end of the chain, we're done. + Matcher *N = MatcherPtr.get(); + if (N == 0) return; + + // If this is not a push node, just scan for one. + ScopeMatcher *Scope = dyn_cast<ScopeMatcher>(N); + if (Scope == 0) + return FactorNodes(N->getNextPtr()); + + // Okay, pull together the children of the scope node into a vector so we can + // inspect it more easily. While we're at it, bucket them up by the hash + // code of their first predicate. + SmallVector<Matcher*, 32> OptionsToMatch; + + for (unsigned i = 0, e = Scope->getNumChildren(); i != e; ++i) { + // Factor the subexpression. + OwningPtr<Matcher> Child(Scope->takeChild(i)); + FactorNodes(Child); + + if (Matcher *N = Child.take()) + OptionsToMatch.push_back(N); + } + + SmallVector<Matcher*, 32> NewOptionsToMatch; + + // Loop over options to match, merging neighboring patterns with identical + // starting nodes into a shared matcher. + for (unsigned OptionIdx = 0, e = OptionsToMatch.size(); OptionIdx != e;) { + // Find the set of matchers that start with this node. + Matcher *Optn = OptionsToMatch[OptionIdx++]; + + if (OptionIdx == e) { + NewOptionsToMatch.push_back(Optn); + continue; + } + + // See if the next option starts with the same matcher. If the two + // neighbors *do* start with the same matcher, we can factor the matcher out + // of at least these two patterns. See what the maximal set we can merge + // together is. + SmallVector<Matcher*, 8> EqualMatchers; + EqualMatchers.push_back(Optn); + + // Factor all of the known-equal matchers after this one into the same + // group. + while (OptionIdx != e && OptionsToMatch[OptionIdx]->isEqual(Optn)) + EqualMatchers.push_back(OptionsToMatch[OptionIdx++]); + + // If we found a non-equal matcher, see if it is contradictory with the + // current node. If so, we know that the ordering relation between the + // current sets of nodes and this node don't matter. Look past it to see if + // we can merge anything else into this matching group. + unsigned Scan = OptionIdx; + while (1) { + while (Scan != e && Optn->isContradictory(OptionsToMatch[Scan])) + ++Scan; + + // Ok, we found something that isn't known to be contradictory. If it is + // equal, we can merge it into the set of nodes to factor, if not, we have + // to cease factoring. + if (Scan == e || !Optn->isEqual(OptionsToMatch[Scan])) break; + + // If is equal after all, add the option to EqualMatchers and remove it + // from OptionsToMatch. + EqualMatchers.push_back(OptionsToMatch[Scan]); + OptionsToMatch.erase(OptionsToMatch.begin()+Scan); + --e; + } + + if (Scan != e && + // Don't print it's obvious nothing extra could be merged anyway. + Scan+1 != e) { + DEBUG(errs() << "Couldn't merge this:\n"; + Optn->print(errs(), 4); + errs() << "into this:\n"; + OptionsToMatch[Scan]->print(errs(), 4); + if (Scan+1 != e) + OptionsToMatch[Scan+1]->printOne(errs()); + if (Scan+2 < e) + OptionsToMatch[Scan+2]->printOne(errs()); + errs() << "\n"); + } + + // If we only found one option starting with this matcher, no factoring is + // possible. + if (EqualMatchers.size() == 1) { + NewOptionsToMatch.push_back(EqualMatchers[0]); + continue; + } + + // Factor these checks by pulling the first node off each entry and + // discarding it. Take the first one off the first entry to reuse. + Matcher *Shared = Optn; + Optn = Optn->takeNext(); + EqualMatchers[0] = Optn; + + // Remove and delete the first node from the other matchers we're factoring. + for (unsigned i = 1, e = EqualMatchers.size(); i != e; ++i) { + Matcher *Tmp = EqualMatchers[i]->takeNext(); + delete EqualMatchers[i]; + EqualMatchers[i] = Tmp; + } + + Shared->setNext(new ScopeMatcher(&EqualMatchers[0], EqualMatchers.size())); + + // Recursively factor the newly created node. + FactorNodes(Shared->getNextPtr()); + + NewOptionsToMatch.push_back(Shared); + } + + // If we're down to a single pattern to match, then we don't need this scope + // anymore. + if (NewOptionsToMatch.size() == 1) { + MatcherPtr.reset(NewOptionsToMatch[0]); + return; + } + + if (NewOptionsToMatch.empty()) { + MatcherPtr.reset(0); + return; + } + + // If our factoring failed (didn't achieve anything) see if we can simplify in + // other ways. + + // Check to see if all of the leading entries are now opcode checks. If so, + // we can convert this Scope to be a OpcodeSwitch instead. + bool AllOpcodeChecks = true, AllTypeChecks = true; + for (unsigned i = 0, e = NewOptionsToMatch.size(); i != e; ++i) { + if (!isa<CheckOpcodeMatcher>(NewOptionsToMatch[i])) { +#if 0 + if (i > 3 && AllOpcodeChecks) { + errs() << "FAILING OPC #" << i << "\n"; + NewOptionsToMatch[i]->dump(); + } +#endif + AllOpcodeChecks = false; + } + + if (!isa<CheckTypeMatcher>(NewOptionsToMatch[i]) || + // iPTR