diff options
93 files changed, 1809 insertions, 2650 deletions
diff --git a/ObsoleteFiles.inc b/ObsoleteFiles.inc index 4fa244d..271cbef 100644 --- a/ObsoleteFiles.inc +++ b/ObsoleteFiles.inc @@ -38,6 +38,60 @@ # xargs -n1 | sort | uniq -d; # done +# 20150118: new clang import which bumps version from 3.5.0 to 3.5.1. +OLD_FILES+=usr/include/clang/3.5.0/__wmmintrin_aes.h +OLD_FILES+=usr/include/clang/3.5.0/__wmmintrin_pclmul.h +OLD_FILES+=usr/include/clang/3.5.0/altivec.h +OLD_FILES+=usr/include/clang/3.5.0/ammintrin.h +OLD_FILES+=usr/include/clang/3.5.0/arm_acle.h +OLD_FILES+=usr/include/clang/3.5.0/arm_neon.h +OLD_FILES+=usr/include/clang/3.5.0/avx2intrin.h +OLD_FILES+=usr/include/clang/3.5.0/avxintrin.h +OLD_FILES+=usr/include/clang/3.5.0/bmi2intrin.h +OLD_FILES+=usr/include/clang/3.5.0/bmiintrin.h +OLD_FILES+=usr/include/clang/3.5.0/cpuid.h +OLD_FILES+=usr/include/clang/3.5.0/emmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/f16cintrin.h +OLD_FILES+=usr/include/clang/3.5.0/fma4intrin.h +OLD_FILES+=usr/include/clang/3.5.0/fmaintrin.h +OLD_FILES+=usr/include/clang/3.5.0/ia32intrin.h +OLD_FILES+=usr/include/clang/3.5.0/immintrin.h +OLD_FILES+=usr/include/clang/3.5.0/lzcntintrin.h +OLD_FILES+=usr/include/clang/3.5.0/mm3dnow.h +OLD_FILES+=usr/include/clang/3.5.0/mm_malloc.h +OLD_FILES+=usr/include/clang/3.5.0/mmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/module.modulemap +OLD_FILES+=usr/include/clang/3.5.0/nmmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/pmmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/popcntintrin.h +OLD_FILES+=usr/include/clang/3.5.0/prfchwintrin.h +OLD_FILES+=usr/include/clang/3.5.0/rdseedintrin.h +OLD_FILES+=usr/include/clang/3.5.0/rtmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/shaintrin.h +OLD_FILES+=usr/include/clang/3.5.0/smmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/tbmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/tmmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/wmmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/x86intrin.h +OLD_FILES+=usr/include/clang/3.5.0/xmmintrin.h +OLD_FILES+=usr/include/clang/3.5.0/xopintrin.h +OLD_DIRS+=usr/include/clang/3.5.0 +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.asan-i386.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.asan-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.asan_cxx-i386.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.asan_cxx-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.profile-arm.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.profile-i386.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.profile-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.san-i386.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.san-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.ubsan-i386.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.ubsan-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.ubsan_cxx-i386.a +OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.ubsan_cxx-x86_64.a +OLD_DIRS+=usr/lib/clang/3.5.0/lib/freebsd +OLD_DIRS+=usr/lib/clang/3.5.0/lib +OLD_DIRS+=usr/lib/clang/3.5.0 # 20150102: removal of texinfo OLD_FILES+=usr/bin/info OLD_FILES+=usr/bin/infokey @@ -31,6 +31,12 @@ NOTE TO PEOPLE WHO THINK THAT FreeBSD 11.x IS SLOW: disable the most expensive debugging functionality run "ln -s 'abort:false,junk:false' /etc/malloc.conf".) +20150118: + Clang and llvm have been upgraded to 3.5.1 release. This is a bugfix + only release, no new features have been added. Please see the 20141231 + entry below for information about prerequisites and upgrading, if you + are not already using 3.5.0. + 20150107: ELF tools addr2line, elfcopy (strip), nm, size, and strings are now taken from the ELF Tool Chain project rather than GNU binutils. They diff --git a/contrib/llvm/include/llvm/Analysis/AliasSetTracker.h b/contrib/llvm/include/llvm/Analysis/AliasSetTracker.h index 6117d91..e32b6d6 100644 --- a/contrib/llvm/include/llvm/Analysis/AliasSetTracker.h +++ b/contrib/llvm/include/llvm/Analysis/AliasSetTracker.h @@ -253,13 +253,16 @@ private: const MDNode *TBAAInfo, bool KnownMustAlias = false); void addUnknownInst(Instruction *I, AliasAnalysis &AA); - void removeUnknownInst(Instruction *I) { + void removeUnknownInst(AliasSetTracker &AST, Instruction *I) { + bool WasEmpty = UnknownInsts.empty(); for (size_t i = 0, e = UnknownInsts.size(); i != e; ++i) if (UnknownInsts[i] == I) { UnknownInsts[i] = UnknownInsts.back(); UnknownInsts.pop_back(); --i; --e; // Revisit the moved entry. } + if (!WasEmpty && UnknownInsts.empty()) + dropRef(AST); } void setVolatile() { Volatile = true; } diff --git a/contrib/llvm/include/llvm/CodeGen/CallingConvLower.h b/contrib/llvm/include/llvm/CodeGen/CallingConvLower.h index 04af4bd..abe00a1 100644 --- a/contrib/llvm/include/llvm/CodeGen/CallingConvLower.h +++ b/contrib/llvm/include/llvm/CodeGen/CallingConvLower.h @@ -31,18 +31,25 @@ class TargetRegisterInfo; class CCValAssign { public: enum LocInfo { - Full, // The value fills the full location. - SExt, // The value is sign extended in the location. - ZExt, // The value is zero extended in the location. - AExt, // The value is extended with undefined upper bits. - BCvt, // The value is bit-converted in the location. - VExt, // The value is vector-widened in the location. - // FIXME: Not implemented yet. Code that uses AExt to mean - // vector-widen should be fixed to use VExt instead. - FPExt, // The floating-point value is fp-extended in the location. - Indirect // The location contains pointer to the value. + Full, // The value fills the full location. + SExt, // The value is sign extended in the location. + ZExt, // The value is zero extended in the location. + AExt, // The value is extended with undefined upper bits. + BCvt, // The value is bit-converted in the location. + VExt, // The value is vector-widened in the location. + // FIXME: Not implemented yet. Code that uses AExt to mean + // vector-widen should be fixed to use VExt instead. + FPExt, // The floating-point value is fp-extended in the location. + Indirect, // The location contains pointer to the value. + SExtUpper, // The value is in the upper bits of the location and should be + // sign extended when retrieved. + ZExtUpper, // The value is in the upper bits of the location and should be + // zero extended when retrieved. + AExtUpper // The value is in the upper bits of the location and should be + // extended with undefined upper bits when retrieved. // TODO: a subset of the value is in the location. }; + private: /// ValNo - This is the value number begin assigned (e.g. an argument number). unsigned ValNo; @@ -146,6 +153,9 @@ public: return (HTP == AExt || HTP == SExt || HTP == ZExt); } + bool isUpperBitsInLoc() const { + return HTP == AExtUpper || HTP == SExtUpper || HTP == ZExtUpper; + } }; /// CCAssignFn - This function assigns a location for Val, updating State to @@ -208,10 +218,10 @@ private: // while "%t" goes to the stack: it wouldn't be described in ByValRegs. // // Supposed use-case for this collection: - // 1. Initially ByValRegs is empty, InRegsParamsProceed is 0. + // 1. Initially ByValRegs is empty, InRegsParamsProcessed is 0. // 2. HandleByVal fillups ByValRegs. // 3. Argument analysis (LowerFormatArguments, for example). After - // some byval argument was analyzed, InRegsParamsProceed is increased. + // some byval argument was analyzed, InRegsParamsProcessed is increased. struct ByValInfo { ByValInfo(unsigned B, unsigned E, bool IsWaste = false) : Begin(B), End(E), Waste(IsWaste) {} @@ -229,9 +239,9 @@ private: }; SmallVector<ByValInfo, 4 > ByValRegs; - // InRegsParamsProceed - shows how many instances of ByValRegs was proceed + // InRegsParamsProcessed - shows how many instances of ByValRegs was proceed // during argument analysis. - unsigned InRegsParamsProceed; + unsigned InRegsParamsProcessed; protected: ParmContext CallOrPrologue; @@ -412,7 +422,7 @@ public: unsigned getInRegsParamsCount() const { return ByValRegs.size(); } // Returns count of byval in-regs arguments proceed. - unsigned getInRegsParamsProceed() const { return InRegsParamsProceed; } + unsigned getInRegsParamsProcessed() const { return InRegsParamsProcessed; } // Get information about N-th byval parameter that is stored in registers. // Here "ByValParamIndex" is N. @@ -436,20 +446,20 @@ public: // Returns false, if end is reached. bool nextInRegsParam() { unsigned e = ByValRegs.size(); - if (InRegsParamsProceed < e) - ++InRegsParamsProceed; - return InRegsParamsProceed < e; + if (InRegsParamsProcessed < e) + ++InRegsParamsProcessed; + return InRegsParamsProcessed < e; } // Clear byval registers tracking info. void clearByValRegsInfo() { - InRegsParamsProceed = 0; + InRegsParamsProcessed = 0; ByValRegs.clear(); } // Rewind byval registers tracking info. void rewindByValRegsInfo() { - InRegsParamsProceed = 0; + InRegsParamsProcessed = 0; } ParmContext getCallOrPrologue() const { return CallOrPrologue; } diff --git a/contrib/llvm/include/llvm/Target/TargetCallingConv.td b/contrib/llvm/include/llvm/Target/TargetCallingConv.td index 8f31e08..2e766c4 100644 --- a/contrib/llvm/include/llvm/Target/TargetCallingConv.td +++ b/contrib/llvm/include/llvm/Target/TargetCallingConv.td @@ -67,6 +67,9 @@ class CCIfSplit<CCAction A> : CCIf<"ArgFlags.isSplit()", A> {} /// the specified action. class CCIfSRet<CCAction A> : CCIf<"ArgFlags.isSRet()", A> {} +/// CCIfVarArg - If the current function is vararg - apply the action +class CCIfVarArg<CCAction A> : CCIf<"State.isVarArg()", A> {} + /// CCIfNotVarArg - If the current function is not vararg - apply the action class CCIfNotVarArg<CCAction A> : CCIf<"!State.isVarArg()", A> {} @@ -119,6 +122,12 @@ class CCPromoteToType<ValueType destTy> : CCAction { ValueType DestTy = destTy; } +/// CCPromoteToUpperBitsInType - If applied, this promotes the specified current +/// value to the specified type and shifts the value into the upper bits. +class CCPromoteToUpperBitsInType<ValueType destTy> : CCAction { + ValueType DestTy = destTy; +} + /// CCBitConvertToType - If applied, this bitconverts the specified current /// value to the specified type. class CCBitConvertToType<ValueType destTy> : CCAction { @@ -141,6 +150,13 @@ class CCDelegateTo<CallingConv cc> : CCAction { /// that the target supports. class CallingConv<list<CCAction> actions> { list<CCAction> Actions = actions; + bit Custom = 0; +} + +/// CustomCallingConv - An instance of this is used to declare calling +/// conventions that are implemented using a custom function of the same name. +class CustomCallingConv : CallingConv<[]> { + let Custom = 1; } /// CalleeSavedRegs - A list of callee saved registers for a given calling diff --git a/contrib/llvm/lib/Analysis/AliasSetTracker.cpp b/contrib/llvm/lib/Analysis/AliasSetTracker.cpp index a45fe23..75b6be0 100644 --- a/contrib/llvm/lib/Analysis/AliasSetTracker.cpp +++ b/contrib/llvm/lib/Analysis/AliasSetTracker.cpp @@ -55,10 +55,13 @@ void AliasSet::mergeSetIn(AliasSet &AS, AliasSetTracker &AST) { AliasTy = MayAlias; } + bool ASHadUnknownInsts = !AS.UnknownInsts.empty(); if (UnknownInsts.empty()) { // Merge call sites... - if (!AS.UnknownInsts.empty()) + if (ASHadUnknownInsts) { std::swap(UnknownInsts, AS.UnknownInsts); - } else if (!AS.UnknownInsts.empty()) { + addRef(); + } + } else if (ASHadUnknownInsts) { UnknownInsts.insert(UnknownInsts.end(), AS.UnknownInsts.begin(), AS.UnknownInsts.end()); AS.UnknownInsts.clear(); } @@ -76,6 +79,8 @@ void AliasSet::mergeSetIn(AliasSet &AS, AliasSetTracker &AST) { AS.PtrListEnd = &AS.PtrList; assert(*AS.PtrListEnd == nullptr && "End of list is not null?"); } + if (ASHadUnknownInsts) + AS.dropRef(AST); } void AliasSetTracker::removeAliasSet(AliasSet *AS) { @@ -123,6 +128,8 @@ void AliasSet::addPointer(AliasSetTracker &AST, PointerRec &Entry, } void AliasSet::addUnknownInst(Instruction *I, AliasAnalysis &AA) { + if (UnknownInsts.empty()) + addRef(); UnknownInsts.push_back(I); if (!I->mayWriteToMemory()) { @@ -218,13 +225,14 @@ AliasSet *AliasSetTracker::findAliasSetForPointer(const Value *Ptr, uint64_t Size, const MDNode *TBAAInfo) { AliasSet *FoundSet = nullptr; - for (iterator I = begin(), E = end(); I != E; ++I) { - if (I->Forward || !I->aliasesPointer(Ptr, Size, TBAAInfo, AA)) continue; + for (iterator I = begin(), E = end(); I != E;) { + iterator Cur = I++; + if (Cur->Forward || !Cur->aliasesPointer(Ptr, Size, TBAAInfo, AA)) continue; if (!FoundSet) { // If this is the first alias set ptr can go into. - FoundSet = I; // Remember it. + FoundSet = Cur; // Remember it. } else { // Otherwise, we must merge the sets. - FoundSet->mergeSetIn(*I, *this); // Merge in contents. + FoundSet->mergeSetIn(*Cur, *this); // Merge in contents. } } @@ -246,14 +254,14 @@ bool AliasSetTracker::containsPointer(Value *Ptr, uint64_t Size, AliasSet *AliasSetTracker::findAliasSetForUnknownInst(Instruction *Inst) { AliasSet *FoundSet = nullptr; - for (iterator I = begin(), E = end(); I != E; ++I) { - if (I->Forward || !I->aliasesUnknownInst(Inst, AA)) + for (iterator I = begin(), E = end(); I != E;) { + iterator Cur = I++; + if (Cur->Forward || !Cur->aliasesUnknownInst(Inst, AA)) continue; - if (!FoundSet) // If this is the first alias set ptr can go into. - FoundSet = I; // Remember it. - else if (!I->Forward) // Otherwise, we must merge the sets. - FoundSet->mergeSetIn(*I, *this); // Merge in contents. + FoundSet = Cur; // Remember it. + else if (!Cur->Forward) // Otherwise, we must merge the sets. + FoundSet->mergeSetIn(*Cur, *this); // Merge in contents. } return FoundSet; } @@ -393,6 +401,8 @@ void AliasSetTracker::add(const AliasSetTracker &AST) { /// tracker. void AliasSetTracker::remove(AliasSet &AS) { // Drop all call sites. + if (!AS.UnknownInsts.empty()) + AS.dropRef(*this); AS.UnknownInsts.clear(); // Clear the alias set. @@ -489,10 +499,10 @@ void AliasSetTracker::deleteValue(Value *PtrVal) { if (Instruction *Inst = dyn_cast<Instruction>(PtrVal)) { if (Inst->mayReadOrWriteMemory()) { // Scan all the alias sets to see if this call site is contained. - for (iterator I = begin(), E = end(); I != E; ++I) { - if (I->Forward) continue; - - I->removeUnknownInst(Inst); + for (iterator I = begin(), E = end(); I != E;) { + iterator Cur = I++; + if (!Cur->Forward) + Cur->removeUnknownInst(*this, Inst); } } } diff --git a/contrib/llvm/lib/Analysis/BlockFrequencyInfoImpl.cpp b/contrib/llvm/lib/Analysis/BlockFrequencyInfoImpl.cpp index 3203c37..723332d 100644 --- a/contrib/llvm/lib/Analysis/BlockFrequencyInfoImpl.cpp +++ b/contrib/llvm/lib/Analysis/BlockFrequencyInfoImpl.cpp @@ -14,7 +14,7 @@ #include "llvm/Analysis/BlockFrequencyInfoImpl.h" #include "llvm/ADT/SCCIterator.h" #include "llvm/Support/raw_ostream.h" -#include <deque> +#include <numeric> using namespace llvm; using namespace llvm::bfi_detail; @@ -123,8 +123,12 @@ static void combineWeight(Weight &W, const Weight &OtherW) { } assert(W.Type == OtherW.Type); assert(W.TargetNode == OtherW.TargetNode); - assert(W.Amount < W.Amount + OtherW.Amount && "Unexpected overflow"); - W.Amount += OtherW.Amount; + assert(OtherW.Amount && "Expected non-zero weight"); + if (W.Amount > W.Amount + OtherW.Amount) + // Saturate on overflow. + W.Amount = UINT64_MAX; + else + W.Amount += OtherW.Amount; } static void combineWeightsBySorting(WeightList &Weights) { // Sort so edges to the same node are adjacent. @@ -207,11 +211,19 @@ void Distribution::normalize() { Shift = 33 - countLeadingZeros(Total); // Early exit if nothing needs to be scaled. - if (!Shift) + if (!Shift) { + // If we didn't overflow then combineWeights() shouldn't have changed the + // sum of the weights, but let's double-check. + assert(Total == std::accumulate(Weights.begin(), Weights.end(), UINT64_C(0), + [](uint64_t Sum, const Weight &W) { + return Sum + W.Amount; + }) && + "Expected total to be correct"); return; + } // Recompute the total through accumulation (rather than shifting it) so that - // it's accurate after shifting. + // it's accurate after shifting and any changes combineWeights() made above. Total = 0; // Sum the weights to each node and shift right if necessary. diff --git a/contrib/llvm/lib/Analysis/ValueTracking.cpp b/contrib/llvm/lib/Analysis/ValueTracking.cpp index e6d09f4..b6a3c36 100644 --- a/contrib/llvm/lib/Analysis/ValueTracking.cpp +++ b/contrib/llvm/lib/Analysis/ValueTracking.cpp @@ -1987,23 +1987,31 @@ bool llvm::isSafeToSpeculativelyExecute(const Value *V, default: return true; case Instruction::UDiv: - case Instruction::URem: - // x / y is undefined if y == 0, but calculations like x / 3 are safe. - return isKnownNonZero(Inst->getOperand(1), TD); + case Instruction::URem: { + // x / y is undefined if y == 0. + const APInt *V; + if (match(Inst->getOperand(1), m_APInt(V))) + return *V != 0; + return false; + } case Instruction::SDiv: case Instruction::SRem: { - Value *Op = Inst->getOperand(1); - // x / y is undefined if y == 0 - if (!isKnownNonZero(Op, TD)) - return false; - // x / y might be undefined if y == -1 - unsigned BitWidth = getBitWidth(Op->getType(), TD); - if (BitWidth == 0) - return false; - APInt KnownZero(BitWidth, 0); - APInt KnownOne(BitWidth, 0); - computeKnownBits(Op, KnownZero, KnownOne, TD); - return !!KnownZero; + // x / y is undefined if y == 0 or x == INT_MIN and y == -1 + const APInt *X, *Y; + if (match(Inst->getOperand(1), m_APInt(Y))) { + if (*Y != 0) { + if (*Y == -1) { + // The numerator can't be MinSignedValue if the denominator is -1. + if (match(Inst->getOperand(0), m_APInt(X))) + return !Y->isMinSignedValue(); + // The numerator *might* be MinSignedValue. + return false; + } + // The denominator is not 0 or -1, it's safe to proceed. + return true; + } + } + return false; } case Instruction::Load: { const LoadInst *LI = cast<LoadInst>(Inst); diff --git a/contrib/llvm/lib/MC/MCObjectFileInfo.cpp b/contrib/llvm/lib/MC/MCObjectFileInfo.cpp index 6b16caa..1b52062 100644 --- a/contrib/llvm/lib/MC/MCObjectFileInfo.cpp +++ b/contrib/llvm/lib/MC/MCObjectFileInfo.cpp @@ -341,6 +341,8 @@ void MCObjectFileInfo::InitELFMCObjectFileInfo(Triple T) { break; case Triple::mips: case Triple::mipsel: + case Triple::mips64: + case Triple::mips64el: // MIPS uses indirect pointer to refer personality functions, so that the // eh_frame section can be read-only. DW.ref.personality will be generated // for relocation. diff --git a/contrib/llvm/lib/Target/ARM/ARMFrameLowering.cpp b/contrib/llvm/lib/Target/ARM/ARMFrameLowering.cpp index a67b360..3624377 100644 --- a/contrib/llvm/lib/Target/ARM/ARMFrameLowering.cpp +++ b/contrib/llvm/lib/Target/ARM/ARMFrameLowering.cpp @@ -566,11 +566,59 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const { AFI->setShouldRestoreSPFromFP(true); } +// Resolve TCReturn pseudo-instruction +void ARMFrameLowering::fixTCReturn(MachineFunction &MF, + MachineBasicBlock &MBB) const { + MachineBasicBlock::iterator MBBI = MBB.getLastNonDebugInstr(); + assert(MBBI->isReturn() && "Can only insert epilog into returning blocks"); + unsigned RetOpcode = MBBI->getOpcode(); + DebugLoc dl = MBBI->getDebugLoc(); + const ARMBaseInstrInfo &TII = + *MF.getTarget().getSubtarget<ARMSubtarget>().getInstrInfo(); + + if (!(RetOpcode == ARM::TCRETURNdi || RetOpcode == ARM::TCRETURNri)) + return; + + // Tail call return: adjust the stack pointer and jump to callee. + MBBI = MBB.getLastNonDebugInstr(); + MachineOperand &JumpTarget = MBBI->getOperand(0); + + // Jump to label or value in register. + if (RetOpcode == ARM::TCRETURNdi) { + unsigned TCOpcode = STI.isThumb() ? + (STI.isTargetMachO() ? ARM::tTAILJMPd : ARM::tTAILJMPdND) : + ARM::TAILJMPd; + MachineInstrBuilder MIB = BuildMI(MBB, MBBI, dl, TII.get(TCOpcode)); + if (JumpTarget.isGlobal()) + MIB.addGlobalAddress(JumpTarget.getGlobal(), JumpTarget.getOffset(), + JumpTarget.getTargetFlags()); + else { + assert(JumpTarget.isSymbol()); + MIB.addExternalSymbol(JumpTarget.getSymbolName(), + JumpTarget.getTargetFlags()); + } + + // Add the default predicate in Thumb mode. + if (STI.isThumb()) MIB.addImm(ARMCC::AL).addReg(0); + } else if (RetOpcode == ARM::TCRETURNri) { + BuildMI(MBB, MBBI, dl, + TII.get(STI.isThumb() ? ARM::tTAILJMPr : ARM::TAILJMPr)). + addReg(JumpTarget.getReg(), RegState::Kill); + } + + MachineInstr *NewMI = std::prev(MBBI); + for (unsigned i = 1, e = MBBI->getNumOperands(); i != e; ++i) + NewMI->addOperand(MBBI->getOperand(i)); + + // Delete the pseudo instruction TCRETURN. + MBB.erase(MBBI); + MBBI = NewMI; +} + void ARMFrameLowering::emitEpilogue(MachineFunction &MF, MachineBasicBlock &MBB) const { MachineBasicBlock::iterator MBBI = MBB.getLastNonDebugInstr(); assert(MBBI->isReturn() && "Can only insert epilog into returning blocks"); - unsigned RetOpcode = MBBI->getOpcode(); DebugLoc dl = MBBI->getDebugLoc(); MachineFrameInfo *MFI = MF.getFrameInfo(); ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>(); @@ -588,8 +636,10 @@ void ARMFrameLowering::emitEpilogue(MachineFunction &MF, // All calls are tail calls in GHC calling conv, and functions have no // prologue/epilogue. - if (MF.getFunction()->getCallingConv() == CallingConv::GHC) + if (MF.getFunction()->getCallingConv() == CallingConv::GHC) { + fixTCReturn(MF, MBB); return; + } if (!AFI->hasStackFrame()) { if (NumBytes - ArgRegsSaveSize != 0) @@ -661,42 +711,7 @@ void ARMFrameLowering::emitEpilogue(MachineFunction &MF, if (AFI->getGPRCalleeSavedArea1Size()) MBBI++; } - if (RetOpcode == ARM::TCRETURNdi || RetOpcode == ARM::TCRETURNri) { - // Tail call return: adjust the stack pointer and jump to callee. - MBBI = MBB.getLastNonDebugInstr(); - MachineOperand &JumpTarget = MBBI->getOperand(0); - - // Jump to label or value in register. - if (RetOpcode == ARM::TCRETURNdi) { - unsigned TCOpcode = STI.isThumb() ? - (STI.isTargetMachO() ? ARM::tTAILJMPd : ARM::tTAILJMPdND) : - ARM::TAILJMPd; - MachineInstrBuilder MIB = BuildMI(MBB, MBBI, dl, TII.get(TCOpcode)); - if (JumpTarget.isGlobal()) - MIB.addGlobalAddress(JumpTarget.getGlobal(), JumpTarget.getOffset(), - JumpTarget.getTargetFlags()); - else { - assert(JumpTarget.isSymbol()); - MIB.addExternalSymbol(JumpTarget.getSymbolName(), - JumpTarget.getTargetFlags()); - } - - // Add the default predicate in Thumb mode. - if (STI.isThumb()) MIB.addImm(ARMCC::AL).addReg(0); - } else if (RetOpcode == ARM::TCRETURNri) { - BuildMI(MBB, MBBI, dl, - TII.get(STI.isThumb() ? ARM::tTAILJMPr : ARM::TAILJMPr)). - addReg(JumpTarget.getReg(), RegState::Kill); - } - - MachineInstr *NewMI = std::prev(MBBI); - for (unsigned i = 1, e = MBBI->getNumOperands(); i != e; ++i) - NewMI->addOperand(MBBI->getOperand(i)); - - // Delete the pseudo instruction TCRETURN. - MBB.erase(MBBI); - MBBI = NewMI; - } + fixTCReturn(MF, MBB); if (ArgRegsSaveSize) emitSPUpdate(isARM, MBB, MBBI, dl, TII, ArgRegsSaveSize); diff --git a/contrib/llvm/lib/Target/ARM/ARMFrameLowering.h b/contrib/llvm/lib/Target/ARM/ARMFrameLowering.h index 709afbc..74629c3 100644 --- a/contrib/llvm/lib/Target/ARM/ARMFrameLowering.h +++ b/contrib/llvm/lib/Target/ARM/ARMFrameLowering.h @@ -31,6 +31,8 @@ public: void emitPrologue(MachineFunction &MF) const override; void emitEpilogue(MachineFunction &MF, MachineBasicBlock &MBB) const override; + void fixTCReturn(MachineFunction &MF, MachineBasicBlock &MBB) const; + bool spillCalleeSavedRegisters(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI, const std::vector<CalleeSavedInfo> &CSI, diff --git a/contrib/llvm/lib/Target/ARM/ARMISelLowering.cpp b/contrib/llvm/lib/Target/ARM/ARMISelLowering.cpp index a76531a..f5baf2e 100644 --- a/contrib/llvm/lib/Target/ARM/ARMISelLowering.cpp +++ b/contrib/llvm/lib/Target/ARM/ARMISelLowering.cpp @@ -1521,7 +1521,7 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, // True if this byval aggregate will be split between registers // and memory. unsigned ByValArgsCount = CCInfo.getInRegsParamsCount(); - unsigned CurByValIdx = CCInfo.getInRegsParamsProceed(); + unsigned CurByValIdx = CCInfo.getInRegsParamsProcessed(); if (CurByValIdx < ByValArgsCount) { @@ -2962,7 +2962,7 @@ ARMTargetLowering::LowerFormalArguments(SDValue Chain, if (Flags.isByVal()) { unsigned ExtraArgRegsSize; unsigned ExtraArgRegsSaveSize; - computeRegArea(CCInfo, MF, CCInfo.getInRegsParamsProceed(), + computeRegArea(CCInfo, MF, CCInfo.getInRegsParamsProcessed(), Flags.getByValSize(), ExtraArgRegsSize, ExtraArgRegsSaveSize); @@ -3086,7 +3086,7 @@ ARMTargetLowering::LowerFormalArguments(SDValue Chain, // Since they could be overwritten by lowering of arguments in case of // a tail call. if (Flags.isByVal()) { - unsigned CurByValIndex = CCInfo.getInRegsParamsProceed(); + unsigned CurByValIndex = CCInfo.getInRegsParamsProcessed(); ByValStoreOffset = RoundUpToAlignment(ByValStoreOffset, Flags.getByValAlign()); int FrameIndex = StoreByValRegs( diff --git a/contrib/llvm/lib/Target/Mips/AsmParser/MipsAsmParser.cpp b/contrib/llvm/lib/Target/Mips/AsmParser/MipsAsmParser.cpp index 53b30f9..2d65b08 100644 --- a/contrib/llvm/lib/Target/Mips/AsmParser/MipsAsmParser.cpp +++ b/contrib/llvm/lib/Target/Mips/AsmParser/MipsAsmParser.cpp @@ -200,14 +200,14 @@ class MipsAsmParser : public MCTargetAsmParser { // Example: INSERT.B $w0[n], $1 => 16 > n >= 0 bool validateMSAIndex(int Val, int RegKind); - void setFeatureBits(unsigned Feature, StringRef FeatureString) { + void setFeatureBits(uint64_t Feature, StringRef FeatureString) { if (!