diff options
Diffstat (limited to 'contrib/llvm/lib/LTO')
-rw-r--r-- | contrib/llvm/lib/LTO/LTOCodeGenerator.cpp | 600 | ||||
-rw-r--r-- | contrib/llvm/lib/LTO/LTOModule.cpp | 658 |
2 files changed, 1258 insertions, 0 deletions
diff --git a/contrib/llvm/lib/LTO/LTOCodeGenerator.cpp b/contrib/llvm/lib/LTO/LTOCodeGenerator.cpp new file mode 100644 index 0000000..6baaaa4 --- /dev/null +++ b/contrib/llvm/lib/LTO/LTOCodeGenerator.cpp @@ -0,0 +1,600 @@ +//===-LTOCodeGenerator.cpp - LLVM Link Time Optimizer ---------------------===// +// +// 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 Link Time Optimization library. This library is +// intended to be used by linker to optimize code at link time. +// +//===----------------------------------------------------------------------===// + +#include "llvm/LTO/LTOCodeGenerator.h" +#include "llvm/ADT/StringExtras.h" +#include "llvm/Analysis/Passes.h" +#include "llvm/Analysis/TargetLibraryInfo.h" +#include "llvm/Analysis/TargetTransformInfo.h" +#include "llvm/Bitcode/ReaderWriter.h" +#include "llvm/CodeGen/ParallelCG.h" +#include "llvm/CodeGen/RuntimeLibcalls.h" +#include "llvm/Config/config.h" +#include "llvm/IR/Constants.h" +#include "llvm/IR/DataLayout.h" +#include "llvm/IR/DerivedTypes.h" +#include "llvm/IR/DiagnosticInfo.h" +#include "llvm/IR/DiagnosticPrinter.h" +#include "llvm/IR/LLVMContext.h" +#include "llvm/IR/LegacyPassManager.h" +#include "llvm/IR/Mangler.h" +#include "llvm/IR/Module.h" +#include "llvm/IR/Verifier.h" +#include "llvm/InitializePasses.h" +#include "llvm/LTO/LTOModule.h" +#include "llvm/Linker/Linker.h" +#include "llvm/MC/MCAsmInfo.h" +#include "llvm/MC/MCContext.h" +#include "llvm/MC/SubtargetFeature.h" +#include "llvm/Support/CommandLine.h" +#include "llvm/Support/FileSystem.h" +#include "llvm/Support/Host.h" +#include "llvm/Support/MemoryBuffer.h" +#include "llvm/Support/Signals.h" +#include "llvm/Support/TargetRegistry.h" +#include "llvm/Support/TargetSelect.h" +#include "llvm/Support/ToolOutputFile.h" +#include "llvm/Support/raw_ostream.h" +#include "llvm/Target/TargetLowering.h" +#include "llvm/Target/TargetOptions.h" +#include "llvm/Target/TargetRegisterInfo.h" +#include "llvm/Target/TargetSubtargetInfo.h" +#include "llvm/Transforms/IPO.h" +#include "llvm/Transforms/IPO/PassManagerBuilder.h" +#include "llvm/Transforms/ObjCARC.h" +#include <system_error> +using namespace llvm; + +const char* LTOCodeGenerator::getVersionString() { +#ifdef LLVM_VERSION_INFO + return PACKAGE_NAME " version " PACKAGE_VERSION ", " LLVM_VERSION_INFO; +#else + return PACKAGE_NAME " version " PACKAGE_VERSION; +#endif +} + +LTOCodeGenerator::LTOCodeGenerator(LLVMContext &Context) + : Context(Context), MergedModule(new Module("ld-temp.o", Context)), + TheLinker(new Linker(*MergedModule)) { + initializeLTOPasses(); +} + +LTOCodeGenerator::~LTOCodeGenerator() {} + +// Initialize LTO passes. Please keep this function in sync with +// PassManagerBuilder::populateLTOPassManager(), and make sure all LTO +// passes are initialized. +void LTOCodeGenerator::initializeLTOPasses() { + PassRegistry &R = *PassRegistry::getPassRegistry(); + + initializeInternalizePassPass(R); + initializeIPSCCPPass(R); + initializeGlobalOptPass(R); + initializeConstantMergePass(R); + initializeDAHPass(R); + initializeInstructionCombiningPassPass(R); + initializeSimpleInlinerPass(R); + initializePruneEHPass(R); + initializeGlobalDCEPass(R); + initializeArgPromotionPass(R); + initializeJumpThreadingPass(R); + initializeSROALegacyPassPass(R); + initializeSROA_DTPass(R); + initializeSROA_SSAUpPass(R); + initializeFunctionAttrsPass(R); + initializeGlobalsAAWrapperPassPass(R); + initializeLICMPass(R); + initializeMergedLoadStoreMotionPass(R); + initializeGVNPass(R); + initializeMemCpyOptPass(R); + initializeDCEPass(R); + initializeCFGSimplifyPassPass(R); +} + +bool LTOCodeGenerator::addModule(LTOModule *Mod) { + assert(&Mod->getModule().getContext() == &Context && + "Expected module in same context"); + + bool ret = TheLinker->linkInModule(Mod->takeModule()); + + const std::vector<const char *> &undefs = Mod->getAsmUndefinedRefs(); + for (int i = 0, e = undefs.size(); i != e; ++i) + AsmUndefinedRefs[undefs[i]] = 1; + + return !ret; +} + +void LTOCodeGenerator::setModule(std::unique_ptr<LTOModule> Mod) { + assert(&Mod->getModule().getContext() == &Context && + "Expected module in same context"); + + AsmUndefinedRefs.clear(); + + MergedModule = Mod->takeModule(); + TheLinker = make_unique<Linker>(*MergedModule); + + const std::vector<const char*> &Undefs = Mod->getAsmUndefinedRefs(); + for (int I = 0, E = Undefs.size(); I != E; ++I) + AsmUndefinedRefs[Undefs[I]] = 1; +} + +void LTOCodeGenerator::setTargetOptions(TargetOptions Options) { + this->Options = Options; +} + +void LTOCodeGenerator::setDebugInfo(lto_debug_model Debug) { + switch (Debug) { + case LTO_DEBUG_MODEL_NONE: + EmitDwarfDebugInfo = false; + return; + + case LTO_DEBUG_MODEL_DWARF: + EmitDwarfDebugInfo = true; + return; + } + llvm_unreachable("Unknown debug