checks could alias any other case without us knowing, don't + // bother with them. + cast<CheckTypeMatcher>(NewOptionsToMatch[i])->getType() == MVT::iPTR) { +#if 0 + if (i > 3 && AllTypeChecks) { + errs() << "FAILING TYPE #" << i << "\n"; + NewOptionsToMatch[i]->dump(); + } +#endif + AllTypeChecks = false; + } + } + // TODO: Can also do CheckChildNType. + + // If all the options are CheckOpcode's, we can form the SwitchOpcode, woot. + if (AllOpcodeChecks) { + StringSet<> Opcodes; + SmallVector<std::pair<const SDNodeInfo*, Matcher*>, 8> Cases; + for (unsigned i = 0, e = NewOptionsToMatch.size(); i != e; ++i) { + CheckOpcodeMatcher *COM = cast<CheckOpcodeMatcher>(NewOptionsToMatch[i]); + assert(Opcodes.insert(COM->getOpcode().getEnumName()) && + "Duplicate opcodes not factored?"); + Cases.push_back(std::make_pair(&COM->getOpcode(), COM->getNext())); + } + + MatcherPtr.reset(new SwitchOpcodeMatcher(&Cases[0], Cases.size())); + return; + } + + // If all the options are CheckType's, we can form the SwitchType, woot. + if (AllTypeChecks) { + DenseSet<unsigned> Types; + SmallVector<std::pair<MVT::SimpleValueType, Matcher*>, 8> Cases; + for (unsigned i = 0, e = NewOptionsToMatch.size(); i != e; ++i) { + CheckTypeMatcher *CTM = cast<CheckTypeMatcher>(NewOptionsToMatch[i]); + assert(Types.insert(CTM->getType()).second && + "Duplicate types not factored?"); + Cases.push_back(std::make_pair(CTM->getType(), CTM->getNext())); + } + + MatcherPtr.reset(new SwitchTypeMatcher(&Cases[0], Cases.size())); + return; + } + + + // Reassemble the Scope node with the adjusted children. + Scope->setNumChildren(NewOptionsToMatch.size()); + for (unsigned i = 0, e = NewOptionsToMatch.size(); i != e; ++i) + Scope->resetChild(i, NewOptionsToMatch[i]); +} + +Matcher *llvm::OptimizeMatcher(Matcher *TheMatcher, + const CodeGenDAGPatterns &CGP) { + OwningPtr<Matcher> MatcherPtr(TheMatcher); + ContractNodes(MatcherPtr, CGP); + SinkPatternPredicates(MatcherPtr); + FactorNodes(MatcherPtr); + return MatcherPtr.take(); +} diff --git a/utils/TableGen/InstrEnumEmitter.cpp b/utils/TableGen/InstrEnumEmitter.cpp index f382d34..d1e7f3d 100644 --- a/utils/TableGen/InstrEnumEmitter.cpp +++ b/utils/TableGen/InstrEnumEmitter.cpp @@ -29,7 +29,7 @@ void InstrEnumEmitter::run(raw_ostream &OS) { std::string Namespace; for (CodeGenTarget::inst_iterator II = Target.inst_begin(), E = Target.inst_end(); II != E; ++II) { - if (II->second.Namespace != "TargetInstrInfo") { + if (II->second.Namespace != "TargetOpcode") { Namespace = II->second.Namespace; break; } diff --git a/utils/TableGen/LLVMCConfigurationEmitter.cpp b/utils/TableGen/LLVMCConfigurationEmitter.cpp index 2abc94b..db59053 100644 --- a/utils/TableGen/LLVMCConfigurationEmitter.cpp +++ b/utils/TableGen/LLVMCConfigurationEmitter.cpp @@ -43,6 +43,7 @@ const unsigned TabWidth = 4; const unsigned Indent1 = TabWidth*1; const unsigned Indent2 = TabWidth*2; const unsigned Indent3 = TabWidth*3; +const unsigned Indent4 = TabWidth*4; // Default help string. const char * const DefaultHelpString = "NO HELP MESSAGE PROVIDED"; @@ -229,7 +230,7 @@ namespace OptionDescriptionFlags { enum OptionDescriptionFlags { Required = 0x1, Hidden = 0x2, ReallyHidden = 0x4, Extern = 0x8, OneOrMore = 0x10, Optional = 0x20, - CommaSeparated = 0x40 }; + CommaSeparated = 0x40, ForwardNotSplit = 0x80 }; } /// OptionDescription - Represents data contained in a single @@ -271,6 +272,9 @@ struct OptionDescription { bool isExtern() const; void setExtern(); + bool isForwardNotSplit() const; + void setForwardNotSplit(); + bool isRequired() const; void setRequired(); @@ -327,6 +331,13 @@ void OptionDescription::setCommaSeparated() { Flags |= OptionDescriptionFlags::CommaSeparated; } +bool OptionDescription::isForwardNotSplit() const { + return Flags & OptionDescriptionFlags::ForwardNotSplit; +} +void OptionDescription::setForwardNotSplit() { + Flags |= OptionDescriptionFlags::ForwardNotSplit; +} + bool OptionDescription::isExtern() const { return Flags & OptionDescriptionFlags::Extern; } @@ -586,6 +597,8 @@ public: AddHandler("required", &CollectOptionProperties::onRequired); AddHandler("optional", &CollectOptionProperties::onOptional); AddHandler("comma_separated", &CollectOptionProperties::onCommaSeparated); + AddHandler("forward_not_split", + &CollectOptionProperties::onForwardNotSplit); staticMembersInitialized_ = true; } @@ -629,6 +642,13 @@ private: optDesc_.setCommaSeparated(); } + void onForwardNotSplit (const DagInit& d) { + CheckNumberOfArguments(d, 0); + if (!optDesc_.isParameter()) + throw "'forward_not_split' is valid only for parameter options!"; + optDesc_.setForwardNotSplit(); + } + void onRequired (const DagInit& d) { CheckNumberOfArguments(d, 