(STI.getFeatureBits() & Feature)) { setAvailableFeatures( ComputeAvailableFeatures(STI.ToggleFeature(FeatureString))); } } - void clearFeatureBits(unsigned Feature, StringRef FeatureString) { + void clearFeatureBits(uint64_t Feature, StringRef FeatureString) { if (STI.getFeatureBits() & Feature) { setAvailableFeatures( ComputeAvailableFeatures(STI.ToggleFeature(FeatureString))); diff --git a/contrib/llvm/lib/Target/Mips/Disassembler/MipsDisassembler.cpp b/contrib/llvm/lib/Target/Mips/Disassembler/MipsDisassembler.cpp index f35a8de..53fb7a1 100644 --- a/contrib/llvm/lib/Target/Mips/Disassembler/MipsDisassembler.cpp +++ b/contrib/llvm/lib/Target/Mips/Disassembler/MipsDisassembler.cpp @@ -250,6 +250,11 @@ static DecodeStatus DecodeMem(MCInst &Inst, uint64_t Address, const void *Decoder); +static DecodeStatus DecodeCacheOp(MCInst &Inst, + unsigned Insn, + uint64_t Address, + const void *Decoder); + static DecodeStatus DecodeMSA128Mem(MCInst &Inst, unsigned Insn, uint64_t Address, const void *Decoder); @@ -267,6 +272,14 @@ static DecodeStatus DecodeFMem(MCInst &Inst, unsigned Insn, uint64_t Address, const void *Decoder); +static DecodeStatus DecodeFMem2(MCInst &Inst, unsigned Insn, + uint64_t Address, + const void *Decoder); + +static DecodeStatus DecodeFMem3(MCInst &Inst, unsigned Insn, + uint64_t Address, + const void *Decoder); + static DecodeStatus DecodeSpecial3LlSc(MCInst &Inst, unsigned Insn, uint64_t Address, @@ -451,7 +464,7 @@ static DecodeStatus DecodeAddiGroupBranch(MCInst &MI, InsnType insn, InsnType Rs = fieldFromInstruction(insn, 21, 5); InsnType Rt = fieldFromInstruction(insn, 16, 5); - InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) << 2; + InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) * 4; bool HasRs = false; if (Rs >= Rt) { @@ -490,7 +503,7 @@ static DecodeStatus DecodeDaddiGroupBranch(MCInst &MI, InsnType insn, InsnType Rs = fieldFromInstruction(insn, 21, 5); InsnType Rt = fieldFromInstruction(insn, 16, 5); - InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) << 2; + InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) * 4; bool HasRs = false; if (Rs >= Rt) { @@ -530,7 +543,7 @@ static DecodeStatus DecodeBlezlGroupBranch(MCInst &MI, InsnType insn, InsnType Rs = fieldFromInstruction(insn, 21, 5); InsnType Rt = fieldFromInstruction(insn, 16, 5); - InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) << 2; + InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) * 4; bool HasRs = false; if (Rt == 0) @@ -575,7 +588,7 @@ static DecodeStatus DecodeBgtzlGroupBranch(MCInst &MI, InsnType insn, InsnType Rs = fieldFromInstruction(insn, 21, 5); InsnType Rt = fieldFromInstruction(insn, 16, 5); - InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) << 2; + InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) * 4; if (Rt == 0) return MCDisassembler::Fail; @@ -617,7 +630,7 @@ static DecodeStatus DecodeBgtzGroupBranch(MCInst &MI, InsnType insn, InsnType Rs = fieldFromInstruction(insn, 21, 5); InsnType Rt = fieldFromInstruction(insn, 16, 5); - InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) << 2; + InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) * 4; bool HasRs = false; bool HasRt = false; @@ -666,7 +679,7 @@ static DecodeStatus DecodeBlezGroupBranch(MCInst &MI, InsnType insn, InsnType Rs = fieldFromInstruction(insn, 21, 5); InsnType Rt = fieldFromInstruction(insn, 16, 5); - InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) << 2; + InsnType Imm = SignExtend64(fieldFromInstruction(insn, 0, 16), 16) * 4; bool HasRs = false; if (Rt == 0) @@ -964,6 +977,23 @@ static DecodeStatus DecodeMem(MCInst &Inst, return MCDisassembler::Success; } +static DecodeStatus DecodeCacheOp(MCInst &Inst, + unsigned Insn, + uint64_t Address, + const void *Decoder) { + int Offset = SignExtend32<16>(Insn & 0xffff); + unsigned Hint = fieldFromInstruction(Insn, 16, 5); + unsigned Base = fieldFromInstruction(Insn, 21, 5); + + Base = getReg(Decoder, Mips::GPR32RegClassID, Base); + + Inst.addOperand(MCOperand::CreateReg(Base)); + Inst.addOperand(MCOperand::CreateImm(Offset)); + Inst.addOperand(MCOperand::CreateImm(Hint)); + + return MCDisassembler::Success; +} + static DecodeStatus DecodeMSA128Mem(MCInst &Inst, unsigned Insn, uint64_t Address, const void *Decoder) { int Offset = SignExtend32<10>(fieldFromInstruction(Insn, 16, 10)); @@ -995,15 +1025,15 @@ static DecodeStatus DecodeMSA128Mem(MCInst &Inst, unsigned Insn, break; case Mips::LD_H: case Mips::ST_H: - Inst.addOperand(MCOperand::CreateImm(Offset << 1)); + Inst.addOperand(MCOperand::CreateImm(Offset * 2)); break; case Mips::LD_W: case Mips::ST_W: - Inst.addOperand(MCOperand::CreateImm(Offset << 2)); + Inst.addOperand(MCOperand::CreateImm(Offset * 4)); break; case Mips::LD_D: case Mips::ST_D: - Inst.addOperand(MCOperand::CreateImm(Offset << 3)); + Inst.addOperand(MCOperand::CreateImm(Offset * 8)); break; } @@ -1067,6 +1097,42 @@ static DecodeStatus DecodeFMem(MCInst &Inst, return MCDisassembler::Success; } +static DecodeStatus DecodeFMem2(MCInst &Inst, + unsigned Insn, + uint64_t Address, + const void *Decoder) { + int Offset = SignExtend32<16>(Insn & 0xffff); + unsigned Reg = fieldFromInstruction(Insn, 16, 5); + unsigned Base = fieldFromInstruction(Insn, 21, 5); + + Reg = getReg(Decoder, Mips::COP2RegClassID, Reg); + Base = getReg(Decoder, Mips::GPR32RegClassID, Base); + + Inst.addOperand(MCOperand::CreateReg(Reg)); + Inst.addOperand(MCOperand::CreateReg(Base)); + Inst.addOperand(MCOperand::CreateImm(Offset)); + + return MCDisassembler::Success; +} + +static DecodeStatus DecodeFMem3(MCInst &Inst, + unsigned Insn, + uint64_t Address, + const void *Decoder) { + int Offset = SignExtend32<16>(Insn & 0xffff); + unsigned Reg = fieldFromInstruction(Insn, 16, 5); + unsigned Base = fieldFromInstruction(Insn, 21, 5); + + Reg = getReg(Decoder, Mips::COP3RegClassID, Reg); + Base = getReg(Decoder, Mips::GPR32RegClassID, Base); + + Inst.addOperand(MCOperand::CreateReg(Reg)); + Inst.addOperand(MCOperand::CreateReg(Base)); + Inst.addOperand(MCOperand::CreateImm(Offset)); + + return MCDisassembler::Success; +} + static DecodeStatus DecodeSpecial3LlSc(MCInst &Inst, unsigned Insn, uint64_t Address, @@ -1225,7 +1291,7 @@ static DecodeStatus DecodeBranchTarget(MCInst &Inst, unsigned Offset, uint64_t Address, const void *Decoder) { - int32_t BranchOffset = (SignExtend32<16>(Offset) << 2) + 4; + int32_t BranchOffset = (SignExtend32<16>(Offset) * 4) + 4; Inst.addOperand(MCOperand::CreateImm(BranchOffset)); return MCDisassembler::Success; } @@ -1244,7 +1310,7 @@ static DecodeStatus DecodeBranchTarget21(MCInst &Inst, unsigned Offset, uint64_t Address, const void *Decoder) { - int32_t BranchOffset = SignExtend32<21>(Offset) << 2; + int32_t BranchOffset = SignExtend32<21>(Offset) * 4; Inst.addOperand(MCOperand::CreateImm(BranchOffset)); return MCDisassembler::Success; @@ -1254,7 +1320,7 @@ static DecodeStatus DecodeBranchTarget26(MCInst &Inst, unsigned Offset, uint64_t Address, const void *Decoder) { - int32_t BranchOffset = SignExtend32<26>(Offset) << 2; + int32_t BranchOffset = SignExtend32<26>(Offset) * 4; Inst.addOperand(MCOperand::CreateImm(BranchOffset)); return MCDisassembler::Success; @@ -1264,7 +1330,7 @@ static DecodeStatus DecodeBranchTargetMM(MCInst &Inst, unsigned Offset, uint64_t Address, const void *Decoder) { - int32_t BranchOffset = SignExtend32<16>(Offset) << 1; + int32_t BranchOffset = SignExtend32<16>(Offset) * 2; Inst.addOperand(MCOperand::CreateImm(BranchOffset)); return MCDisassembler::Success; } @@ -1317,12 +1383,12 @@ static DecodeStatus DecodeExtSize(MCInst &Inst, static DecodeStatus DecodeSimm19Lsl2(MCInst &Inst, unsigned Insn, uint64_t Address, const void *Decoder) { - Inst.addOperand(MCOperand::CreateImm(SignExtend32<19>(Insn) << 2)); + Inst.addOperand(MCOperand::CreateImm(SignExtend32<19>(Insn) * 4)); return MCDisassembler::Success; } static DecodeStatus DecodeSimm18Lsl3(MCInst &Inst, unsigned Insn, uint64_t Address, const void *Decoder) { - Inst.addOperand(MCOperand::CreateImm(SignExtend32<18>(Insn) << 3)); + Inst.addOperand(MCOperand::CreateImm(SignExtend32<18>(Insn) * 8)); return MCDisassembler::Success; } diff --git a/contrib/llvm/lib/Target/Mips/Mips.td b/contrib/llvm/lib/Target/Mips/Mips.td index dd3bc9b..3e1d047 100644 --- a/contrib/llvm/lib/Target/Mips/Mips.td +++ b/contrib/llvm/lib/Target/Mips/Mips.td @@ -57,6 +57,8 @@ def MipsInstrInfo : InstrInfo; // Mips Subtarget features // //===----------------------------------------------------------------------===// +def FeatureNoABICalls : SubtargetFeature<"noabicalls", "NoABICalls", "true", + "Disable SVR4-style position-independent code.">; def FeatureGP64Bit : SubtargetFeature<"gp64", "IsGP64bit", "true", "General Purpose Registers are 64-bit wide.">; def FeatureFP64Bit : SubtargetFeature<"fp64", "IsFP64bit", "true", @@ -67,13 +69,13 @@ def FeatureNaN2008 : SubtargetFeature<"nan2008", "IsNaN2008bit", "true", "IEEE 754-2008 NaN encoding.">; def FeatureSingleFloat : SubtargetFeature<"single-float", "IsSingleFloat", "true", "Only supports single precision float">; -def FeatureO32 : SubtargetFeature<"o32", "MipsABI", "O32", +def FeatureO32 : SubtargetFeature<"o32", "ABI", "MipsABIInfo::O32()", "Enable o32 ABI">; -def FeatureN32 : SubtargetFeature<"n32", "MipsABI", "N32", +def FeatureN32 : SubtargetFeature<"n32", "ABI", "MipsABIInfo::N32()", "Enable n32 ABI">; -def FeatureN64 : SubtargetFeature<"n64", "MipsABI", "N64", +def FeatureN64 : SubtargetFeature<"n64", "ABI", "MipsABIInfo::N64()", "Enable n64 ABI">; -def FeatureEABI : SubtargetFeature<"eabi", "MipsABI", "EABI", +def FeatureEABI : SubtargetFeature<"eabi", "ABI", "MipsABIInfo::EABI()", "Enable eabi ABI">; def FeatureNoOddSPReg : SubtargetFeature<"nooddspreg", "UseOddSPReg", "false", "Disable odd numbered single-precision " diff --git a/contrib/llvm/lib/Target/Mips/Mips16ISelLowering.cpp b/contrib/llvm/lib/Target/Mips/Mips16ISelLowering.cpp index 587925d..9576fd4 100644 --- a/contrib/llvm/lib/Target/Mips/Mips16ISelLowering.cpp +++ b/contrib/llvm/lib/Target/Mips/Mips16ISelLowering.cpp @@ -241,10 +241,9 @@ Mips16TargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI, } } -bool Mips16TargetLowering:: -isEligibleForTailCallOptimization(const MipsCC &MipsCCInfo, - unsigned NextStackOffset, - const MipsFunctionInfo& FI) const { +bool Mips16TargetLowering::isEligibleForTailCallOptimization( + const CCState &CCInfo, unsigned NextStackOffset, + const MipsFunctionInfo &FI) const { // No tail call optimization for mips16. return false; } diff --git a/contrib/llvm/lib/Target/Mips/Mips16ISelLowering.h b/contrib/llvm/lib/Target/Mips/Mips16ISelLowering.h index e7e4d7f..1649fe6 100644 --- a/contrib/llvm/lib/Target/Mips/Mips16ISelLowering.h +++ b/contrib/llvm/lib/Target/Mips/Mips16ISelLowering.h @@ -30,9 +30,9 @@ namespace llvm { MachineBasicBlock *MBB) const override; private: - bool isEligibleForTailCallOptimization(const MipsCC &MipsCCInfo, - unsigned NextStackOffset, - const MipsFunctionInfo& FI) const override; + bool isEligibleForTailCallOptimization( + const CCState &CCInfo, unsigned NextStackOffset, + const MipsFunctionInfo &FI) const override; void setMips16HardFloatLibCalls(); diff --git a/contrib/llvm/lib/Target/Mips/Mips64InstrInfo.td b/contrib/llvm/lib/Target/Mips/Mips64InstrInfo.td index f0b6814..f2072a6 100644 --- a/contrib/llvm/lib/Target/Mips/Mips64InstrInfo.td +++ b/contrib/llvm/lib/Target/Mips/Mips64InstrInfo.td @@ -419,6 +419,10 @@ defm : SetgePats<GPR64, SLT64, SLTu64>; defm : SetgeImmPats<GPR64, SLTi64, SLTiu64>; // truncate +def : MipsPat<(trunc (assertsext GPR64:$src)), + (EXTRACT_SUBREG GPR64:$src, sub_32)>; +def : MipsPat<(trunc (assertzext GPR64:$src)), + (EXTRACT_SUBREG GPR64:$src, sub_32)>; def : MipsPat<(i32 (trunc GPR64:$src)), (SLL (EXTRACT_SUBREG GPR64:$src, sub_32), 0)>; diff --git a/contrib/llvm/lib/Target/Mips/MipsABIInfo.cpp b/contrib/llvm/lib/Target/Mips/MipsABIInfo.cpp new file mode 100644 index 0000000..f885369 --- /dev/null +++ b/contrib/llvm/lib/Target/Mips/MipsABIInfo.cpp @@ -0,0 +1,45 @@ +//===---- MipsABIInfo.cpp - Information about MIPS ABI's ------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "MipsABIInfo.h" +#include "MipsRegisterInfo.h" + +using namespace llvm; + +namespace { +static const MCPhysReg O32IntRegs[4] = {Mips::A0, Mips::A1, Mips::A2, Mips::A3}; + +static const MCPhysReg Mips64IntRegs[8] = { + Mips::A0_64, Mips::A1_64, Mips::A2_64, Mips::A3_64, + Mips::T0_64, Mips::T1_64, Mips::T2_64, Mips::T3_64}; +} + +const ArrayRef<MCPhysReg> MipsABIInfo::GetByValArgRegs() const { + if (IsO32()) + return makeArrayRef(O32IntRegs); + if (IsN32() || IsN64()) + return makeArrayRef(Mips64IntRegs); + llvm_unreachable("Unhandled ABI"); +} + +const ArrayRef<MCPhysReg> MipsABIInfo::GetVarArgRegs() const { + if (IsO32()) + return makeArrayRef(O32IntRegs); + if (IsN32() || IsN64()) + return makeArrayRef(Mips64IntRegs); + llvm_unreachable("Unhandled ABI"); +} + +unsigned MipsABIInfo::GetCalleeAllocdArgSizeInBytes(CallingConv::ID CC) const { + if (IsO32()) + return CC != CallingConv::Fast ? 16 : 0; + if (IsN32() || IsN64() || IsEABI()) + return 0; + llvm_unreachable("Unhandled ABI"); +} diff --git a/contrib/llvm/lib/Target/Mips/MipsABIInfo.h b/contrib/llvm/lib/Target/Mips/MipsABIInfo.h new file mode 100644 index 0000000..bea585e --- /dev/null +++ b/contrib/llvm/lib/Target/Mips/MipsABIInfo.h @@ -0,0 +1,61 @@ +//===---- MipsABIInfo.h - Information about MIPS ABI's --------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#ifndef MIPSABIINFO_H +#define MIPSABIINFO_H + +#include "llvm/ADT/ArrayRef.h" +#include "llvm/MC/MCRegisterInfo.h" +#include "llvm/IR/CallingConv.h" + +namespace llvm { + +class MipsABIInfo { +public: + enum class ABI { Unknown, O32, N32, N64, EABI }; + +protected: + ABI ThisABI; + +public: + MipsABIInfo(ABI ThisABI) : ThisABI(ThisABI) {} + + static MipsABIInfo Unknown() { return MipsABIInfo(ABI::Unknown); } + static MipsABIInfo O32() { return MipsABIInfo(ABI::O32); } + static MipsABIInfo N32() { return MipsABIInfo(ABI::N32); } + static MipsABIInfo N64() { return MipsABIInfo(ABI::N64); } + static MipsABIInfo EABI() { return MipsABIInfo(ABI::EABI); } + + bool IsKnown() const { return ThisABI != ABI::Unknown; } + bool IsO32() const { return ThisABI == ABI::O32; } + bool IsN32() const { return ThisABI == ABI::N32; } + bool IsN64() const { return ThisABI == ABI::N64; } + bool IsEABI() const { return ThisABI == ABI::EABI; } + ABI GetEnumValue() const { return ThisABI; } + + /// The registers to use for byval arguments. + const ArrayRef<MCPhysReg> GetByValArgRegs() const; + + /// The registers to use for the variable argument list. + const ArrayRef<MCPhysReg> GetVarArgRegs() const; + + /// Obtain the size of the area allocated by the callee for arguments. + /// CallingConv::FastCall affects the value for O32. + unsigned GetCalleeAllocdArgSizeInBytes(CallingConv::ID CC) const; + + /// Ordering of ABI's + /// MipsGenSubtargetInfo.inc will use this to resolve conflicts when given + /// multiple ABI options. + bool operator<(const MipsABIInfo Other) const { + return ThisABI < Other.GetEnumValue(); + } +}; +} + +#endif diff --git a/contrib/llvm/lib/Target/Mips/MipsAsmPrinter.cpp b/contrib/llvm/lib/Target/Mips/MipsAsmPrinter.cpp index 7f21d68..14463fc 100644 --- a/contrib/llvm/lib/Target/Mips/MipsAsmPrinter.cpp +++ b/contrib/llvm/lib/Target/Mips/MipsAsmPrinter.cpp @@ -317,11 +317,11 @@ void MipsAsmPrinter::emitFrameDirective() { /// Emit Set directives. const char *MipsAsmPrinter::getCurrentABIString() const { - switch (Subtarget->getTargetABI()) { - case MipsSubtarget::O32: return "abi32"; - case MipsSubtarget::N32: return "abiN32"; - case MipsSubtarget::N64: return "abi64"; - case MipsSubtarget::EABI: return "eabi32"; // TODO: handle eabi64 + switch (Subtarget->getABI().GetEnumValue()) { + case MipsABIInfo::ABI::O32: return "abi32"; + case MipsABIInfo::ABI::N32: return "abiN32"; + case MipsABIInfo::ABI::N64: return "abi64"; + case MipsABIInfo::ABI::EABI: return "eabi32"; // TODO: handle eabi64 default: llvm_unreachable("Unknown Mips ABI"); } } @@ -471,14 +471,12 @@ bool MipsAsmPrinter::PrintAsmOperand(const MachineInstr *MI, unsigned OpNum, return false; case 'z': { // $0 if zero, regular printing otherwise - if (MO.getType() != MachineOperand::MO_Immediate) - return true; - int64_t Val = MO.getImm(); - if (Val) - O << Val; - else + if (MO.getType() == MachineOperand::MO_Immediate && MO.getImm() == 0) { O << "$0"; - return false; + return false; + } + // If not, call printOperand as normal. + break; } case 'D': // Second part of a double word register operand case 'L': // Low order register of a double word register operand @@ -669,9 +667,7 @@ printFCCOperand(const MachineInstr *MI, int opNum, raw_ostream &O, } void MipsAsmPrinter::EmitStartOfAsmFile(Module &M) { - // TODO: Need to add -mabicalls and -mno-abicalls flags. - // Currently we assume that -mabicalls is the default. - bool IsABICalls = true; + bool IsABICalls = Subtarget->isABICalls(); if (IsABICalls) { getTargetStreamer().emitDirectiveAbiCalls(); Reloc::Model RM = TM.getRelocationModel(); diff --git a/contrib/llvm/lib/Target/Mips/MipsCCState.cpp b/contrib/llvm/lib/Target/Mips/MipsCCState.cpp new file mode 100644 index 0000000..e18cc8b --- /dev/null +++ b/contrib/llvm/lib/Target/Mips/MipsCCState.cpp @@ -0,0 +1,142 @@ +//===---- MipsCCState.cpp - CCState with Mips specific extensions ---------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "MipsCCState.h" +#include "MipsSubtarget.h" +#include "llvm/IR/Module.h" + +using namespace llvm; + +/// This function returns true if CallSym is a long double emulation routine. +static bool isF128SoftLibCall(const char *CallSym) { + const char *const LibCalls[] = { + "__addtf3", "__divtf3", "__eqtf2", "__extenddftf2", + "__extendsftf2", "__fixtfdi", "__fixtfsi", "__fixtfti", + "__fixunstfdi", "__fixunstfsi", "__fixunstfti", "__floatditf", + "__floatsitf", "__floattitf", "__floatunditf", "__floatunsitf", + "__floatuntitf", "__getf2", "__gttf2", "__letf2", + "__lttf2", "__multf3", "__netf2", "__powitf2", + "__subtf3", "__trunctfdf2", "__trunctfsf2", "__unordtf2", + "ceill", "copysignl", "cosl", "exp2l", + "expl", "floorl", "fmal", "fmodl", + "log10l", "log2l", "logl", "nearbyintl", + "powl", "rintl", "sinl", "sqrtl", + "truncl"}; + + const char *const *End = LibCalls + array_lengthof(LibCalls); + + // Check that LibCalls is sorted alphabetically. + MipsTargetLowering::LTStr Comp; + +#ifndef NDEBUG + for (const char *const *I = LibCalls; I < End - 1; ++I) + assert(Comp(*I, *(I + 1))); +#endif + + return std::binary_search(LibCalls, End, CallSym, Comp); +} + +/// This function returns true if Ty is fp128, {f128} or i128 which was +/// originally a fp128. +static bool originalTypeIsF128(const Type *Ty, const SDNode *CallNode) { + if (Ty->isFP128Ty()) + return true; + + if (Ty->isStructTy() && Ty->getStructNumElements() == 1 && + Ty->getStructElementType(0)->isFP128Ty()) + return true; + + const ExternalSymbolSDNode *ES = + dyn_cast_or_null<const ExternalSymbolSDNode>(CallNode); + + // If the Ty is i128 and the function being called is a long double emulation + // routine, then the original type is f128. + return (ES && Ty->isIntegerTy(128) && isF128SoftLibCall(ES->getSymbol())); +} + +MipsCCState::SpecialCallingConvType +MipsCCState::getSpecialCallingConvForCallee(const SDNode *Callee, + const MipsSubtarget &Subtarget) { + MipsCCState::SpecialCallingConvType SpecialCallingConv = NoSpecialCallingConv; + if (Subtarget.inMips16HardFloat()) { + if (const GlobalAddressSDNode *G = + dyn_cast<const GlobalAddressSDNode>(Callee)) { + llvm::StringRef Sym = G->getGlobal()->getName(); + Function *F = G->getGlobal()->getParent()->getFunction(Sym); + if (F && F->hasFnAttribute("__Mips16RetHelper")) { + SpecialCallingConv = Mips16RetHelperConv; + } + } + } + return SpecialCallingConv; +} + +void MipsCCState::PreAnalyzeCallResultForF128( + const SmallVectorImpl<ISD::InputArg> &Ins, + const TargetLowering::CallLoweringInfo &CLI) { + for (unsigned i = 0; i < Ins.size(); ++i) { + OriginalArgWasF128.push_back( + originalTypeIsF128(CLI.RetTy, CLI.Callee.getNode())); + OriginalArgWasFloat.push_back(CLI.RetTy->isFloatingPointTy()); + } +} + +/// Identify lowered values that originated from f128 arguments and record +/// this for use by RetCC_MipsN. +void MipsCCState::PreAnalyzeReturnForF128( + const SmallVectorImpl<ISD::OutputArg> &Outs) { + const MachineFunction &MF = getMachineFunction(); + for (unsigned i = 0; i < Outs.size(); ++i) { + OriginalArgWasF128.push_back( + originalTypeIsF128(MF.getFunction()->getReturnType(), nullptr)); + OriginalArgWasFloat.push_back( + MF.getFunction()->getReturnType()->isFloatingPointTy()); + } +} + +/// Identify lowered values that originated from f128 arguments and record +/// this. +void MipsCCState::PreAnalyzeCallOperands( + const SmallVectorImpl<ISD::OutputArg> &Outs, + std::vector<TargetLowering::ArgListEntry> &FuncArgs, + const SDNode *CallNode) { + for (unsigned i = 0; i < Outs.size(); ++i) { + OriginalArgWasF128.push_back( + originalTypeIsF128(FuncArgs[Outs[i].OrigArgIndex].Ty, CallNode)); + OriginalArgWasFloat.push_back( + FuncArgs[Outs[i].OrigArgIndex].Ty->isFloatingPointTy()); + CallOperandIsFixed.push_back(Outs[i].IsFixed); + } +} + +/// Identify lowered values that originated from f128 arguments and record +/// this. +void MipsCCState::PreAnalyzeFormalArgumentsForF128( + const SmallVectorImpl<ISD::InputArg> &Ins) { + const MachineFunction &MF = getMachineFunction(); + for (unsigned i = 0; i < Ins.size(); ++i) { + Function::const_arg_iterator FuncArg = MF.getFunction()->arg_begin(); + + // SRet arguments cannot originate from f128 or {f128} returns so we just + // push false. We have to handle this specially since SRet arguments + // aren't mapped to an original argument. + if (Ins[i].Flags.isSRet()) { + OriginalArgWasF128.push_back(false); + OriginalArgWasFloat.push_back(false); + continue; + } + + assert(Ins[i].OrigArgIndex < MF.getFunction()->arg_size()); + std::advance(FuncArg, Ins[i].OrigArgIndex); + + OriginalArgWasF128.push_back( + originalTypeIsF128(FuncArg->getType(), nullptr)); + OriginalArgWasFloat.push_back(FuncArg->getType()->isFloatingPointTy()); + } +} diff --git a/contrib/llvm/lib/Target/Mips/MipsCCState.h b/contrib/llvm/lib/Target/Mips/MipsCCState.h new file mode 100644 index 0000000..026b7b5 --- /dev/null +++ b/contrib/llvm/lib/Target/Mips/MipsCCState.h @@ -0,0 +1,137 @@ +//===---- MipsCCState.h - CCState with Mips specific extensions -----------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#ifndef MIPSCCSTATE_H +#define MIPSCCSTATE_H + +#include "llvm/ADT/SmallVector.h" +#include "llvm/CodeGen/CallingConvLower.h" +#include "MipsISelLowering.h" + +namespace llvm { +class SDNode; +class MipsSubtarget; + +class MipsCCState : public CCState { +public: + enum SpecialCallingConvType { Mips16RetHelperConv, NoSpecialCallingConv }; + + /// Determine the SpecialCallingConvType for the given callee + static SpecialCallingConvType + getSpecialCallingConvForCallee(const SDNode *Callee, + const MipsSubtarget &Subtarget); + +private: + /// Identify lowered values that originated from f128 arguments and record + /// this for use by RetCC_MipsN. + void PreAnalyzeCallResultForF128(const SmallVectorImpl<ISD::InputArg> &Ins, + const TargetLowering::CallLoweringInfo &CLI); + + /// Identify lowered values that originated from f128 arguments and record + /// this for use by RetCC_MipsN. + void PreAnalyzeReturnForF128(const SmallVectorImpl<ISD::OutputArg> &Outs); + + /// Identify lowered values that originated from f128 arguments and record + /// this. + void + PreAnalyzeCallOperands(const SmallVectorImpl<ISD::OutputArg> &Outs, + std::vector<TargetLowering::ArgListEntry> &FuncArgs, + const SDNode *CallNode); + + /// Identify lowered values that originated from f128 arguments and record + /// this. + void + PreAnalyzeFormalArgumentsForF128(const SmallVectorImpl<ISD::InputArg> &Ins); + + /// Records whether the value has been lowered from an f128. + SmallVector<bool, 4> OriginalArgWasF128; + + /// Records whether the value has been lowered from float. + SmallVector<bool, 4> OriginalArgWasFloat; + + /// Records whether the value was a fixed argument. + /// See ISD::OutputArg::IsFixed, + SmallVector<bool, 4> CallOperandIsFixed; + + // Used to handle MIPS16-specific calling convention tweaks. + // FIXME: This should probably be a fully fledged calling convention. + SpecialCallingConvType SpecialCallingConv; + +public: + MipsCCState(CallingConv::ID CC, bool isVarArg, MachineFunction &MF, + const TargetMachine &TM, SmallVectorImpl<CCValAssign> &locs, + LLVMContext &C, + SpecialCallingConvType SpecialCC = NoSpecialCallingConv) + : CCState(CC, isVarArg, MF, TM, locs, C), SpecialCallingConv(SpecialCC) {} + + void + AnalyzeCallOperands(const SmallVectorImpl<ISD::OutputArg> &Outs, + CCAssignFn Fn, + std::vector<TargetLowering::ArgListEntry> &FuncArgs, + const SDNode *CallNode) { + PreAnalyzeCallOperands(Outs, FuncArgs, CallNode); + CCState::AnalyzeCallOperands(Outs, Fn); + OriginalArgWasF128.clear(); + OriginalArgWasFloat.clear(); + CallOperandIsFixed.clear(); + } + + // The AnalyzeCallOperands in the base class is not usable since we must + // provide a means of accessing ArgListEntry::IsFixed. Delete them from this + // class. This doesn't stop them being used via the base class though. + void AnalyzeCallOperands(const SmallVectorImpl<ISD::OutputArg> &Outs, + CCAssignFn Fn) LLVM_DELETED_FUNCTION; + void AnalyzeCallOperands(const SmallVectorImpl<MVT> &Outs, + SmallVectorImpl<ISD::ArgFlagsTy> &Flags, + CCAssignFn Fn) LLVM_DELETED_FUNCTION; + + void AnalyzeFormalArguments(const SmallVectorImpl<ISD::InputArg> &Ins, + CCAssignFn Fn) { + PreAnalyzeFormalArgumentsForF128(Ins); + CCState::AnalyzeFormalArguments(Ins, Fn); + OriginalArgWasFloat.clear(); + OriginalArgWasF128.clear(); + } + + void AnalyzeCallResult(const SmallVectorImpl<ISD::InputArg> &Ins, + CCAssignFn Fn, + const TargetLowering::CallLoweringInfo &CLI) { + PreAnalyzeCallResultForF128(Ins, CLI); + CCState::AnalyzeCallResult(Ins, Fn); + OriginalArgWasFloat.clear(); + OriginalArgWasF128.clear(); + } + + void AnalyzeReturn(const SmallVectorImpl<ISD::OutputArg> &Outs, + CCAssignFn Fn) { + PreAnalyzeReturnForF128(Outs); + CCState::AnalyzeReturn(Outs, Fn); + OriginalArgWasFloat.clear(); + OriginalArgWasF128.clear(); + } + + bool CheckReturn(const SmallVectorImpl<ISD::OutputArg> &ArgsFlags, + CCAssignFn Fn) { + PreAnalyzeReturnForF128(ArgsFlags); + bool Return = CCState::CheckReturn(ArgsFlags, Fn); + OriginalArgWasFloat.clear(); + OriginalArgWasF128.clear(); + return Return; + } + + bool WasOriginalArgF128(unsigned ValNo) { return OriginalArgWasF128[ValNo]; } + bool WasOriginalArgFloat(unsigned ValNo) { + return OriginalArgWasFloat[ValNo]; + } + bool IsCallOperandFixed(unsigned ValNo) { return CallOperandIsFixed[ValNo]; } + SpecialCallingConvType getSpecialCallingConv() { return SpecialCallingConv; } +}; +} + +#endif diff --git a/contrib/llvm/lib/Target/Mips/MipsCallingConv.td b/contrib/llvm/lib/Target/Mips/MipsCallingConv.td index b1cd3c3..76297dc 100644 --- a/contrib/llvm/lib/Target/Mips/MipsCallingConv.td +++ b/contrib/llvm/lib/Target/Mips/MipsCallingConv.td @@ -10,8 +10,42 @@ //===----------------------------------------------------------------------===// /// CCIfSubtarget - Match if the current subtarget has a feature F. -class CCIfSubtarget<string F, CCAction A>: - CCIf<!strconcat("State.getTarget().getSubtarget<MipsSubtarget>().", F), A>; +class CCIfSubtarget<string F, CCAction A, string Invert = ""> + : CCIf<!strconcat(Invert, + "State.getMachineFunction().getTarget()." + "getSubtarget<const MipsSubtarget>().", + F), + A>; + +// The inverse of CCIfSubtarget +class CCIfSubtargetNot<string F, CCAction A> : CCIfSubtarget<F, A, "!">; + +// For soft-float, f128 values are returned in A0_64 rather than V1_64. +def RetCC_F128SoftFloat : CallingConv<[ + CCAssignToReg<[V0_64, A0_64]> +]>; + +// For hard-float, f128 values are returned as a pair of f64's rather than a +// pair of i64's. +def RetCC_F128HardFloat : CallingConv<[ + CCBitConvertToType<f64>, + + // Contrary to the ABI documentation, a struct containing a long double is + // returned in $f0, and $f1 instead of the usual $f0, and $f2. This is to + // match the de facto ABI as implemented by GCC. + CCIfInReg<CCAssignToReg<[D0_64, D1_64]>>, + + CCAssignToReg<[D0_64, D2_64]> +]>; + +// Handle F128 specially since we can't identify the original type during the +// tablegen-erated code. +def RetCC_F128 : CallingConv<[ + CCIfSubtarget<"abiUsesSoftFloat()", + CCIfType<[i64], CCDelegateTo<RetCC_F128SoftFloat>>>, + CCIfSubtargetNot<"abiUsesSoftFloat()", + CCIfType<[i64], CCDelegateTo<RetCC_F128HardFloat>>> +]>; //===----------------------------------------------------------------------===// // Mips O32 Calling Convention @@ -29,23 +63,43 @@ def RetCC_MipsO32 : CallingConv<[ // f64 arguments are returned in D0_64 and D2_64 in FP64bit mode or // in D0 and D1 in FP32bit mode. CCIfType<[f64], CCIfSubtarget<"isFP64bit()", CCAssignToReg<[D0_64, D2_64]>>>, - CCIfType<[f64], CCIfSubtarget<"isNotFP64bit()", CCAssignToReg<[D0, D1]>>> + CCIfType<[f64], CCIfSubtargetNot<"isFP64bit()", CCAssignToReg<[D0, D1]>>> +]>; + +def CC_MipsO32_FP32 : CustomCallingConv; +def CC_MipsO32_FP64 : CustomCallingConv; + +def CC_MipsO32_FP : CallingConv<[ + CCIfSubtargetNot<"isFP64bit()", CCDelegateTo<CC_MipsO32_FP32>>, + CCIfSubtarget<"isFP64bit()", CCDelegateTo<CC_MipsO32_FP64>> ]>; //===----------------------------------------------------------------------===// // Mips N32/64 Calling Convention //===----------------------------------------------------------------------===// +def CC_MipsN_SoftFloat : CallingConv<[ + CCAssignToRegWithShadow<[A0, A1, A2, A3, + T0, T1, T2, T3], + [D12_64, D13_64, D14_64, D15_64, + D16_64, D17_64, D18_64, D19_64]>, + CCAssignToStack<4, 8> +]>; + def CC_MipsN : CallingConv<[ - // Promote i8/i16 arguments to i32. - CCIfType<[i8, i16], CCPromoteToType<i32>>, + CCIfType<[i8, i16, i32], + CCIfSubtargetNot<"isLittle()", + CCIfInReg<CCPromoteToUpperBitsInType<i64>>>>, - // Integer arguments are passed in integer registers. - CCIfType<[i32], CCAssignToRegWithShadow<[A0, A1, A2, A3, - T0, T1, T2, T3], - [F12, F13, F14, F15, - F16, F17, F18, F19]>>, + // All integers (except soft-float integers) are promoted to 64-bit. + CCIfType<[i8, i16, i32], + CCIf<"!static_cast<MipsCCState *>(&State)->WasOriginalArgFloat(ValNo)", + CCPromoteToType<i64>>>, + + // The only i32's we have left are soft-float arguments. + CCIfSubtarget<"abiUsesSoftFloat()", CCIfType<[i32], CCDelegateTo<CC_MipsN_SoftFloat>>>, + // Integer arguments are passed in integer registers. CCIfType<[i64], CCAssignToRegWithShadow<[A0_64, A1_64, A2_64, A3_64, T0_64, T1_64, T2_64, T3_64], [D12_64, D13_64, D14_64, D15_64, @@ -64,29 +118,49 @@ def CC_MipsN : CallingConv<[ T0_64, T1_64, T2_64, T3_64]>>, // All stack parameter slots become 64-bit doublewords and are 8-byte aligned. - CCIfType<[i32, f32], CCAssignToStack<4, 8>>, + CCIfType<[f32], CCAssignToStack<4, 8>>, CCIfType<[i64, f64], CCAssignToStack<8, 8>> ]>; // N32/64 variable arguments. // All arguments are passed in integer registers. def CC_MipsN_VarArg : CallingConv<[ - // Promote i8/i16 arguments to i32. - CCIfType<[i8, i16], CCPromoteToType<i32>>, + // All integers are promoted to 64-bit. + CCIfType<[i8, i16, i32], CCPromoteToType<i64>>, - CCIfType<[i32, f32], CCAssignToReg<[A0, A1, A2, A3, T0, T1, T2, T3]>>, + CCIfType<[f32], CCAssignToReg<[A0, A1, A2, A3, T0, T1, T2, T3]>>, CCIfType<[i64, f64], CCAssignToReg<[A0_64, A1_64, A2_64, A3_64, T0_64, T1_64, T2_64, T3_64]>>, // All stack parameter slots become 64-bit doublewords and are 8-byte aligned. - CCIfType<[i32, f32], CCAssignToStack<4, 8>>, + CCIfType<[f32], CCAssignToStack<4, 8>>, CCIfType<[i64, f64], CCAssignToStack<8, 8>> ]>; def RetCC_MipsN : CallingConv<[ - // i32 are returned in registers V0, V1 - CCIfType<[i32], CCAssignToReg<[V0, V1]>>, + // f128 needs to be handled similarly to f32 and f64. However, f128 is not + // legal and is lowered to i128 which is further lowered to a pair of i64's. + // This presents us with a problem for the calling convention since hard-float + // still needs to pass them in FPU registers, and soft-float needs to use $v0, + // and $a0 instead of the usual $v0, and $v1. We therefore resort to a + // pre-analyze (see PreAnalyzeReturnForF128()) step to pass information on + // whether the result was originally an f128 into the tablegen-erated code. + // + // f128 should only occur for the N64 ABI where long double is 128-bit. On + // N32, long double is equivalent to double. + CCIfType<[i64], + CCIf<"static_cast<MipsCCState *>(&State)->WasOriginalArgF128(ValNo)", + CCDelegateTo<RetCC_F128>>>, + + // Aggregate returns are positioned at the lowest address in the slot for + // both little and big-endian targets. When passing in registers, this + // requires that big-endian targets shift the value into the upper bits. + CCIfSubtarget<"isLittle()", + CCIfType<[i8, i16, i32, i64], CCIfInReg<CCPromoteToType<i64>>>>, + CCIfSubtargetNot<"isLittle()", + CCIfType<[i8, i16, i32, i64], + CCIfInReg<CCPromoteToUpperBitsInType<i64>>>>, // i64 are returned in registers V0_64, V1_64 CCIfType<[i64], CCAssignToReg<[V0_64, V1_64]>>, @@ -98,12 +172,6 @@ def RetCC_MipsN : CallingConv<[ CCIfType<[f64], CCAssignToReg<[D0_64, D2_64]>> ]>; -// In soft-mode, register A0_64, instead of V1_64, is used to return a long -// double value. -def RetCC_F128Soft : CallingConv<[ - CCIfType<[i64], CCAssignToReg<[V0_64, A0_64]>> -]>; - //===----------------------------------------------------------------------===// // Mips EABI Calling Convention //===----------------------------------------------------------------------===// @@ -119,11 +187,11 @@ def CC_MipsEABI : CallingConv<[ CCIfType<[f32], CCIfSubtarget<"isSingleFloat()", CCAssignToReg<[F12, F13, F14, F15, F16, F17, F18, F19]>>>, - CCIfType<[f32], CCIfSubtarget<"isNotSingleFloat()", + CCIfType<[f32], CCIfSubtargetNot<"isSingleFloat()", CCAssignToReg<[F12, F14, F16, F18]>>>, // The first 4 double fp arguments are passed in single fp registers. - CCIfType<[f64], CCIfSubtarget<"isNotSingleFloat()", + CCIfType<[f64], CCIfSubtargetNot<"isSingleFloat()", CCAssignToReg<[D6, D7, D8, D9]>>>, // Integer values get stored in stack slots that are 4 bytes in @@ -132,7 +200,7 @@ def CC_MipsEABI : CallingConv<[ // Integer values get stored in stack slots that are 8 bytes in // size and 8-byte aligned. - CCIfType<[f64], CCIfSubtarget<"isNotSingleFloat()", CCAssignToStack<8, 8>>> + CCIfType<[f64], CCIfSubtargetNot<"isSingleFloat()", CCAssignToStack<8, 8>>> ]>; def RetCC_MipsEABI : CallingConv<[ @@ -143,7 +211,7 @@ def RetCC_MipsEABI : CallingConv<[ CCIfType<[f32], CCAssignToReg<[F0, F1]>>, // f64 are returned in register D0 - CCIfType<[f64], CCIfSubtarget<"isNotSingleFloat()", CCAssignToReg<[D0]>>> + CCIfType<[f64], CCIfSubtargetNot<"isSingleFloat()", CCAssignToReg<[D0]>>> ]>; //===----------------------------------------------------------------------===// @@ -151,16 +219,20 @@ def RetCC_MipsEABI : CallingConv<[ //===----------------------------------------------------------------------===// def CC_MipsO32_FastCC : CallingConv<[ // f64 arguments are passed in double-precision floating pointer registers. - CCIfType<[f64], CCIfSubtarget<"isNotFP64bit()", - CCAssignToReg<[D0, D1, D2, D3, D4, D5, D6, D7, - D8, D9]>>>, - CCIfType<[f64], CCIfSubtarget<"isFP64bit()", + CCIfType<[f64], CCIfSubtargetNot<"isFP64bit()", + CCAssignToReg<[D0, D1, D2, D3, D4, D5, D6, + D7, D8, D9]>>>, + CCIfType<[f64], CCIfSubtarget<"isFP64bit()", CCIfSubtarget<"useOddSPReg()", CCAssignToReg<[D0_64, D1_64, D2_64, D3_64, D4_64, D5_64, D6_64, D7_64, D8_64, D9_64, D10_64, D11_64, D12_64, D13_64, D14_64, D15_64, D16_64, D17_64, D18_64, - D19_64]>>>, + D19_64]>>>>, + CCIfType<[f64], CCIfSubtarget<"isFP64bit()", CCIfSubtarget<"noOddSPReg()", + CCAssignToReg<[D0_64, D2_64, D4_64, D6_64, + D8_64, D10_64, D12_64, D14_64, + D16_64, D18_64]>>>>, // Stack parameter slots for f64 are 64-bit doublewords and 8-byte aligned. CCIfType<[f64], CCAssignToStack<8, 8>> @@ -192,7 +264,7 @@ def CC_Mips_FastCC : CallingConv<[ // Integer arguments are passed in integer registers. All scratch registers, // except for AT, V0 and T9, are available to be used as argument registers. - CCIfType<[i32], CCIfSubtarget<"isNotTargetNaCl()", + CCIfType<[i32], CCIfSubtargetNot<"isTargetNaCl()", CCAssignToReg<[A0, A1, A2, A3, T0, T1, T2, T3, T4, T5, T6, T7, T8, V1]>>>, // In NaCl, T6, T7 and T8 are reserved and not available as argument @@ -219,13 +291,6 @@ def CC_Mips_FastCC : CallingConv<[ CCDelegateTo<CC_MipsN_FastCC> ]>; -//== - -def CC_Mips16RetHelper : CallingConv<[ - // Integer arguments are passed in integer registers. - CCIfType<[i32], CCAssignToReg<[V0, V1, A0, A1]>> -]>; - //===----------------------------------------------------------------------===// // Mips Calling Convention Dispatch //===----------------------------------------------------------------------===// @@ -237,6 +302,66 @@ def RetCC_Mips : CallingConv<[ CCDelegateTo<RetCC_MipsO32> ]>; +def CC_Mips_ByVal : CallingConv<[ + CCIfSubtarget<"isABI_O32()", CCIfByVal<CCPassByVal<4, 4>>>, + CCIfByVal<CCPassByVal<8, 8>> +]>; + +def CC_Mips16RetHelper : CallingConv<[ + CCIfByVal<CCDelegateTo<CC_Mips_ByVal>>, + + // Integer arguments are passed in integer registers. + CCIfType<[i32], CCAssignToReg<[V0, V1, A0, A1]>> +]>; + +def CC_Mips_FixedArg : CallingConv<[ + // Mips16 needs special handling on some functions. + CCIf<"State.getCallingConv() != CallingConv::Fast", + CCIf<"static_cast<MipsCCState *>(&State)->getSpecialCallingConv() == " + "MipsCCState::Mips16RetHelperConv", + CCDelegateTo<CC_Mips16RetHelper>>>, + + CCIfByVal<CCDelegateTo<CC_Mips_ByVal>>, + + // f128 needs to be handled similarly to f32 and f64 on hard-float. However, + // f128 is not legal and is lowered to i128 which is further lowered to a pair + // of i64's. + // This presents us with a problem for the calling convention since hard-float + // still needs to pass them in FPU registers. We therefore resort to a + // pre-analyze (see PreAnalyzeFormalArgsForF128()) step to pass information on + // whether the argument was originally an f128 into the tablegen-erated code. + // + // f128 should only occur for the N64 ABI where long double is 128-bit. On + // N32, long double is equivalent to double. + CCIfType<[i64], + CCIfSubtargetNot<"abiUsesSoftFloat()", + CCIf<"static_cast<MipsCCState *>(&State)->WasOriginalArgF128(ValNo)", + CCBitConvertToType<f64>>>>, + + CCIfCC<"CallingConv::Fast", CCDelegateTo<CC_Mips_FastCC>>, + + // FIXME: There wasn't an EABI case in the original code and it seems unlikely + // that it's the same as CC_MipsN + CCIfSubtarget<"isABI_O32()", CCDelegateTo<CC_MipsO32_FP>>, + CCDelegateTo<CC_MipsN> +]>; + +def CC_Mips_VarArg : CallingConv<[ + CCIfByVal<CCDelegateTo<CC_Mips_ByVal>>, + + // FIXME: There wasn't an EABI case in the original code and it seems unlikely + // that it's the same as CC_MipsN_VarArg + CCIfSubtarget<"isABI_O32()", CCDelegateTo<CC_MipsO32_FP>>, + CCDelegateTo<CC_MipsN_VarArg> +]>; + +def CC_Mips : CallingConv<[ + CCIfVarArg< + CCIf<"!static_cast<MipsCCState *>(&State)->IsCallOperandFixed(ValNo)", + CCDelegateTo<CC_Mips_VarArg>>>, + CCDelegateTo<CC_Mips_FixedArg> +]>; + //===----------------------------------------------------------------------===// // Callee-saved register lists. //===----------------------------------------------------------------------===// diff --git a/contrib/llvm/lib/Target/Mips/MipsConstantIslandPass.cpp b/contrib/llvm/lib/Target/Mips/MipsConstantIslandPass.cpp index 80bf573..76313b1 100644 --- a/contrib/llvm/lib/Target/Mips/MipsConstantIslandPass.cpp +++ b/contrib/llvm/lib/Target/Mips/MipsConstantIslandPass.cpp @@ -343,7 +343,6 @@ namespace { const TargetMachine &TM; bool IsPIC; - unsigned ABI; const MipsSubtarget *STI; const Mips16InstrInfo *TII; MipsFunctionInfo *MFI; @@ -366,8 +365,7 @@ namespace { static char ID; MipsConstantIslands(TargetMachine &tm) : MachineFunctionPass(ID), TM(tm), - IsPIC(TM.getRelocationModel() == Reloc::PIC_), - ABI(TM.getSubtarget<MipsSubtarget>().getTargetABI()), STI(nullptr), + IsPIC(TM.getRelocationModel() == Reloc::PIC_), STI(nullptr), MF(nullptr), MCP(nullptr), PrescannedForConstants(false) {} const char *getPassName() const override { diff --git a/contrib/llvm/lib/Target/Mips/MipsISelLowering.cpp b/contrib/llvm/lib/Target/Mips/MipsISelLowering.cpp index 40dc8e4..8e40668 100644 --- a/contrib/llvm/lib/Target/Mips/MipsISelLowering.cpp +++ b/contrib/llvm/lib/Target/Mips/MipsISelLowering.cpp @@ -14,6 +14,7 @@ #include "MipsISelLowering.h" #include "InstPrinter/MipsInstPrinter.h" #include "MCTargetDesc/MipsBaseInfo.h" +#include "MipsCCState.h" #include "MipsMachineFunction.h" #include "MipsSubtarget.h" #include "MipsTargetMachine.h" @@ -56,15 +57,6 @@ EnableMipsFastISel("mips-fast-isel", cl::Hidden, cl::desc("Allow mips-fast-isel to be used"), cl::init(false)); -static const MCPhysReg O32IntRegs[4] = { - Mips::A0, Mips::A1, Mips::A2, Mips::A3 -}; - -static const MCPhysReg Mips64IntRegs[8] = { - Mips::A0_64, Mips::A1_64, Mips::A2_64, Mips::A3_64, - Mips::T0_64, Mips::T1_64, Mips::T2_64, Mips::T3_64 -}; - static const MCPhysReg Mips64DPRegs[8] = { Mips::D12_64, Mips::D13_64, Mips::D14_64, Mips::D15_64, Mips::D16_64, Mips::D17_64, Mips::D18_64, Mips::D19_64 @@ -251,7 +243,6 @@ MipsTargetLowering::MipsTargetLowering(MipsTargetMachine &TM, setOperationAction(ISD::SETCC, MVT::f32, Custom); setOperationAction(ISD::SETCC, MVT::f64, Custom); setOperationAction(ISD::BRCOND, MVT::Other, Custom); - setOperationAction(ISD::VASTART, MVT::Other, Custom); setOperationAction(ISD::FCOPYSIGN, MVT::f32, Custom); setOperationAction(ISD::FCOPYSIGN, MVT::f64, Custom); setOperationAction(ISD::FP_TO_SINT, MVT::i32, Custom); @@ -343,7 +334,8 @@ MipsTargetLowering::MipsTargetLowering(MipsTargetMachine &TM, setOperationAction(ISD::EH_RETURN, MVT::Other, Custom); - setOperationAction(ISD::VAARG, MVT::Other, Expand); + setOperationAction(ISD::VASTART, MVT::Other, Custom); + setOperationAction(ISD::VAARG, MVT::Other, Custom); setOperationAction(ISD::VACOPY, MVT::Other, Expand); setOperationAction(ISD::VAEND, MVT::Other, Expand); @@ -392,6 +384,11 @@ MipsTargetLowering::MipsTargetLowering(MipsTargetMachine &TM, setMinFunctionAlignment(Subtarget.isGP64bit() ? 3 : 2); + // The arguments on the stack are defined in terms of 4-byte slots on O32 + // and 8-byte slots on N32/N64. + setMinStackArgumentAlignment( + (Subtarget.isABI_N32() || Subtarget.isABI_N64()) ? 8 : 4); + setStackPointerRegisterToSaveRestore(Subtarget.isABI_N64() ? Mips::SP_64 : Mips::SP); @@ -792,6 +789,7 @@ LowerOperation(SDValue Op, SelectionDAG &DAG) const case ISD::SELECT_CC: return lowerSELECT_CC(Op, DAG); case ISD::SETCC: return lowerSETCC(Op, DAG); case ISD::VASTART: return lowerVASTART(Op, DAG); + case ISD::VAARG: return lowerVAARG(Op, DAG); case ISD::FCOPYSIGN: return lowerFCOPYSIGN(Op, DAG); case ISD::FRAMEADDR: return lowerFRAMEADDR(Op, DAG); case ISD::RETURNADDR: return lowerRETURNADDR(Op, DAG); @@ -1755,6 +1753,65 @@ SDValue MipsTargetLowering::lowerVASTART(SDValue Op, SelectionDAG &DAG) const { MachinePointerInfo(SV), false, false, 0); } +SDValue MipsTargetLowering::lowerVAARG(SDValue Op, SelectionDAG &DAG) const { + SDNode *Node = Op.getNode(); + EVT VT = Node->getValueType(0); + SDValue Chain = Node->getOperand(0); + SDValue VAListPtr = Node->getOperand(1); + unsigned Align = Node->getConstantOperandVal(3); + const Value *SV = cast<SrcValueSDNode>(Node->getOperand(2))->getValue(); + SDLoc DL(Node); + unsigned ArgSlotSizeInBytes = + (Subtarget.isABI_N32() || Subtarget.isABI_N64()) ? 8 : 4; + + SDValue VAListLoad = DAG.getLoad(getPointerTy(), DL, Chain, VAListPtr, + MachinePointerInfo(SV), false, false, false, + 0); + SDValue VAList = VAListLoad; + + // Re-align the pointer if necessary. + // It should only ever be necessary for 64-bit types on O32 since the minimum + // argument alignment is the same as the maximum type alignment for N32/N64. + // + // FIXME: We currently align too often. The code generator doesn't notice + // when the pointer is still aligned from the last va_arg (or pair of + // va_args for the i64 on O32 case). + if (Align > getMinStackArgumentAlignment()) { + assert(((Align & (Align-1)) == 0) && "Expected Align to be a power of 2"); + + VAList = DAG.getNode(ISD::ADD, DL, VAList.getValueType(), VAList, + DAG.getConstant(Align - 1, + VAList.getValueType())); + + VAList = DAG.getNode(ISD::AND, DL, VAList.getValueType(), VAList, + DAG.getConstant(-(int64_t)Align, + VAList.getValueType())); + } + + // Increment the pointer, VAList, to the next vaarg. + unsigned ArgSizeInBytes = getDataLayout()->getTypeAllocSize(VT.getTypeForEVT(*DAG.getContext())); + SDValue Tmp3 = DAG.getNode(ISD::ADD, DL, VAList.getValueType(), VAList, + DAG.getConstant(RoundUpToAlignment(ArgSizeInBytes, ArgSlotSizeInBytes), + VAList.getValueType())); + // Store the incremented VAList to the legalized pointer + Chain = DAG.getStore(VAListLoad.getValue(1), DL, Tmp3, VAListPtr, + MachinePointerInfo(SV), false, false, 0); + + // In big-endian mode we must adjust the pointer when the load size is smaller + // than the argument slot size. We must also reduce the known alignment to + // match. For example in the N64 ABI, we must add 4 bytes to the offset to get + // the correct half of the slot, and reduce the alignment from 8 (slot + // alignment) down to 4 (type alignment). + if (!Subtarget.isLittle() && ArgSizeInBytes < ArgSlotSizeInBytes) { + unsigned Adjustment = ArgSlotSizeInBytes - ArgSizeInBytes; + VAList = DAG.getNode(ISD::ADD, DL, VAListPtr.getValueType(), VAList, + DAG.getIntPtrConstant(Adjustment)); + } + // Load the actual argument out of the pointer VAList + return DAG.getLoad(VT, DL, Chain, VAList, MachinePointerInfo(), false, false, + false, 0); +} + static SDValue lowerFCOPYSIGN32(SDValue Op, SelectionDAG &DAG, bool HasExtractInsert) { EVT TyX = Op.getOperand(0).getValueType(); @@ -2211,6 +2268,9 @@ SDValue MipsTargetLowering::lowerFP_TO_SINT(SDValue Op, static bool CC_MipsO32(unsigned ValNo, MVT ValVT, MVT LocVT, CCValAssign::LocInfo LocInfo, ISD::ArgFlagsTy ArgFlags, CCState &State, const MCPhysReg *F64Regs) { + const MipsSubtarget &Subtarget = + State.getMachineFunction().getTarget() + .getSubtarget<const MipsSubtarget>(); static const unsigned IntRegsSize = 4, FloatRegsSize = 2; @@ -2222,6 +2282,19 @@ static bool CC_MipsO32(unsigned ValNo, MVT ValVT, MVT LocVT, return true; // Promote i8 and i16 + if (ArgFlags.isInReg() && !Subtarget.isLittle()) { + if (LocVT == MVT::i8 || LocVT == MVT::i16 || LocVT == MVT::i32) { + LocVT = MVT::i32; + if (ArgFlags.isSExt()) + LocInfo = CCValAssign::SExtUpper; + else if (ArgFlags.isZExt()) + LocInfo = CCValAssign::ZExtUpper; + else + LocInfo = CCValAssign::AExtUpper; + } + } + + // Promote i8 and i16 if (LocVT == MVT::i8 || LocVT == MVT::i16) { LocVT = MVT::i32; if (ArgFlags.isSExt()) @@ -2407,25 +2480,25 @@ MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, // Analyze operands of the call, assigning locations to each operand. SmallVector<CCValAssign, 16> ArgLocs; - CCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(), - getTargetMachine(), ArgLocs, *DAG.getContext()); - MipsCC::SpecialCallingConvType SpecialCallingConv = - getSpecialCallingConv(Callee); - MipsCC MipsCCInfo(CallConv, Subtarget.isABI_O32(), Subtarget.isFP64bit(), - CCInfo, SpecialCallingConv); + MipsCCState CCInfo( + CallConv, IsVarArg, DAG.getMachineFunction(), + getTargetMachine(), ArgLocs, *DAG.getContext(), + MipsCCState::getSpecialCallingConvForCallee(Callee.getNode(), Subtarget)); - MipsCCInfo.analyzeCallOperands(Outs, IsVarArg, - Subtarget.abiUsesSoftFloat(), - Callee.getNode(), CLI.getArgs()); + // Allocate the reserved argument area. It seems strange to do this from the + // caller side but removing it breaks the frame size calculation. + const MipsABIInfo &ABI = Subtarget.getABI(); + CCInfo.AllocateStack(ABI.GetCalleeAllocdArgSizeInBytes(CallConv), 1); + + CCInfo.AnalyzeCallOperands(Outs, CC_Mips, CLI.getArgs(), Callee.getNode()); // Get a count of how many bytes are to be pushed on the stack. unsigned NextStackOffset = CCInfo.getNextStackOffset(); // Check if it's really possible to do a tail call. if (IsTailCall) - IsTailCall = - isEligibleForTailCallOptimization(MipsCCInfo, NextStackOffset, - *MF.getInfo<MipsFunctionInfo>()); + IsTailCall = isEligibleForTailCallOptimization( + CCInfo, NextStackOffset, *MF.getInfo<MipsFunctionInfo>()); if (!IsTailCall && CLI.CS && CLI.CS->isMustTailCall()) report_fatal_error("failed to perform tail call elimination on a call " @@ -2451,7 +2524,8 @@ MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, // With EABI is it possible to have 16 args on registers. std::deque< std::pair<unsigned, SDValue> > RegsToPass; SmallVector<SDValue, 8> MemOpChains; - MipsCC::byval_iterator ByValArg = MipsCCInfo.byval_begin(); + + CCInfo.rewindByValRegsInfo(); // Walk the register/memloc assignments, inserting copies/loads. for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) { @@ -2459,23 +2533,30 @@ MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, CCValAssign &VA = ArgLocs[i]; MVT ValVT = VA.getValVT(), LocVT = VA.getLocVT(); ISD::ArgFlagsTy Flags = Outs[i].Flags; + bool UseUpperBits = false; // ByVal Arg. if (Flags.isByVal()) { + unsigned FirstByValReg, LastByValReg; + unsigned ByValIdx = CCInfo.getInRegsParamsProcessed(); + CCInfo.getInRegsParamInfo(ByValIdx, FirstByValReg, LastByValReg); + assert(Flags.getByValSize() && "ByVal args of size 0 should have been ignored by front-end."); - assert(ByValArg != MipsCCInfo.byval_end()); + assert(ByValIdx < CCInfo.getInRegsParamsCount()); assert(!IsTailCall && "Do not tail-call optimize if there is a byval argument."); passByValArg(Chain, DL, RegsToPass, MemOpChains, StackPtr, MFI, DAG, Arg, - MipsCCInfo, *ByValArg, Flags, Subtarget.isLittle()); - ++ByValArg; + FirstByValReg, LastByValReg, Flags, Subtarget.isLittle(), + VA); + CCInfo.nextInRegsParam(); continue; } // Promote the value if needed. switch (VA.getLocInfo()) { - default: llvm_unreachable("Unknown loc info!"); + default: + llvm_unreachable("Unknown loc info!"); case CCValAssign::Full: if (VA.isRegLoc()) { if ((ValVT == MVT::f32 && LocVT == MVT::i32) || @@ -2497,17 +2578,37 @@ MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, } } break; + case CCValAssign::BCvt: + Arg = DAG.getNode(ISD::BITCAST, DL, LocVT, Arg); + break; + case CCValAssign::SExtUpper: + UseUpperBits = true; + // Fallthrough case CCValAssign::SExt: Arg = DAG.getNode(ISD::SIGN_EXTEND, DL, LocVT, Arg); break; + case CCValAssign::ZExtUpper: + UseUpperBits = true; + // Fallthrough case CCValAssign::ZExt: Arg = DAG.getNode(ISD::ZERO_EXTEND, DL, LocVT, Arg); break; + case CCValAssign::AExtUpper: + UseUpperBits = true; + // Fallthrough case CCValAssign::AExt: Arg = DAG.getNode(ISD::ANY_EXTEND, DL, LocVT, Arg); break; } + if (UseUpperBits) { + unsigned ValSizeInBits = Outs[i].ArgVT.getSizeInBits(); + unsigned LocSizeInBits = VA.getLocVT().getSizeInBits(); + Arg = DAG.getNode( + ISD::SHL, DL, VA.getLocVT(), Arg, + DAG.getConstant(LocSizeInBits - ValSizeInBits, VA.getLocVT())); + } + // Arguments that can be passed on register must be kept at // RegsToPass vector if (VA.isRegLoc()) { @@ -2595,39 +2696,68 @@ MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, // Handle result values, copying them out of physregs into vregs that we // return. - return LowerCallResult(Chain, InFlag, CallConv, IsVarArg, - Ins, DL, DAG, InVals, CLI.Callee.getNode(), CLI.RetTy); + return LowerCallResult(Chain, InFlag, CallConv, IsVarArg, Ins, DL, DAG, + InVals, CLI); } /// LowerCallResult - Lower the result values of a call into the /// appropriate copies out of appropriate physical registers. -SDValue -MipsTargetLowering::LowerCallResult(SDValue Chain, SDValue InFlag, - CallingConv::ID CallConv, bool IsVarArg, - const SmallVectorImpl<ISD::InputArg> &Ins, - SDLoc DL, SelectionDAG &DAG, - SmallVectorImpl<SDValue> &InVals, - const SDNode *CallNode, - const Type *RetTy) const { +SDValue MipsTargetLowering::LowerCallResult( + SDValue Chain, SDValue InFlag, CallingConv::ID CallConv, bool IsVarArg, + const SmallVectorImpl<ISD::InputArg> &Ins, SDLoc DL, SelectionDAG &DAG, + SmallVectorImpl<SDValue> &InVals, + TargetLowering::CallLoweringInfo &CLI) const { // Assign locations to each value returned by this call. SmallVector<CCValAssign, 16> RVLocs; - CCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(), - getTargetMachine(), RVLocs, *DAG.getContext()); - MipsCC MipsCCInfo(CallConv, Subtarget.isABI_O32(), Subtarget.isFP64bit(), - CCInfo); - - MipsCCInfo.analyzeCallResult(Ins, Subtarget.abiUsesSoftFloat(), - CallNode, RetTy); + MipsCCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(), + getTargetMachine(), RVLocs, *DAG.getContext()); + CCInfo.AnalyzeCallResult(Ins, RetCC_Mips, CLI); // Copy all of the result registers out of their specified physreg. for (unsigned i = 0; i != RVLocs.size(); ++i) { + CCValAssign &VA = RVLocs[i]; + assert(VA.isRegLoc() && "Can only return in registers!"); + SDValue Val = DAG.getCopyFromReg(Chain, DL, RVLocs[i].getLocReg(), RVLocs[i].getLocVT(), InFlag); Chain = Val.getValue(1); InFlag = Val.getValue(2); - if (RVLocs[i].getValVT() != RVLocs[i].getLocVT()) - Val = DAG.getNode(ISD::BITCAST, DL, RVLocs[i].getValVT(), Val); + if (VA.isUpperBitsInLoc()) { + unsigned ValSizeInBits = Ins[i].ArgVT.getSizeInBits(); + unsigned LocSizeInBits = VA.getLocVT().getSizeInBits(); + unsigned Shift = + VA.getLocInfo() == CCValAssign::ZExtUpper ? ISD::SRL : ISD::SRA; + Val = DAG.getNode( + Shift, DL, VA.getLocVT(), Val, + DAG.getConstant(LocSizeInBits - ValSizeInBits, VA.getLocVT())); + } + + switch (VA.getLocInfo()) { + default: + llvm_unreachable("Unknown loc info!"); + case CCValAssign::Full: + break; + case CCValAssign::BCvt: + Val = DAG.getNode(ISD::BITCAST, DL, VA.getValVT(), Val); + break; + case CCValAssign::AExt: + case CCValAssign::AExtUpper: + Val = DAG.getNode(ISD::TRUNCATE, DL, VA.getValVT(), Val); + break; + case CCValAssign::ZExt: + case CCValAssign::ZExtUpper: + Val = DAG.getNode(ISD::AssertZext, DL, VA.getLocVT(), Val, + DAG.getValueType(VA.getValVT())); + Val = DAG.getNode(ISD::TRUNCATE, DL, VA.getValVT(), Val); + break; + case CCValAssign::SExt: + case CCValAssign::SExtUpper: + Val = DAG.getNode(ISD::AssertSext, DL, VA.getLocVT(), Val, + DAG.getValueType(VA.getValVT())); + Val = DAG.getNode(ISD::TRUNCATE, DL, VA.getValVT(), Val); + break; + } InVals.push_back(Val); } @@ -2635,6 +2765,60 @@ MipsTargetLowering::LowerCallResult(SDValue Chain, SDValue InFlag, return Chain; } +static SDValue UnpackFromArgumentSlot(SDValue Val, const CCValAssign &VA, + EVT ArgVT, SDLoc DL, SelectionDAG &DAG) { + MVT LocVT = VA.getLocVT(); + EVT ValVT = VA.getValVT(); + + // Shift into the upper bits if necessary. + switch (VA.getLocInfo()) { + default: + break; + case CCValAssign::AExtUpper: + case CCValAssign::SExtUpper: + case CCValAssign::ZExtUpper: { + unsigned ValSizeInBits = ArgVT.getSizeInBits(); + unsigned LocSizeInBits = VA.getLocVT().getSizeInBits(); + unsigned Opcode = + VA.getLocInfo() == CCValAssign::ZExtUpper ? ISD::SRL : ISD::SRA; + Val = DAG.getNode( + Opcode, DL, VA.getLocVT(), Val, + DAG.getConstant(LocSizeInBits - ValSizeInBits, VA.getLocVT())); + break; + } + } + + // If this is an value smaller than the argument slot size (32-bit for O32, + // 64-bit for N32/N64), it has been promoted in some way to the argument slot + // size. Extract the value and insert any appropriate assertions regarding + // sign/zero extension. + switch (VA.getLocInfo()) { + default: + llvm_unreachable("Unknown loc info!"); + case CCValAssign::Full: + break; + case CCValAssign::AExtUpper: + case CCValAssign::AExt: + Val = DAG.getNode(ISD::TRUNCATE, DL, ValVT, Val); + break; + case CCValAssign::SExtUpper: + case CCValAssign::SExt: + Val = DAG.getNode(ISD::AssertSext, DL, LocVT, Val, DAG.getValueType(ValVT)); + Val = DAG.getNode(ISD::TRUNCATE, DL, ValVT, Val); + break; + case CCValAssign::ZExtUpper: + case CCValAssign::ZExt: + Val = DAG.getNode(ISD::AssertZext, DL, LocVT, Val, DAG.getValueType(ValVT)); + Val = DAG.getNode(ISD::TRUNCATE, DL, ValVT, Val); + break; + case CCValAssign::BCvt: + Val = DAG.getNode(ISD::BITCAST, DL, ValVT, Val); + break; + } + + return Val; +} + //===----------------------------------------------------------------------===// // Formal Arguments Calling Convention Implementation //===----------------------------------------------------------------------===// @@ -2659,20 +2843,19 @@ MipsTargetLowering::LowerFormalArguments(SDValue Chain, // Assign locations to all of the incoming arguments. SmallVector<CCValAssign, 16> ArgLocs; - CCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(), - getTargetMachine(), ArgLocs, *DAG.getContext()); - MipsCC MipsCCInfo(CallConv, Subtarget.isABI_O32(), Subtarget.isFP64bit(), - CCInfo); + MipsCCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(), + getTargetMachine(), ArgLocs, *DAG.getContext()); + const MipsABIInfo &ABI = Subtarget.getABI(); + CCInfo.AllocateStack(ABI.GetCalleeAllocdArgSizeInBytes(CallConv), 1); Function::const_arg_iterator FuncArg = DAG.getMachineFunction().getFunction()->arg_begin(); - bool UseSoftFloat = Subtarget.abiUsesSoftFloat(); - MipsCCInfo.analyzeFormalArguments(Ins, UseSoftFloat, FuncArg); + CCInfo.AnalyzeFormalArguments(Ins, CC_Mips_FixedArg); MipsFI->setFormalArgInfo(CCInfo.getNextStackOffset(), - MipsCCInfo.hasByValArg()); + CCInfo.getInRegsParamsCount() > 0); unsigned CurArgIdx = 0; - MipsCC::byval_iterator ByValArg = MipsCCInfo.byval_begin(); + CCInfo.rewindByValRegsInfo(); for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) { CCValAssign &VA = ArgLocs[i]; @@ -2683,12 +2866,16 @@ MipsTargetLowering::LowerFormalArguments(SDValue Chain, bool IsRegLoc = VA.isRegLoc(); if (Flags.isByVal()) { + unsigned FirstByValReg, LastByValReg; + unsigned ByValIdx = CCInfo.getInRegsParamsProcessed(); + CCInfo.getInRegsParamInfo(ByValIdx, FirstByValReg, LastByValReg); + assert(Flags.getByValSize() && "ByVal args of size 0 should have been ignored by front-end."); - assert(ByValArg != MipsCCInfo.byval_end()); + assert(ByValIdx < CCInfo.getInRegsParamsCount()); copyByValRegs(Chain, DL, OutChains, DAG, Flags, InVals, &*FuncArg, - MipsCCInfo, *ByValArg); - ++ByValArg; + FirstByValReg, LastByValReg, VA, CCInfo); + CCInfo.nextInRegsParam(); continue; } @@ -2703,20 +2890,7 @@ MipsTargetLowering::LowerFormalArguments(SDValue Chain, unsigned Reg = addLiveIn(DAG.getMachineFunction(), ArgReg, 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(ValVT)); - ArgValue = DAG.getNode(ISD::TRUNCATE, DL, ValVT, ArgValue); - } + ArgValue = UnpackFromArgumentSlot(ArgValue, VA, Ins[i].ArgVT, DL, DAG); // Handle floating point arguments passed in integer registers and // long double arguments passed in floating point registers. @@ -2737,21 +2911,34 @@ MipsTargetLowering::LowerFormalArguments(SDValue Chain, InVals.push_back(ArgValue); } else { // VA.isRegLoc() + MVT LocVT = VA.getLocVT(); + + if (Subtarget.isABI_O32()) { + // We ought to be able to use LocVT directly but O32 sets it to i32 + // when allocating floating point values to integer registers. + // This shouldn't influence how we load the value into registers unless + // we are targetting softfloat. + if (VA.getValVT().isFloatingPoint() && !Subtarget.abiUsesSoftFloat()) + LocVT = VA.getValVT(); + } // sanity check assert(VA.isMemLoc()); // The stack pointer offset is relative to the caller stack frame. - int FI = MFI->CreateFixedObject(ValVT.getSizeInBits()/8, + int FI = MFI->CreateFixedObject(LocVT.getSizeInBits() / 8, VA.getLocMemOffset(), true); // Create load nodes to retrieve arguments from the stack SDValue FIN = DAG.getFrameIndex(FI, getPointerTy()); - SDValue Load = DAG.getLoad(ValVT, DL, Chain, FIN, - MachinePointerInfo::getFixedStack(FI), - false, false, false, 0); - InVals.push_back(Load); - OutChains.push_back(Load.getValue(1)); + SDValue ArgValue = DAG.getLoad(LocVT, DL, Chain, FIN, + MachinePointerInfo::getFixedStack(FI), + false, false, false, 0); + OutChains.push_back(ArgValue.getValue(1)); + + ArgValue = UnpackFromArgumentSlot(ArgValue, VA, Ins[i].ArgVT, DL, DAG); + + InVals.push_back(ArgValue); } } @@ -2773,7 +2960,7 @@ MipsTargetLowering::LowerFormalArguments(SDValue Chain, } if (IsVarArg) - writeVarArgRegs(OutChains, MipsCCInfo, Chain, DL, DAG); + writeVarArgRegs(OutChains, Chain, DL, DAG, CCInfo); // All stores are grouped in one node to allow the matching between // the size of Ins and InVals. This only happens when on varg functions @@ -2795,8 +2982,8 @@ MipsTargetLowering::CanLowerReturn(CallingConv::ID CallConv, const SmallVectorImpl<ISD::OutputArg> &Outs, LLVMContext &Context) const { SmallVector<CCValAssign, 16> RVLocs; - CCState CCInfo(CallConv, IsVarArg, MF, getTargetMachine(), - RVLocs, Context); + MipsCCState CCInfo(CallConv, IsVarArg, MF, getTargetMachine(), + RVLocs, Context); return CCInfo.CheckReturn(Outs, RetCC_Mips); } @@ -2812,14 +2999,11 @@ MipsTargetLowering::LowerReturn(SDValue Chain, MachineFunction &MF = DAG.getMachineFunction(); // CCState - Info about the registers and stack slot. - CCState CCInfo(CallConv, IsVarArg, MF, getTargetMachine(), RVLocs, - *DAG.getContext()); - MipsCC MipsCCInfo(CallConv, Subtarget.isABI_O32(), Subtarget.isFP64bit(), - CCInfo); + MipsCCState CCInfo(CallConv, IsVarArg, MF, getTargetMachine(), RVLocs, + *DAG.getContext()); // Analyze return values. - MipsCCInfo.analyzeReturn(Outs, Subtarget.abiUsesSoftFloat(), - MF.getFunction()->getReturnType()); + CCInfo.AnalyzeReturn(Outs, RetCC_Mips); SDValue Flag; SmallVector<SDValue, 4> RetOps(1, Chain); @@ -2829,9 +3013,43 @@ MipsTargetLowering::LowerReturn(SDValue Chain, SDValue Val = OutVals[i]; CCValAssign &VA = RVLocs[i]; assert(VA.isRegLoc() && "Can only return in registers!"); + bool UseUpperBits = false; - if (RVLocs[i].getValVT() != RVLocs[i].getLocVT()) - Val = DAG.getNode(ISD::BITCAST, DL, RVLocs[i].getLocVT(), Val); + switch (VA.getLocInfo()) { + default: + llvm_unreachable("Unknown loc info!"); + case CCValAssign::Full: + break; + case CCValAssign::BCvt: + Val = DAG.getNode(ISD::BITCAST, DL, VA.getLocVT(), Val); + break; + case CCValAssign::AExtUpper: + UseUpperBits = true; + // Fallthrough + case CCValAssign::AExt: + Val = DAG.getNode(ISD::ANY_EXTEND, DL, VA.getLocVT(), Val); + break; + case CCValAssign::ZExtUpper: + UseUpperBits = true; + // Fallthrough + case CCValAssign::ZExt: + Val = DAG.getNode(ISD::ZERO_EXTEND, DL, VA.getLocVT(), Val); + break; + case CCValAssign::SExtUpper: + UseUpperBits = true; + // Fallthrough + case CCValAssign::SExt: + Val = DAG.getNode(ISD::SIGN_EXTEND, DL, VA.getLocVT(), Val); + break; + } + + if (UseUpperBits) { + unsigned ValSizeInBits = Outs[i].ArgVT.getSizeInBits(); + unsigned LocSizeInBits = VA.getLocVT().getSizeInBits(); + Val = DAG.getNode( + ISD::SHL, DL, VA.getLocVT(), Val, + DAG.getConstant(LocSizeInBits - ValSizeInBits, VA.getLocVT())); + } Chain = DAG.getCopyToReg(Chain, DL, VA.getLocReg(), Val, Flag); @@ -3267,285 +3485,27 @@ unsigned MipsTargetLowering::getJumpTableEncoding() const { return TargetLowering::getJumpTableEncoding(); } -/// This function returns true if CallSym is a long double emulation routine. -static bool isF128SoftLibCall(const char *CallSym) { - const char *const LibCalls[] = - {"__addtf3", "__divtf3", "__eqtf2", "__extenddftf2", "__extendsftf2", - "__fixtfdi", "__fixtfsi", "__fixtfti", "__fixunstfdi", "__fixunstfsi", - "__fixunstfti", "__floatditf", "__floatsitf", "__floattitf", - "__floatunditf", "__floatunsitf", "__floatuntitf", "__getf2", "__gttf2", - "__letf2", "__lttf2", "__multf3", "__netf2", "__powitf2", "__subtf3", - "__trunctfdf2", "__trunctfsf2", "__unordtf2", - "ceill", "copysignl", "cosl", "exp2l", "expl", "floorl", "fmal", "fmodl", - "log10l", "log2l", "logl", "nearbyintl", "powl", "rintl", "sinl", "sqrtl", - "truncl"}; - - const char *const *End = LibCalls + array_lengthof(LibCalls); - - // Check that LibCalls is sorted alphabetically. - MipsTargetLowering::LTStr Comp; - -#ifndef NDEBUG - for (const char *const *I = LibCalls; I < End - 1; ++I) - assert(Comp(*I, *(I + 1))); -#endif - - return std::binary_search(LibCalls, End, CallSym, Comp); -} - -/// This function returns true if Ty is fp128 or i128 which was originally a -/// fp128. -static bool originalTypeIsF128(const Type *Ty, const SDNode *CallNode) { - if (Ty->isFP128Ty()) - return true; - - const ExternalSymbolSDNode *ES = - dyn_cast_or_null<const ExternalSymbolSDNode>(CallNode); - - // If the Ty is i128 and the function being called is a long double emulation - // routine, then the original type is f128. - return (ES && Ty->isIntegerTy(128) && isF128SoftLibCall(ES->getSymbol())); -} - -MipsTargetLowering::MipsCC::SpecialCallingConvType - MipsTargetLowering::getSpecialCallingConv(SDValue Callee) const { - MipsCC::SpecialCallingConvType SpecialCallingConv = - MipsCC::NoSpecialCallingConv; - if (Subtarget.inMips16HardFloat()) { - if (GlobalAddressSDNode *G = dyn_cast<GlobalAddressSDNode>(Callee)) { - llvm::StringRef Sym = G->getGlobal()->getName(); - Function *F = G->getGlobal()->getParent()->getFunction(Sym); - if (F && F->hasFnAttribute("__Mips16RetHelper")) { - SpecialCallingConv = MipsCC::Mips16RetHelperConv; - } - } - } - return SpecialCallingConv; -} - -MipsTargetLowering::MipsCC::MipsCC( - CallingConv::ID CC, bool IsO32_, bool IsFP64_, CCState &Info, - MipsCC::SpecialCallingConvType SpecialCallingConv_) - : CCInfo(Info), CallConv(CC), IsO32(IsO32_), IsFP64(IsFP64_), - SpecialCallingConv(SpecialCallingConv_){ - // Pre-allocate reserved argument area. - CCInfo.AllocateStack(reservedArgArea(), 1); -} - - -void MipsTargetLowering::MipsCC:: -analyzeCallOperands(const SmallVectorImpl<ISD::OutputArg> &Args, - bool IsVarArg, bool IsSoftFloat, const SDNode *CallNode, - std::vector<ArgListEntry> &FuncArgs) { - assert((CallConv != CallingConv::Fast || !IsVarArg) && - "CallingConv::Fast shouldn't be used for vararg functions."); - - unsigned NumOpnds = Args.size(); - llvm::CCAssignFn *FixedFn = fixedArgFn(), *VarFn = varArgFn(); - - for (unsigned I = 0; I != NumOpnds; ++I) { - MVT ArgVT = Args[I].VT; - ISD::ArgFlagsTy ArgFlags = Args[I].Flags; - bool R; - - if (ArgFlags.isByVal()) { - handleByValArg(I, ArgVT, ArgVT, CCValAssign::Full, ArgFlags); - continue; - } - - if (IsVarArg && !Args[I].IsFixed) - R = VarFn(I, ArgVT, ArgVT, CCValAssign::Full, ArgFlags, CCInfo); - else { - MVT RegVT = getRegVT(ArgVT, FuncArgs[Args[I].OrigArgIndex].Ty, CallNode, - IsSoftFloat); - R = FixedFn(I, ArgVT, RegVT, CCValAssign::Full, ArgFlags, CCInfo); - } - - if (R) { -#ifndef NDEBUG - dbgs() << "Call operand #" << I << " has unhandled type " - << EVT(ArgVT).getEVTString(); -#endif - llvm_unreachable(nullptr); - } - } -} - -void MipsTargetLowering::MipsCC:: -analyzeFormalArguments(const SmallVectorImpl<ISD::InputArg> &Args, - bool IsSoftFloat, Function::const_arg_iterator FuncArg) { - unsigned NumArgs = Args.size(); - llvm::CCAssignFn *FixedFn = fixedArgFn(); - unsigned CurArgIdx = 0; - - for (unsigned I = 0; I != NumArgs; ++I) { - MVT ArgVT = Args[I].VT; - ISD::ArgFlagsTy ArgFlags = Args[I].Flags; - std::advance(FuncArg, Args[I].OrigArgIndex - CurArgIdx); - CurArgIdx = Args[I].OrigArgIndex; - - if (ArgFlags.isByVal()) { - handleByValArg(I, ArgVT, ArgVT, CCValAssign::Full, ArgFlags); - continue; - } - - MVT RegVT = getRegVT(ArgVT, FuncArg->getType(), nullptr, IsSoftFloat); - - if (!FixedFn(I, ArgVT, RegVT, CCValAssign::Full, ArgFlags, CCInfo)) - continue; - -#ifndef NDEBUG - dbgs() << "Formal Arg #" << I << " has unhandled type " - << EVT(ArgVT).getEVTString(); -#endif - llvm_unreachable(nullptr); - } -} - -template<typename Ty> -void MipsTargetLowering::MipsCC:: -analyzeReturn(const SmallVectorImpl<Ty> &RetVals, bool IsSoftFloat, - const SDNode *CallNode, const Type *RetTy) const { - CCAssignFn *Fn; - - if (IsSoftFloat && originalTypeIsF128(RetTy, CallNode)) - Fn = RetCC_F128Soft; - else - Fn = RetCC_Mips; - - for (unsigned I = 0, E = RetVals.size(); I < E; ++I) { - MVT VT = RetVals[I].VT; - ISD::ArgFlagsTy Flags = RetVals[I].Flags; - MVT RegVT = this->getRegVT(VT, RetTy, CallNode, IsSoftFloat); - - if (Fn(I, VT, RegVT, CCValAssign::Full, Flags, this->CCInfo)) { -#ifndef NDEBUG - dbgs() << "Call result #" << I << " has unhandled type " - << EVT(VT).getEVTString() << '\n'; -#endif - llvm_unreachable(nullptr); - } - } -} - -void MipsTargetLowering::MipsCC:: -analyzeCallResult(const SmallVectorImpl<ISD::InputArg> &Ins, bool IsSoftFloat, - const SDNode *CallNode, const Type *RetTy) const { - analyzeReturn(Ins, IsSoftFloat, CallNode, RetTy); -} - -void MipsTargetLowering::MipsCC:: -analyzeReturn(const SmallVectorImpl<ISD::OutputArg> &Outs, bool IsSoftFloat, - const Type *RetTy) const { - analyzeReturn(Outs, IsSoftFloat, nullptr, RetTy); -} - -void MipsTargetLowering::MipsCC::handleByValArg(unsigned ValNo, MVT ValVT, - MVT LocVT, - CCValAssign::LocInfo LocInfo, - ISD::ArgFlagsTy ArgFlags) { - assert(ArgFlags.getByValSize() && "Byval argument's size shouldn't be 0."); - - struct ByValArgInfo ByVal; - unsigned RegSize = regSize(); - unsigned ByValSize = RoundUpToAlignment(ArgFlags.getByValSize(), RegSize); - unsigned Align = std::min(std::max(ArgFlags.getByValAlign(), RegSize), - RegSize * 2); - - if (useRegsForByval()) - allocateRegs(ByVal, ByValSize, Align); - - // Allocate space on caller's stack. - ByVal.Address = CCInfo.AllocateStack(ByValSize - RegSize * ByVal.NumRegs, - Align); - CCInfo.addLoc(CCValAssign::getMem(ValNo, ValVT, ByVal.Address, LocVT, - LocInfo)); - ByValArgs.push_back(ByVal); -} - -unsigned MipsTargetLowering::MipsCC::numIntArgRegs() const { - return IsO32 ? array_lengthof(O32IntRegs) : array_lengthof(Mips64IntRegs); -} - -unsigned MipsTargetLowering::MipsCC::reservedArgArea() const { - return (IsO32 && (CallConv != CallingConv::Fast)) ? 16 : 0; -} - -const MCPhysReg *MipsTargetLowering::MipsCC::intArgRegs() const { - return IsO32 ? O32IntRegs : Mips64IntRegs; -} - -llvm::CCAssignFn *MipsTargetLowering::MipsCC::fixedArgFn() const { - if (CallConv == CallingConv::Fast) - return CC_Mips_FastCC; - - if (SpecialCallingConv == Mips16RetHelperConv) - return CC_Mips16RetHelper; - return IsO32 ? (IsFP64 ? CC_MipsO32_FP64 : CC_MipsO32_FP32) : CC_MipsN; -} - -llvm::CCAssignFn *MipsTargetLowering::MipsCC::varArgFn() const { - return IsO32 ? (IsFP64 ? CC_MipsO32_FP64 : CC_MipsO32_FP32) : CC_MipsN_VarArg; -} - -const MCPhysReg *MipsTargetLowering::MipsCC::shadowRegs() const { - return IsO32 ? O32IntRegs : Mips64DPRegs; -} - -void MipsTargetLowering::MipsCC::allocateRegs(ByValArgInfo &ByVal, - unsigned ByValSize, - unsigned Align) { - unsigned RegSize = regSize(), NumIntArgRegs = numIntArgRegs(); - const MCPhysReg *IntArgRegs = intArgRegs(), *ShadowRegs = shadowRegs(); - assert(!(ByValSize % RegSize) && !(Align % RegSize) && - "Byval argument's size and alignment should be a multiple of" - "RegSize."); - - ByVal.FirstIdx = CCInfo.getFirstUnallocated(IntArgRegs, NumIntArgRegs); - - // If Align > RegSize, the first arg register must be even. - if ((Align > RegSize) && (ByVal.FirstIdx % 2)) { - CCInfo.AllocateReg(IntArgRegs[ByVal.FirstIdx], ShadowRegs[ByVal.FirstIdx]); - ++ByVal.FirstIdx; - } - - // Mark the registers allocated. - for (unsigned I = ByVal.FirstIdx; ByValSize && (I < NumIntArgRegs); - ByValSize -= RegSize, ++I, ++ByVal.NumRegs) - CCInfo.AllocateReg(IntArgRegs[I], ShadowRegs[I]); -} - -MVT MipsTargetLowering::MipsCC::getRegVT(MVT VT, const Type *OrigTy, - const SDNode *CallNode, - bool IsSoftFloat) const { - if (IsSoftFloat || IsO32) - return VT; - - // Check if the original type was fp128. - if (originalTypeIsF128(OrigTy, CallNode)) { - assert(VT == MVT::i64); - return MVT::f64; - } - - return VT; -} - -void MipsTargetLowering:: -copyByValRegs(SDValue Chain, SDLoc DL, std::vector<SDValue> &OutChains, - SelectionDAG &DAG, const ISD::ArgFlagsTy &Flags, - SmallVectorImpl<SDValue> &InVals, const Argument *FuncArg, - const MipsCC &CC, const ByValArgInfo &ByVal) const { +void MipsTargetLowering::copyByValRegs( + SDValue Chain, SDLoc DL, std::vector<SDValue> &OutChains, SelectionDAG &DAG, + const ISD::ArgFlagsTy &Flags, SmallVectorImpl<SDValue> &InVals, + const Argument *FuncArg, unsigned FirstReg, unsigned LastReg, + const CCValAssign &VA, MipsCCState &State) const { MachineFunction &MF = DAG.getMachineFunction(); MachineFrameInfo *MFI = MF.getFrameInfo(); - unsigned RegAreaSize = ByVal.NumRegs * CC.regSize(); + unsigned GPRSizeInBytes = Subtarget.getGPRSizeInBytes(); + unsigned NumRegs = LastReg - FirstReg; + unsigned RegAreaSize = NumRegs * GPRSizeInBytes; unsigned FrameObjSize = std::max(Flags.getByValSize(), RegAreaSize); int FrameObjOffset; + const MipsABIInfo &ABI = Subtarget.getABI(); + ArrayRef<MCPhysReg> ByValArgRegs = ABI.GetByValArgRegs(); if (RegAreaSize) - FrameObjOffset = (int)CC.reservedArgArea() - - (int)((CC.numIntArgRegs() - ByVal.FirstIdx) * CC.regSize()); + FrameObjOffset = + (int)ABI.GetCalleeAllocdArgSizeInBytes(State.getCallingConv()) - + (int)((ByValArgRegs.size() - FirstReg) * GPRSizeInBytes); else - FrameObjOffset = ByVal.Address; + FrameObjOffset = VA.getLocMemOffset(); // Create frame object. EVT PtrTy = getPointerTy(); @@ -3553,17 +3513,17 @@ copyByValRegs(SDValue Chain, SDLoc DL, std::vector<SDValue> &OutChains, SDValue FIN = DAG.getFrameIndex(FI, PtrTy); InVals.push_back(FIN); - if (!ByVal.NumRegs) + if (!NumRegs) return; // Copy arg registers. - MVT RegTy = MVT::getIntegerVT(CC.regSize() * 8); + MVT RegTy = MVT::getIntegerVT(GPRSizeInBytes * 8); const TargetRegisterClass *RC = getRegClassFor(RegTy); - for (unsigned I = 0; I < ByVal.NumRegs; ++I) { - unsigned ArgReg = CC.intArgRegs()[ByVal.FirstIdx + I]; + for (unsigned I = 0; I < NumRegs; ++I) { + unsigned ArgReg = ByValArgRegs[FirstReg + I]; unsigned VReg = addLiveIn(MF, ArgReg, RC); - unsigned Offset = I * CC.regSize(); + unsigned Offset = I * GPRSizeInBytes; SDValue StorePtr = DAG.getNode(ISD::ADD, DL, PtrTy, FIN, DAG.getConstant(Offset, PtrTy)); SDValue Store = DAG.getStore(Chain, DL, DAG.getRegister(VReg, RegTy), @@ -3574,34 +3534,34 @@ copyByValRegs(SDValue Chain, SDLoc DL, std::vector<SDValue> &OutChains, } // Copy byVal arg to registers and stack. -void MipsTargetLowering:: -passByValArg(SDValue Chain, SDLoc DL, - std::deque< std::pair<unsigned, SDValue> > &RegsToPass, - SmallVectorImpl<SDValue> &MemOpChains, SDValue StackPtr, - MachineFrameInfo *MFI, SelectionDAG &DAG, SDValue Arg, - const MipsCC &CC, const ByValArgInfo &ByVal, - const ISD::ArgFlagsTy &Flags, bool isLittle) const { +void MipsTargetLowering::passByValArg( + SDValue Chain, SDLoc DL, + std::deque<std::pair<unsigned, SDValue>> &RegsToPass, + SmallVectorImpl<SDValue> &MemOpChains, SDValue StackPtr, + MachineFrameInfo *MFI, SelectionDAG &DAG, SDValue Arg, unsigned FirstReg, + unsigned LastReg, const ISD::ArgFlagsTy &Flags, bool isLittle, + const CCValAssign &VA) const { unsigned ByValSizeInBytes = Flags.getByValSize(); unsigned OffsetInBytes = 0; // From beginning of struct - unsigned RegSizeInBytes = CC.regSize(); + unsigned RegSizeInBytes = Subtarget.getGPRSizeInBytes(); unsigned Alignment = std::min(Flags.getByValAlign(), RegSizeInBytes); EVT PtrTy = getPointerTy(), RegTy = MVT::getIntegerVT(RegSizeInBytes * 8); + unsigned NumRegs = LastReg - FirstReg; - if (ByVal.NumRegs) { - const MCPhysReg *ArgRegs = CC.intArgRegs(); - bool LeftoverBytes = (ByVal.NumRegs * RegSizeInBytes > ByValSizeInBytes); + if (NumRegs) { + const ArrayRef<MCPhysReg> ArgRegs = Subtarget.getABI().GetByValArgRegs(); + bool LeftoverBytes = (NumRegs * RegSizeInBytes > ByValSizeInBytes); unsigned I = 0; // Copy words to registers. - for (; I < ByVal.NumRegs - LeftoverBytes; - ++I, OffsetInBytes += RegSizeInBytes) { + for (; I < NumRegs - LeftoverBytes; ++I, OffsetInBytes += RegSizeInBytes) { SDValue LoadPtr = DAG.getNode(ISD::ADD, DL, PtrTy, Arg, DAG.getConstant(OffsetInBytes, PtrTy)); SDValue LoadVal = DAG.getLoad(RegTy, DL, Chain, LoadPtr, MachinePointerInfo(), false, false, false, Alignment); MemOpChains.push_back(LoadVal.getValue(1)); - unsigned ArgReg = ArgRegs[ByVal.FirstIdx + I]; + unsigned ArgReg = ArgRegs[FirstReg + I]; RegsToPass.push_back(std::make_pair(ArgReg, LoadVal)); } @@ -3611,9 +3571,6 @@ passByValArg(SDValue Chain, SDLoc DL, // Copy the remainder of the byval argument with sub-word loads and shifts. if (LeftoverBytes) { - assert((ByValSizeInBytes > OffsetInBytes) && - (ByValSizeInBytes < OffsetInBytes + RegSizeInBytes) && - "Size of the remainder should be smaller than RegSizeInBytes."); SDValue Val; for (unsigned LoadSizeInBytes = RegSizeInBytes / 2, TotalBytesLoaded = 0; @@ -3652,7 +3609,7 @@ passByValArg(SDValue Chain, SDLoc DL, Alignment = std::min(Alignment, LoadSizeInBytes); } - unsigned ArgReg = ArgRegs[ByVal.FirstIdx + I]; + unsigned ArgReg = ArgRegs[FirstReg + I]; RegsToPass.push_back(std::make_pair(ArgReg, Val)); return; } @@ -3663,7 +3620,7 @@ passByValArg(SDValue Chain, SDLoc DL, SDValue Src = DAG.getNode(ISD::ADD, DL, PtrTy, Arg, DAG.getConstant(OffsetInBytes, PtrTy)); SDValue Dst = DAG.getNode(ISD::ADD, DL, PtrTy, StackPtr, - DAG.getIntPtrConstant(ByVal.Address)); + DAG.getIntPtrConstant(VA.getLocMemOffset())); Chain = DAG.getMemcpy(Chain, DL, Dst, Src, DAG.getConstant(MemCpySize, PtrTy), Alignment, /*isVolatile=*/false, /*AlwaysInline=*/false, MachinePointerInfo(), MachinePointerInfo()); @@ -3671,14 +3628,13 @@ passByValArg(SDValue Chain, SDLoc DL, } void MipsTargetLowering::writeVarArgRegs(std::vector<SDValue> &OutChains, - const MipsCC &CC, SDValue Chain, - SDLoc DL, SelectionDAG &DAG) const { - unsigned NumRegs = CC.numIntArgRegs(); - const MCPhysReg *ArgRegs = CC.intArgRegs(); - const CCState &CCInfo = CC.getCCInfo(); - unsigned Idx = CCInfo.getFirstUnallocated(ArgRegs, NumRegs); - unsigned RegSize = CC.regSize(); - MVT RegTy = MVT::getIntegerVT(RegSize * 8); + SDValue Chain, SDLoc DL, + SelectionDAG &DAG, + CCState &State) const { + const ArrayRef<MCPhysReg> ArgRegs = Subtarget.getABI().GetVarArgRegs(); + unsigned Idx = State.getFirstUnallocated(ArgRegs.data(), ArgRegs.size()); + unsigned RegSizeInBytes = Subtarget.getGPRSizeInBytes(); + MVT RegTy = MVT::getIntegerVT(RegSizeInBytes * 8); const TargetRegisterClass *RC = getRegClassFor(RegTy); MachineFunction &MF = DAG.getMachineFunction(); MachineFrameInfo *MFI = MF.getFrameInfo(); @@ -3687,24 +3643,30 @@ void MipsTargetLowering::writeVarArgRegs(std::vector<SDValue> &OutChains, // Offset of the first variable argument from stack pointer. int VaArgOffset; - if (NumRegs == Idx) - VaArgOffset = RoundUpToAlignment(CCInfo.getNextStackOffset(), RegSize); - else - VaArgOffset = (int)CC.reservedArgArea() - (int)(RegSize * (NumRegs - Idx)); + if (ArgRegs.size() == Idx) + VaArgOffset = + RoundUpToAlignment(State.getNextStackOffset(), RegSizeInBytes); + else { + const MipsABIInfo &ABI = Subtarget.getABI(); + VaArgOffset = + (int)ABI.GetCalleeAllocdArgSizeInBytes(State.getCallingConv()) - + (int)(RegSizeInBytes * (ArgRegs.size() - Idx)); + } // Record the frame index of the first variable argument // which is a value necessary to VASTART. - int FI = MFI->CreateFixedObject(RegSize, VaArgOffset, true); + int FI = MFI->CreateFixedObject(RegSizeInBytes, VaArgOffset, true); MipsFI->setVarArgsFrameIndex(FI); // Copy the integer registers that have not been used for argument passing // to the argument register save area. For O32, the save area is allocated // in the caller's stack frame, while for N32/64, it is allocated in the // callee's stack frame. - for (unsigned I = Idx; I < NumRegs; ++I, VaArgOffset += RegSize) { + for (unsigned I = Idx; I < ArgRegs.size(); + ++I, VaArgOffset += RegSizeInBytes) { unsigned Reg = addLiveIn(MF, ArgRegs[I], RC); SDValue ArgValue = DAG.getCopyFromReg(Chain, DL, Reg, RegTy); - FI = MFI->CreateFixedObject(RegSize, VaArgOffset, true); + FI = MFI->CreateFixedObject(RegSizeInBytes, VaArgOffset, true); SDValue PtrOff = DAG.getFrameIndex(FI, getPointerTy()); SDValue Store = DAG.getStore(Chain, DL, ArgValue, PtrOff, MachinePointerInfo(), false, false, 0); @@ -3713,3 +3675,49 @@ void MipsTargetLowering::writeVarArgRegs(std::vector<SDValue> &OutChains, OutChains.push_back(Store); } } + +void MipsTargetLowering::HandleByVal(CCState *State, unsigned &Size, + unsigned Align) const { + MachineFunction &MF = State->getMachineFunction(); + const TargetFrameLowering *TFL = MF.getTarget().getFrameLowering(); + + assert(Size && "Byval argument's size shouldn't be 0."); + + Align = std::min(Align, TFL->getStackAlignment()); + + unsigned FirstReg = 0; + unsigned NumRegs = 0; + + if (State->getCallingConv() != CallingConv::Fast) { + unsigned RegSizeInBytes = Subtarget.getGPRSizeInBytes(); + const ArrayRef<MCPhysReg> IntArgRegs = Subtarget.getABI().GetByValArgRegs(); + // FIXME: The O32 case actually describes no shadow registers. + const MCPhysReg *ShadowRegs = + Subtarget.isABI_O32() ? IntArgRegs.data() : Mips64DPRegs; + + // We used to check the size as well but we can't do that anymore since + // CCState::HandleByVal() rounds up the size after calling this function. + assert(!