format!"); +} + +void LTOCodeGenerator::setOptLevel(unsigned Level) { + OptLevel = Level; + switch (OptLevel) { + case 0: + CGOptLevel = CodeGenOpt::None; + break; + case 1: + CGOptLevel = CodeGenOpt::Less; + break; + case 2: + CGOptLevel = CodeGenOpt::Default; + break; + case 3: + CGOptLevel = CodeGenOpt::Aggressive; + break; + } +} + +bool LTOCodeGenerator::writeMergedModules(const char *Path) { + if (!determineTarget()) + return false; + + // mark which symbols can not be internalized + applyScopeRestrictions(); + + // create output file + std::error_code EC; + tool_output_file Out(Path, EC, sys::fs::F_None); + if (EC) { + std::string ErrMsg = "could not open bitcode file for writing: "; + ErrMsg += Path; + emitError(ErrMsg); + return false; + } + + // write bitcode to it + WriteBitcodeToFile(MergedModule.get(), Out.os(), ShouldEmbedUselists); + Out.os().close(); + + if (Out.os().has_error()) { + std::string ErrMsg = "could not write bitcode file: "; + ErrMsg += Path; + emitError(ErrMsg); + Out.os().clear_error(); + return false; + } + + Out.keep(); + return true; +} + +bool LTOCodeGenerator::compileOptimizedToFile(const char **Name) { + // make unique temp output file to put generated code + SmallString<128> Filename; + int FD; + + const char *Extension = + (FileType == TargetMachine::CGFT_AssemblyFile ? "s" : "o"); + + std::error_code EC = + sys::fs::createTemporaryFile("lto-llvm", Extension, FD, Filename); + if (EC) { + emitError(EC.message()); + return false; + } + + // generate object file + tool_output_file objFile(Filename.c_str(), FD); + + bool genResult = compileOptimized(&objFile.os()); + objFile.os().close(); + if (objFile.os().has_error()) { + objFile.os().clear_error(); + sys::fs::remove(Twine(Filename)); + return false; + } + + objFile.keep(); + if (!genResult) { + sys::fs::remove(Twine(Filename)); + return false; + } + + NativeObjectPath = Filename.c_str(); + *Name = NativeObjectPath.c_str(); + return true; +} + +std::unique_ptr<MemoryBuffer> +LTOCodeGenerator::compileOptimized() { + const char *name; + if (!compileOptimizedToFile(&name)) + return nullptr; + + // read .o file into memory buffer + ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr = + MemoryBuffer::getFile(name, -1, false); + if (std::error_code EC = BufferOrErr.getError()) { + emitError(EC.message()); + sys::fs::remove(NativeObjectPath); + return nullptr; + } + + // remove temp files + sys::fs::remove(NativeObjectPath); + + return std::move(*BufferOrErr); +} + +bool LTOCodeGenerator::compile_to_file(const char **Name, bool DisableVerify, + bool DisableInline, + bool DisableGVNLoadPRE, + bool DisableVectorization) { + if (!optimize(DisableVerify, DisableInline, DisableGVNLoadPRE, + DisableVectorization)) + return false; + + return compileOptimizedToFile(Name); +} + +std::unique_ptr<MemoryBuffer> +LTOCodeGenerator::compile(bool DisableVerify, bool DisableInline, + bool DisableGVNLoadPRE, bool DisableVectorization) { + if (!optimize(DisableVerify, DisableInline, DisableGVNLoadPRE, + DisableVectorization)) + return nullptr; + + return compileOptimized(); +} + +bool LTOCodeGenerator::determineTarget() { + if (TargetMach) + return true; + + std::string TripleStr = MergedModule->getTargetTriple(); + if (TripleStr.empty()) { + TripleStr = sys::getDefaultTargetTriple(); + MergedModule->setTargetTriple(TripleStr); + } + llvm::Triple Triple(TripleStr); + + // create target machine from info for merged modules + std::string ErrMsg; + const Target *march = TargetRegistry::lookupTarget(TripleStr, ErrMsg); + if (!march) { + emitError(ErrMsg); + return false; + } + + // Construct LTOModule, hand over ownership of module and target. Use MAttr as + // the default set of features. + SubtargetFeatures Features(MAttr); + Features.getDefaultSubtargetFeatures(Triple); + FeatureStr = Features.getString(); + // Set a default CPU for Darwin triples. + if (MCpu.empty() && Triple.isOSDarwin()) { + if (Triple.getArch() == llvm::Triple::x86_64) + MCpu = "core2"; + else if (Triple.getArch() == llvm::Triple::x86) + MCpu = "yonah"; + else if (Triple.getArch() == llvm::Triple::aarch64) + MCpu = "cyclone"; + } + + TargetMach.reset(march->createTargetMachine(TripleStr, MCpu, FeatureStr, + Options, RelocModel, + CodeModel::Default, CGOptLevel)); + return true; +} + +void LTOCodeGenerator:: +applyRestriction(GlobalValue &GV, + ArrayRef<StringRef> Libcalls, + std::vector<const char*> &MustPreserveList, + SmallPtrSetImpl<GlobalValue*> &AsmUsed, + Mangler &Mangler) { + // There are no restrictions to apply to declarations. + if (GV.isDeclaration()) + return; + + // There is nothing more restrictive than private linkage. + if (GV.hasPrivateLinkage()) + return; + + SmallString<64> Buffer; + TargetMach->getNameWithPrefix(Buffer, &GV, Mangler); + + if (MustPreserveSymbols.count(Buffer)) + MustPreserveList.push_back(GV.getName().data()); + if (AsmUndefinedRefs.count(Buffer)) + AsmUsed.insert(&GV); + + // Conservatively append user-supplied runtime library functions to + // llvm.compiler.used. These