0); if (optDesc_.isOneOrMore() || optDesc_.isOptional()) @@ -761,9 +781,12 @@ struct ToolDescription : public RefCountedBase<ToolDescription> { Init* CmdLine; Init* Actions; StrVector InLanguage; + std::string InFileOption; + std::string OutFileOption; std::string OutLanguage; std::string OutputSuffix; unsigned Flags; + const Init* OnEmpty; // Various boolean properties void setSink() { Flags |= ToolFlags::Sink; } @@ -773,9 +796,13 @@ struct ToolDescription : public RefCountedBase<ToolDescription> { // Default ctor here is needed because StringMap can only store // DefaultConstructible objects - ToolDescription() : CmdLine(0), Actions(0), Flags(0) {} + ToolDescription () + : CmdLine(0), Actions(0), OutFileOption("-o"), + Flags(0), OnEmpty(0) + {} ToolDescription (const std::string& n) - : Name(n), CmdLine(0), Actions(0), Flags(0) + : Name(n), CmdLine(0), Actions(0), OutFileOption("-o"), + Flags(0), OnEmpty(0) {} }; @@ -806,12 +833,17 @@ public: if (!staticMembersInitialized_) { AddHandler("actions", &CollectToolProperties::onActions); - AddHandler("cmd_line", &CollectToolProperties::onCmdLine); + AddHandler("command", &CollectToolProperties::onCommand); AddHandler("in_language", &CollectToolProperties::onInLanguage); AddHandler("join", &CollectToolProperties::onJoin); AddHandler("out_language", &CollectToolProperties::onOutLanguage); + + AddHandler("out_file_option", &CollectToolProperties::onOutFileOption); + AddHandler("in_file_option", &CollectToolProperties::onInFileOption); + AddHandler("output_suffix", &CollectToolProperties::onOutputSuffix); AddHandler("sink", &CollectToolProperties::onSink); + AddHandler("works_on_empty", &CollectToolProperties::onWorksOnEmpty); staticMembersInitialized_ = true; } @@ -836,7 +868,7 @@ private: toolDesc_.Actions = Case; } - void onCmdLine (const DagInit& d) { + void onCommand (const DagInit& d) { CheckNumberOfArguments(d, 1); toolDesc_.CmdLine = d.getArg(0); } @@ -878,6 +910,16 @@ private: toolDesc_.OutLanguage = InitPtrToString(d.getArg(0)); } + void onOutFileOption (const DagInit& d) { + CheckNumberOfArguments(d, 1); + toolDesc_.OutFileOption = InitPtrToString(d.getArg(0)); + } + + void onInFileOption (const DagInit& d) { + CheckNumberOfArguments(d, 1); + toolDesc_.InFileOption = InitPtrToString(d.getArg(0)); + } + void onOutputSuffix (const DagInit& d) { CheckNumberOfArguments(d, 1); toolDesc_.OutputSuffix = InitPtrToString(d.getArg(0)); @@ -888,6 +930,10 @@ private: toolDesc_.setSink(); } + void onWorksOnEmpty (const DagInit& d) { + toolDesc_.OnEmpty = d.getArg(0); + } + }; /// CollectToolDescriptions - Gather information about tool properties @@ -1490,7 +1536,7 @@ public: /// EmitCaseConstructHandler - Emit code that handles the 'case' /// construct. Takes a function object that should emit code for every case /// clause. Implemented on top of WalkCase. -/// Callback's type is void F(Init* Statement, unsigned IndentLevel, +/// Callback's type is void F(const Init* Statement, unsigned IndentLevel, /// raw_ostream& O). /// EmitElseIf parameter controls the type of condition that is emitted ('if /// (..) {..} else if (..) {} .. else {..}' vs. 'if (..) {..} if(..) {..} @@ -1699,82 +1745,41 @@ void EmitCmdLineVecFill(const Init* CmdLine, const std::string& ToolName, if (StrVec.empty()) throw "Tool '" + ToolName + "' has empty command line!"; - StrVector::const_iterator I = StrVec.begin(), E = StrVec.end(); + StrVector::const_iterator B = StrVec.begin(), E = StrVec.end(); - // If there is a hook invocation on the place of the first command, skip it. + // Emit the command itself. assert(!StrVec[0].empty()); + O.indent(IndentLevel) << "cmd = "; if (StrVec[0][0] == '$') { - while (I != E && (*I)[0] != ')' ) - ++I; - - // Skip the ')' symbol. - ++I; + B = SubstituteSpecialCommands(B, E, IsJoin, O); + ++B; } else { - ++I; + O << '"' << StrVec[0] << '"'; + ++B; } + O << ";\n"; - bool hasINFILE = false; + // Go through the command arguments. + assert(B <= E); + for (; B != E; ++B) { + const std::string& cmd = *B; - for (; I != E; ++I) { - const std::string& cmd = *I; assert(!cmd.empty()); O.indent(IndentLevel); + if (cmd.at(0) == '$') { - if (cmd == "$INFILE") { - hasINFILE = true; - if (IsJoin) { - O << "for (PathVector::const_iterator B = inFiles.begin()" - << ", E = inFiles.end();\n"; - O.indent(IndentLevel) << "B != E; ++B)\n"; - O.indent(IndentLevel + Indent1) << "vec.push_back(B->str());\n"; - } - else { - O << "vec.push_back(inFile.str());\n"; - } - } - else if (cmd == "$OUTFILE") { - O << "vec.push_back(\"\");\n"; - O.indent(IndentLevel) << "out_file_index = vec.size()-1;\n"; - } - else { - O << "vec.push_back("; - I = SubstituteSpecialCommands(I, E, IsJoin, O); - O << ");\n"; - } + O << "vec.push_back(std::make_pair(0, "; + B = SubstituteSpecialCommands(B, E, IsJoin, O); + O << "));\n"; } else { - O << "vec.push_back(\"" << cmd << "\");\n"; + O << "vec.push_back(std::make_pair(0, \"" << cmd << "\"));\n"; } } - if (!hasINFILE) - throw "Tool '" + ToolName + "' doesn't take any input!"; - O.indent(IndentLevel) << "cmd = "; - if (StrVec[0][0] == '$') - SubstituteSpecialCommands(StrVec.begin(), StrVec.end(), IsJoin, O); - else - O << '"' << StrVec[0] << '"'; - O << ";\n"; } -/// EmitCmdLineVecFillCallback - A function object wrapper around -/// EmitCmdLineVecFill(). Used by EmitGenerateActionMethod() as an -/// argument to EmitCaseConstructHandler(). -class EmitCmdLineVecFillCallback { - bool IsJoin; - const std::string& ToolName; - public: - EmitCmdLineVecFillCallback(bool J, const std::string& TN) - : IsJoin(J), ToolName(TN) {} - - void operator()(const Init* Statement, unsigned IndentLevel, - raw_ostream& O) const - { - EmitCmdLineVecFill(Statement, ToolName, IsJoin, IndentLevel, O); - } -}; - /// EmitForwardOptionPropertyHandlingCode - Helper function used to /// implement EmitActionHandler. Emits code for /// handling the (forward) and (forward_as) option properties. @@ -1789,15 +1794,30 @@ void EmitForwardOptionPropertyHandlingCode (const OptionDescription& D, switch (D.Type) { case OptionType::Switch: - O.indent(IndentLevel) << "vec.push_back(\"" << Name << "\");\n"; + O.indent(IndentLevel) + << "vec.push_back(std::make_pair(" << D.GenVariableName() + << ".getPosition(), \"" << Name << "\"));\n"; break; case OptionType::Parameter: - O.indent(IndentLevel) << "vec.push_back(\"" << Name << "\");\n"; - O.indent(IndentLevel) << "vec.push_back(" << D.GenVariableName() << ");\n"; + O.indent(IndentLevel) << "vec.push_back(std::make_pair(" + << D.GenVariableName() + <<".getPosition(), \"" << Name; + + if (!D.isForwardNotSplit()) { + O << "\"));\n"; + O.indent(IndentLevel) << "vec.push_back(std::make_pair(" + << D.GenVariableName() << ".getPosition(), " + << D.GenVariableName() << "));\n"; + } + else { + O << "=\" + " << D.GenVariableName() << "));\n"; + } break; case OptionType::Prefix: - O.indent(IndentLevel) << "vec.push_back(\"" << Name << "\" + " - << D.GenVariableName() << ");\n"; + O.indent(IndentLevel) << "vec.push_back(std::make_pair(" + << D.GenVariableName() << ".getPosition(), \"" + << Name << "\" + " + << D.GenVariableName() << "));\n"; break; case OptionType::PrefixList: O.indent(IndentLevel) @@ -1805,11 +1825,15 @@ void EmitForwardOptionPropertyHandlingCode (const OptionDescription& D, << "::iterator B = " << D.GenVariableName() << ".begin(),\n"; O.indent(IndentLevel) << "E = " << D.GenVariableName() << ".end(); B != E;) {\n"; - O.indent(IndentLevel1) << "vec.push_back(\"" << Name << "\" + " << "*B);\n"; + O.indent(IndentLevel1) << "unsigned pos = " << D.GenVariableName() + << ".getPosition(B - " << D.GenVariableName() + << ".begin());\n"; + O.indent(IndentLevel1) << "vec.push_back(std::make_pair(pos, \"" + << Name << "\" + " << "*B));\n"; O.indent(IndentLevel1) << "++B;\n"; for (int i = 1, j = D.MultiVal; i < j; ++i) { - O.indent(IndentLevel1) << "vec.push_back(*B);\n"; + O.indent(IndentLevel1) << "vec.push_back(std::make_pair(pos, *B));\n"; O.indent(IndentLevel1) << "++B;\n"; } @@ -1821,10 +1845,14 @@ void EmitForwardOptionPropertyHandlingCode (const OptionDescription& D, << D.GenVariableName() << ".begin(),\n"; O.indent(IndentLevel) << "E = " << D.GenVariableName() << ".end() ; B != E;) {\n"; - O.indent(IndentLevel1) << "vec.push_back(\"" << Name << "\");\n"; + O.indent(IndentLevel1) << "unsigned pos = " << D.GenVariableName() + << ".getPosition(B - " << D.GenVariableName() + << ".begin());\n"; + O.indent(IndentLevel1) << "vec.push_back(std::make_pair(pos, \"" + << Name << "\"));\n"; for (int i = 0, j = D.MultiVal; i < j; ++i) { - O.indent(IndentLevel1) << "vec.push_back(*B);\n"; + O.indent(IndentLevel1) << "vec.push_back(std::make_pair(pos, *B));\n"; O.indent(IndentLevel1) << "++B;\n"; } @@ -1906,7 +1934,8 @@ class EmitActionHandlersCallback : { CheckNumberOfArguments(Dag, 1); this->EmitHookInvocation(InitPtrToString(Dag.getArg(0)), - "vec.push_back(", ");\n", IndentLevel, O); + "vec.push_back(std::make_pair(65536, ", "));\n", + IndentLevel, O); } void onForward (const DagInit& Dag, @@ -1936,13 +1965,23 @@ class EmitActionHandlersCallback : const OptionDescription& D = OptDescs.FindListOrParameter(Name); if (D.isParameter()) { - O.indent(IndentLevel) << "vec.push_back(" - << D.GenVariableName() << ");\n"; + O.indent(IndentLevel) << "vec.push_back(std::make_pair(" + << D.GenVariableName() << ".getPosition(), " + << D.GenVariableName() << "));\n"; } else { - O.indent(IndentLevel) << "std::copy(" << D.GenVariableName() - << ".begin(), " << D.GenVariableName() - << ".end(), std::back_inserter(vec));\n"; + O.indent(IndentLevel) << "for (cl::list<std::string>::iterator B = " + << D.GenVariableName() << ".begin(), \n"; + O.indent(IndentLevel + Indent1) << " E = " << D.GenVariableName() + << ".end(); B != E; ++B)\n"; + O.indent(IndentLevel) << "{\n"; + O.indent(IndentLevel + Indent1) + << "unsigned pos = " << D.GenVariableName() + << ".getPosition(B - " << D.GenVariableName() + << ".begin());\n"; + O.indent(IndentLevel + Indent1) + << "vec.push_back(std::make_pair(pos, *B));\n"; + O.indent(IndentLevel) << "}\n"; } } @@ -1954,9 +1993,18 @@ class EmitActionHandlersCallback : const std::string& Hook = InitPtrToString(Dag.getArg(1)); const OptionDescription& D = OptDescs.FindListOrParameter(Name); - O.indent(IndentLevel) << "vec.push_back(" << "hooks::" - << Hook << "(" << D.GenVariableName() - << (D.isParameter() ? ".c_str()" : "") << "));\n"; + O.indent(IndentLevel) << "vec.push_back(std::make_pair(" + << D.GenVariableName() << ".getPosition(" + << (D.isList() ? "0" : "") << "), " + << "hooks::" << Hook << "(" << D.GenVariableName() + << (D.isParameter() ? ".c_str()" : "") << ")));\n"; + } + + void onNoOutFile (const DagInit& Dag, + unsigned IndentLevel, raw_ostream& O) const + { + CheckNumberOfArguments(Dag, 0); + O.indent(IndentLevel) << "no_out_file = true;\n"; } void onOutputSuffix (const DagInit& Dag, @@ -1995,12 +2043,15 @@ class EmitActionHandlersCallback : AddHandler("forward_value", &EmitActionHandlersCallback::onForwardValue); AddHandler("forward_transformed_value", &EmitActionHandlersCallback::onForwardTransformedValue); + AddHandler("no_out_file", + &EmitActionHandlersCallback::onNoOutFile); AddHandler("output_suffix", &EmitActionHandlersCallback::onOutputSuffix); AddHandler("stop_compilation", &EmitActionHandlersCallback::onStopCompilation); AddHandler("unpack_values", &EmitActionHandlersCallback::onUnpackValues); + staticMembersInitialized_ = true; } } @@ -2012,58 +2063,6 @@ class EmitActionHandlersCallback : } }; -bool IsOutFileIndexCheckRequiredStr (const Init* CmdLine) { - StrVector StrVec; - TokenizeCmdLine(InitPtrToString(CmdLine), StrVec); - - for (StrVector::const_iterator I = StrVec.begin(), E = StrVec.end(); - I != E; ++I) { - if (*I == "$OUTFILE") - return false; - } - - return true; -} - -class IsOutFileIndexCheckRequiredStrCallback { - bool* ret_; - -public: - IsOutFileIndexCheckRequiredStrCallback(bool* ret) : ret_(ret) - {} - - void operator()(const Init* CmdLine) { - // Ignore nested 'case' DAG. - if (typeid(*CmdLine) == typeid(DagInit)) - return; - - if (IsOutFileIndexCheckRequiredStr(CmdLine)) - *ret_ = true; - } - - void operator()(const DagInit* Test, unsigned, bool) { - this->operator()(Test); - } - void operator()(const Init* Statement, unsigned) { - this->operator()(Statement); - } -}; - -bool IsOutFileIndexCheckRequiredCase (Init* CmdLine) { - bool ret = false; - WalkCase(CmdLine, Id(), IsOutFileIndexCheckRequiredStrCallback(&ret)); - return ret; -} - -/// IsOutFileIndexCheckRequired - Should we emit an "out_file_index != -1" check -/// in EmitGenerateActionMethod() ? -bool IsOutFileIndexCheckRequired (Init* CmdLine) { - if (typeid(*CmdLine) == typeid(StringInit)) - return IsOutFileIndexCheckRequiredStr(CmdLine); - else - return IsOutFileIndexCheckRequiredCase(CmdLine); -} - void EmitGenerateActionMethodHeader(const ToolDescription& D, bool IsJoin, raw_ostream& O) { @@ -2078,8 +2077,10 @@ void EmitGenerateActionMethodHeader(const ToolDescription& D, O.indent(Indent2) << "const LanguageMap& LangMap) const\n"; O.indent(Indent1) << "{\n"; O.indent(Indent2) << "std::string cmd;\n"; - O.indent(Indent2) << "std::vector<std::string> vec;\n"; + O.indent(Indent2) << "std::string out_file;\n"; + O.indent(Indent2) << "std::vector<std::pair<unsigned, std::string> > vec;\n"; O.indent(Indent2) << "bool stop_compilation = !HasChildren;\n"; + O.indent(Indent2) << "bool no_out_file = false;\n"; O.indent(Indent2) << "const char* output_suffix = \"" << D.OutputSuffix << "\";\n"; } @@ -2095,46 +2096,67 @@ void EmitGenerateActionMethod (const ToolDescription& D, if (!D.CmdLine) throw "Tool " + D.Name + " has no cmd_line property!"; - bool IndexCheckRequired = IsOutFileIndexCheckRequired(D.CmdLine); - O.indent(Indent2) << "int out_file_index" - << (IndexCheckRequired ? " = -1" : "") - << ";\n\n"; - - // Process the cmd_line property. - if (typeid(*D.CmdLine) == typeid(StringInit)) - EmitCmdLineVecFill(D.CmdLine, D.Name, IsJoin, Indent2, O); - else - EmitCaseConstructHandler(D.CmdLine, Indent2, - EmitCmdLineVecFillCallback(IsJoin, D.Name), - true, OptDescs, O); + // Process the 'command' property. + O << '\n'; + EmitCmdLineVecFill(D.CmdLine, D.Name, IsJoin, Indent2, O); + O << '\n'; - // For every understood option, emit handling code. + // Process the 'actions' list of this tool. if (D.Actions) EmitCaseConstructHandler(D.Actions, Indent2, EmitActionHandlersCallback(OptDescs), false, OptDescs, O); - O << '\n'; - O.indent(Indent2) - << "std::string out_file = OutFilename(" - << (IsJoin ? "sys::Path(),\n" : "inFile,\n"); - O.indent(Indent3) << "TempDir, stop_compilation, output_suffix).str();\n\n"; - if (IndexCheckRequired) - O.indent(Indent2) << "if (out_file_index != -1)\n"; - O.indent(IndexCheckRequired ? Indent3 : Indent2) - << "vec[out_file_index] = out_file;\n"; + // Input file (s) + if (!D.InFileOption.empty()) { + O.indent(Indent2) + << "vec.push_back(std::make_pair(InputFilenames.getPosition(0), \"" + << D.InFileOption << "\");\n"; + } + + if (IsJoin) { + O.indent(Indent2) + << "for (PathVector::const_iterator B = inFiles.begin(),\n"; + O.indent(Indent3) << "E = inFiles.end(); B != E; ++B)\n"; + O.indent(Indent2) << "{\n"; + O.indent(Indent3) << "vec.push_back(std::make_pair(" + << "InputFilenames.getPosition(B - inFiles.begin()), " + << "B->str()));\n"; + O.indent(Indent2) << "}\n"; + } + else { + O.indent(Indent2) << "vec.push_back(std::make_pair(" + << "InputFilenames.getPosition(0), inFile.str()));\n"; + } + + // Output file + O.indent(Indent2) << "if (!no_out_file) {\n"; + if (!D.OutFileOption.empty()) + O.indent(Indent3) << "vec.push_back(std::make_pair(65536, \"" + << D.OutFileOption << "\"));\n"; + + O.indent(Indent3) << "out_file = this->OutFilename(" + << (IsJoin ? "sys::Path(),\n" : "inFile,\n"); + O.indent(Indent4) << "TempDir, stop_compilation, output_suffix).str();\n\n"; + O.indent(Indent3) << "vec.push_back(std::make_pair(65536, out_file));\n"; + + O.indent(Indent2) << "}\n\n"; // Handle the Sink property. if (D.isSink()) { O.indent(Indent2) << "if (!" << SinkOptionName << ".empty()) {\n"; - O.indent(Indent3) << "vec.insert(vec.end(), " - << SinkOptionName << ".begin(), " << SinkOptionName - << ".end());\n"; + O.indent(Indent3) << "for (cl::list<std::string>::iterator B = " + << SinkOptionName << ".begin(), E = " << SinkOptionName + << ".end(); B != E; ++B)\n"; + O.indent(Indent4) << "vec.push_back(std::make_pair(" << SinkOptionName + << ".getPosition(B - " << SinkOptionName + << ".begin()), *B));\n"; O.indent(Indent2) << "}\n"; } - O.indent(Indent2) << "return Action(cmd, vec, stop_compilation, out_file);\n"; + O.indent(Indent2) << "return Action(cmd, this->SortArgs(vec), " + << "stop_compilation, out_file);\n"; O.indent(Indent1) << "}\n\n"; } @@ -2194,6 +2216,29 @@ void EmitIsJoinMethod (const ToolDescription& D, raw_ostream& O) { O.indent(Indent1) << "}\n\n"; } +/// EmitWorksOnEmptyCallback - Callback used by EmitWorksOnEmptyMethod in +/// conjunction with EmitCaseConstructHandler. +void EmitWorksOnEmptyCallback (const Init* Value, + unsigned IndentLevel, raw_ostream& O) { + CheckBooleanConstant(Value); + O.indent(IndentLevel) << "return " << Value->getAsString() << ";\n"; +} + +/// EmitWorksOnEmptyMethod - Emit the WorksOnEmpty() method for a given Tool +/// class. +void EmitWorksOnEmptyMethod (const ToolDescription& D, + const OptionDescriptions& OptDescs, + raw_ostream& O) +{ + O.indent(Indent1) << "bool WorksOnEmpty() const {\n"; + if (D.OnEmpty == 0) + O.indent(Indent2) << "return false;\n"; + else + EmitCaseConstructHandler(D.OnEmpty, Indent2, EmitWorksOnEmptyCallback, + /*EmitElseIf = */ true, OptDescs, O); + O.indent(Indent1) << "}\n\n"; +} + /// EmitStaticMemberDefinitions - Emit static member definitions for a /// given Tool class. void EmitStaticMemberDefinitions(const ToolDescription& D, raw_ostream& O) { @@ -2228,6 +2273,7 @@ void EmitToolClassDefinition (const ToolDescription& D, EmitNameMethod(D, O); EmitInOutLanguageMethods(D, O); EmitIsJoinMethod(D, O); + EmitWorksOnEmptyMethod(D, OptDescs, O); EmitGenerateActionMethods(D, OptDescs, O); // Close class definition @@ -2981,7 +3027,7 @@ void LLVMCConfigurationEmitter::run (raw_ostream &O) { CollectPluginData(Records, Data); CheckPluginData(Data); - EmitSourceFileHeader("LLVMC Configuration Library", O); + this->EmitSourceFileHeader("LLVMC Configuration Library", O); EmitPluginCode(Data, O); } catch (std::exception& Error) { diff --git a/utils/TableGen/Record.h b/utils/TableGen/Record.h index 45f3072..90096e9 100644 --- a/utils/TableGen/Record.h +++ b/utils/TableGen/Record.h @@ -1225,6 +1225,10 @@ public: ID(LastID++), Name(N), Loc(loc) {} ~Record() {} + + static unsigned getNewUID() { return LastID++; } + + unsigned getID() const { return ID; } const std::string &getName() const { return Name; } diff --git a/utils/TableGen/X86RecognizableInstr.cpp