(Align % RegSizeInBytes) && + "Byval argument's alignment should be a multiple of" + "RegSizeInBytes."); + + FirstReg = State->getFirstUnallocated(IntArgRegs.data(), IntArgRegs.size()); + + // If Align > RegSizeInBytes, the first arg register must be even. + // FIXME: This condition happens to do the right thing but it's not the + // right way to test it. We want to check that the stack frame offset + // of the register is aligned. + if ((Align > RegSizeInBytes) && (FirstReg % 2)) { + State->AllocateReg(IntArgRegs[FirstReg], ShadowRegs[FirstReg]); + ++FirstReg; + } + + // Mark the registers allocated. + Size = RoundUpToAlignment(Size, RegSizeInBytes); + for (unsigned I = FirstReg; Size > 0 && (I < IntArgRegs.size()); + Size -= RegSizeInBytes, ++I, ++NumRegs) + State->AllocateReg(IntArgRegs[I], ShadowRegs[I]); + } + + State->addInRegsParamInfo(FirstReg, FirstReg + NumRegs); +} diff --git a/contrib/llvm/lib/Target/Mips/MipsISelLowering.h b/contrib/llvm/lib/Target/Mips/MipsISelLowering.h index 10e4e0b..23163b7 100644 --- a/contrib/llvm/lib/Target/Mips/MipsISelLowering.h +++ b/contrib/llvm/lib/Target/Mips/MipsISelLowering.h @@ -210,6 +210,7 @@ namespace llvm { //===--------------------------------------------------------------------===// class MipsFunctionInfo; class MipsSubtarget; + class MipsCCState; class MipsTargetLowering : public TargetLowering { bool isMicroMips; @@ -259,6 +260,8 @@ namespace llvm { } }; + void HandleByVal(CCState *, unsigned &, unsigned) const override; + protected: SDValue getGlobalReg(SelectionDAG &DAG, EVT Ty) const; @@ -338,101 +341,6 @@ namespace llvm { bool IsPICCall, bool GlobalOrExternal, bool InternalLinkage, CallLoweringInfo &CLI, SDValue Callee, SDValue Chain) const; - /// ByValArgInfo - Byval argument information. - struct ByValArgInfo { - unsigned FirstIdx; // Index of the first register used. - unsigned NumRegs; // Number of registers used for this argument. - unsigned Address; // Offset of the stack area used to pass this argument. - - ByValArgInfo() : FirstIdx(0), NumRegs(0), Address(0) {} - }; - - /// MipsCC - This class provides methods used to analyze formal and call - /// arguments and inquire about calling convention information. - class MipsCC { - public: - enum SpecialCallingConvType { - Mips16RetHelperConv, NoSpecialCallingConv - }; - - MipsCC(CallingConv::ID CallConv, bool IsO32, bool IsFP64, CCState &Info, - SpecialCallingConvType SpecialCallingConv = NoSpecialCallingConv); - - - void analyzeCallOperands(const SmallVectorImpl<ISD::OutputArg> &Outs, - bool IsVarArg, bool IsSoftFloat, - const SDNode *CallNode, - std::vector<ArgListEntry> &FuncArgs); - void analyzeFormalArguments(const SmallVectorImpl<ISD::InputArg> &Ins, - bool IsSoftFloat, - Function::const_arg_iterator FuncArg); - - void analyzeCallResult(const SmallVectorImpl<ISD::InputArg> &Ins, - bool IsSoftFloat, const SDNode *CallNode, - const Type *RetTy) const; - - void analyzeReturn(const SmallVectorImpl<ISD::OutputArg> &Outs, - bool IsSoftFloat, const Type *RetTy) const; - - const CCState &getCCInfo() const { return CCInfo; } - - /// hasByValArg - Returns true if function has byval arguments. - bool hasByValArg() const { return !ByValArgs.empty(); } - - /// regSize - Size (in number of bits) of integer registers. - unsigned regSize() const { return IsO32 ? 4 : 8; } - - /// numIntArgRegs - Number of integer registers available for calls. - unsigned numIntArgRegs() const; - - /// reservedArgArea - The size of the area the caller reserves for - /// register arguments. This is 16-byte if ABI is O32. - unsigned reservedArgArea() const; - - /// Return pointer to array of integer argument registers. - const MCPhysReg *intArgRegs() const; - - typedef SmallVectorImpl<ByValArgInfo>::const_iterator byval_iterator; - byval_iterator byval_begin() const { return ByValArgs.begin(); } - byval_iterator byval_end() const { return ByValArgs.end(); } - - private: - void handleByValArg(unsigned ValNo, MVT ValVT, MVT LocVT, - CCValAssign::LocInfo LocInfo, - ISD::ArgFlagsTy ArgFlags); - - /// useRegsForByval - Returns true if the calling convention allows the - /// use of registers to pass byval arguments. - bool useRegsForByval() const { return CallConv != CallingConv::Fast; } - - /// Return the function that analyzes fixed argument list functions. - llvm::CCAssignFn *fixedArgFn() const; - - /// Return the function that analyzes variable argument list functions. - llvm::CCAssignFn *varArgFn() const; - - const MCPhysReg *shadowRegs() const; - - void allocateRegs(ByValArgInfo &ByVal, unsigned ByValSize, - unsigned Align); - - /// Return the type of the register which is used to pass an argument or - /// return a value. This function returns f64 if the argument is an i64 - /// value which has been generated as a result of softening an f128 value. - /// Otherwise, it just returns VT. - MVT getRegVT(MVT VT, const Type *OrigTy, const SDNode *CallNode, - bool IsSoftFloat) const; - - template<typename Ty> - void analyzeReturn(const SmallVectorImpl<Ty> &RetVals, bool IsSoftFloat, - const SDNode *CallNode, const Type *RetTy) const; - - CCState &CCInfo; - CallingConv::ID CallConv; - bool IsO32, IsFP64; - SpecialCallingConvType SpecialCallingConv; - SmallVector<ByValArgInfo, 2> ByValArgs; - }; protected: SDValue lowerLOAD(SDValue Op, SelectionDAG &DAG) const; SDValue lowerSTORE(SDValue Op, SelectionDAG &DAG) const; @@ -461,14 +369,12 @@ namespace llvm { SDValue getTargetNode(ConstantPoolSDNode *N, EVT Ty, SelectionDAG &DAG, unsigned Flag) const; - MipsCC::SpecialCallingConvType getSpecialCallingConv(SDValue Callee) const; // Lower Operand helpers SDValue LowerCallResult(SDValue Chain, SDValue InFlag, CallingConv::ID CallConv, bool isVarArg, - const SmallVectorImpl<ISD::InputArg> &Ins, - SDLoc dl, SelectionDAG &DAG, - SmallVectorImpl<SDValue> &InVals, - const SDNode *CallNode, const Type *RetTy) const; + const SmallVectorImpl<ISD::InputArg> &Ins, SDLoc dl, + SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals, + TargetLowering::CallLoweringInfo &CLI) const; // Lower Operand specifics SDValue lowerBR_JT(SDValue Op, SelectionDAG &DAG) const; @@ -482,6 +388,7 @@ namespace llvm { SDValue lowerSELECT_CC(SDValue Op, SelectionDAG &DAG) const; SDValue lowerSETCC(SDValue Op, SelectionDAG &DAG) const; SDValue lowerVASTART(SDValue Op, SelectionDAG &DAG) const; + SDValue lowerVAARG(SDValue Op, SelectionDAG &DAG) const; SDValue lowerFCOPYSIGN(SDValue Op, SelectionDAG &DAG) const; SDValue lowerFABS(SDValue Op, SelectionDAG &DAG) const; SDValue lowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const; @@ -497,33 +404,34 @@ namespace llvm { /// isEligibleForTailCallOptimization - Check whether the call is eligible /// for tail call optimization. virtual bool - isEligibleForTailCallOptimization(const MipsCC &MipsCCInfo, + isEligibleForTailCallOptimization(const CCState &CCInfo, unsigned NextStackOffset, - const MipsFunctionInfo& FI) const = 0; + const MipsFunctionInfo &FI) const = 0; /// copyByValArg - Copy argument registers which were used to pass a byval /// argument to the stack. Create a stack frame object for the byval /// argument. - void copyByValRegs(SDValue Chain, SDLoc DL, - std::vector<SDValue> &OutChains, SelectionDAG &DAG, - const ISD::ArgFlagsTy &Flags, + void copyByValRegs(SDValue Chain, SDLoc DL, std::vector<SDValue> &OutChains, + SelectionDAG &DAG, const ISD::ArgFlagsTy &Flags, SmallVectorImpl<SDValue> &InVals, - const Argument *FuncArg, - const MipsCC &CC, const ByValArgInfo &ByVal) const; + const Argument *FuncArg, unsigned FirstReg, + unsigned LastReg, const CCValAssign &VA, + MipsCCState &State) const; /// passByValArg - Pass a byval argument in registers or on stack. void passByValArg(SDValue Chain, SDLoc DL, - std::deque< std::pair<unsigned, SDValue> > &RegsToPass, + std::deque<std::pair<unsigned, SDValue>> &RegsToPass, SmallVectorImpl<SDValue> &MemOpChains, SDValue StackPtr, MachineFrameInfo *MFI, SelectionDAG &DAG, SDValue Arg, - const MipsCC &CC, const ByValArgInfo &ByVal, - const ISD::ArgFlagsTy &Flags, bool isLittle) const; + unsigned FirstReg, unsigned LastReg, + const ISD::ArgFlagsTy &Flags, bool isLittle, + const CCValAssign &VA) const; /// writeVarArgRegs - Write variable function arguments passed in registers /// to the stack. Also create a stack frame object for the first variable /// argument. - void writeVarArgRegs(std::vector<SDValue> &OutChains, const MipsCC &CC, - SDValue Chain, SDLoc DL, SelectionDAG &DAG) const; + void writeVarArgRegs(std::vector<SDValue> &OutChains, SDValue Chain, + SDLoc DL, SelectionDAG &DAG, CCState &State) const; SDValue LowerFormalArguments(SDValue Chain, diff --git a/contrib/llvm/lib/Target/Mips/MipsInstrFPU.td b/contrib/llvm/lib/Target/Mips/MipsInstrFPU.td index 29d8e30..1100e1e 100644 --- a/contrib/llvm/lib/Target/Mips/MipsInstrFPU.td +++ b/contrib/llvm/lib/Target/Mips/MipsInstrFPU.td @@ -178,6 +178,38 @@ class SW_FT<string opstr, RegisterOperand RC, InstrItinClass Itin, let mayStore = 1; } +class SW_FT2<string opstr, RegisterOperand RC, InstrItinClass Itin, + SDPatternOperator OpNode= null_frag> : + InstSE<(outs), (ins RC:$rt, mem:$addr), !strconcat(opstr, "\t$rt, $addr"), + [(OpNode RC:$rt, addrDefault:$addr)], Itin, FrmFI, opstr> { + let DecoderMethod = "DecodeFMem2"; + let mayStore = 1; +} + +class LW_FT2<string opstr, RegisterOperand RC, InstrItinClass Itin, + SDPatternOperator OpNode= null_frag> : + InstSE<(outs RC:$rt), (ins mem:$addr), !strconcat(opstr, "\t$rt, $addr"), + [(set RC:$rt, (OpNode addrDefault:$addr))], Itin, FrmFI, opstr> { + let DecoderMethod = "DecodeFMem2"; + let mayLoad = 1; +} + +class SW_FT3<string opstr, RegisterOperand RC, InstrItinClass Itin, + SDPatternOperator OpNode= null_frag> : + InstSE<(outs), (ins RC:$rt, mem:$addr), !strconcat(opstr, "\t$rt, $addr"), + [(OpNode RC:$rt, addrDefault:$addr)], Itin, FrmFI, opstr> { + let DecoderMethod = "DecodeFMem3"; + let mayStore = 1; +} + +class LW_FT3<string opstr, RegisterOperand RC, InstrItinClass Itin, + SDPatternOperator OpNode= null_frag> : + InstSE<(outs RC:$rt), (ins mem:$addr), !strconcat(opstr, "\t$rt, $addr"), + [(set RC:$rt, (OpNode addrDefault:$addr))], Itin, FrmFI, opstr> { + let DecoderMethod = "DecodeFMem3"; + let mayLoad = 1; +} + class MADDS_FT<string opstr, RegisterOperand RC, InstrItinClass Itin, SDPatternOperator OpNode = null_frag> : InstSE<(outs RC:$fd), (ins RC:$fr, RC:$fs, RC:$ft), @@ -407,24 +439,24 @@ def SDC1 : MMRel, SW_FT<"sdc1", AFGR64Opnd, II_SDC1, store>, LW_FM<0x3d>, // Cop2 Memory Instructions // FIXME: These aren't really FPU instructions and as such don't belong in this // file -def LWC2 : LW_FT<"lwc2", COP2Opnd, NoItinerary, load>, LW_FM<0x32>, +def LWC2 : LW_FT2<"lwc2", COP2Opnd, NoItinerary, load>, LW_FM<0x32>, ISA_MIPS1_NOT_32R6_64R6; -def SWC2 : SW_FT<"swc2", COP2Opnd, NoItinerary, store>, LW_FM<0x3a>, +def SWC2 : SW_FT2<"swc2", COP2Opnd, NoItinerary, store>, LW_FM<0x3a>, ISA_MIPS1_NOT_32R6_64R6; -def LDC2 : LW_FT<"ldc2", COP2Opnd, NoItinerary, load>, LW_FM<0x36>, +def LDC2 : LW_FT2<"ldc2", COP2Opnd, NoItinerary, load>, LW_FM<0x36>, ISA_MIPS2_NOT_32R6_64R6; -def SDC2 : SW_FT<"sdc2", COP2Opnd, NoItinerary, store>, LW_FM<0x3e>, +def SDC2 : SW_FT2<"sdc2", COP2Opnd, NoItinerary, store>, LW_FM<0x3e>, ISA_MIPS2_NOT_32R6_64R6; // Cop3 Memory Instructions // FIXME: These aren't really FPU instructions and as such don't belong in this // file let DecoderNamespace = "COP3_" in { - def LWC3 : LW_FT<"lwc3", COP3Opnd, NoItinerary, load>, LW_FM<0x33>; - def SWC3 : SW_FT<"swc3", COP3Opnd, NoItinerary, store>, LW_FM<0x3b>; - def LDC3 : LW_FT<"ldc3", COP3Opnd, NoItinerary, load>, LW_FM<0x37>, + def LWC3 : LW_FT3<"lwc3", COP3Opnd, NoItinerary, load>, LW_FM<0x33>; + def SWC3 : SW_FT3<"swc3", COP3Opnd, NoItinerary, store>, LW_FM<0x3b>; + def LDC3 : LW_FT3<"ldc3", COP3Opnd, NoItinerary, load>, LW_FM<0x37>, ISA_MIPS2; - def SDC3 : SW_FT<"sdc3", COP3Opnd, NoItinerary, store>, LW_FM<0x3f>, + def SDC3 : SW_FT3<"sdc3", COP3Opnd, NoItinerary, store>, LW_FM<0x3f>, ISA_MIPS2; } diff --git a/contrib/llvm/lib/Target/Mips/MipsInstrInfo.td b/contrib/llvm/lib/Target/Mips/MipsInstrInfo.td index 8e9472c..26ecfbf 100644 --- a/contrib/llvm/lib/Target/Mips/MipsInstrInfo.td +++ b/contrib/llvm/lib/Target/Mips/MipsInstrInfo.td @@ -1414,13 +1414,15 @@ def TLBR : TLB<"tlbr">, COP0_TLB_FM<0x01>; def TLBWI : TLB<"tlbwi">, COP0_TLB_FM<0x02>; def TLBWR : TLB<"tlbwr">, COP0_TLB_FM<0x06>; -class CacheOp<string instr_asm, Operand MemOpnd, RegisterOperand GPROpnd> : +class CacheOp<string instr_asm, Operand MemOpnd> : InstSE<(outs), (ins MemOpnd:$addr, uimm5:$hint), - !strconcat(instr_asm, "\t$hint, $addr"), [], NoItinerary, FrmOther>; + !strconcat(instr_asm, "\t$hint, $addr"), [], NoItinerary, FrmOther> { + let DecoderMethod = "DecodeCacheOp"; +} -def CACHE : CacheOp<"cache", mem, GPR32Opnd>, CACHEOP_FM<0b101111>, +def CACHE : CacheOp<"cache", mem>, CACHEOP_FM<0b101111>, INSN_MIPS3_32_NOT_32R6_64R6; -def PREF : CacheOp<"pref", mem, GPR32Opnd>, CACHEOP_FM<0b110011>, +def PREF : CacheOp<"pref", mem>, CACHEOP_FM<0b110011>, INSN_MIPS3_32_NOT_32R6_64R6; //===----------------------------------------------------------------------===// diff --git a/contrib/llvm/lib/Target/Mips/MipsLongBranch.cpp b/contrib/llvm/lib/Target/Mips/MipsLongBranch.cpp index 27110b6..5790ed6 100644 --- a/contrib/llvm/lib/Target/Mips/MipsLongBranch.cpp +++ b/contrib/llvm/lib/Target/Mips/MipsLongBranch.cpp @@ -64,8 +64,8 @@ namespace { MipsLongBranch(TargetMachine &tm) : MachineFunctionPass(ID), TM(tm), IsPIC(TM.getRelocationModel() == Reloc::PIC_), - ABI(TM.getSubtarget<MipsSubtarget>().getTargetABI()), - LongBranchSeqSize(!IsPIC ? 2 : (ABI == MipsSubtarget::N64 ? 10 : + ABI(TM.getSubtarget<MipsSubtarget>().getABI()), + LongBranchSeqSize(!IsPIC ? 2 : (ABI.IsN64() ? 10 : (!TM.getSubtarget<MipsSubtarget>().isTargetNaCl() ? 9 : 10))) {} const char *getPassName() const override { @@ -86,7 +86,7 @@ namespace { MachineFunction *MF; SmallVector<MBBInfo, 16> MBBInfos; bool IsPIC; - unsigned ABI; + MipsABIInfo ABI; unsigned LongBranchSeqSize; }; @@ -273,7 +273,7 @@ void MipsLongBranch::expandToLongBranch(MBBInfo &I) { const MipsSubtarget &Subtarget = TM.getSubtarget<MipsSubtarget>(); unsigned BalOp = Subtarget.hasMips32r6() ? Mips::BAL : Mips::BAL_BR; - if (ABI != MipsSubtarget::N64) { + if (!ABI.IsN64()) { // $longbr: // addiu $sp, $sp, -8 // sw $ra, 0($sp) diff --git a/contrib/llvm/lib/Target/Mips/MipsRegisterInfo.cpp b/contrib/llvm/lib/Target/Mips/MipsRegisterInfo.cpp index 084449b..a616970 100644 --- a/contrib/llvm/lib/Target/Mips/MipsRegisterInfo.cpp +++ b/contrib/llvm/lib/Target/Mips/MipsRegisterInfo.cpp @@ -149,6 +149,12 @@ getReservedRegs(const MachineFunction &MF) const { for (unsigned I = 0; I < array_lengthof(ReservedGPR64); ++I) Reserved.set(ReservedGPR64[I]); + // For mno-abicalls, GP is a program invariant! + if (!Subtarget.isABICalls()) { + Reserved.set(Mips::GP); + Reserved.set(Mips::GP_64); + } + if (Subtarget.isFP64bit()) { // Reserve all registers in AFGR64. for (RegIter Reg = Mips::AFGR64RegClass.begin(), diff --git a/contrib/llvm/lib/Target/Mips/MipsSEFrameLowering.cpp b/contrib/llvm/lib/Target/Mips/MipsSEFrameLowering.cpp index d0a17cd..4825e1e 100644 --- a/contrib/llvm/lib/Target/Mips/MipsSEFrameLowering.cpp +++ b/contrib/llvm/lib/Target/Mips/MipsSEFrameLowering.cpp @@ -325,6 +325,8 @@ bool ExpandPseudo::expandBuildPairF64(MachineBasicBlock &MBB, // We re-use the same spill slot each time so that the stack frame doesn't // grow too much in functions with a large number of moves. int FI = MF.getInfo<MipsFunctionInfo>()->getMoveF64ViaSpillFI(RC2); + if (!Subtarget.isLittle()) + std::swap(LoReg, HiReg); TII.storeRegToStack(MBB, I, LoReg, I->getOperand(1).isKill(), FI, RC, &TRI, 0); TII.storeRegToStack(MBB, I, HiReg, I->getOperand(2).isKill(), FI, RC, &TRI, @@ -369,6 +371,7 @@ bool ExpandPseudo::expandExtractElementF64(MachineBasicBlock &MBB, unsigned DstReg = I->getOperand(0).getReg(); unsigned SrcReg = I->getOperand(1).getReg(); unsigned N = I->getOperand(2).getImm(); + int64_t Offset = 4 * (Subtarget.isLittle() ? N : (1 - N)); // It should be impossible to have FGR64 on MIPS-II or MIPS32r1 (which are // the cases where mfhc1 is not available). 64-bit architectures and @@ -385,7 +388,7 @@ bool ExpandPseudo::expandExtractElementF64(MachineBasicBlock &MBB, int FI = MF.getInfo<MipsFunctionInfo>()->getMoveF64ViaSpillFI(RC); TII.storeRegToStack(MBB, I, SrcReg, I->getOperand(1).isKill(), FI, RC, &TRI, 0); - TII.loadRegFromStack(MBB, I, DstReg, FI, RC2, &TRI, N * 4); + TII.loadRegFromStack(MBB, I, DstReg, FI, RC2, &TRI, Offset); return true; } diff --git a/contrib/llvm/lib/Target/Mips/MipsSEISelLowering.cpp b/contrib/llvm/lib/Target/Mips/MipsSEISelLowering.cpp index 8173615..d7452e2 100644 --- a/contrib/llvm/lib/Target/Mips/MipsSEISelLowering.cpp +++ b/contrib/llvm/lib/Target/Mips/MipsSEISelLowering.cpp @@ -1167,15 +1167,14 @@ MipsSETargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI, } } -bool MipsSETargetLowering:: -isEligibleForTailCallOptimization(const MipsCC &MipsCCInfo, - unsigned NextStackOffset, - const MipsFunctionInfo& FI) const { +bool MipsSETargetLowering::isEligibleForTailCallOptimization( + const CCState &CCInfo, unsigned NextStackOffset, + const MipsFunctionInfo &FI) const { if (!EnableMipsTailCalls) return false; // Return false if either the callee or caller has a byval argument. - if (MipsCCInfo.hasByValArg() || FI.hasByvalArg()) + if (CCInfo.getInRegsParamsCount() > 0 || FI.hasByvalArg()) return false; // Return true if the callee's argument area is no larger than the diff --git a/contrib/llvm/lib/Target/Mips/MipsSEISelLowering.h b/contrib/llvm/lib/Target/Mips/MipsSEISelLowering.h index 00d8683..2121b3b 100644 --- a/contrib/llvm/lib/Target/Mips/MipsSEISelLowering.h +++ b/contrib/llvm/lib/Target/Mips/MipsSEISelLowering.h @@ -50,9 +50,9 @@ namespace llvm { const TargetRegisterClass *getRepRegClassFor(MVT VT) const override; private: - bool isEligibleForTailCallOptimization(const MipsCC &MipsCCInfo, - unsigned NextStackOffset, - const MipsFunctionInfo& FI) const override; + bool isEligibleForTailCallOptimization( + const CCState &CCInfo, unsigned NextStackOffset, + const MipsFunctionInfo &FI) const override; void getOpndList(SmallVectorImpl<SDValue> &Ops, diff --git a/contrib/llvm/lib/Target/Mips/MipsSubtarget.cpp b/contrib/llvm/lib/Target/Mips/MipsSubtarget.cpp index 5bf875d..9a1c409 100644 --- a/contrib/llvm/lib/Target/Mips/MipsSubtarget.cpp +++ b/contrib/llvm/lib/Target/Mips/MipsSubtarget.cpp @@ -106,13 +106,14 @@ MipsSubtarget::MipsSubtarget(const std::string &TT, const std::string &CPU, const std::string &FS, bool little, MipsTargetMachine *_TM) : MipsGenSubtargetInfo(TT, CPU, FS), MipsArchVersion(Mips32), - MipsABI(UnknownABI), IsLittle(little), IsSingleFloat(false), - IsFPXX(false), IsFP64bit(false), UseOddSPReg(true), IsNaN2008bit(false), - IsGP64bit(false), HasVFPU(false), HasCnMips(false), IsLinux(true), - HasMips3_32(false), HasMips3_32r2(false), HasMips4_32(false), - HasMips4_32r2(false), HasMips5_32r2(false), InMips16Mode(false), - InMips16HardFloat(Mips16HardFloat), InMicroMipsMode(false), HasDSP(false), - HasDSPR2(false), AllowMixed16_32(Mixed16_32 | Mips_Os16), Os16(Mips_Os16), + ABI(MipsABIInfo::Unknown()), IsLittle(little), IsSingleFloat(false), + IsFPXX(false), NoABICalls(false), IsFP64bit(false), UseOddSPReg(true), + IsNaN2008bit(false), IsGP64bit(false), HasVFPU(false), HasCnMips(false), + IsLinux(true), HasMips3_32(false), HasMips3_32r2(false), + HasMips4_32(false), HasMips4_32r2(false), HasMips5_32r2(false), + InMips16Mode(false), InMips16HardFloat(Mips16HardFloat), + InMicroMipsMode(false), HasDSP(false), HasDSPR2(false), + AllowMixed16_32(Mixed16_32 | Mips_Os16), Os16(Mips_Os16), HasMSA(false), TM(_TM), TargetTriple(TT), DL(computeDataLayout(initializeSubtargetDependencies(CPU, FS, TM))), TSInfo(DL), JITInfo(), InstrInfo(MipsInstrInfo::create(*this)), @@ -135,7 +136,7 @@ MipsSubtarget::MipsSubtarget(const std::string &TT, const std::string &CPU, report_fatal_error("Code generation for MIPS-V is not implemented", false); // Assert exactly one ABI was chosen. - assert(MipsABI != UnknownABI); + assert(ABI.IsKnown()); assert((((getFeatureBits() & Mips::FeatureO32) != 0) + ((getFeatureBits() & Mips::FeatureEABI) != 0) + ((getFeatureBits() & Mips::FeatureN32) != 0) + diff --git a/contrib/llvm/lib/Target/Mips/MipsSubtarget.h b/contrib/llvm/lib/Target/Mips/MipsSubtarget.h index f326462..c5b9a3e 100644 --- a/contrib/llvm/lib/Target/Mips/MipsSubtarget.h +++ b/contrib/llvm/lib/Target/Mips/MipsSubtarget.h @@ -23,6 +23,7 @@ #include "llvm/MC/MCInstrItineraries.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Target/TargetSubtargetInfo.h" +#include "MipsABIInfo.h" #include <string> #define GET_SUBTARGETINFO_HEADER @@ -36,13 +37,6 @@ class MipsTargetMachine; class MipsSubtarget : public MipsGenSubtargetInfo { virtual void anchor(); -public: - // NOTE: O64 will not be supported. - enum MipsABIEnum { - UnknownABI, O32, N32, N64, EABI - }; - -protected: enum MipsArchEnum { Mips1, Mips2, Mips32, Mips32r2, Mips32r6, Mips3, Mips4, Mips5, Mips64, Mips64r2, Mips64r6 @@ -51,8 +45,8 @@ protected: // Mips architecture version MipsArchEnum MipsArchVersion; - // Mips supported ABIs - MipsABIEnum MipsABI; + // Selected ABI + MipsABIInfo ABI; // IsLittle - The target is Little Endian bool IsLittle; @@ -65,6 +59,9 @@ protected: // IsFPXX - MIPS O32 modeless ABI. bool IsFPXX; + // NoABICalls - Disable SVR4-style position-independent code. + bool NoABICalls; + // IsFP64bit - The target processor has 64-bit floating point registers. bool IsFP64bit; @@ -157,12 +154,12 @@ public: CodeGenOpt::Level getOptLevelToEnablePostRAScheduler() const override; /// Only O32 and EABI supported right now. - bool isABI_EABI() const { return MipsABI == EABI; } - bool isABI_N64() const { return MipsABI == N64; } - bool isABI_N32() const { return MipsABI == N32; } - bool isABI_O32() const { return MipsABI == O32; } + bool isABI_EABI() const { return ABI.IsEABI(); } + bool isABI_N64() const { return ABI.IsN64(); } + bool isABI_N32() const { return ABI.IsN32(); } + bool isABI_O32() const { return ABI.IsO32(); } bool isABI_FPXX() const { return isABI_O32() && IsFPXX; } - unsigned getTargetABI() const { return MipsABI; } + const MipsABIInfo &getABI() const { return ABI; } /// This constructor initializes the data members to match that /// of the specified triple. @@ -200,16 +197,16 @@ public: bool hasCnMips() const { return HasCnMips; } bool isLittle() const { return IsLittle; } + bool isABICalls() const { return !NoABICalls; } bool isFPXX() const { return IsFPXX; } bool isFP64bit() const { return IsFP64bit; } bool useOddSPReg() const { return UseOddSPReg; } bool noOddSPReg() const { return !UseOddSPReg; } bool isNaN2008() const { return IsNaN2008bit; } - bool isNotFP64bit() const { return !IsFP64bit; } bool isGP64bit() const { return IsGP64bit; } bool isGP32bit() const { return !IsGP64bit; } + unsigned getGPRSizeInBytes() const { return isGP64bit() ? 