could be internalized and deleted by + // optimizations like -globalopt, causing problems when later optimizations + // add new library calls (e.g., llvm.memset => memset and printf => puts). + // Leave it to the linker to remove any dead code (e.g. with -dead_strip). + if (isa<Function>(GV) && + std::binary_search(Libcalls.begin(), Libcalls.end(), GV.getName())) + AsmUsed.insert(&GV); +} + +static void findUsedValues(GlobalVariable *LLVMUsed, + SmallPtrSetImpl<GlobalValue*> &UsedValues) { + if (!LLVMUsed) return; + + ConstantArray *Inits = cast<ConstantArray>(LLVMUsed->getInitializer()); + for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i) + if (GlobalValue *GV = + dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts())) + UsedValues.insert(GV); +} + +// Collect names of runtime library functions. User-defined functions with the +// same names are added to llvm.compiler.used to prevent them from being +// deleted by optimizations. +static void accumulateAndSortLibcalls(std::vector<StringRef> &Libcalls, + const TargetLibraryInfo& TLI, + const Module &Mod, + const TargetMachine &TM) { + // TargetLibraryInfo has info on C runtime library calls on the current + // target. + for (unsigned I = 0, E = static_cast<unsigned>(LibFunc::NumLibFuncs); + I != E; ++I) { + LibFunc::Func F = static_cast<LibFunc::Func>(I); + if (TLI.has(F)) + Libcalls.push_back(TLI.getName(F)); + } + + SmallPtrSet<const TargetLowering *, 1> TLSet; + + for (const Function &F : Mod) { + const TargetLowering *Lowering = + TM.getSubtargetImpl(F)->getTargetLowering(); + + if (Lowering && TLSet.insert(Lowering).second) + // TargetLowering has info on library calls that CodeGen expects to be + // available, both from the C runtime and compiler-rt. + for (unsigned I = 0, E = static_cast<unsigned>(RTLIB::UNKNOWN_LIBCALL); + I != E; ++I) + if (const char *Name = + Lowering->getLibcallName(static_cast<RTLIB::Libcall>(I))) + Libcalls.push_back(Name); + } + + array_pod_sort(Libcalls.begin(), Libcalls.end()); + Libcalls.erase(std::unique(Libcalls.begin(), Libcalls.end()), + Libcalls.end()); +} + +void LTOCodeGenerator::applyScopeRestrictions() { + if (ScopeRestrictionsDone || !ShouldInternalize) + return; + + // Start off with a verification pass. + legacy::PassManager passes; + passes.add(createVerifierPass()); + + // mark which symbols can not be internalized + Mangler Mangler; + std::vector<const char*> MustPreserveList; + SmallPtrSet<GlobalValue*, 8> AsmUsed; + std::vector<StringRef> Libcalls; + TargetLibraryInfoImpl TLII(Triple(TargetMach->getTargetTriple())); + TargetLibraryInfo TLI(TLII); + + accumulateAndSortLibcalls(Libcalls, TLI, *MergedModule, *TargetMach); + + for (Function &f : *MergedModule) + applyRestriction(f, Libcalls, MustPreserveList, AsmUsed, Mangler); + for (GlobalVariable &v : MergedModule->globals()) + applyRestriction(v, Libcalls, MustPreserveList, AsmUsed, Mangler); + for (GlobalAlias &a : MergedModule->aliases()) + applyRestriction(a, Libcalls, MustPreserveList, AsmUsed, Mangler); + + GlobalVariable *LLVMCompilerUsed = + MergedModule->getGlobalVariable("llvm.compiler.used"); + findUsedValues(LLVMCompilerUsed, AsmUsed); + if (LLVMCompilerUsed) + LLVMCompilerUsed->eraseFromParent(); + + if (!AsmUsed.empty()) { + llvm::Type *i8PTy = llvm::Type::getInt8PtrTy(Context); + std::vector<Constant*> asmUsed2; + for (auto *GV : AsmUsed) { + Constant *c = ConstantExpr::getBitCast(GV, i8PTy); + asmUsed2.push_back(c); + } + + llvm::ArrayType *ATy = llvm::ArrayType::get(i8PTy, asmUsed2.size()); + LLVMCompilerUsed = + new llvm::GlobalVariable(*MergedModule, ATy, false, + llvm::GlobalValue::AppendingLinkage, + llvm::ConstantArray::get(ATy, asmUsed2), + "llvm.compiler.used"); + + LLVMCompilerUsed->setSection("llvm.metadata"); + } + + passes.add(createInternalizePass(MustPreserveList)); + + // apply scope restrictions + passes.run(*MergedModule); + + ScopeRestrictionsDone = true; +} + +/// Optimize merged modules using various IPO passes +bool LTOCodeGenerator::optimize(bool DisableVerify, bool DisableInline, + bool DisableGVNLoadPRE, + bool DisableVectorization) { + if (!this->determineTarget()) + return false; + + // Mark which symbols can not be internalized + this->applyScopeRestrictions(); + + // Instantiate the pass manager to organize the passes. + legacy::PassManager passes; + + // Add an appropriate DataLayout instance for this module... + MergedModule->setDataLayout(TargetMach->createDataLayout()); + + passes.add( + createTargetTransformInfoWrapperPass(TargetMach->getTargetIRAnalysis())); + + Triple TargetTriple(TargetMach->getTargetTriple()); + PassManagerBuilder PMB; + PMB.DisableGVNLoadPRE = DisableGVNLoadPRE; + PMB.LoopVectorize = !DisableVectorization; + PMB.SLPVectorize = !DisableVectorization; + if (!DisableInline) + PMB.Inliner = createFunctionInliningPass(); + PMB.LibraryInfo = new TargetLibraryInfoImpl(TargetTriple); + PMB.OptLevel = OptLevel; + PMB.VerifyInput = !DisableVerify; + PMB.VerifyOutput = !DisableVerify; + + PMB.populateLTOPassManager(passes); + + // Run