b/utils/TableGen/X86RecognizableInstr.cpp index 3843e56..ea78d41 100644 --- a/utils/TableGen/X86RecognizableInstr.cpp +++ b/utils/TableGen/X86RecognizableInstr.cpp @@ -282,6 +282,10 @@ RecognizableInstr::filter_ret RecognizableInstr::filter() const { IsCodeGenOnly) return FILTER_STRONG; + if (Form == X86Local::MRMInitReg) + return FILTER_STRONG; + + // Filter out instructions with a LOCK prefix; // prefer forms that do not have the prefix if (HasLockPrefix) @@ -353,9 +357,6 @@ RecognizableInstr::filter_ret RecognizableInstr::filter() const { if (AsmString.find("subreg") != AsmString.npos) return FILTER_STRONG; - assert(Form != X86Local::MRMInitReg && - "FORMAT_MRMINITREG instruction not skipped"); - if (HasFROperands && Name.find("MOV") != Name.npos && ((Name.find("2") != Name.npos && Name.find("32") == Name.npos) || (Name.find("to") != Name.npos))) diff --git a/utils/buildit/GNUmakefile b/utils/buildit/GNUmakefile index 57aac43..8d8504c 100644 --- a/utils/buildit/GNUmakefile +++ b/utils/buildit/GNUmakefile @@ -34,9 +34,9 @@ DSTROOT = $(OBJROOT)/../dst PREFIX = /usr/local -# Unless assertions are forced on in the GMAKE command line, enable them. +# Unless assertions are forced on in the GMAKE command line, disable them. ifndef ENABLE_ASSERTIONS -ENABLE_ASSERTIONS := yes +ENABLE_ASSERTIONS := no endif # Default is optimized build. diff --git a/utils/git/find-rev b/utils/git/find-rev new file mode 100755 index 0000000..a6161db --- /dev/null +++ b/utils/git/find-rev @@ -0,0 +1,50 @@ +#!/usr/bin/python + +import os, sys, subprocess + +def main(): + from optparse import OptionParser, OptionGroup + parser = OptionParser("usage: %prog [options] <repo> <revision>") + parser.add_option("", "--dump-section-data", dest="dumpSectionData", + help="Dump the contents of sections", + action="store_true", default=False) + (opts, args) = parser.parse_args() + + if len(args) != 2: + parser.error("invalid number of arguments") + + repo,rev = args + + try: + rev = int(rev) + except: + parser.error("invalid revision argument (not an integer)") + + os.chdir(repo) + p = subprocess.Popen(['git', 'rev-list', 'git-svn', '--pretty'], + stdout=subprocess.PIPE) + + bestRev = bestCommit = None + lastCommit = None + for ln in p.stdout: + if ln.startswith('commit '): + lastCommit = ln.split(' ',2)[1] + elif ln.startswith(' git-svn-id: '): + _,repo,_ = ln.strip().split(' ') + _,lrev = repo.rsplit('@',1) + lrev = int(lrev) + if lrev<=rev: + if bestRev is None or lrev>bestRev: + assert lastCommit + bestCommit = lastCommit + bestRev = lrev + if lrev == rev: + break + + if bestCommit is not None: + print bestCommit + sys.exit(0) + sys.exit(1) + +if __name__=='__main__': + main() diff --git a/utils/lit/lit/ExampleTests/LLVM.InTree/test/site.exp b/utils/lit/lit/ExampleTests/LLVM.InTree/test/site.exp index 1d9c743..efa839e 100644 --- a/utils/lit/lit/ExampleTests/LLVM.InTree/test/site.exp +++ b/utils/lit/lit/ExampleTests/LLVM.InTree/test/site.exp @@ -4,7 +4,6 @@ set target_triplet "x86_64-apple-darwin10" set TARGETS_TO_BUILD "X86 Sparc PowerPC Alpha ARM Mips CellSPU PIC16 XCore MSP430 SystemZ Blackfin CBackend MSIL CppBackend" set llvmgcc_langs "c,c++,objc,obj-c++" -set llvmgcc_version "4.2.1" set prcontext "/usr/bin/tclsh8.4 /Volumes/Data/ddunbar/llvm/test/Scripts/prcontext.tcl" set llvmtoolsdir "/Users/ddunbar/llvm.obj.64/Debug/bin" set llvmlibsdir "/Users/ddunbar/llvm.obj.64/Debug/lib" @@ -19,7 +18,6 @@ set compile_cxx " /usr/bin/g++ -arch x86_64 -I/Users/ddunbar/llvm.obj.64/include set link " /usr/bin/g++ -arch x86_64 -I/Users/ddunbar/llvm.obj.64/include -I/Users/ddunbar/llvm.obj.64/test -I/Volumes/Data/ddunbar/llvm.obj.64/include -I/Volumes/Data/ddunbar/llvm/include -I/Volumes/Data/ddunbar/llvm/test -D_DEBUG -D_GNU_SOURCE -D__STDC_LIMIT_MACROS -D__STDC_CONSTANT_MACROS -g -fno-exceptions -fno-common -Woverloaded-virtual -m64 -pedantic -Wno-long-long -Wall -W -Wno-unused-parameter -Wwrite-strings -g -L/Users/ddunbar/llvm.obj.64/Debug/lib -L/Volumes/Data/ddunbar/llvm.obj.64/Debug/lib " set llvmgcc "/Users/ddunbar/llvm-gcc/install/bin/llvm-gcc -m64 " set llvmgxx "/Users/ddunbar/llvm-gcc/install/bin/llvm-gcc -m64 " -set llvmgccmajvers "4" set bugpoint_topts "-gcc-tool-args -m64" set shlibext ".dylib" set ocamlopt "/sw/bin/ocamlopt -cc \"g++ -Wall -D_FILE_OFFSET_BITS=64 -D_REENTRANT\" -I /Users/ddunbar/llvm.obj.64/Debug/lib/ocaml" diff --git a/utils/lit/lit/ExampleTests/LLVM.OutOfTree/obj/test/site.exp b/utils/lit/lit/ExampleTests/LLVM.OutOfTree/obj/test/site.exp index 1d9c743..efa839e 100644 --- a/utils/lit/lit/ExampleTests/LLVM.OutOfTree/obj/test/site.exp +++ b/utils/lit/lit/ExampleTests/LLVM.OutOfTree/obj/test/site.exp @@ -4,7 +4,6 @@ set target_triplet "x86_64-apple-darwin10" set