8 : 4; } bool isSingleFloat() const { return IsSingleFloat; } - bool isNotSingleFloat() const { return !IsSingleFloat; } bool hasVFPU() const { return HasVFPU; } bool inMips16Mode() const { return InMips16Mode; } bool inMips16ModeDefault() const { @@ -248,7 +245,6 @@ public: bool os16() const { return Os16;}; bool isTargetNaCl() const { return TargetTriple.isOSNaCl(); } - bool isNotTargetNaCl() const { return !TargetTriple.isOSNaCl(); } // for now constant islands are on for the whole compilation unit but we only // really use them if in addition we are in mips16 mode diff --git a/contrib/llvm/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp b/contrib/llvm/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp index 771b6f5..628819d 100644 --- a/contrib/llvm/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp +++ b/contrib/llvm/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp @@ -261,7 +261,7 @@ void PPCInstPrinter::printAbsBranchOperand(const MCInst *MI, unsigned OpNo, if (!MI->getOperand(OpNo).isImm()) return printOperand(MI, OpNo, O); - O << (int)MI->getOperand(OpNo).getImm()*4; + O << SignExtend32<32>((unsigned)MI->getOperand(OpNo).getImm() << 2); } diff --git a/contrib/llvm/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.cpp b/contrib/llvm/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.cpp index 4c6780f..1bb5173 100644 --- a/contrib/llvm/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.cpp +++ b/contrib/llvm/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.cpp @@ -184,6 +184,23 @@ public: if ((Flags & ELF::EF_PPC64_ABI) == 0) MCA.setELFHeaderEFlags(Flags | 2); } + void emitAssignment(MCSymbol *Symbol, const MCExpr *Value) override { + // When encoding an assignment to set symbol A to symbol B, also copy + // the st_other bits encoding the local entry point offset. + if (Value->getKind() != MCExpr::SymbolRef) + return; + const MCSymbol &RhsSym = + static_cast<const MCSymbolRefExpr *>(Value)->getSymbol(); + MCSymbolData &Data = getStreamer().getOrCreateSymbolData(&RhsSym); + MCSymbolData &SymbolData = getStreamer().getOrCreateSymbolData(Symbol); + // The "other" values are stored in the last 6 bits of the second byte. + // The traditional defines for STO values assume the full byte and thus + // the shift to pack it. + unsigned Other = MCELF::getOther(SymbolData) << 2; + Other &= ~ELF::STO_PPC64_LOCAL_MASK; + Other |= (MCELF::getOther(Data) << 2) & ELF::STO_PPC64_LOCAL_MASK; + MCELF::setOther(SymbolData, Other >> 2); + } }; class PPCTargetMachOStreamer : public PPCTargetStreamer { diff --git a/contrib/llvm/lib/Target/PowerPC/PPCAsmPrinter.cpp b/contrib/llvm/lib/Target/PowerPC/PPCAsmPrinter.cpp index 41cba1f..de7547f 100644 --- a/contrib/llvm/lib/Target/PowerPC/PPCAsmPrinter.cpp +++ b/contrib/llvm/lib/Target/PowerPC/PPCAsmPrinter.cpp @@ -400,7 +400,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineInstr *MI) { const MachineOperand &MO = MI->getOperand(1); // Map symbol -> label of TOC entry - assert(MO.isGlobal() || MO.isCPI() || MO.isJTI()); + assert(MO.isGlobal() || MO.isCPI() || MO.isJTI() || MO.isBlockAddress()); MCSymbol *MOSymbol = nullptr; if (MO.isGlobal()) MOSymbol = getSymbol(MO.getGlobal()); @@ -408,6 +408,8 @@ void PPCAsmPrinter::EmitInstruction(const MachineInstr *MI) { MOSymbol = GetCPISymbol(MO.getIndex()); else if (MO.isJTI()) MOSymbol = GetJTISymbol(MO.getIndex()); + else if (MO.isBlockAddress()) + MOSymbol = GetBlockAddressSymbol(MO.getBlockAddress()); if (PL == PICLevel::Small) { const MCExpr *Exp = @@ -431,6 +433,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineInstr *MI) { } case PPC::LDtocJTI: case PPC::LDtocCPT: + case PPC::LDtocBA: case PPC::LDtoc: { // Transform %X3 = LDtoc <ga:@min1>, %X2 LowerPPCMachineInstrToMCInst(MI, TmpInst, *this, isDarwin); @@ -441,7 +444,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineInstr *MI) { const MachineOperand &MO = MI->getOperand(1); // Map symbol -> label of TOC entry - assert(MO.isGlobal() || MO.isCPI() || MO.isJTI()); + assert(MO.isGlobal() || MO.isCPI() || MO.isJTI() || MO.isBlockAddress()); MCSymbol *MOSymbol = nullptr; if (MO.isGlobal()) MOSymbol = getSymbol(MO.getGlobal()); @@ -449,6 +452,8 @@ void PPCAsmPrinter::EmitInstruction(const MachineInstr *MI) { MOSymbol = GetCPISymbol(MO.getIndex()); else if (MO.isJTI()) MOSymbol = GetJTISymbol(MO.getIndex()); + else if (MO.isBlockAddress()) + MOSymbol = GetBlockAddressSymbol(MO.getBlockAddress()); MCSymbol *TOCEntry = lookUpOrCreateTOCEntry(MOSymbol); @@ -470,7 +475,8 @@ void PPCAsmPrinter::EmitInstruction(const MachineInstr *MI) { // reference the symbol directly. TmpInst.setOpcode(PPC::ADDIS8); const MachineOperand &MO = MI->getOperand(2); - assert((MO.isGlobal() || MO.isCPI() || MO.isJTI()) && + assert((MO.isGlobal() || MO.isCPI() || MO.isJTI() || + MO.isBlockAddress()) && "Invalid operand for ADDIStocHA!"); MCSymbol *MOSymbol = nullptr; bool IsExternal = false; @@ -490,9 +496,12 @@ void PPCAsmPrinter::EmitInstruction(const MachineInstr *MI) { MOSymbol = GetCPISymbol(MO.getIndex()); else if (MO.isJTI()) MOSymbol = GetJTISymbol(MO.getIndex()); + else if (MO.isBlockAddress()) + MOSymbol = GetBlockAddressSymbol(MO.getBlockAddress()); if (IsExternal || IsNonLocalFunction || IsCommon || IsAvailExt || - MO.isJTI() || TM.getCodeModel() == CodeModel::Large) + MO.isJTI() || MO.isBlockAddress() || + TM.getCodeModel() == CodeModel::Large) MOSymbol = lookUpOrCreateTOCEntry(MOSymbol); const MCExpr *Exp = @@ -511,12 +520,17 @@ void PPCAsmPrinter::EmitInstruction(const MachineInstr *MI) { // associated TOC entry. Otherwise reference the symbol directly. TmpInst.setOpcode(PPC::LD); const MachineOperand &MO = MI->getOperand(1); - assert((MO.isGlobal() || MO.isJTI() || MO.isCPI()) && + assert((MO.isGlobal() || MO.isCPI() || MO.isJTI() || + MO.isBlockAddress()) && "Invalid operand for LDtocL!"); MCSymbol *MOSymbol = nullptr; if (MO.isJTI()) MOSymbol = lookUpOrCreateTOCEntry(GetJTISymbol(MO.getIndex())); + else if (MO.isBlockAddress()) { + MOSymbol = GetBlockAddressSymbol(MO.getBlockAddress()); + MOSymbol = lookUpOrCreateTOCEntry(MOSymbol); + } else if (MO.isCPI()) { MOSymbol = GetCPISymbol(MO.getIndex()); if (TM.getCodeModel() == CodeModel::Large) @@ -941,7 +955,7 @@ bool PPCLinuxAsmPrinter::doFinalization(Module &M) { for (MapVector<MCSymbol*, MCSymbol*>::iterator I = TOC.begin(), E = TOC.end(); I != E; ++I) { OutStreamer.EmitLabel(I->second); - MCSymbol *S = OutContext.GetOrCreateSymbol(I->first->getName()); + MCSymbol *S = I->first; if (isPPC64) TS.emitTCEntry(*S); else diff --git a/contrib/llvm/lib/Target/PowerPC/PPCFastISel.cpp b/contrib/llvm/lib/Target/PowerPC/PPCFastISel.cpp index 2e524d6..9c55a29 100644 --- a/contrib/llvm/lib/Target/PowerPC/PPCFastISel.cpp +++ b/contrib/llvm/lib/Target/PowerPC/PPCFastISel.cpp @@ -153,7 +153,7 @@ class PPCFastISel final : public FastISel { unsigned DestReg, bool IsZExt); unsigned PPCMaterializeFP(const ConstantFP *CFP, MVT VT); unsigned PPCMaterializeGV(const GlobalValue *GV, MVT VT); - unsigned PPCMaterializeInt(const Constant *C, MVT VT); + unsigned PPCMaterializeInt(const Constant *C, MVT VT, bool UseSExt = true); unsigned PPCMaterialize32BitInt(int64_t Imm, const TargetRegisterClass *RC); unsigned PPCMaterialize64BitInt(int64_t Imm, @@ -865,7 +865,7 @@ bool PPCFastISel::SelectFPTrunc(const Instruction *I) { } // Move an i32 or i64 value in a GPR to an f64 value in an FPR. -// FIXME: When direct register moves are implemented (see PowerISA 2.08), +// FIXME: When direct register moves are implemented (see PowerISA 2.07), // those should be used instead of moving via a stack slot when the // subtarget permits. // FIXME: The code here is sloppy for the 4-byte case. Can use a 4-byte @@ -898,10 +898,10 @@ unsigned PPCFastISel::PPCMoveToFPReg(MVT SrcVT, unsigned SrcReg, if (SrcVT == MVT::i32) { if (!IsSigned) { LoadOpc = PPC::LFIWZX; - Addr.Offset = 4; + Addr.Offset = (PPCSubTarget->isLittleEndian()) ? 0 : 4; } else if (PPCSubTarget->hasLFIWAX()) { LoadOpc = PPC::LFIWAX; - Addr.Offset = 4; + Addr.Offset = (PPCSubTarget->isLittleEndian()) ? 0 : 4; } } @@ -985,7 +985,7 @@ bool PPCFastISel::SelectIToFP(const Instruction *I, bool IsSigned) { // Move the floating-point value in SrcReg into an integer destination // register, and return the register (or zero if we can't handle it). -// FIXME: When direct register moves are implemented (see PowerISA 2.08), +// FIXME: When direct register moves are implemented (see PowerISA 2.07), // those should be used instead of moving via a stack slot when the // subtarget permits. unsigned PPCFastISel::PPCMoveToIntReg(const Instruction *I, MVT VT, @@ -1548,13 +1548,23 @@ bool PPCFastISel::SelectRet(const Instruction *I) { // Special case for returning a constant integer of any size. // Materialize the constant as an i64 and copy it to the return - // register. This avoids an unnecessary extend or truncate. + // register. We still need to worry about properly extending the sign. E.g: + // If the constant has only one bit, it means it is a boolean. Therefore + // we can't use PPCMaterializeInt because it extends the sign which will + // cause negations of the returned value to be incorrect as they are + // implemented as the flip of the least significant bit. if (isa<ConstantInt>(*RV)) { const Constant *C = cast<Constant>(RV); - unsigned SrcReg = PPCMaterializeInt(C, MVT::i64); - unsigned RetReg = ValLocs[0].getLocReg(); + + CCValAssign &VA = ValLocs[0]; + + unsigned RetReg = VA.getLocReg(); + unsigned SrcReg = PPCMaterializeInt(C, MVT::i64, + VA.getLocInfo() == CCValAssign::SExt); + BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, - TII.get(TargetOpcode::COPY), RetReg).addReg(SrcReg); + TII.get(TargetOpcode::COPY), RetReg).addReg(SrcReg); + RetRegs.push_back(RetReg); } else { @@ -2014,7 +2024,8 @@ unsigned PPCFastISel::PPCMaterialize64BitInt(int64_t Imm, // Materialize an integer constant into a register, and return // the register number (or zero if we failed to handle it). -unsigned PPCFastISel::PPCMaterializeInt(const Constant *C, MVT VT) { +unsigned PPCFastISel::PPCMaterializeInt(const Constant *C, MVT VT, + bool UseSExt) { // If we're using CR bit registers for i1 values, handle that as a special // case first. if (VT == MVT::i1 && PPCSubTarget->useCRBits()) { @@ -2038,7 +2049,7 @@ unsigned PPCFastISel::PPCMaterializeInt(const Constant *C, MVT VT) { unsigned Opc = (VT == MVT::i64) ? PPC::LI8 : PPC::LI; unsigned ImmReg = createResultReg(RC); BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(Opc), ImmReg) - .addImm(CI->getSExtValue()); + .addImm( (UseSExt) ? CI->getSExtValue() : CI->getZExtValue() ); return ImmReg; } diff --git a/contrib/llvm/lib/Target/PowerPC/PPCFrameLowering.cpp b/contrib/llvm/lib/Target/PowerPC/PPCFrameLowering.cpp index b2577a9..2a02dfc 100644 --- a/contrib/llvm/lib/Target/PowerPC/PPCFrameLowering.cpp +++ b/contrib/llvm/lib/Target/PowerPC/PPCFrameLowering.cpp @@ -505,7 +505,7 @@ void PPCFrameLowering::emitPrologue(MachineFunction &MF) const { MachineModuleInfo &MMI = MF.getMMI(); const MCRegisterInfo *MRI = MMI.getContext().getRegisterInfo(); DebugLoc dl; - bool needsFrameMoves = MMI.hasDebugInfo() || + bool needsCFI = MMI.hasDebugInfo() || MF.getFunction()->needsUnwindTableEntry(); bool isPIC = MF.getTarget().getRelocationModel() == Reloc::PIC_; @@ -726,17 +726,28 @@ void PPCFrameLowering::emitPrologue(MachineFunction &MF) const { .addReg(ScratchReg); } - // Add the "machine moves" for the instructions we generated above, but in - // reverse order. - if (needsFrameMoves) { - // Show update of SP. - assert(NegFrameSize); - unsigned CFIIndex = MMI.addFrameInst( - MCCFIInstruction::createDefCfaOffset(nullptr, NegFrameSize)); + // Add Call Frame Information for the instructions we generated above. + if (needsCFI) { + unsigned CFIIndex; + + if (HasBP) { + // Define CFA in terms of BP. Do this in preference to using FP/SP, + // because if the stack needed aligning then CFA won't be at a fixed + // offset from FP/SP. + unsigned Reg = MRI->getDwarfRegNum(BPReg, true); + CFIIndex = MMI.addFrameInst( + MCCFIInstruction::createDefCfaRegister(nullptr, Reg)); + } else { + // Adjust the definition of CFA to account for the change in SP. + assert(NegFrameSize); + CFIIndex = MMI.addFrameInst( + MCCFIInstruction::createDefCfaOffset(nullptr, NegFrameSize)); + } BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION)) .addCFIIndex(CFIIndex); if (HasFP) { + // Describe where FP was saved, at a fixed offset from CFA. unsigned Reg = MRI->getDwarfRegNum(FPReg, true); CFIIndex = MMI.addFrameInst( MCCFIInstruction::createOffset(nullptr, Reg, FPOffset)); @@ -745,6 +756,7 @@ void PPCFrameLowering::emitPrologue(MachineFunction &MF) const { } if (HasBP) { + // Describe where BP was saved, at a fixed offset from CFA. unsigned Reg = MRI->getDwarfRegNum(BPReg, true); CFIIndex = MMI.addFrameInst( MCCFIInstruction::createOffset(nullptr, Reg, BPOffset)); @@ -753,6 +765,7 @@ void PPCFrameLowering::emitPrologue(MachineFunction &MF) const { } if (MustSaveLR) { + // Describe where LR was saved, at a fixed offset from CFA. unsigned Reg = MRI->getDwarfRegNum(LRReg, true); CFIIndex = MMI.addFrameInst( MCCFIInstruction::createOffset(nullptr, Reg, LROffset)); @@ -767,8 +780,9 @@ void PPCFrameLowering::emitPrologue(MachineFunction &MF) const { .addReg(SPReg) .addReg(SPReg); - if (needsFrameMoves) { - // Mark effective beginning of when frame pointer is ready. + if (!HasBP && needsCFI) { + // Change the definition of CFA from SP+offset to FP+offset, because SP + // will change at every alloca. unsigned Reg = MRI->getDwarfRegNum(FPReg, true); unsigned CFIIndex = MMI.addFrameInst( MCCFIInstruction::createDefCfaRegister(nullptr, Reg)); @@ -778,8 +792,9 @@ void PPCFrameLowering::emitPrologue(MachineFunction &MF) const { } } - if (needsFrameMoves) { - // Add callee saved registers to move list. + if (needsCFI) { + // Describe where callee saved registers were saved, at fixed offsets from + // CFA. const std::vector<CalleeSavedInfo> &CSI = MFI->getCalleeSavedInfo(); for (unsigned I = 0, E = CSI.size(); I != E; ++I) { unsigned Reg = CSI[I].getReg(); diff --git a/contrib/llvm/lib/Target/PowerPC/PPCISelDAGToDAG.cpp b/contrib/llvm/lib/Target/PowerPC/PPCISelDAGToDAG.cpp index 09c0c31..d037b6a 100644 --- a/contrib/llvm/lib/Target/PowerPC/PPCISelDAGToDAG.cpp +++ b/contrib/llvm/lib/Target/PowerPC/PPCISelDAGToDAG.cpp @@ -173,10 +173,20 @@ namespace { /// a register. The case of adding a (possibly relocatable) constant to a /// register can be improved, but it is wrong to substitute Reg+Reg for /// Reg in an asm, because the load or store opcode would have to change. - bool SelectInlineAsmMemoryOperand(const SDValue &Op, + bool SelectInlineAsmMemoryOperand(const SDValue &Op, char ConstraintCode, std::vector<SDValue> &OutOps) override { - OutOps.push_back(Op); + // We need to make sure that this one operand does not end up in r0 + // (because we might end up lowering this as 0(%op)). + const TargetRegisterInfo *TRI = TM.getRegisterInfo(); + const TargetRegisterClass *TRC = TRI->getPointerRegClass(*MF, /*Kind=*/1); + SDValue RC = CurDAG->getTargetConstant(TRC->getID(), MVT::i32); + SDValue NewOp = + SDValue(CurDAG->getMachineNode(TargetOpcode::COPY_TO_REGCLASS, + SDLoc(Op), Op.getValueType(), + Op, RC), 0); + + OutOps.push_back(NewOp); return false; } @@ -1446,7 +1456,7 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) { // For medium and large code model, we generate two instructions as // described below. Otherwise we allow SelectCodeCommon to handle this, - // selecting one of LDtoc, LDtocJTI, and LDtocCPT. + // selecting one of LDtoc, LDtocJTI, LDtocCPT, and LDtocBA. CodeModel::Model CModel = TM.getCodeModel(); if (CModel != CodeModel::Medium && CModel != CodeModel::Large) break; @@ -1463,7 +1473,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) { SDNode *Tmp = CurDAG->getMachineNode(PPC::ADDIStocHA, dl, MVT::i64, TOCbase, GA); - if (isa<JumpTableSDNode>(GA) || CModel == CodeModel::Large) + if (isa<JumpTableSDNode>(GA) || isa<BlockAddressSDNode>(GA) || + CModel == CodeModel::Large) return CurDAG->getMachineNode(PPC::LDtocL, dl, MVT::i64, GA, SDValue(Tmp, 0)); diff --git a/contrib/llvm/lib/Target/PowerPC/PPCISelLowering.cpp b/contrib/llvm/lib/Target/PowerPC/PPCISelLowering.cpp index 68bc61d..16ddaf1 100644 --- a/contrib/llvm/lib/Target/PowerPC/PPCISelLowering.cpp +++ b/contrib/llvm/lib/Target/PowerPC/PPCISelLowering.cpp @@ -1631,8 +1631,16 @@ SDValue PPCTargetLowering::LowerJumpTable(SDValue Op, SelectionDAG &DAG) const { SDValue PPCTargetLowering::LowerBlockAddress(SDValue Op, SelectionDAG &DAG) const { EVT PtrVT = Op.getValueType(); + BlockAddressSDNode *BASDN = cast<BlockAddressSDNode>(Op); + const BlockAddress *BA = BASDN->getBlockAddress(); - const BlockAddress *BA = cast<BlockAddressSDNode>(Op)->getBlockAddress(); + // 64-bit SVR4 ABI code is always position-independent. + // The actual BlockAddress is stored in the TOC. + if (Subtarget.isSVR4ABI() && Subtarget.isPPC64()) { + SDValue GA = DAG.getTargetBlockAddress(BA, PtrVT, BASDN->getOffset()); + return DAG.getNode(PPCISD::TOC_ENTRY, SDLoc(BASDN), MVT::i64, GA, + DAG.getRegister(PPC::X2, MVT::i64)); + } unsigned MOHiFlag, MOLoFlag; bool isPIC = GetLabelAccessInfo(DAG.getTarget(), MOHiFlag, MOLoFlag); @@ -2695,7 +2703,7 @@ PPCTargetLowering::LowerFormalArguments_64SVR4( int FI; if (HasParameterArea || ArgSize + ArgOffset > LinkageSize + Num_GPR_Regs * PtrByteSize) - FI = MFI->CreateFixedObject(ArgSize, ArgOffset, true); + FI = MFI->CreateFixedObject(ArgSize, ArgOffset, false); else FI = MFI->CreateStackObject(ArgSize, Align, false); SDValue FIN = DAG.getFrameIndex(FI, PtrVT); @@ -3061,7 +3069,7 @@ PPCTargetLowering::LowerFormalArguments_Darwin( CurArgOffset = CurArgOffset + (4 - ObjSize); } // The value of the object is its address. - int FI = MFI->CreateFixedObject(ObjSize, CurArgOffset, true); + int FI = MFI->CreateFixedObject(ObjSize, CurArgOffset, false); SDValue FIN = DAG.getFrameIndex(FI, PtrVT); InVals.push_back(FIN); if (ObjSize==1 || ObjSize==2) { @@ -8974,6 +8982,12 @@ PPCTargetLowering::getRegForInlineAsmConstraint(const std::string &Constraint, &PPC::G8RCRegClass); } + // GCC accepts 'cc' as an alias for 'cr0', and we need to do the same. + if (!R.second && StringRef("{cc}").equals_lower(Constraint)) { + R.first = PPC::CR0; + R.second = &PPC::CRRCRegClass; + } + return R; } @@ -9002,37 +9016,42 @@ void PPCTargetLowering::LowerAsmOperandForConstraint(SDValue Op, case 'P': { ConstantSDNode *CST = dyn_cast<ConstantSDNode>(Op); if (!CST) return; // Must be an immediate to match. - unsigned Value = CST->getZExtValue(); + int64_t Value = CST->getSExtValue(); + EVT TCVT = MVT::i64; // All constants taken to be 64 bits so that negative + // numbers are printed as such. switch (Letter) { default: llvm_unreachable("Unknown constraint letter!"); case 'I': // "I" is a signed 16-bit constant. - if ((short)Value == (int)Value) - Result = DAG.getTargetConstant(Value, Op.getValueType()); + if (isInt<16>(Value)) + Result = DAG.getTargetConstant(Value, TCVT); break; case 'J': // "J" is a constant with only the high-order 16 bits nonzero. + if (isShiftedUInt<16, 16>(Value)) + Result = DAG.getTargetConstant(Value, TCVT); + break; case 'L': // "L" is a signed 16-bit constant shifted left 16 bits. - if ((short)Value == 0) - Result = DAG.getTargetConstant(Value, Op.getValueType()); + if (isShiftedInt<16, 16>(Value)) + Result = DAG.getTargetConstant(Value, TCVT); break; case 'K': // "K" is a constant with only the low-order 16 bits nonzero. - if ((Value >> 16) == 0) - Result = DAG.getTargetConstant(Value, Op.getValueType()); + if (isUInt<16>(Value)) + Result = DAG.getTargetConstant(Value, TCVT); break; case 'M': // "M" is a constant that is greater than 31. if (Value > 31) - Result = DAG.getTargetConstant(Value, Op.getValueType()); + Result = DAG.getTargetConstant(Value, TCVT); break; case 'N': // "N" is a positive constant that is an exact power of two. - if ((int)Value > 0 && isPowerOf2_32(Value)) - Result = DAG.getTargetConstant(Value, Op.getValueType()); + if (Value > 0 && isPowerOf2_64(Value)) + Result = DAG.getTargetConstant(Value, TCVT); break; case 'O': // "O" is the constant zero. if (Value == 0) - Result = DAG.getTargetConstant(Value, Op.getValueType()); + Result = DAG.getTargetConstant(Value, TCVT); break; case 'P': // "P" is a constant whose negation is a signed 16-bit constant. - if ((short)-Value == (int)-Value) - Result = DAG.getTargetConstant(Value, Op.getValueType()); + if (isInt<16>(-Value)) + Result = DAG.getTargetConstant(Value, TCVT); break; } break; diff --git a/contrib/llvm/lib/Target/PowerPC/PPCInstr64Bit.td b/contrib/llvm/lib/Target/PowerPC/PPCInstr64Bit.td index 8a0ce46..b85d9c0 100644 --- a/contrib/llvm/lib/Target/PowerPC/PPCInstr64Bit.td +++ b/contrib/llvm/lib/Target/PowerPC/PPCInstr64Bit.td @@ -789,7 +789,7 @@ let canFoldAsLoad = 1, PPC970_Unit = 2 in { def LD : DSForm_1<58, 0, (outs g8rc:$rD), (ins memrix:$src), "ld $rD, $src", IIC_LdStLD, [(set i64:$rD, (aligned4load ixaddr:$src))]>, isPPC64; -// The following three definitions are selected for small code model only. +// The following four definitions are selected for small code model only. // Otherwise, we need to create two instructions to form a 32-bit offset, // so we have a custom matcher for TOC_ENTRY in PPCDAGToDAGIsel::Select(). def LDtoc: Pseudo<(outs g8rc:$rD), (ins tocentry:$disp, g8rc:$reg), @@ -804,6 +804,10 @@ def LDtocCPT: Pseudo<(outs g8rc:$rD), (ins tocentry:$disp, g8rc:$reg), "#LDtocCPT", [(set i64:$rD, (PPCtoc_entry tconstpool:$disp, i64:$reg))]>, isPPC64; +def LDtocBA: Pseudo<(outs g8rc:$rD), (ins tocentry:$disp, g8rc:$reg), + "#LDtocCPT", + [(set i64:$rD, + (PPCtoc_entry tblockaddress:$disp, i64:$reg))]>, isPPC64; let hasSideEffects = 1, isCodeGenOnly = 1, RST = 2, Defs = [X2] in def LDinto_toc: DSForm_1<58, 0, (outs), (ins memrix:$src), diff --git a/contrib/llvm/lib/Target/PowerPC/PPCRegisterInfo.td b/contrib/llvm/lib/Target/PowerPC/PPCRegisterInfo.td index b3d145b..0442a94 100644 --- a/contrib/llvm/lib/Target/PowerPC/PPCRegisterInfo.td +++ b/contrib/llvm/lib/Target/PowerPC/PPCRegisterInfo.td @@ -188,13 +188,6 @@ def CR6 : CR<6, "cr6", [CR6LT, CR6GT, CR6EQ, CR6UN]>, DwarfRegNum<[74, 74]>; def CR7 : CR<7, "cr7", [CR7LT, CR7GT, CR7EQ, CR7UN]>, DwarfRegNum<[75, 75]>; } -// The full condition-code register. This is not modeled fully, but defined -// here primarily, for compatibility with gcc, to allow the inline asm "cc" -// clobber specification to work. -def CC : PPCReg<"cc">, DwarfRegAlias<CR0> { - let Aliases = [CR0, CR1, CR2, CR3, CR4, CR5, CR6, CR7]; -} - // Link register def LR : SPR<8, "lr">, DwarfRegNum<[-2, 65]>; //let Aliases = [LR] in @@ -308,7 +301,3 @@ def CARRYRC : RegisterClass<"PPC", [i32], 32, (add CARRY)> { let CopyCost = -1; } -def CCRC : RegisterClass<"PPC", [i32], 32, (add CC)> { - let isAllocatable = 0; -} - diff --git a/contrib/llvm/lib/Target/X86/X86FastISel.cpp b/contrib/llvm/lib/Target/X86/X86FastISel.cpp index 2d494b4..ae3eea4 100644 --- a/contrib/llvm/lib/Target/X86/X86FastISel.cpp +++ b/contrib/llvm/lib/Target/X86/X86FastISel.cpp @@ -2829,6 +2829,9 @@ bool X86FastISel::FastLowerCall(CallLoweringInfo &CLI) { // VExt has not been implemented, so this should be impossible to reach // for now. However, fallback to Selection DAG isel once implemented. return false; + case CCValAssign::AExtUpper: + case CCValAssign::SExtUpper: + case CCValAssign::ZExtUpper: case CCValAssign::FPExt: llvm_unreachable("Unexpected loc info!"); case CCValAssign::Indirect: diff --git a/contrib/llvm/lib/Transforms/Scalar/IndVarSimplify.cpp b/contrib/llvm/lib/Transforms/Scalar/IndVarSimplify.cpp index e83a5c42..9cf0ca0 100644 --- a/contrib/llvm/lib/Transforms/Scalar/IndVarSimplify.cpp +++ b/contrib/llvm/lib/Transforms/Scalar/IndVarSimplify.cpp @@ -1623,15 +1623,30 @@ LinearFunctionTestReplace(Loop *L, // compare against the post-incremented value, otherwise we must compare // against the preincremented value. if (L->getExitingBlock() == L->getLoopLatch()) { - // Add one to the "backedge-taken" count to get the trip count. - // This addition may overflow, which is valid as long as the comparison is - // truncated to BackedgeTakenCount->getType(). - IVCount = SE->getAddExpr(BackedgeTakenCount, - SE->getConstant(BackedgeTakenCount->getType(), 1)); // The BackedgeTaken expression contains the number of times that the // backedge branches to the loop header. This is one less than the // number of times the loop executes, so use the incremented indvar. - CmpIndVar = IndVar->getIncomingValueForBlock(L->getExitingBlock()); + llvm::Value *IncrementedIndvar = + IndVar->getIncomingValueForBlock(L->getExitingBlock()); + const auto *IncrementedIndvarSCEV = + cast<SCEVAddRecExpr>(SE->getSCEV(IncrementedIndvar)); + // It is unsafe to use the incremented indvar if it has a wrapping flag, we + // don't want to compare against a poison value. Check the SCEV that + // corresponds to the incremented indvar, the SCEVExpander will only insert + // flags in the IR if the SCEV originally had wrapping flags. + // FIXME: In theory, SCEV could drop flags even though they exist in IR. + // A more robust solution would involve getting a new expression for + // CmpIndVar by applying non-NSW/NUW AddExprs. + if (!ScalarEvolution::maskFlags(IncrementedIndvarSCEV->getNoWrapFlags(), + SCEV::FlagNUW | SCEV::FlagNSW)) { + // Add one to the "backedge-taken" count to get the trip count. + // This addition may overflow, which is valid as long as the comparison is + // truncated to BackedgeTakenCount->getType(). + IVCount = + SE->getAddExpr(BackedgeTakenCount, + SE->getConstant(BackedgeTakenCount->getType(), 1)); + CmpIndVar = IncrementedIndvar; + } } Value *ExitCnt = genLoopLimit(IndVar, IVCount, L, Rewriter, SE); diff --git a/contrib/llvm/lib/Transforms/Scalar/MergedLoadStoreMotion.cpp b/contrib/llvm/lib/Transforms/Scalar/MergedLoadStoreMotion.cpp index a7e8024..c2467fe 100644 --- a/contrib/llvm/lib/Transforms/Scalar/MergedLoadStoreMotion.cpp +++ b/contrib/llvm/lib/Transforms/Scalar/MergedLoadStoreMotion.cpp @@ -27,18 +27,18 @@ // // header: // br %cond, label %if.then, label %if.else -// / \ -// / \ -// / \ +// + + +// + + +// + + // if.then: if.else: // %lt = load %addr_l %le = load %addr_l // <use %lt> <use %le> // <...> <...> // store %st, %addr_s store %se, %addr_s // br label %if.end br label %if.end -// \ / -// \ / -// \ / +// + + +// + + +// + + // if.end ("footer"): // <...> // @@ -47,16 +47,16 @@ // header: // %l = load %addr_l // br %cond, label %if.then, label %if.else -// / \ -// / \ -// / \ +// + + +// + + +// + + // if.then: if.else: // <use %l> <use %l> // <...> <...> // br label %if.end br label %if.end -// \ / -// \ / -// \ / +// + + +// + + +// + + // if.end ("footer"): // %s.sink = phi [%st, if.then], [%se, if.else] // <...