our queue of passes all at once now, efficiently. + passes.run(*MergedModule); + + return true; +} + +bool LTOCodeGenerator::compileOptimized(ArrayRef<raw_pwrite_stream *> Out) { + if (!this->determineTarget()) + return false; + + legacy::PassManager preCodeGenPasses; + + // If the bitcode files contain ARC code and were compiled with optimization, + // the ObjCARCContractPass must be run, so do it unconditionally here. + preCodeGenPasses.add(createObjCARCContractPass()); + preCodeGenPasses.run(*MergedModule); + + // Do code generation. We need to preserve the module in case the client calls + // writeMergedModules() after compilation, but we only need to allow this at + // parallelism level 1. This is achieved by having splitCodeGen return the + // original module at parallelism level 1 which we then assign back to + // MergedModule. + MergedModule = + splitCodeGen(std::move(MergedModule), Out, MCpu, FeatureStr, Options, + RelocModel, CodeModel::Default, CGOptLevel, FileType); + + return true; +} + +/// setCodeGenDebugOptions - Set codegen debugging options to aid in debugging +/// LTO problems. +void LTOCodeGenerator::setCodeGenDebugOptions(const char *Options) { + for (std::pair<StringRef, StringRef> o = getToken(Options); !o.first.empty(); + o = getToken(o.second)) + CodegenOptions.push_back(o.first); +} + +void LTOCodeGenerator::parseCodeGenDebugOptions() { + // if options were requested, set them + if (!CodegenOptions.empty()) { + // ParseCommandLineOptions() expects argv[0] to be program name. + std::vector<const char *> CodegenArgv(1, "libLLVMLTO"); + for (std::string &Arg : CodegenOptions) + CodegenArgv.push_back(Arg.c_str()); + cl::ParseCommandLineOptions(CodegenArgv.size(), CodegenArgv.data()); + } +} + +void LTOCodeGenerator::DiagnosticHandler(const DiagnosticInfo &DI, + void *Context) { + ((LTOCodeGenerator *)Context)->DiagnosticHandler2(DI); +} + +void LTOCodeGenerator::DiagnosticHandler2(const DiagnosticInfo &DI) { + // Map the LLVM internal diagnostic severity to the LTO diagnostic severity. + lto_codegen_diagnostic_severity_t Severity; + switch (DI.getSeverity()) { + case DS_Error: + Severity = LTO_DS_ERROR; + break; + case DS_Warning: + Severity = LTO_DS_WARNING; + break; + case DS_Remark: + Severity = LTO_DS_REMARK; + break; + case DS_Note: + Severity = LTO_DS_NOTE; + break; + } + // Create the string that will be reported to the external diagnostic handler. + std::string MsgStorage; + raw_string_ostream Stream(MsgStorage); + DiagnosticPrinterRawOStream DP(Stream); + DI.print(DP); + Stream.flush(); + + // If this method has been called it means someone has set up an external + // diagnostic handler. Assert on that. + assert(DiagHandler && "Invalid diagnostic handler"); + (*DiagHandler)(Severity, MsgStorage.c_str(), DiagContext); +} + +void +LTOCodeGenerator::setDiagnosticHandler(lto_diagnostic_handler_t DiagHandler, + void *Ctxt) { + this->DiagHandler = DiagHandler; + this->DiagContext = Ctxt; + if (!DiagHandler) + return Context.setDiagnosticHandler(nullptr, nullptr); + // Register the LTOCodeGenerator stub in the LLVMContext to forward the + // diagnostic to the external DiagHandler. + Context.setDiagnosticHandler(LTOCodeGenerator::DiagnosticHandler, this, + /* RespectFilters */ true); +} + +namespace { +class LTODiagnosticInfo : public DiagnosticInfo { + const Twine &Msg; +public: + LTODiagnosticInfo(const Twine &DiagMsg, DiagnosticSeverity Severity=DS_Error) + : DiagnosticInfo(DK_Linker, Severity), Msg(DiagMsg) {} + void print(DiagnosticPrinter &DP) const override { DP << Msg; } +}; +} + +void LTOCodeGenerator::emitError(const std::string &ErrMsg) { + if (DiagHandler) + (*DiagHandler)(LTO_DS_ERROR, ErrMsg.c_str(), DiagContext); + else + Context.diagnose(LTODiagnosticInfo(ErrMsg)); +} diff --git a/contrib/llvm/lib/LTO/LTOModule.cpp b/contrib/llvm/lib/LTO/LTOModule.cpp new file mode 100644 index 0000000..409b949 --- /dev/null +++ b/contrib/llvm/lib/LTO/LTOModule.cpp @@ -0,0 +1,658 @@ +//===-- LTOModule.cpp - LLVM Link Time Optimizer --------------------------===// +// +// 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 Link Time Optimization library. This library is +// intended to be used by linker to optimize code at link time. +// +//===----------------------------------------------------------------------===// + +#include "llvm/LTO/LTOModule.h" +#include "llvm/ADT/Triple.h" +#include "llvm/Bitcode/ReaderWriter.h" +#include "llvm/CodeGen/Analysis.h" +#include "llvm/IR/Constants.h" +#include "llvm/IR/DiagnosticPrinter.h" +#include "llvm/IR/LLVMContext.h" +#include "llvm/IR/Mangler.h" +#include "llvm/IR/Metadata.h" +#include "llvm/IR/Module.h" +#include "llvm/MC/MCExpr.h" +#include "llvm/MC/MCInst.h" +#include "llvm/MC/MCInstrInfo.h" +#include "llvm/MC/MCParser/MCAsmParser.h" +#include "llvm/MC/MCSection.h" +#include "llvm/MC/MCSubtargetInfo.h" +#include "llvm/MC/MCSymbol.h" +#include "llvm/MC/MCTargetAsmParser.h" +#include "llvm/MC/SubtargetFeature.h" +#include "llvm/Object/IRObjectFile.h" +#include "llvm/Object/ObjectFile.h" +#include "llvm/Support/CommandLine.h" +#include "llvm/Support/FileSystem.h" +#include "llvm/Support/Host.h" +#include "llvm/Support/MemoryBuffer.h" +#include "llvm/Support/Path.h" +#include "llvm/Support/SourceMgr.h" +#include "llvm/Support/TargetRegistry.h" +#include "llvm/Support/TargetSelect.h" +#include "llvm/Target/TargetLowering.h" +#include "llvm/Target/TargetLoweringObjectFile.h" +#include "llvm/Target/TargetRegisterInfo.h" +#include "llvm/Target/TargetSubtargetInfo.h" +#include "llvm/Transforms/Utils/GlobalStatus.h" +#include <system_error> +using namespace llvm; +using namespace llvm::object; + +LTOModule::LTOModule(std::unique_ptr<object::IRObjectFile> Obj, + llvm::TargetMachine *TM) + : IRFile(std::move(Obj)), _target(TM) {} + +LTOModule::LTOModule(std::unique_ptr<object::IRObjectFile> Obj, + llvm::TargetMachine *TM, + std::unique_ptr<LLVMContext> Context) + : OwnedContext(std::move(Context)), IRFile(std::move(Obj)), _target(TM) {} + +LTOModule::~LTOModule() {} + +/// isBitcodeFile - Returns 'true' if the file (or memory contents) is LLVM +/// bitcode. +bool LTOModule::isBitcodeFile(const void *Mem, size_t Length) { + ErrorOr<MemoryBufferRef> BCData = IRObjectFile::findBitcodeInMemBuffer( + MemoryBufferRef(StringRef((const char *)Mem, Length), "<mem>")); + return bool(BCData); +} + +bool LTOModule::isBitcodeFile(const char *Path) { + ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr = + MemoryBuffer::getFile(Path); + if (!BufferOrErr) + return false; + + ErrorOr<MemoryBufferRef> BCData = IRObjectFile::findBitcodeInMemBuffer( + BufferOrErr.get()->getMemBufferRef()); + return bool(BCData); +} + +bool LTOModule::isBitcodeForTarget(MemoryBuffer *Buffer, + StringRef TriplePrefix) { + ErrorOr<MemoryBufferRef> BCOrErr = + IRObjectFile::findBitcodeInMemBuffer(Buffer->getMemBufferRef()); + if (!BCOrErr) + return false; + LLVMContext Context; + std::string Triple = getBitcodeTargetTriple(*BCOrErr, Context); + return StringRef(Triple).startswith(TriplePrefix); +} + +std::string LTOModule::getProducerString(MemoryBuffer *Buffer) { + ErrorOr<MemoryBufferRef> BCOrErr = + IRObjectFile::findBitcodeInMemBuffer(Buffer->getMemBufferRef()); + if (!BCOrErr) + return ""; + LLVMContext Context; + return getBitcodeProducerString(*BCOrErr, Context); +} + +ErrorOr<std::unique_ptr<LTOModule>> +LTOModule::createFromFile(LLVMContext &Context, const char *path, + TargetOptions options) { + ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr = + MemoryBuffer::getFile(path); + if (std::error_code EC = BufferOrErr.getError()) + return EC; + std::unique_ptr<MemoryBuffer> Buffer = std::move(BufferOrErr.get()); + return makeLTOModule(Buffer->getMemBufferRef(), options, &Context); +} + +ErrorOr<std::unique_ptr<LTOModule>> +LTOModule::createFromOpenFile(LLVMContext &Context, int fd, const char *path, + size_t size, TargetOptions options) { + return createFromOpenFileSlice(Context, fd, path, size, 0, options); +} + +ErrorOr<std::unique_ptr<LTOModule>> +LTOModule::createFromOpenFileSlice(LLVMContext &Context, int fd, + const char *path, size_t map_size, + off_t offset, TargetOptions options) { + ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr = + MemoryBuffer::getOpenFileSlice(fd, path, map_size, offset); + if (std::error_code EC = BufferOrErr.getError()) + return EC; + std::unique_ptr<MemoryBuffer> Buffer = std::move(BufferOrErr.get()); + return makeLTOModule(Buffer->getMemBufferRef(), options, &Context); +} + +ErrorOr<std::unique_ptr<LTOModule>> +LTOModule::createFromBuffer(LLVMContext &Context, const void *mem, + size_t length, TargetOptions options, + StringRef path) { + return createInContext(mem, length, options, path, &Context); +} + +ErrorOr<std::unique_ptr<LTOModule>> +LTOModule::createInLocalContext(const void *mem, size_t length, + TargetOptions options, StringRef path) { + return createInContext(mem, length, options, path, nullptr); +} + +ErrorOr<std::unique_ptr<LTOModule>> +LTOModule::createInContext(const void *mem, size_t length, + TargetOptions options, StringRef path, + LLVMContext *Context) { + StringRef Data((const char *)mem, length); + MemoryBufferRef Buffer(Data, path); + return makeLTOModule(Buffer, options, Context); +} + +static ErrorOr<std::unique_ptr<Module>> +parseBitcodeFileImpl(MemoryBufferRef Buffer, LLVMContext &Context, + bool ShouldBeLazy) { + + // Find the buffer. + ErrorOr<MemoryBufferRef> MBOrErr = + IRObjectFile::findBitcodeInMemBuffer(Buffer); + if (std::error_code EC = MBOrErr.getError()) + return EC; + + if (!ShouldBeLazy) { + // Parse the full file. + ErrorOr<std::unique_ptr<Module>> M = parseBitcodeFile(*MBOrErr, Context); + if (std::error_code EC = M.getError()) + return EC; + return std::move(*M); + } + + // Parse lazily. + std::unique_ptr<MemoryBuffer> LightweightBuf = + MemoryBuffer::getMemBuffer(*MBOrErr, false); + ErrorOr<std::unique_ptr<Module>> M = getLazyBitcodeModule( + std::move(LightweightBuf), Context, true /*ShouldLazyLoadMetadata*/); + if (std::error_code EC = M.getError()) + return EC; + return std::move(*M); +} + +ErrorOr<std::unique_ptr<LTOModule>> +LTOModule::makeLTOModule(MemoryBufferRef Buffer, TargetOptions options, + LLVMContext *Context) { + std::unique_ptr<LLVMContext> OwnedContext; + if (!Context) { + OwnedContext = llvm::make_unique<LLVMContext>(); + Context = OwnedContext.get(); + } + + // If we own a context, we know this is being used only for symbol + // extraction, not linking. Be lazy in that case. + ErrorOr<std::unique_ptr<Module>> MOrErr = + parseBitcodeFileImpl(Buffer, *Context, + /* ShouldBeLazy */ static_cast<bool>(OwnedContext)); + if (std::error_code EC = MOrErr.getError()) + return EC; + std::unique_ptr<Module> &M = *MOrErr; + + std::string TripleStr = M->getTargetTriple(); + if (TripleStr.empty()) + TripleStr = sys::getDefaultTargetTriple(); + llvm::Triple Triple(TripleStr); + + // find machine architecture for this module + std::string errMsg; + const Target *march = TargetRegistry::lookupTarget(TripleStr, errMsg); + if (!march) + return std::unique_ptr<LTOModule>(nullptr); + + // construct LTOModule, hand over ownership of module and target + SubtargetFeatures Features; + Features.getDefaultSubtargetFeatures(Triple); + std::string FeatureStr = Features.getString(); + // Set a default CPU for Darwin triples. + std::string CPU; + if (Triple.isOSDarwin()) { + if (Triple.getArch() == llvm::Triple::x86_64) + CPU = "core2"; + else if (Triple.getArch() == llvm::Triple::x86) + CPU = "yonah"; + else if (Triple.getArch() == llvm::Triple::aarch64) + CPU = "cyclone"; + } + + TargetMachine *target = march->createTargetMachine(TripleStr, CPU, FeatureStr, + options); + M->setDataLayout(target->createDataLayout()); + + std::unique_ptr<object::IRObjectFile> IRObj( + new object::IRObjectFile(Buffer, std::move(M))); + + std::unique_ptr<LTOModule> Ret; + if (OwnedContext) + Ret.reset(new LTOModule(std::move(IRObj), target, std::move(OwnedContext))); + else + Ret.reset(new LTOModule(std::move(IRObj), target)); + + Ret->parseSymbols(); + Ret->parseMetadata(); + + return std::move(Ret); +} + +/// Create a MemoryBuffer from a memory range with an optional name. +std::unique_ptr<MemoryBuffer> +LTOModule::makeBuffer(const void *mem, size_t length, StringRef name) { + const char *startPtr = (const char*)mem; + return MemoryBuffer::getMemBuffer(StringRef(startPtr, length), name, false); +} + +/// objcClassNameFromExpression - Get string that the data pointer points to. +bool +LTOModule::objcClassNameFromExpression(const Constant *c, std::string &name) { + if (const ConstantExpr *ce = dyn_cast<ConstantExpr>(c)) { + Constant *op = ce->getOperand(0); + if (GlobalVariable *gvn = dyn_cast<GlobalVariable>(op)) { + Constant *cn = gvn->getInitializer(); + if (ConstantDataArray *ca = dyn_cast<ConstantDataArray>(cn)) { + if (ca->isCString()) { + name = (".objc_class_name_" + ca->getAsCString()).str(); + return true; + } + } + } + } + return false; +} + +/// addObjCClass - Parse i386/ppc ObjC class data structure. +void LTOModule::addObjCClass(const GlobalVariable *clgv) { + const ConstantStruct *c = dyn_cast<ConstantStruct>(clgv->getInitializer()); + if (!c) return; + + // second slot in __OBJC,__class is pointer to superclass name + std::string superclassName; + if (objcClassNameFromExpression(c->getOperand(1), superclassName)) { + auto IterBool = + _undefines.insert(std::make_pair(superclassName, NameAndAttributes())); + if (IterBool.second) { + NameAndAttributes &info = IterBool.first->second; + info.name = IterBool.first->first().data(); + info.attributes = LTO_SYMBOL_DEFINITION_UNDEFINED; + info.isFunction = false; + info.symbol = clgv; + } + } + + // third slot in __OBJC,__class is pointer to class name + std::string className; + if (objcClassNameFromExpression(c->getOperand(2), className)) { + auto Iter = _defines.insert(className).first; + + NameAndAttributes info; + info.name = Iter->first().data(); + info.attributes = LTO_SYMBOL_PERMISSIONS_DATA | + LTO_SYMBOL_DEFINITION_REGULAR | LTO_SYMBOL_SCOPE_DEFAULT; + info.isFunction = false; + info.symbol = clgv; + _symbols.push_back(info); + } +} + +/// addObjCCategory - Parse i386/ppc ObjC category data structure. +void LTOModule::addObjCCategory(const GlobalVariable *clgv) { + const ConstantStruct *c = dyn_cast<ConstantStruct>(clgv->getInitializer()); + if (!c) return; + + // second slot in __OBJC,__category is pointer to target class name + std::string targetclassName; + if (!objcClassNameFromExpression(c->getOperand(1), targetclassName)) + return; + + auto IterBool = + _undefines.insert(std::make_pair(targetclassName, NameAndAttributes())); + + if (!IterBool.second) + return; + + NameAndAttributes &info = IterBool.first->second; + info.name = IterBool.first->first().data(); + info.attributes = LTO_SYMBOL_DEFINITION_UNDEFINED; + info.isFunction = false; + info.symbol = clgv; +} + +/// addObjCClassRef - Parse i386/ppc ObjC class list data structure. +void LTOModule::addObjCClassRef(const GlobalVariable *clgv) { + std::string