TARGETS_TO_BUILD "X86 Sparc PowerPC Alpha ARM Mips CellSPU PIC16 XCore MSP430 SystemZ Blackfin CBackend MSIL CppBackend" set llvmgcc_langs "c,c++,objc,obj-c++" -set llvmgcc_version "4.2.1" set prcontext "/usr/bin/tclsh8.4 /Volumes/Data/ddunbar/llvm/test/Scripts/prcontext.tcl" set llvmtoolsdir "/Users/ddunbar/llvm.obj.64/Debug/bin" set llvmlibsdir "/Users/ddunbar/llvm.obj.64/Debug/lib" @@ -19,7 +18,6 @@ set compile_cxx " /usr/bin/g++ -arch x86_64 -I/Users/ddunbar/llvm.obj.64/include set link " /usr/bin/g++ -arch x86_64 -I/Users/ddunbar/llvm.obj.64/include -I/Users/ddunbar/llvm.obj.64/test -I/Volumes/Data/ddunbar/llvm.obj.64/include -I/Volumes/Data/ddunbar/llvm/include -I/Volumes/Data/ddunbar/llvm/test -D_DEBUG -D_GNU_SOURCE -D__STDC_LIMIT_MACROS -D__STDC_CONSTANT_MACROS -g -fno-exceptions -fno-common -Woverloaded-virtual -m64 -pedantic -Wno-long-long -Wall -W -Wno-unused-parameter -Wwrite-strings -g -L/Users/ddunbar/llvm.obj.64/Debug/lib -L/Volumes/Data/ddunbar/llvm.obj.64/Debug/lib " set llvmgcc "/Users/ddunbar/llvm-gcc/install/bin/llvm-gcc -m64 " set llvmgxx "/Users/ddunbar/llvm-gcc/install/bin/llvm-gcc -m64 " -set llvmgccmajvers "4" set bugpoint_topts "-gcc-tool-args -m64" set shlibext ".dylib" set ocamlopt "/sw/bin/ocamlopt -cc \"g++ -Wall -D_FILE_OFFSET_BITS=64 -D_REENTRANT\" -I /Users/ddunbar/llvm.obj.64/Debug/lib/ocaml" diff --git a/utils/llvm.grm b/utils/llvm.grm index 86a707a..d391e2a 100644 --- a/utils/llvm.grm +++ b/utils/llvm.grm @@ -162,6 +162,7 @@ FuncAttr ::= noreturn | readnone | readonly | inlinehint + | alignstack | noinline | alwaysinline | optsize diff --git a/utils/vim/llvm.vim b/utils/vim/llvm.vim index 6e4a207..518aa04 100644 --- a/utils/vim/llvm.vim +++ b/utils/vim/llvm.vim @@ -1,7 +1,7 @@ " Vim syntax file " Language: llvm " Maintainer: The LLVM team, http://llvm.org/ -" Updated: 2003-06-02 +" Version: $Revision: 97271 $ if version < 600 syntax clear @@ -52,11 +52,12 @@ syn keyword llvmKeyword x86_stdcallcc x86_fastcallcc syn keyword llvmKeyword signext zeroext inreg sret nounwind noreturn syn keyword llvmKeyword nocapture byval nest readnone readonly noalias syn keyword llvmKeyword inlinehint noinline alwaysinline optsize ssp sspreq -syn keyword llvmKeyword noredzone noimplicitfloat naked +syn keyword llvmKeyword noredzone noimplicitfloat naked alignstack syn keyword llvmKeyword module asm align tail to syn keyword llvmKeyword addrspace section alias sideeffect c gc syn keyword llvmKeyword target datalayout triple syn keyword llvmKeyword blockaddress +syn keyword llvmKeyword union " Obsolete keywords. syn keyword llvmError uninitialized implementation diff --git a/utils/vim/tablegen.vim b/utils/vim/tablegen.vim index ad35872..11a4d80 100644 --- a/utils/vim/tablegen.vim +++ b/utils/vim/tablegen.vim @@ -1,7 +1,7 @@ " Vim syntax file " Language: TableGen " Maintainer: The LLVM team, http://llvm.org/ -" Updated: 2003-08-11 +" Version: $Revision: 97271 $ if version < 600 syntax clear diff --git a/utils/vim/vimrc b/utils/vim/vimrc index 4909f60..63108f2 100644 --- a/utils/vim/vimrc +++ b/utils/vim/vimrc @@ -1,4 +1,5 @@ " LLVM coding guidelines conformance for VIM +" $Revision: 97273 $ " " Maintainer: The LLVM Team, http://llvm.org " WARNING: Read before you source in all these commands and macros! Some @@ -10,17 +11,30 @@ " It's VIM, not VI set nocompatible -" Wrap text at 80 cols -set textwidth=80 - " A tab produces a 2-space indentation set softtabstop=2 set shiftwidth=2 set expandtab -" Highlight trailing whitespace +" Highlight trailing whitespace and lines longer than 80 columns. +highlight LongLine ctermbg=DarkYellow guibg=DarkYellow highlight WhitespaceEOL ctermbg=DarkYellow guibg=DarkYellow -match WhitespaceEOL /\s\+$/ +if v:version >= 702 + " Lines longer than 80 columns. + au BufWinEnter * let w:m0=matchadd('LongLine', '\%>80v.\+', -1) + + " Whitespace at the end of a line. This little dance suppresses + " whitespace that has just been typed. + au BufWinEnter * let w:m1=matchadd('WhitespaceEOL', '\s\+$', -1) + au InsertEnter * call matchdelete(w:m1) + au InsertEnter * let w:m2=matchadd('WhitespaceEOL', '\s\+\%#\@<!$', -1) + au InsertLeave * call matchdelete(w:m2) + au InsertLeave * let w:m1=matchadd('WhitespaceEOL', '\s\+$', -1) +else + au BufRead,BufNewFile * syntax match LongLine /\%>80v.\+/ + au InsertEnter * syntax match WhitespaceEOL /\s\+\%#\@<!$/ + au InsertLeave * syntax match WhitespaceEOL /\s\+$/ +endif " Enable filetype detection filetype on @@ -70,3 +84,10 @@ augroup END augroup filetype au! BufRead,BufNewFile *.td set filetype=tablegen augroup END + +" Additional vim features to optionally uncomment. +"set showcmd +"set showmatch +"set showmode +"set incsearch +"set ruler |