> diff --git a/contrib/llvm/lib/Transforms/Scalar/SROA.cpp b/contrib/llvm/lib/Transforms/Scalar/SROA.cpp index 8c7f253..f902eb2 100644 --- a/contrib/llvm/lib/Transforms/Scalar/SROA.cpp +++ b/contrib/llvm/lib/Transforms/Scalar/SROA.cpp @@ -2697,7 +2697,10 @@ private: // the old pointer, which necessarily must be in the right position to // dominate the PHI. IRBuilderTy PtrBuilder(IRB); - PtrBuilder.SetInsertPoint(OldPtr); + if (isa<PHINode>(OldPtr)) + PtrBuilder.SetInsertPoint(OldPtr->getParent()->getFirstInsertionPt()); + else + PtrBuilder.SetInsertPoint(OldPtr); PtrBuilder.SetCurrentDebugLocation(OldPtr->getDebugLoc()); Value *NewPtr = getNewAllocaSlicePtr(PtrBuilder, OldPtr->getType()); diff --git a/contrib/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/contrib/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp index c1e8df9..79fcb09 100644 --- a/contrib/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp +++ b/contrib/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp @@ -3357,7 +3357,7 @@ void InnerLoopVectorizer::updateAnalysis() { DT->addNewBlock(LoopMiddleBlock, LoopBypassBlocks[1]); DT->addNewBlock(LoopScalarPreHeader, LoopBypassBlocks[0]); DT->changeImmediateDominator(LoopScalarBody, LoopScalarPreHeader); - DT->changeImmediateDominator(LoopExitBlock, LoopMiddleBlock); + DT->changeImmediateDominator(LoopExitBlock, LoopBypassBlocks[0]); DEBUG(DT->verifyDomTree()); } @@ -5201,7 +5201,13 @@ LoopVectorizationLegality::isInductionVariable(PHINode *Phi) { return IK_NoInduction; assert(PhiTy->isPointerTy() && "The PHI must be a pointer"); - uint64_t Size = DL->getTypeAllocSize(PhiTy->getPointerElementType()); + Type *PointerElementType = PhiTy->getPointerElementType(); + // The pointer stride cannot be determined if the pointer element type is not + // sized. + if (!PointerElementType->isSized()) + return IK_NoInduction; + + uint64_t Size = DL->getTypeAllocSize(PointerElementType); if (C->getValue()->equalsInt(Size)) return IK_PtrInduction; else if (C->getValue()->equalsInt(0 - Size)) diff --git a/contrib/llvm/patches/README.TXT b/contrib/llvm/patches/README.TXT index 86e3c2d..ad18595 100644 --- a/contrib/llvm/patches/README.TXT +++ b/contrib/llvm/patches/README.TXT @@ -1,11 +1,11 @@ This is a set of individual patches, which contain all the customizations to llvm/clang currently in the FreeBSD base system. These can be applied in -alphabetical order to a pristine llvm/clang 3.5.0 release source tree, for +alphabetical order to a pristine llvm/clang 3.5.1 release source tree, for example by doing: -svn co https://llvm.org/svn/llvm-project/llvm/tags/RELEASE_350/final llvm-3.5.0-final -svn co https://llvm.org/svn/llvm-project/cfe/tags/RELEASE_350/final llvm-3.5.0-final/tools/clang -cd llvm-3.5.0-final +svn co https://llvm.org/svn/llvm-project/llvm/tags/RELEASE_351/final llvm-3.5.1-final +svn co https://llvm.org/svn/llvm-project/cfe/tags/RELEASE_351/final llvm-3.5.1-final/tools/clang +cd llvm-3.5.1-final for p in /usr/src/contrib/llvm/patches/patch-*.diff; do patch -p0 -f -F0 -E -i $p -s || break done diff --git a/contrib/llvm/patches/patch-07-llvm-r213960-ppc32-tls.diff b/contrib/llvm/patches/patch-07-llvm-r213960-ppc32-tls.diff deleted file mode 100644 index 14d3517..0000000 --- a/contrib/llvm/patches/patch-07-llvm-r213960-ppc32-tls.diff +++ /dev/null @@ -1,547 +0,0 @@ -Pull in r213960 from upstream llvm trunk (by Hal Finkel): - - [PowerPC] Support TLS on PPC32/ELF - - Patch by Justin Hibbits! - -Introduced here: http://svnweb.freebsd.org/changeset/base/270147 - -Index: lib/Target/PowerPC/PPCISelLowering.h -=================================================================== ---- lib/Target/PowerPC/PPCISelLowering.h -+++ lib/Target/PowerPC/PPCISelLowering.h -@@ -181,6 +181,10 @@ namespace llvm { - /// on PPC32. - PPC32_GOT, - -+ /// GPRC = address of _GLOBAL_OFFSET_TABLE_. Used by general dynamic and -+ /// local dynamic TLS on PPC32. -+ PPC32_PICGOT, -+ - /// G8RC = ADDIS_GOT_TPREL_HA %X2, Symbol - Used by the initial-exec - /// TLS model, produces an ADDIS8 instruction that adds the GOT - /// base to sym\@got\@tprel\@ha. -Index: lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp -=================================================================== ---- lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp -+++ lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp -@@ -17,6 +17,7 @@ - #include "llvm/MC/MCExpr.h" - #include "llvm/MC/MCInst.h" - #include "llvm/MC/MCInstrInfo.h" -+#include "llvm/MC/MCSymbol.h" - #include "llvm/Support/CommandLine.h" - #include "llvm/Support/raw_ostream.h" - #include "llvm/Target/TargetOpcodes.h" -@@ -308,10 +309,16 @@ void PPCInstPrinter::printMemRegReg(const MCInst * - - void PPCInstPrinter::printTLSCall(const MCInst *MI, unsigned OpNo, - raw_ostream &O) { -- printBranchOperand(MI, OpNo, O); -+ // On PPC64, VariantKind is VK_None, but on PPC32, it's VK_PLT, and it must -+ // come at the _end_ of the expression. -+ const MCOperand &Op = MI->getOperand(OpNo); -+ const MCSymbolRefExpr &refExp = cast<MCSymbolRefExpr>(*Op.getExpr()); -+ O << refExp.getSymbol().getName(); - O << '('; - printOperand(MI, OpNo+1, O); - O << ')'; -+ if (refExp.getKind() != MCSymbolRefExpr::VK_None) -+ O << '@' << MCSymbolRefExpr::getVariantKindName(refExp.getKind()); - } - - -Index: lib/Target/PowerPC/PPCInstrInfo.td -=================================================================== ---- lib/Target/PowerPC/PPCInstrInfo.td -+++ lib/Target/PowerPC/PPCInstrInfo.td -@@ -588,6 +588,12 @@ def tlsreg32 : Operand<i32> { - let EncoderMethod = "getTLSRegEncoding"; - let ParserMatchClass = PPCTLSRegOperand; - } -+def tlsgd32 : Operand<i32> {} -+def tlscall32 : Operand<i32> { -+ let PrintMethod = "printTLSCall"; -+ let MIOperandInfo = (ops calltarget:$func, tlsgd32:$sym); -+ let EncoderMethod = "getTLSCallEncoding"; -+} - - // PowerPC Predicate operand. - def pred : Operand<OtherVT> { -@@ -1071,6 +1077,8 @@ let isCall = 1, PPC970_Unit = 7, Defs = [LR] in { - "bla $func", IIC_BrB, [(PPCcall (i32 imm:$func))]>; - - let isCodeGenOnly = 1 in { -+ def BL_TLS : IForm<18, 0, 1, (outs), (ins tlscall32:$func), -+ "bl $func", IIC_BrB, []>; - def BCCL : BForm<16, 0, 1, (outs), (ins pred:$cond, condbrtarget:$dst), - "b${cond:cc}l${cond:pm} ${cond:reg}, $dst">; - def BCCLA : BForm<16, 1, 1, (outs), (ins pred:$cond, abscondbrtarget:$dst), -@@ -2396,13 +2404,45 @@ def : Pat<(add i32:$in, (PPChi tblockaddress:$g, 0 - def PPC32GOT: Pseudo<(outs gprc:$rD), (ins), "#PPC32GOT", - [(set i32:$rD, (PPCppc32GOT))]>; - -+// Get the _GLOBAL_OFFSET_TABLE_ in PIC mode. -+// This uses two output registers, the first as the real output, the second as a -+// temporary register, used internally in code generation. -+def PPC32PICGOT: Pseudo<(outs gprc:$rD, gprc:$rT), (ins), "#PPC32PICGOT", -+ []>, NoEncode<"$rT">; -+ - def LDgotTprelL32: Pseudo<(outs gprc:$rD), (ins s16imm:$disp, gprc_nor0:$reg), -- "#LDgotTprelL32", -- [(set i32:$rD, -- (PPCldGotTprelL tglobaltlsaddr:$disp, i32:$reg))]>; -+ "#LDgotTprelL32", -+ [(set i32:$rD, -+ (PPCldGotTprelL tglobaltlsaddr:$disp, i32:$reg))]>; - def : Pat<(PPCaddTls i32:$in, tglobaltlsaddr:$g), - (ADD4TLS $in, tglobaltlsaddr:$g)>; - -+def ADDItlsgdL32 : Pseudo<(outs gprc:$rD), (ins gprc_nor0:$reg, s16imm:$disp), -+ "#ADDItlsgdL32", -+ [(set i32:$rD, -+ (PPCaddiTlsgdL i32:$reg, tglobaltlsaddr:$disp))]>; -+def GETtlsADDR32 : Pseudo<(outs gprc:$rD), (ins gprc:$reg, tlsgd32:$sym), -+ "#GETtlsADDR32", -+ [(set i32:$rD, -+ (PPCgetTlsAddr i32:$reg, tglobaltlsaddr:$sym))]>; -+def ADDItlsldL32 : Pseudo<(outs gprc:$rD), (ins gprc_nor0:$reg, s16imm:$disp), -+ "#ADDItlsldL32", -+ [(set i32:$rD, -+ (PPCaddiTlsldL i32:$reg, tglobaltlsaddr:$disp))]>; -+def GETtlsldADDR32 : Pseudo<(outs gprc:$rD), (ins gprc:$reg, tlsgd32:$sym), -+ "#GETtlsldADDR32", -+ [(set i32:$rD, -+ (PPCgetTlsldAddr i32:$reg, tglobaltlsaddr:$sym))]>; -+def ADDIdtprelL32 : Pseudo<(outs gprc:$rD), (ins gprc_nor0:$reg, s16imm:$disp), -+ "#ADDIdtprelL32", -+ [(set i32:$rD, -+ (PPCaddiDtprelL i32:$reg, tglobaltlsaddr:$disp))]>; -+def ADDISdtprelHA32 : Pseudo<(outs gprc:$rD), (ins gprc_nor0:$reg, s16imm:$disp), -+ "#ADDISdtprelHA32", -+ [(set i32:$rD, -+ (PPCaddisDtprelHA i32:$reg, -+ tglobaltlsaddr:$disp))]>; -+ - // Support for Position-independent code - def LWZtoc: Pseudo<(outs gprc:$rD), (ins tocentry32:$disp, gprc:$reg), - "#LWZtoc", -Index: lib/Target/PowerPC/PPCISelLowering.cpp -=================================================================== ---- lib/Target/PowerPC/PPCISelLowering.cpp -+++ lib/Target/PowerPC/PPCISelLowering.cpp -@@ -1685,47 +1685,61 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S - - if (Model == TLSModel::GeneralDynamic) { - SDValue TGA = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, 0); -- SDValue GOTReg = DAG.getRegister(PPC::X2, MVT::i64); -- SDValue GOTEntryHi = DAG.getNode(PPCISD::ADDIS_TLSGD_HA, dl, PtrVT, -- GOTReg, TGA); -+ SDValue GOTPtr; -+ if (is64bit) { -+ SDValue GOTReg = DAG.getRegister(PPC::X2, MVT::i64); -+ GOTPtr = DAG.getNode(PPCISD::ADDIS_TLSGD_HA, dl, PtrVT, -+ GOTReg, TGA); -+ } else { -+ GOTPtr = DAG.getNode(PPCISD::PPC32_PICGOT, dl, PtrVT); -+ } - SDValue GOTEntry = DAG.getNode(PPCISD::ADDI_TLSGD_L, dl, PtrVT, -- GOTEntryHi, TGA); -+ GOTPtr, TGA); - - // We need a chain node, and don't have one handy. The underlying - // call has no side effects, so using the function entry node - // suffices. - SDValue Chain = DAG.getEntryNode(); -- Chain = DAG.getCopyToReg(Chain, dl, PPC::X3, GOTEntry); -- SDValue ParmReg = DAG.getRegister(PPC::X3, MVT::i64); -+ Chain = DAG.getCopyToReg(Chain, dl, -+ is64bit ? PPC::X3 : PPC::R3, GOTEntry); -+ SDValue ParmReg = DAG.getRegister(is64bit ? PPC::X3 : PPC::R3, -+ is64bit ? MVT::i64 : MVT::i32); - SDValue TLSAddr = DAG.getNode(PPCISD::GET_TLS_ADDR, dl, - PtrVT, ParmReg, TGA); - // The return value from GET_TLS_ADDR really is in X3 already, but - // some hacks are needed here to tie everything together. The extra - // copies dissolve during subsequent transforms. -- Chain = DAG.getCopyToReg(Chain, dl, PPC::X3, TLSAddr); -- return DAG.getCopyFromReg(Chain, dl, PPC::X3, PtrVT); -+ Chain = DAG.getCopyToReg(Chain, dl, is64bit ? PPC::X3 : PPC::R3, TLSAddr); -+ return DAG.getCopyFromReg(Chain, dl, is64bit ? PPC::X3 : PPC::R3, PtrVT); - } - - if (Model == TLSModel::LocalDynamic) { - SDValue TGA = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, 0); -- SDValue GOTReg = DAG.getRegister(PPC::X2, MVT::i64); -- SDValue GOTEntryHi = DAG.getNode(PPCISD::ADDIS_TLSLD_HA, dl, PtrVT, -- GOTReg, TGA); -+ SDValue GOTPtr; -+ if (is64bit) { -+ SDValue GOTReg = DAG.getRegister(PPC::X2, MVT::i64); -+ GOTPtr = DAG.getNode(PPCISD::ADDIS_TLSLD_HA, dl, PtrVT, -+ GOTReg, TGA); -+ } else { -+ GOTPtr = DAG.getNode(PPCISD::PPC32_PICGOT, dl, PtrVT); -+ } - SDValue GOTEntry = DAG.getNode(PPCISD::ADDI_TLSLD_L, dl, PtrVT, -- GOTEntryHi, TGA); -+ GOTPtr, TGA); - - // We need a chain node, and don't have one handy. The underlying - // call has no side effects, so using the function entry node - // suffices. - SDValue Chain = DAG.getEntryNode(); -- Chain = DAG.getCopyToReg(Chain, dl, PPC::X3, GOTEntry); -- SDValue ParmReg = DAG.getRegister(PPC::X3, MVT::i64); -+ Chain = DAG.getCopyToReg(Chain, dl, -+ is64bit ? PPC::X3 : PPC::R3, GOTEntry); -+ SDValue ParmReg = DAG.getRegister(is64bit ? PPC::X3 : PPC::R3, -+ is64bit ? MVT::i64 : MVT::i32); - SDValue TLSAddr = DAG.getNode(PPCISD::GET_TLSLD_ADDR, dl, - PtrVT, ParmReg, TGA); - // The return value from GET_TLSLD_ADDR really is in X3 already, but - // some hacks are needed here to tie everything together. The extra - // copies dissolve during subsequent transforms. -- Chain = DAG.getCopyToReg(Chain, dl, PPC::X3, TLSAddr); -+ Chain = DAG.getCopyToReg(Chain, dl, is64bit ? PPC::X3 : PPC::R3, TLSAddr); - SDValue DtvOffsetHi = DAG.getNode(PPCISD::ADDIS_DTPREL_HA, dl, PtrVT, - Chain, ParmReg, TGA); - return DAG.getNode(PPCISD::ADDI_DTPREL_L, dl, PtrVT, DtvOffsetHi, TGA); -Index: lib/Target/PowerPC/MCTargetDesc/PPCELFObjectWriter.cpp -=================================================================== ---- lib/Target/PowerPC/MCTargetDesc/PPCELFObjectWriter.cpp -+++ lib/Target/PowerPC/MCTargetDesc/PPCELFObjectWriter.cpp -@@ -236,7 +236,10 @@ unsigned PPCELFObjectWriter::getRelocTypeInner(con - Type = ELF::R_PPC64_DTPREL16_HIGHESTA; - break; - case MCSymbolRefExpr::VK_PPC_GOT_TLSGD: -- Type = ELF::R_PPC64_GOT_TLSGD16; -+ if (is64Bit()) -+ Type = ELF::R_PPC64_GOT_TLSGD16; -+ else -+ Type = ELF::R_PPC_GOT_TLSGD16; - break; - case MCSymbolRefExpr::VK_PPC_GOT_TLSGD_LO: - Type = ELF::R_PPC64_GOT_TLSGD16_LO; -@@ -248,7 +251,10 @@ unsigned PPCELFObjectWriter::getRelocTypeInner(con - Type = ELF::R_PPC64_GOT_TLSGD16_HA; - break; - case MCSymbolRefExpr::VK_PPC_GOT_TLSLD: -- Type = ELF::R_PPC64_GOT_TLSLD16; -+ if (is64Bit()) -+ Type = ELF::R_PPC64_GOT_TLSLD16; -+ else -+ Type = ELF::R_PPC_GOT_TLSLD16; - break; - case MCSymbolRefExpr::VK_PPC_GOT_TLSLD_LO: - Type = ELF::R_PPC64_GOT_TLSLD16_LO; -@@ -344,13 +350,22 @@ unsigned PPCELFObjectWriter::getRelocTypeInner(con - switch (Modifier) { - default: llvm_unreachable("Unsupported Modifier"); - case MCSymbolRefExpr::VK_PPC_TLSGD: -- Type = ELF::R_PPC64_TLSGD; -+ if (is64Bit()) -+ Type = ELF::R_PPC64_TLSGD; -+ else -+ Type = ELF::R_PPC_TLSGD; - break; - case MCSymbolRefExpr::VK_PPC_TLSLD: -- Type = ELF::R_PPC64_TLSLD; -+ if (is64Bit()) -+ Type = ELF::R_PPC64_TLSLD; -+ else -+ Type = ELF::R_PPC_TLSLD; - break; - case MCSymbolRefExpr::VK_PPC_TLS: -- Type = ELF::R_PPC64_TLS; -+ if (is64Bit()) -+ Type = ELF::R_PPC64_TLS; -+ else -+ Type = ELF::R_PPC_TLS; - break; - } - break; -Index: lib/Target/PowerPC/PPCAsmPrinter.cpp -=================================================================== ---- lib/Target/PowerPC/PPCAsmPrinter.cpp -+++ lib/Target/PowerPC/PPCAsmPrinter.cpp -@@ -573,6 +573,34 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - return; - } - -+ case PPC::PPC32PICGOT: { -+ MCSymbol *GOTSymbol = OutContext.GetOrCreateSymbol(StringRef("_GLOBAL_OFFSET_TABLE_")); -+ MCSymbol *GOTRef = OutContext.CreateTempSymbol(); -+ MCSymbol *NextInstr = OutContext.CreateTempSymbol(); -+ -+ EmitToStreamer(OutStreamer, MCInstBuilder(PPC::BL) -+ // FIXME: We would like an efficient form for this, so we don't have to do -+ // a lot of extra uniquing. -+ .addExpr(MCSymbolRefExpr::Create(NextInstr, OutContext))); -+ const MCExpr *OffsExpr = -+ MCBinaryExpr::CreateSub(MCSymbolRefExpr::Create(GOTSymbol, OutContext), -+ MCSymbolRefExpr::Create(GOTRef, OutContext), -+ OutContext); -+ OutStreamer.EmitLabel(GOTRef); -+ OutStreamer.EmitValue(OffsExpr, 4); -+ OutStreamer.EmitLabel(NextInstr); -+ EmitToStreamer(OutStreamer, MCInstBuilder(PPC::MFLR) -+ .addReg(MI->getOperand(0).getReg())); -+ EmitToStreamer(OutStreamer, MCInstBuilder(PPC::LWZ) -+ .addReg(MI->getOperand(1).getReg()) -+ .addImm(0) -+ .addReg(MI->getOperand(0).getReg())); -+ EmitToStreamer(OutStreamer, MCInstBuilder(PPC::ADD4) -+ .addReg(MI->getOperand(0).getReg()) -+ .addReg(MI->getOperand(1).getReg()) -+ .addReg(MI->getOperand(0).getReg())); -+ return; -+ } - case PPC::PPC32GOT: { - MCSymbol *GOTSymbol = OutContext.GetOrCreateSymbol(StringRef("_GLOBAL_OFFSET_TABLE_")); - const MCExpr *SymGotTlsL = -@@ -606,31 +634,43 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - .addExpr(SymGotTlsGD)); - return; - } -- case PPC::ADDItlsgdL: { -+ case PPC::ADDItlsgdL: - // Transform: %Xd = ADDItlsgdL %Xs, <ga:@sym> - // Into: %Xd = ADDI8 %Xs, sym@got@tlsgd@l -- assert(Subtarget.isPPC64() && "Not supported for 32-bit PowerPC"); -+ case PPC::ADDItlsgdL32: { -+ // Transform: %Rd = ADDItlsgdL32 %Rs, <ga:@sym> -+ // Into: %Rd = ADDI %Rs, sym@got@tlsgd - const MachineOperand &MO = MI->getOperand(2); - const GlobalValue *GValue = MO.getGlobal(); - MCSymbol *MOSymbol = getSymbol(GValue); - const MCExpr *SymGotTlsGD = -- MCSymbolRefExpr::Create(MOSymbol, MCSymbolRefExpr::VK_PPC_GOT_TLSGD_LO, -+ MCSymbolRefExpr::Create(MOSymbol, Subtarget.isPPC64() ? -+ MCSymbolRefExpr::VK_PPC_GOT_TLSGD_LO : -+ MCSymbolRefExpr::VK_PPC_GOT_TLSGD, - OutContext); -- EmitToStreamer(OutStreamer, MCInstBuilder(PPC::ADDI8) -- .addReg(MI->getOperand(0).getReg()) -- .addReg(MI->getOperand(1).getReg()) -- .addExpr(SymGotTlsGD)); -+ EmitToStreamer(OutStreamer, -+ MCInstBuilder(Subtarget.isPPC64() ? PPC::ADDI8 : PPC::ADDI) -+ .addReg(MI->getOperand(0).getReg()) -+ .addReg(MI->getOperand(1).getReg()) -+ .addExpr(SymGotTlsGD)); - return; - } -- case PPC::GETtlsADDR: { -+ case PPC::GETtlsADDR: - // Transform: %X3 = GETtlsADDR %X3, <ga:@sym> - // Into: BL8_NOP_TLS __tls_get_addr(sym@tlsgd) -- assert(Subtarget.isPPC64() && "Not supported for 32-bit PowerPC"); -+ case PPC::GETtlsADDR32: { -+ // Transform: %R3 = GETtlsADDR32 %R3, <ga:@sym> -+ // Into: BL_TLS __tls_get_addr(sym@tlsgd)@PLT - - StringRef Name = "__tls_get_addr"; - MCSymbol *TlsGetAddr = OutContext.GetOrCreateSymbol(Name); -+ MCSymbolRefExpr::VariantKind Kind = MCSymbolRefExpr::VK_None; -+ -+ if (!Subtarget.isPPC64() && !Subtarget.isDarwin() && -+ TM.getRelocationModel() == Reloc::PIC_) -+ Kind = MCSymbolRefExpr::VK_PLT; - const MCSymbolRefExpr *TlsRef = -- MCSymbolRefExpr::Create(TlsGetAddr, MCSymbolRefExpr::VK_None, OutContext); -+ MCSymbolRefExpr::Create(TlsGetAddr, Kind, OutContext); - const MachineOperand &MO = MI->getOperand(2); - const GlobalValue *GValue = MO.getGlobal(); - MCSymbol *MOSymbol = getSymbol(GValue); -@@ -637,9 +677,11 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - const MCExpr *SymVar = - MCSymbolRefExpr::Create(MOSymbol, MCSymbolRefExpr::VK_PPC_TLSGD, - OutContext); -- EmitToStreamer(OutStreamer, MCInstBuilder(PPC::BL8_NOP_TLS) -- .addExpr(TlsRef) -- .addExpr(SymVar)); -+ EmitToStreamer(OutStreamer, -+ MCInstBuilder(Subtarget.isPPC64() ? -+ PPC::BL8_NOP_TLS : PPC::BL_TLS) -+ .addExpr(TlsRef) -+ .addExpr(SymVar)); - return; - } - case PPC::ADDIStlsldHA: { -@@ -658,31 +700,44 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - .addExpr(SymGotTlsLD)); - return; - } -- case PPC::ADDItlsldL: { -+ case PPC::ADDItlsldL: - // Transform: %Xd = ADDItlsldL %Xs, <ga:@sym> - // Into: %Xd = ADDI8 %Xs, sym@got@tlsld@l -- assert(Subtarget.isPPC64() && "Not supported for 32-bit PowerPC"); -+ case PPC::ADDItlsldL32: { -+ // Transform: %Rd = ADDItlsldL32 %Rs, <ga:@sym> -+ // Into: %Rd = ADDI %Rs, sym@got@tlsld - const MachineOperand &MO = MI->getOperand(2); - const GlobalValue *GValue = MO.getGlobal(); - MCSymbol *MOSymbol = getSymbol(GValue); - const MCExpr *SymGotTlsLD = -- MCSymbolRefExpr::Create(MOSymbol, MCSymbolRefExpr::VK_PPC_GOT_TLSLD_LO, -+ MCSymbolRefExpr::Create(MOSymbol, Subtarget.isPPC64() ? -+ MCSymbolRefExpr::VK_PPC_GOT_TLSLD_LO : -+ MCSymbolRefExpr::VK_PPC_GOT_TLSLD, - OutContext); -- EmitToStreamer(OutStreamer, MCInstBuilder(PPC::ADDI8) -- .addReg(MI->getOperand(0).getReg()) -- .addReg(MI->getOperand(1).getReg()) -- .addExpr(SymGotTlsLD)); -+ EmitToStreamer(OutStreamer, -+ MCInstBuilder(Subtarget.isPPC64() ? PPC::ADDI8 : PPC::ADDI) -+ .addReg(MI->getOperand(0).getReg()) -+ .addReg(MI->getOperand(1).getReg()) -+ .addExpr(SymGotTlsLD)); - return; - } -- case PPC::GETtlsldADDR: { -+ case PPC::GETtlsldADDR: - // Transform: %X3 = GETtlsldADDR %X3, <ga:@sym> - // Into: BL8_NOP_TLS __tls_get_addr(sym@tlsld) -- assert(Subtarget.isPPC64() && "Not supported for 32-bit PowerPC"); -+ case PPC::GETtlsldADDR32: { -+ // Transform: %R3 = GETtlsldADDR32 %R3, <ga:@sym> -+ // Into: BL_TLS __tls_get_addr(sym@tlsld)@PLT - - StringRef Name = "__tls_get_addr"; - MCSymbol *TlsGetAddr = OutContext.GetOrCreateSymbol(Name); -+ MCSymbolRefExpr::VariantKind Kind = MCSymbolRefExpr::VK_None; -+ -+ if (!Subtarget.isPPC64() && !Subtarget.isDarwin() && -+ TM.getRelocationModel() == Reloc::PIC_) -+ Kind = MCSymbolRefExpr::VK_PLT; -+ - const MCSymbolRefExpr *TlsRef = -- MCSymbolRefExpr::Create(TlsGetAddr, MCSymbolRefExpr::VK_None, OutContext); -+ MCSymbolRefExpr::Create(TlsGetAddr, Kind, OutContext); - const MachineOperand &MO = MI->getOperand(2); - const GlobalValue *GValue = MO.getGlobal(); - MCSymbol *MOSymbol = getSymbol(GValue); -@@ -689,15 +744,19 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - const MCExpr *SymVar = - MCSymbolRefExpr::Create(MOSymbol, MCSymbolRefExpr::VK_PPC_TLSLD, - OutContext); -- EmitToStreamer(OutStreamer, MCInstBuilder(PPC::BL8_NOP_TLS) -- .addExpr(TlsRef) -- .addExpr(SymVar)); -+ EmitToStreamer(OutStreamer, -+ MCInstBuilder(Subtarget.isPPC64() ? -+ PPC::BL8_NOP_TLS : PPC::BL_TLS) -+ .addExpr(TlsRef) -+ .addExpr(SymVar)); - return; - } -- case PPC::ADDISdtprelHA: { -+ case PPC::ADDISdtprelHA: - // Transform: %Xd = ADDISdtprelHA %X3, <ga:@sym> - // Into: %Xd = ADDIS8 %X3, sym@dtprel@ha -- assert(Subtarget.isPPC64() && "Not supported for 32-bit PowerPC"); -+ case PPC::ADDISdtprelHA32: { -+ // Transform: %Rd = ADDISdtprelHA32 %R3, <ga:@sym> -+ // Into: %Rd = ADDIS %R3, sym@dtprel@ha - const MachineOperand &MO = MI->getOperand(2); - const GlobalValue *GValue = MO.getGlobal(); - MCSymbol *MOSymbol = getSymbol(GValue); -@@ -704,16 +763,19 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - const MCExpr *SymDtprel = - MCSymbolRefExpr::Create(MOSymbol, MCSymbolRefExpr::VK_PPC_DTPREL_HA, - OutContext); -- EmitToStreamer(OutStreamer, MCInstBuilder(PPC::ADDIS8) -- .addReg(MI->getOperand(0).getReg()) -- .addReg(PPC::X3) -- .addExpr(SymDtprel)); -+ EmitToStreamer(OutStreamer, -+ MCInstBuilder(Subtarget.isPPC64() ? PPC::ADDIS8 : PPC::ADDIS) -+ .addReg(MI->getOperand(0).getReg()) -+ .addReg(Subtarget.isPPC64() ? PPC::X3 : PPC::R3) -+ .addExpr(SymDtprel)); - return; - } -- case PPC::ADDIdtprelL: { -+ case PPC::ADDIdtprelL: - // Transform: %Xd = ADDIdtprelL %Xs, <ga:@sym> - // Into: %Xd = ADDI8 %Xs, sym@dtprel@l -- assert(Subtarget.isPPC64() && "Not supported for 32-bit PowerPC"); -+ case PPC::ADDIdtprelL32: { -+ // Transform: %Rd = ADDIdtprelL32 %Rs, <ga:@sym> -+ // Into: %Rd = ADDI %Rs, sym@dtprel@l - const MachineOperand &MO = MI->getOperand(2); - const GlobalValue *GValue = MO.getGlobal(); - MCSymbol *MOSymbol = getSymbol(GValue); -@@ -720,10 +782,11 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - const MCExpr *SymDtprel = - MCSymbolRefExpr::Create(MOSymbol, MCSymbolRefExpr::VK_PPC_DTPREL_LO, - OutContext); -- EmitToStreamer(OutStreamer, MCInstBuilder(PPC::ADDI8) -- .addReg(MI->getOperand(0).getReg()) -- .addReg(MI->getOperand(1).getReg()) -- .addExpr(SymDtprel)); -+ EmitToStreamer(OutStreamer, -+ MCInstBuilder(Subtarget.isPPC64() ? PPC::ADDI8 : PPC::ADDI) -+ .addReg(MI->getOperand(0).getReg()) -+ .addReg(MI->getOperand(1).getReg()) -+ .addExpr(SymDtprel)); - return; - } - case PPC::MFOCRF: -Index: lib/Target/PowerPC/PPCISelDAGToDAG.cpp -=================================================================== ---- lib/Target/PowerPC/PPCISelDAGToDAG.cpp -+++ lib/Target/PowerPC/PPCISelDAGToDAG.cpp -@@ -1473,6 +1473,12 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) { - return CurDAG->getMachineNode(PPC::ADDItocL, dl, MVT::i64, - SDValue(Tmp, 0), GA); - } -+ case PPCISD::PPC32_PICGOT: { -+ // Generate a PIC-safe GOT reference. -+ assert(!PPCSubTarget->isPPC64() && PPCSubTarget->isSVR4ABI() && -+ "PPCISD::PPC32_PICGOT is only supported for 32-bit SVR4"); -+ return CurDAG->SelectNodeTo(N, PPC::PPC32PICGOT, PPCLowering->getPointerTy(), MVT::i32); -+ } - case PPCISD::VADD_SPLAT: { - // This expands into one of three sequences, depending on whether - // the first operand is odd or even, positive or negative. -Index: test/CodeGen/PowerPC/tls-pic.ll -=================================================================== ---- test/CodeGen/PowerPC/tls-pic.ll -+++ test/CodeGen/PowerPC/tls-pic.ll -@@ -1,5 +1,7 @@ - ; RUN: llc -march=ppc64 -mcpu=pwr7 -O0 -relocation-model=pic < %s | FileCheck -check-prefix=OPT0 %s - ; RUN: llc -march=ppc64 -mcpu=pwr7 -O1 -relocation-model=pic < %s | FileCheck -check-prefix=OPT1 %s -+; RUN: llc -march=ppc32 -O0 -relocation-model=pic < %s | FileCheck -check-prefix=OPT0-32 %s -+; RUN: llc -march=ppc32 -O1 -relocation-model=pic < %s | FileCheck -check-prefix=OPT1-32 %s - - target triple = "powerpc64-unknown-linux-gnu" - ; Test correct assembly code generation for thread-local storage using -@@ -22,6 +24,16 @@ entry: - ; OPT0-NEXT: nop - ; OPT0: addis [[REG2:[0-9]+]], 3, a@dtprel@ha - ; OPT0-NEXT: addi {{[0-9]+}}, [[REG2]], a@dtprel@l -+; OPT0-32-LABEL: main -+; OPT0-32: addi {{[0-9]+}}, {{[0-9]+}}, a@got@tlsld -+; OPT0-32: bl __tls_get_addr(a@tlsld)@PLT -+; OPT0-32: addis [[REG:[0-9]+]], 3, a@dtprel@ha -+; OPT0-32-NEXT: addi {{[0-9]+}}, [[REG]], a@dtprel@l -+; OPT1-32-LABEL: main -+; OPT1-32: addi 3, {{[0-9]+}}, a@got@tlsld -+; OPT1-32: bl __tls_get_addr(a@tlsld)@PLT -+; OPT1-32: addis [[REG:[0-9]+]], 3, a@dtprel@ha -+; OPT1-32-NEXT: addi {{[0-9]+}}, [[REG]], a@dtprel@l - - ; Test peephole optimization for thread-local storage using the - ; local dynamic model. -@@ -52,4 +64,6 @@ entry: - ; OPT1-NEXT: addi 3, [[REG]], a2@got@tlsgd@l - ; OPT1: bl __tls_get_addr(a2@tlsgd) - ; OPT1-NEXT: nop -- -+; OPT1-32-LABEL: main2 -+; OPT1-32: addi 3, {{[0-9]+}}, a2@got@tlsgd -+; OPT1-32: bl __tls_get_addr(a2@tlsgd)@PLT diff --git a/contrib/llvm/patches/patch-08-llvm-r216989-r216990-fix-movw-armv6.diff b/contrib/llvm/patches/patch-07-llvm-r216989-r216990-fix-movw-armv6.diff index 9861d9f..9861d9f 100644 --- a/contrib/llvm/patches/patch-08-llvm-r216989-r216990-fix-movw-armv6.diff +++ b/contrib/llvm/patches/patch-07-llvm-r216989-r216990-fix-movw-armv6.diff diff --git a/contrib/llvm/patches/patch-09-clang-r217410-i386-garbage-float.diff b/contrib/llvm/patches/patch-08-clang-r217410-i386-garbage-float.diff index 26cd2bf..26cd2bf 100644 --- a/contrib/llvm/patches/patch-09-clang-r217410-i386-garbage-float.diff +++ b/contrib/llvm/patches/patch-08-clang-r217410-i386-garbage-float.diff diff --git a/contrib/llvm/patches/patch-10-llvm-r221709-debug-oom.diff b/contrib/llvm/patches/patch-09-llvm-r221709-debug-oom.diff index a00b9f1..a00b9f1 100644 --- a/contrib/llvm/patches/patch-10-llvm-r221709-debug-oom.diff +++ b/contrib/llvm/patches/patch-09-llvm-r221709-debug-oom.diff diff --git a/contrib/llvm/patches/patch-11-llvm-r222562-loop-rotate.diff b/contrib/llvm/patches/patch-10-llvm-r222562-loop-rotate.diff index bc09f27..bc09f27 100644 --- a/contrib/llvm/patches/patch-11-llvm-r222562-loop-rotate.diff +++ b/contrib/llvm/patches/patch-10-llvm-r222562-loop-rotate.diff diff --git a/contrib/llvm/patches/patch-12-add-llvm-gvn-option.diff b/contrib/llvm/patches/patch-11-add-llvm-gvn-option.diff index fcd051f..fcd051f 100644 --- a/contrib/llvm/patches/patch-12-add-llvm-gvn-option.diff +++ b/contrib/llvm/patches/patch-11-add-llvm-gvn-option.diff diff --git a/contrib/llvm/patches/patch-13-llvm-r218241-dwarf2-warning.diff b/contrib/llvm/patches/patch-12-llvm-r218241-dwarf2-warning.diff index 6127c78..6127c78 100644 --- a/contrib/llvm/patches/patch-13-llvm-r218241-dwarf2-warning.diff +++ b/contrib/llvm/patches/patch-12-llvm-r218241-dwarf2-warning.diff diff --git a/contrib/llvm/patches/patch-14-llvm-r215352-aarch64-dyn-loader.diff b/contrib/llvm/patches/patch-13-llvm-r215352-aarch64-dyn-loader.diff index d4c3880..d4c3880 100644 --- a/contrib/llvm/patches/patch-14-llvm-r215352-aarch64-dyn-loader.diff +++ b/contrib/llvm/patches/patch-13-llvm-r215352-aarch64-dyn-loader.diff diff --git a/contrib/llvm/patches/patch-15-llvm-r216571-dynamiclib-usability.diff b/contrib/llvm/patches/patch-14-llvm-r216571-dynamiclib-usability.diff index ff0a6d1..ff0a6d1 100644 --- a/contrib/llvm/patches/patch-15-llvm-r216571-dynamiclib-usability.diff +++ b/contrib/llvm/patches/patch-14-llvm-r216571-dynamiclib-usability.diff diff --git a/contrib/llvm/patches/patch-16-clang-r221900-freebsd-aarch64.diff b/contrib/llvm/patches/patch-15-clang-r221900-freebsd-aarch64.diff index 1c9eab1..1c9eab1 100644 --- a/contrib/llvm/patches/patch-16-clang-r221900-freebsd-aarch64.diff +++ b/contrib/llvm/patches/patch-15-clang-r221900-freebsd-aarch64.diff diff --git a/contrib/llvm/patches/patch-17-llvm-r222856-libapr-miscompile.diff b/contrib/llvm/patches/patch-16-llvm-r222856-libapr-miscompile.diff index c9e64d9..c9e64d9 100644 --- a/contrib/llvm/patches/patch-17-llvm-r222856-libapr-miscompile.diff +++ b/contrib/llvm/patches/patch-16-llvm-r222856-libapr-miscompile.diff diff --git a/contrib/llvm/patches/patch-18-llvm-r214802-armv6-cp10-cp11.diff