targetclassName; + if (!objcClassNameFromExpression(clgv->getInitializer(), targetclassName)) + return; + + auto IterBool = + _undefines.insert(std::make_pair(targetclassName, NameAndAttributes())); + + if (!IterBool.second) + return; + + NameAndAttributes &info = IterBool.first->second; + info.name = IterBool.first->first().data(); + info.attributes = LTO_SYMBOL_DEFINITION_UNDEFINED; + info.isFunction = false; + info.symbol = clgv; +} + +void LTOModule::addDefinedDataSymbol(const object::BasicSymbolRef &Sym) { + SmallString<64> Buffer; + { + raw_svector_ostream OS(Buffer); + Sym.printName(OS); + } + + const GlobalValue *V = IRFile->getSymbolGV(Sym.getRawDataRefImpl()); + addDefinedDataSymbol(Buffer.c_str(), V); +} + +void LTOModule::addDefinedDataSymbol(const char *Name, const GlobalValue *v) { + // Add to list of defined symbols. + addDefinedSymbol(Name, v, false); + + if (!v->hasSection() /* || !isTargetDarwin */) + return; + + // Special case i386/ppc ObjC data structures in magic sections: + // The issue is that the old ObjC object format did some strange + // contortions to avoid real linker symbols. For instance, the + // ObjC class data structure is allocated statically in the executable + // that defines that class. That data structures contains a pointer to + // its superclass. But instead of just initializing that part of the + // struct to the address of its superclass, and letting the static and + // dynamic linkers do the rest, the runtime works by having that field + // instead point to a C-string that is the name of the superclass. + // At runtime the objc initialization updates that pointer and sets + // it to point to the actual super class. As far as the linker + // knows it is just a pointer to a string. But then someone wanted the + // linker to issue errors at build time if the superclass was not found. + // So they figured out a way in mach-o object format to use an absolute + // symbols (.objc_class_name_Foo = 0) and a floating reference + // (.reference .objc_class_name_Bar) to cause the linker into erroring when + // a class was missing. + // The following synthesizes the implicit .objc_* symbols for the linker + // from the ObjC data structures generated by the front end. + + // special case if this data blob is an ObjC class definition + std::string Section = v->getSection(); + if (Section.compare(0, 15, "__OBJC,__class,") == 0) { + if (const GlobalVariable *gv = dyn_cast<GlobalVariable>(v)) { + addObjCClass(gv); + } + } + + // special case if this data blob is an ObjC category definition + else if (Section.compare(0, 18, "__OBJC,__category,") == 0) { + if (const GlobalVariable *gv = dyn_cast<GlobalVariable>(v)) { + addObjCCategory(gv); + } + } + + // special case if this data blob is the list of referenced classes + else if (Section.compare(0, 18, "__OBJC,__cls_refs,") == 0) { + if (const GlobalVariable *gv = dyn_cast<GlobalVariable>(v)) { + addObjCClassRef(gv); + } + } +} + +void LTOModule::addDefinedFunctionSymbol(const object::BasicSymbolRef &Sym) { + SmallString<64> Buffer; + { + raw_svector_ostream OS(Buffer); + Sym.printName(OS); + } + + const Function *F = + cast<Function>(IRFile->getSymbolGV(Sym.getRawDataRefImpl())); + addDefinedFunctionSymbol(Buffer.c_str(), F); +} + +void LTOModule::addDefinedFunctionSymbol(const char *Name, const Function *F) { + // add to list of defined symbols + addDefinedSymbol(Name, F, true); +} + +void LTOModule::addDefinedSymbol(const char *Name, const GlobalValue *def, + bool isFunction) { + // set alignment part log2() can have rounding errors + uint32_t align = def->getAlignment(); + uint32_t attr = align ? countTrailingZeros(align) : 0; + + // set permissions part + if (isFunction) { + attr |= LTO_SYMBOL_PERMISSIONS_CODE; + } else { + const GlobalVariable *gv = dyn_cast<GlobalVariable>(def); + if (gv && gv->isConstant()) + attr |= LTO_SYMBOL_PERMISSIONS_RODATA; + else + attr |= LTO_SYMBOL_PERMISSIONS_DATA; + } + + // set definition part + if (def->hasWeakLinkage() || def->hasLinkOnceLinkage()) + attr |= LTO_SYMBOL_DEFINITION_WEAK; + else if (def->hasCommonLinkage()) + attr |= LTO_SYMBOL_DEFINITION_TENTATIVE; + else + attr |= LTO_SYMBOL_DEFINITION_REGULAR; + + // set scope part + if (def->hasLocalLinkage()) + // Ignore visibility if linkage is local. + attr |= LTO_SYMBOL_SCOPE_INTERNAL; + else if (def->hasHiddenVisibility()) + attr |= LTO_SYMBOL_SCOPE_HIDDEN; + else if (def->hasProtectedVisibility()) + attr |= LTO_SYMBOL_SCOPE_PROTECTED; + else if (canBeOmittedFromSymbolTable(def)) + attr |= LTO_SYMBOL_SCOPE_DEFAULT_CAN_BE_HIDDEN; + else + attr |= LTO_SYMBOL_SCOPE_DEFAULT; + + if (def->hasComdat()) + attr |= LTO_SYMBOL_COMDAT; + + if (isa<GlobalAlias>(def)) + attr |= LTO_SYMBOL_ALIAS; + + auto Iter = _defines.insert(Name).first; + + // fill information structure + NameAndAttributes info; + StringRef NameRef = Iter->first(); + info.name = NameRef.data(); + assert(info.name[NameRef.size()] == '\0'); + info.attributes = attr; + info.isFunction = isFunction; + info.symbol = def; + + // add to table of symbols + _symbols.push_back(info); +} + +/// addAsmGlobalSymbol - Add a global symbol from module-level ASM to the +/// defined list. +void LTOModule::addAsmGlobalSymbol(const char *name, + lto_symbol_attributes scope) { + auto IterBool = _defines.insert(name); + + // only add new define if not already defined + if (!IterBool.second) + return; + + NameAndAttributes &info = _undefines[IterBool.first->first().data()]; + + if (info.symbol == nullptr) { + // FIXME: This is trying to take care of module ASM like this: + // + // module asm ".zerofill __FOO, __foo, _bar_baz_qux, 0" + // + // but is gross and its mother dresses it funny. Have the ASM parser give us + // more details for this type of situation so that we're not guessing so + // much. + + // fill information structure + info.name = IterBool.first->first().data(); + info.attributes = + LTO_SYMBOL_PERMISSIONS_DATA | LTO_SYMBOL_DEFINITION_REGULAR | scope; + info.isFunction = false; + info.symbol = nullptr; + + // add to table of symbols + _symbols.push_back(info); + return; + } + + if (info.isFunction) + addDefinedFunctionSymbol(info.name, cast<Function>(info.symbol)); + else + addDefinedDataSymbol(info.name, info.symbol); + + _symbols.back().attributes &= ~LTO_SYMBOL_SCOPE_MASK; + _symbols.back().attributes |= scope; +} + +/// addAsmGlobalSymbolUndef - Add a global symbol from module-level ASM to the +/// undefined list. +void LTOModule::addAsmGlobalSymbolUndef(const char *name) { + auto IterBool = _undefines.insert(std::make_pair(name, NameAndAttributes())); + + _asm_undefines.push_back(IterBool.first->first().data()); + + // we already have the symbol + if (!IterBool.second) + return; + + uint32_t attr = LTO_SYMBOL_DEFINITION_UNDEFINED; + attr |= LTO_SYMBOL_SCOPE_DEFAULT; + NameAndAttributes &info = IterBool.first->second; + info.name = IterBool.first->first().data(); + info.attributes = attr; + info.isFunction = false; + info.symbol = nullptr; +} + +/// Add a symbol which isn't defined just yet to a list to be resolved later. +void LTOModule::addPotentialUndefinedSymbol(const object::BasicSymbolRef &Sym, + bool isFunc) { + SmallString<64> name; + { + raw_svector_ostream OS(name); + Sym.printName(OS); + } + + auto IterBool = _undefines.insert(std::make_pair(name, NameAndAttributes())); + + // we already have the symbol + if (!IterBool.second) + return; + + NameAndAttributes &info = IterBool.first->second; + + info.name = IterBool.first->first().data(); + + const GlobalValue *decl = IRFile->getSymbolGV(Sym.getRawDataRefImpl()); + + if (decl->hasExternalWeakLinkage()) + info.attributes = LTO_SYMBOL_DEFINITION_WEAKUNDEF; + else + info.attributes = LTO_SYMBOL_DEFINITION_UNDEFINED; + + info.isFunction = isFunc; + info.symbol = decl; +} + +void LTOModule::parseSymbols() { + for (auto &Sym : IRFile->symbols()) { + const GlobalValue *GV = IRFile->getSymbolGV(Sym.getRawDataRefImpl()); + uint32_t Flags = Sym.getFlags(); + if (Flags & object::BasicSymbolRef::SF_FormatSpecific) + continue; + + bool IsUndefined = Flags & object::BasicSymbolRef::SF_Undefined; + + if (!GV) { + SmallString<64> Buffer; + { + raw_svector_ostream OS(Buffer); + Sym.printName(OS); + } + const char *Name = Buffer.c_str(); + + if (IsUndefined) + addAsmGlobalSymbolUndef(Name); + else if (Flags & object::BasicSymbolRef::SF_Global) + addAsmGlobalSymbol(Name, LTO_SYMBOL_SCOPE_DEFAULT); + else + addAsmGlobalSymbol(Name, LTO_SYMBOL_SCOPE_INTERNAL); + continue; + } + + auto *F = dyn_cast<Function>(GV); + if (IsUndefined) { + addPotentialUndefinedSymbol(Sym, F != nullptr); + continue; + } + + if (F) { + addDefinedFunctionSymbol(Sym); + continue; + } + + if (isa<GlobalVariable>(GV)) { + addDefinedDataSymbol(Sym); + continue; + } + + assert(isa<GlobalAlias>(GV)); + addDefinedDataSymbol(Sym); + } + + // make symbols for all undefines + for (StringMap<NameAndAttributes>::iterator u =_undefines.begin(), + e = _undefines.end(); u != e; ++u) { + // If this symbol also has a definition, then don't make an undefine because + // it is a tentative definition. + if (_defines.count(u->getKey())) continue; + NameAndAttributes info = u->getValue(); + _symbols.push_back(info); + } +} + +/// parseMetadata - Parse metadata from the module +void LTOModule::parseMetadata() { + raw_string_ostream OS(LinkerOpts); + + // Linker Options + if (Metadata *Val = getModule().getModuleFlag("Linker Options")) { + MDNode *LinkerOptions = cast<MDNode>(Val); + for (unsigned i = 0, e = LinkerOptions->getNumOperands(); i != e; ++i) { + MDNode *MDOptions = cast<MDNode>(LinkerOptions->getOperand(i)); + for (unsigned ii = 0, ie = MDOptions->getNumOperands(); ii != ie; ++ii) { + MDString *MDOption = cast<MDString>(MDOptions->getOperand(ii)); + OS << " " << MDOption->getString(); + } + } + } + + // Globals + Mangler Mang; + for (const NameAndAttributes &Sym : _symbols) { + if (!Sym.symbol) + continue; + _target->getObjFileLowering()->emitLinkerFlagsForGlobal(OS, Sym.symbol, + Mang); + } + + // Add other interesting metadata here. +} |