b/contrib/llvm/patches/patch-17-llvm-r214802-armv6-cp10-cp11.diff index 17e1c75..17e1c75 100644 --- a/contrib/llvm/patches/patch-18-llvm-r214802-armv6-cp10-cp11.diff +++ b/contrib/llvm/patches/patch-17-llvm-r214802-armv6-cp10-cp11.diff diff --git a/contrib/llvm/patches/patch-19-llvm-r215811-arm-fpu-directive.diff b/contrib/llvm/patches/patch-18-llvm-r215811-arm-fpu-directive.diff index c11365c..c11365c 100644 --- a/contrib/llvm/patches/patch-19-llvm-r215811-arm-fpu-directive.diff +++ b/contrib/llvm/patches/patch-18-llvm-r215811-arm-fpu-directive.diff diff --git a/contrib/llvm/patches/patch-20-enable-armv6-clrex.diff b/contrib/llvm/patches/patch-19-enable-armv6-clrex.diff index fd77f45..fd77f45 100644 --- a/contrib/llvm/patches/patch-20-enable-armv6-clrex.diff +++ b/contrib/llvm/patches/patch-19-enable-armv6-clrex.diff diff --git a/contrib/llvm/patches/patch-22-llvm-r223147-arm-cpu-directive.diff b/contrib/llvm/patches/patch-20-llvm-r223147-arm-cpu-directive.diff index e97dc2e..e97dc2e 100644 --- a/contrib/llvm/patches/patch-22-llvm-r223147-arm-cpu-directive.diff +++ b/contrib/llvm/patches/patch-20-llvm-r223147-arm-cpu-directive.diff diff --git a/contrib/llvm/patches/patch-23-llvm-r221170-ppc-vaarg.diff b/contrib/llvm/patches/patch-21-llvm-r221170-ppc-vaarg.diff index 7820d6b..7820d6b 100644 --- a/contrib/llvm/patches/patch-23-llvm-r221170-ppc-vaarg.diff +++ b/contrib/llvm/patches/patch-21-llvm-r221170-ppc-vaarg.diff diff --git a/contrib/llvm/patches/patch-21-llvm-r223171-fix-vectorizer.diff b/contrib/llvm/patches/patch-21-llvm-r223171-fix-vectorizer.diff deleted file mode 100644 index 306f950..0000000 --- a/contrib/llvm/patches/patch-21-llvm-r223171-fix-vectorizer.diff +++ /dev/null @@ -1,617 +0,0 @@ -Pull in r223170 from upstream llvm trunk (by Michael Zolotukhin): - - Apply loop-rotate to several vectorizer tests. - - Such loops shouldn't be vectorized due to the loops form. - After applying loop-rotate (+simplifycfg) the tests again start to check - what they are intended to check. - -Pull in r223171 from upstream llvm trunk (by Michael Zolotukhin): - - PR21302. Vectorize only bottom-tested loops. - - rdar://problem/18886083 - -This fixes a bug in the llvm vectorizer, which could sometimes cause -vectorized loops to perform an additional iteration, leading to possible -buffer overruns. Symptoms of this, which are usually segfaults, were -first noticed when building gcc ports, here: - -https://lists.freebsd.org/pipermail/freebsd-ports/2014-September/095466.html -https://lists.freebsd.org/pipermail/freebsd-toolchain/2014-September/001211.html - -Note: because this is applied on top of llvm/clang 3.5.0, this fix is -slightly different from the one just checked into head in r275633. - -Introduced here: http://svnweb.freebsd.org/changeset/base/275635 - -Index: lib/Transforms/Vectorize/LoopVectorize.cpp -=================================================================== ---- lib/Transforms/Vectorize/LoopVectorize.cpp -+++ lib/Transforms/Vectorize/LoopVectorize.cpp -@@ -3466,6 +3466,15 @@ bool LoopVectorizationLegality::canVectorize() { - return false; - } - -+ // We only handle bottom-tested loops, i.e. loop in which the condition is -+ // checked at the end of each iteration. With that we can assume that all -+ // instructions in the loop are executed the same number of times. -+ if (TheLoop->getExitingBlock() != TheLoop->getLoopLatch()) { -+ emitAnalysis( -+ Report() << "loop control flow is not understood by vectorizer"); -+ return false; -+ } -+ - // We need to have a loop header. - DEBUG(dbgs() << "LV: Found a loop: " << - TheLoop->getHeader()->getName() << '\n'); -Index: test/Transforms/LoopVectorize/vect.stats.ll -=================================================================== ---- test/Transforms/LoopVectorize/vect.stats.ll -+++ test/Transforms/LoopVectorize/vect.stats.ll -@@ -13,53 +13,47 @@ target triple = "x86_64-unknown-linux-gnu" - - define void @vectorized(float* nocapture %a, i64 %size) { - entry: -- %cmp1 = icmp sgt i64 %size, 0 -- br i1 %cmp1, label %for.header, label %for.end -+ %cmp1 = icmp sle i64 %size, 0 -+ %cmp21 = icmp sgt i64 0, %size -+ %or.cond = or i1 %cmp1, %cmp21 -+ br i1 %or.cond, label %for.end, label %for.body - --for.header: -- %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ] -- %cmp2 = icmp sgt i64 %indvars.iv, %size -- br i1 %cmp2, label %for.end, label %for.body -- --for.body: -- -- %arrayidx = getelementptr inbounds float* %a, i64 %indvars.iv -+for.body: ; preds = %entry, %for.body -+ %indvars.iv2 = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ] -+ %arrayidx = getelementptr inbounds float* %a, i64 %indvars.iv2 - %0 = load float* %arrayidx, align 4 - %mul = fmul float %0, %0 - store float %mul, float* %arrayidx, align 4 -+ %indvars.iv.next = add nuw nsw i64 %indvars.iv2, 1 -+ %cmp2 = icmp sgt i64 %indvars.iv.next, %size -+ br i1 %cmp2, label %for.end, label %for.body - -- %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1 -- br label %for.header -- --for.end: -+for.end: ; preds = %entry, %for.body - ret void - } - - define void @not_vectorized(float* nocapture %a, i64 %size) { - entry: -- %cmp1 = icmp sgt i64 %size, 0 -- br i1 %cmp1, label %for.header, label %for.end -+ %cmp1 = icmp sle i64 %size, 0 -+ %cmp21 = icmp sgt i64 0, %size -+ %or.cond = or i1 %cmp1, %cmp21 -+ br i1 %or.cond, label %for.end, label %for.body - --for.header: -- %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ] -- %cmp2 = icmp sgt i64 %indvars.iv, %size -- br i1 %cmp2, label %for.end, label %for.body -- --for.body: -- -- %0 = add nsw i64 %indvars.iv, -5 -+for.body: ; preds = %entry, %for.body -+ %indvars.iv2 = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ] -+ %0 = add nsw i64 %indvars.iv2, -5 - %arrayidx = getelementptr inbounds float* %a, i64 %0 - %1 = load float* %arrayidx, align 4 -- %2 = add nsw i64 %indvars.iv, 2 -+ %2 = add nsw i64 %indvars.iv2, 2 - %arrayidx2 = getelementptr inbounds float* %a, i64 %2 - %3 = load float* %arrayidx2, align 4 - %mul = fmul float %1, %3 -- %arrayidx4 = getelementptr inbounds float* %a, i64 %indvars.iv -+ %arrayidx4 = getelementptr inbounds float* %a, i64 %indvars.iv2 - store float %mul, float* %arrayidx4, align 4 -+ %indvars.iv.next = add nuw nsw i64 %indvars.iv2, 1 -+ %cmp2 = icmp sgt i64 %indvars.iv.next, %size -+ br i1 %cmp2, label %for.end, label %for.body - -- %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1 -- br label %for.header -- --for.end: -+for.end: ; preds = %entry, %for.body - ret void --} -\ No newline at end of file -+} -Index: test/Transforms/LoopVectorize/loop-form.ll -=================================================================== ---- test/Transforms/LoopVectorize/loop-form.ll -+++ test/Transforms/LoopVectorize/loop-form.ll -@@ -0,0 +1,31 @@ -+; RUN: opt -S -loop-vectorize < %s | FileCheck %s -+target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128" -+ -+; Check that we vectorize only bottom-tested loops. -+; This is a reduced testcase from PR21302. -+; -+; rdar://problem/18886083 -+ -+%struct.X = type { i32, i16 } -+; CHECK-LABEL: @foo( -+; CHECK-NOT: vector.body -+ -+define void @foo(i32 %n) { -+entry: -+ br label %for.cond -+ -+for.cond: -+ %i = phi i32 [ 0, %entry ], [ %inc, %for.body ] -+ %cmp = icmp slt i32 %i, %n -+ br i1 %cmp, label %for.body, label %if.end -+ -+for.body: -+ %iprom = sext i32 %i to i64 -+ %b = getelementptr inbounds %struct.X* undef, i64 %iprom, i32 1 -+ store i16 0, i16* %b, align 4 -+ %inc = add nsw i32 %i, 1 -+ br label %for.cond -+ -+if.end: -+ ret void -+} -Index: test/Transforms/LoopVectorize/runtime-check-readonly-address-space.ll -=================================================================== ---- test/Transforms/LoopVectorize/runtime-check-readonly-address-space.ll -+++ test/Transforms/LoopVectorize/runtime-check-readonly-address-space.ll -@@ -8,26 +8,24 @@ define void @add_ints_1_1_1(i32 addrspace(1)* %a, - ; CHECK-LABEL: @add_ints_1_1_1( - ; CHECK: <4 x i32> - ; CHECK: ret -+ - entry: -- br label %for.cond -+ br label %for.body - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp ult i64 %i.0, 200 -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.0 -+for.body: ; preds = %entry, %for.body -+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -+ %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.01 - %0 = load i32 addrspace(1)* %arrayidx, align 4 -- %arrayidx1 = getelementptr inbounds i32 addrspace(1)* %c, i64 %i.0 -+ %arrayidx1 = getelementptr inbounds i32 addrspace(1)* %c, i64 %i.01 - %1 = load i32 addrspace(1)* %arrayidx1, align 4 - %add = add nsw i32 %0, %1 -- %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %i.0 -+ %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %i.01 - store i32 %add, i32 addrspace(1)* %arrayidx2, align 4 -- %inc = add i64 %i.0, 1 -- br label %for.cond -+ %inc = add i64 %i.01, 1 -+ %cmp = icmp ult i64 %inc, 200 -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body - ret void - } - -@@ -35,26 +33,24 @@ define void @add_ints_as_1_0_0(i32 addrspace(1)* % - ; CHECK-LABEL: @add_ints_as_1_0_0( - ; CHECK-NOT: <4 x i32> - ; CHECK: ret -+ - entry: -- br label %for.cond -+ br label %for.body - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp ult i64 %i.0, 200 -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %arrayidx = getelementptr inbounds i32* %b, i64 %i.0 -+for.body: ; preds = %entry, %for.body -+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -+ %arrayidx = getelementptr inbounds i32* %b, i64 %i.01 - %0 = load i32* %arrayidx, align 4 -- %arrayidx1 = getelementptr inbounds i32* %c, i64 %i.0 -+ %arrayidx1 = getelementptr inbounds i32* %c, i64 %i.01 - %1 = load i32* %arrayidx1, align 4 - %add = add nsw i32 %0, %1 -- %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %i.0 -+ %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %i.01 - store i32 %add, i32 addrspace(1)* %arrayidx2, align 4 -- %inc = add i64 %i.0, 1 -- br label %for.cond -+ %inc = add i64 %i.01, 1 -+ %cmp = icmp ult i64 %inc, 200 -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body - ret void - } - -@@ -62,26 +58,24 @@ define void @add_ints_as_0_1_0(i32* %a, i32 addrsp - ; CHECK-LABEL: @add_ints_as_0_1_0( - ; CHECK-NOT: <4 x i32> - ; CHECK: ret -+ - entry: -- br label %for.cond -+ br label %for.body - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp ult i64 %i.0, 200 -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.0 -+for.body: ; preds = %entry, %for.body -+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -+ %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.01 - %0 = load i32 addrspace(1)* %arrayidx, align 4 -- %arrayidx1 = getelementptr inbounds i32* %c, i64 %i.0 -+ %arrayidx1 = getelementptr inbounds i32* %c, i64 %i.01 - %1 = load i32* %arrayidx1, align 4 - %add = add nsw i32 %0, %1 -- %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.0 -+ %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.01 - store i32 %add, i32* %arrayidx2, align 4 -- %inc = add i64 %i.0, 1 -- br label %for.cond -+ %inc = add i64 %i.01, 1 -+ %cmp = icmp ult i64 %inc, 200 -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body - ret void - } - -@@ -89,26 +83,24 @@ define void @add_ints_as_0_1_1(i32* %a, i32 addrsp - ; CHECK-LABEL: @add_ints_as_0_1_1( - ; CHECK-NOT: <4 x i32> - ; CHECK: ret -+ - entry: -- br label %for.cond -+ br label %for.body - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp ult i64 %i.0, 200 -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.0 -+for.body: ; preds = %entry, %for.body -+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -+ %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.01 - %0 = load i32 addrspace(1)* %arrayidx, align 4 -- %arrayidx1 = getelementptr inbounds i32 addrspace(1)* %c, i64 %i.0 -+ %arrayidx1 = getelementptr inbounds i32 addrspace(1)* %c, i64 %i.01 - %1 = load i32 addrspace(1)* %arrayidx1, align 4 - %add = add nsw i32 %0, %1 -- %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.0 -+ %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.01 - store i32 %add, i32* %arrayidx2, align 4 -- %inc = add i64 %i.0, 1 -- br label %for.cond -+ %inc = add i64 %i.01, 1 -+ %cmp = icmp ult i64 %inc, 200 -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body - ret void - } - -@@ -116,26 +108,24 @@ define void @add_ints_as_0_1_2(i32* %a, i32 addrsp - ; CHECK-LABEL: @add_ints_as_0_1_2( - ; CHECK-NOT: <4 x i32> - ; CHECK: ret -+ - entry: -- br label %for.cond -+ br label %for.body - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp ult i64 %i.0, 200 -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.0 -+for.body: ; preds = %entry, %for.body -+ %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] -+ %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %i.01 - %0 = load i32 addrspace(1)* %arrayidx, align 4 -- %arrayidx1 = getelementptr inbounds i32 addrspace(2)* %c, i64 %i.0 -+ %arrayidx1 = getelementptr inbounds i32 addrspace(2)* %c, i64 %i.01 - %1 = load i32 addrspace(2)* %arrayidx1, align 4 - %add = add nsw i32 %0, %1 -- %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.0 -+ %arrayidx2 = getelementptr inbounds i32* %a, i64 %i.01 - store i32 %add, i32* %arrayidx2, align 4 -- %inc = add i64 %i.0, 1 -- br label %for.cond -+ %inc = add i64 %i.01, 1 -+ %cmp = icmp ult i64 %inc, 200 -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body - ret void - } - -Index: test/Transforms/LoopVectorize/runtime-check-address-space.ll -=================================================================== ---- test/Transforms/LoopVectorize/runtime-check-address-space.ll -+++ test/Transforms/LoopVectorize/runtime-check-address-space.ll -@@ -31,25 +31,23 @@ define void @foo(i32 addrspace(1)* %a, i32 addrspa - ; CHECK: ret - - entry: -- br label %for.cond -+ %cmp1 = icmp slt i32 0, %n -+ br i1 %cmp1, label %for.body, label %for.end - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp slt i32 %i.0, %n -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %idxprom = sext i32 %i.0 to i64 -+for.body: ; preds = %entry, %for.body -+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ] -+ %idxprom = sext i32 %i.02 to i64 - %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %idxprom - %0 = load i32 addrspace(1)* %arrayidx, align 4 - %mul = mul nsw i32 %0, 3 -- %idxprom1 = sext i32 %i.0 to i64 -+ %idxprom1 = sext i32 %i.02 to i64 - %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %idxprom1 - store i32 %mul, i32 addrspace(1)* %arrayidx2, align 4 -- %inc = add nsw i32 %i.0, 1 -- br label %for.cond -+ %inc = add nsw i32 %i.02, 1 -+ %cmp = icmp slt i32 %inc, %n -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body, %entry - ret void - } - -@@ -60,25 +58,23 @@ define void @bar0(i32* %a, i32 addrspace(1)* %b, i - ; CHECK: ret - - entry: -- br label %for.cond -+ %cmp1 = icmp slt i32 0, %n -+ br i1 %cmp1, label %for.body, label %for.end - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp slt i32 %i.0, %n -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %idxprom = sext i32 %i.0 to i64 -+for.body: ; preds = %entry, %for.body -+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ] -+ %idxprom = sext i32 %i.02 to i64 - %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %idxprom - %0 = load i32 addrspace(1)* %arrayidx, align 4 - %mul = mul nsw i32 %0, 3 -- %idxprom1 = sext i32 %i.0 to i64 -+ %idxprom1 = sext i32 %i.02 to i64 - %arrayidx2 = getelementptr inbounds i32* %a, i64 %idxprom1 - store i32 %mul, i32* %arrayidx2, align 4 -- %inc = add nsw i32 %i.0, 1 -- br label %for.cond -+ %inc = add nsw i32 %i.02, 1 -+ %cmp = icmp slt i32 %inc, %n -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body, %entry - ret void - } - -@@ -89,25 +85,23 @@ define void @bar1(i32 addrspace(1)* %a, i32* %b, i - ; CHECK: ret - - entry: -- br label %for.cond -+ %cmp1 = icmp slt i32 0, %n -+ br i1 %cmp1, label %for.body, label %for.end - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp slt i32 %i.0, %n -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %idxprom = sext i32 %i.0 to i64 -+for.body: ; preds = %entry, %for.body -+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ] -+ %idxprom = sext i32 %i.02 to i64 - %arrayidx = getelementptr inbounds i32* %b, i64 %idxprom - %0 = load i32* %arrayidx, align 4 - %mul = mul nsw i32 %0, 3 -- %idxprom1 = sext i32 %i.0 to i64 -+ %idxprom1 = sext i32 %i.02 to i64 - %arrayidx2 = getelementptr inbounds i32 addrspace(1)* %a, i64 %idxprom1 - store i32 %mul, i32 addrspace(1)* %arrayidx2, align 4 -- %inc = add nsw i32 %i.0, 1 -- br label %for.cond -+ %inc = add nsw i32 %i.02, 1 -+ %cmp = icmp slt i32 %inc, %n -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body, %entry - ret void - } - -@@ -119,25 +113,23 @@ define void @bar2(i32* noalias %a, i32 addrspace(1 - ; CHECK: ret - - entry: -- br label %for.cond -+ %cmp1 = icmp slt i32 0, %n -+ br i1 %cmp1, label %for.body, label %for.end - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp slt i32 %i.0, %n -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %idxprom = sext i32 %i.0 to i64 -+for.body: ; preds = %entry, %for.body -+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ] -+ %idxprom = sext i32 %i.02 to i64 - %arrayidx = getelementptr inbounds i32 addrspace(1)* %b, i64 %idxprom - %0 = load i32 addrspace(1)* %arrayidx, align 4 - %mul = mul nsw i32 %0, 3 -- %idxprom1 = sext i32 %i.0 to i64 -+ %idxprom1 = sext i32 %i.02 to i64 - %arrayidx2 = getelementptr inbounds i32* %a, i64 %idxprom1 - store i32 %mul, i32* %arrayidx2, align 4 -- %inc = add nsw i32 %i.0, 1 -- br label %for.cond -+ %inc = add nsw i32 %i.02, 1 -+ %cmp = icmp slt i32 %inc, %n -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body, %entry - ret void - } - -@@ -149,25 +141,23 @@ define void @arst0(i32* %b, i32 %n) #0 { - ; CHECK: ret - - entry: -- br label %for.cond -+ %cmp1 = icmp slt i32 0, %n -+ br i1 %cmp1, label %for.body, label %for.end - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp slt i32 %i.0, %n -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %idxprom = sext i32 %i.0 to i64 -+for.body: ; preds = %entry, %for.body -+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ] -+ %idxprom = sext i32 %i.02 to i64 - %arrayidx = getelementptr inbounds i32* %b, i64 %idxprom - %0 = load i32* %arrayidx, align 4 - %mul = mul nsw i32 %0, 3 -- %idxprom1 = sext i32 %i.0 to i64 -+ %idxprom1 = sext i32 %i.02 to i64 - %arrayidx2 = getelementptr inbounds [1024 x i32] addrspace(1)* @g_as1, i64 0, i64 %idxprom1 - store i32 %mul, i32 addrspace(1)* %arrayidx2, align 4 -- %inc = add nsw i32 %i.0, 1 -- br label %for.cond -+ %inc = add nsw i32 %i.02, 1 -+ %cmp = icmp slt i32 %inc, %n -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body, %entry - ret void - } - -@@ -180,25 +170,23 @@ define void @arst1(i32* %b, i32 %n) #0 { - ; CHECK: ret - - entry: -- br label %for.cond -+ %cmp1 = icmp slt i32 0, %n -+ br i1 %cmp1, label %for.body, label %for.end - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp slt i32 %i.0, %n -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %idxprom = sext i32 %i.0 to i64 -+for.body: ; preds = %entry, %for.body -+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ] -+ %idxprom = sext i32 %i.02 to i64 - %arrayidx = getelementptr inbounds [1024 x i32] addrspace(1)* @g_as1, i64 0, i64 %idxprom - %0 = load i32 addrspace(1)* %arrayidx, align 4 - %mul = mul nsw i32 %0, 3 -- %idxprom1 = sext i32 %i.0 to i64 -+ %idxprom1 = sext i32 %i.02 to i64 - %arrayidx2 = getelementptr inbounds i32* %b, i64 %idxprom1 - store i32 %mul, i32* %arrayidx2, align 4 -- %inc = add nsw i32 %i.0, 1 -- br label %for.cond -+ %inc = add nsw i32 %i.02, 1 -+ %cmp = icmp slt i32 %inc, %n -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body, %entry - ret void - } - -@@ -210,25 +198,23 @@ define void @aoeu(i32 %n) #0 { - ; CHECK: ret - - entry: -- br label %for.cond -+ %cmp1 = icmp slt i32 0, %n -+ br i1 %cmp1, label %for.body, label %for.end - --for.cond: ; preds = %for.body, %entry -- %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ] -- %cmp = icmp slt i32 %i.0, %n -- br i1 %cmp, label %for.body, label %for.end -- --for.body: ; preds = %for.cond -- %idxprom = sext i32 %i.0 to i64 -+for.body: ; preds = %entry, %for.body -+ %i.02 = phi i32 [ %inc, %for.body ], [ 0, %entry ] -+ %idxprom = sext i32 %i.02 to i64 - %arrayidx = getelementptr inbounds [1024 x i32] addrspace(2)* @q_as2, i64 0, i64 %idxprom - %0 = load i32 addrspace(2)* %arrayidx, align 4 - %mul = mul nsw i32 %0, 3 -- %idxprom1 = sext i32 %i.0 to i64 -+ %idxprom1 = sext i32 %i.02 to i64 - %arrayidx2 = getelementptr inbounds [1024 x i32] addrspace(1)* @g_as1, i64 0, i64 %idxprom1 - store i32 %mul, i32 addrspace(1)* %arrayidx2, align 4 -- %inc = add nsw i32 %i.0, 1 -- br label %for.cond -+ %inc = add nsw i32 %i.02, 1 -+ %cmp = icmp slt i32 %inc, %n -+ br i1 %cmp, label %for.body, label %for.end - --for.end: ; preds = %for.cond -+for.end: ; preds = %for.body, %entry - ret void - } - diff --git a/contrib/llvm/patches/patch-24-llvm-r221791-ppc-small-pic.diff b/contrib/llvm/patches/patch-22-llvm-r221791-ppc-small-pic.diff index e69954af..b066b13 100644 --- a/contrib/llvm/patches/patch-24-llvm-r221791-ppc-small-pic.diff +++ b/contrib/llvm/patches/patch-22-llvm-r221791-ppc-small-pic.diff @@ -318,9 +318,9 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp // Change the opcode to LWZ, and the global address operand to be a // reference to the GOT entry we will synthesize later. -@@ -382,16 +409,23 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - else if (MO.isJTI()) - MOSymbol = GetJTISymbol(MO.getIndex()); +@@ -384,16 +411,23 @@ void PPCAsmPrinter::EmitInstruction(const MachineI + else if (MO.isBlockAddress()) + MOSymbol = GetBlockAddressSymbol(MO.getBlockAddress()); - MCSymbol *TOCEntry = lookUpOrCreateTOCEntry(MOSymbol); + if (PL == PICLevel::Small) { @@ -351,8 +351,8 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp EmitToStreamer(OutStreamer, TmpInst); return; } -@@ -399,7 +433,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - case PPC::LDtocCPT: +@@ -402,7 +436,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI + case PPC::LDtocBA: case PPC::LDtoc: { // Transform %X3 = LDtoc <ga:@min1>, %X2 - LowerPPCMachineInstrToMCInst(MI, TmpInst, *this, Subtarget.isDarwin()); @@ -360,7 +360,7 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp // Change the opcode to LD, and the global address operand to be a // reference to the TOC entry we will synthesize later. -@@ -428,7 +462,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI +@@ -433,7 +467,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI case PPC::ADDIStocHA: { // Transform %Xd = ADDIStocHA %X2, <ga:@sym> @@ -369,7 +369,7 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp // Change the opcode to ADDIS8. If the global address is external, has // common linkage, is a non-local function address, or is a jump table -@@ -470,7 +504,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI +@@ -479,7 +513,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI } case PPC::LDtocL: { // Transform %Xd = LDtocL <ga:@sym>, %Xs @@ -378,7 +378,7 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp // Change the opcode to LD. If the global address is external, has // common linkage, or is a jump table address, then reference the -@@ -507,7 +541,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI +@@ -521,7 +555,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI } case PPC::ADDItocL: { // Transform %Xd = ADDItocL %Xs, <ga:@sym> @@ -387,7 +387,7 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp // Change the opcode to ADDI8. If the global address is external, then // generate a TOC entry and reference that. Otherwise reference the -@@ -558,7 +592,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI +@@ -572,7 +606,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI case PPC::LDgotTprelL: case PPC::LDgotTprelL32: { // Transform %Xd = LDgotTprelL <ga:@sym>, %Xs @@ -396,7 +396,7 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp // Change the opcode to LD. TmpInst.setOpcode(isPPC64 ? PPC::LD : PPC::LWZ); -@@ -841,7 +875,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI +@@ -796,7 +830,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineI } } @@ -405,7 +405,7 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp EmitToStreamer(OutStreamer, TmpInst); } -@@ -857,16 +891,14 @@ void PPCLinuxAsmPrinter::EmitStartOfAsmFile(Module +@@ -812,16 +846,14 @@ void PPCLinuxAsmPrinter::EmitStartOfAsmFile(Module if (Subtarget.isPPC64() || TM.getRelocationModel() != Reloc::PIC_) return AsmPrinter::EmitStartOfAsmFile(M); @@ -426,7 +426,7 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp MCSymbol *CurrentPos = OutContext.CreateTempSymbol(); OutStreamer.EmitLabel(CurrentPos); -@@ -885,7 +917,9 @@ void PPCLinuxAsmPrinter::EmitStartOfAsmFile(Module +@@ -840,7 +872,9 @@ void PPCLinuxAsmPrinter::EmitStartOfAsmFile(Module void PPCLinuxAsmPrinter::EmitFunctionEntryLabel() { // linux/ppc32 - Normal entry label. @@ -437,7 +437,7 @@ Index: lib/Target/PowerPC/PPCAsmPrinter.cpp return AsmPrinter::EmitFunctionEntryLabel(); if (!Subtarget.isPPC64()) { -@@ -897,7 +931,7 @@ void PPCLinuxAsmPrinter::EmitFunctionEntryLabel() +@@ -852,7 +886,7 @@ void PPCLinuxAsmPrinter::EmitFunctionEntryLabel() const MCExpr *OffsExpr = MCBinaryExpr::CreateSub( @@ -458,7 +458,7 @@ Index: lib/Target/PowerPC/PPCISelDAGToDAG.cpp #include "llvm/Support/CommandLine.h" #include "llvm/Support/Debug.h" #include "llvm/Support/ErrorHandling.h" -@@ -273,23 +274,29 @@ SDNode *PPCDAGToDAGISel::getGlobalBaseReg() { +@@ -283,23 +284,29 @@ SDNode *PPCDAGToDAGISel::getGlobalBaseReg() { // Insert the set of GlobalBaseReg into the first MBB of the function MachineBasicBlock &FirstMBB = MF->front(); MachineBasicBlock::iterator MBBI = FirstMBB.begin(); @@ -499,7 +499,7 @@ Index: lib/Target/PowerPC/PPCISelDAGToDAG.cpp } } else { GlobalBaseReg = RegInfo->createVirtualRegister(&PPC::G8RC_NOX0RegClass); -@@ -1429,13 +1436,13 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) { +@@ -1439,13 +1446,13 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) { return CurDAG->SelectNodeTo(N, Reg, MVT::Other, Chain); } case PPCISD::TOC_ENTRY: { @@ -519,7 +519,7 @@ Index: lib/Target/PowerPC/PPCISelLowering.cpp =================================================================== --- lib/Target/PowerPC/PPCISelLowering.cpp +++ lib/Target/PowerPC/PPCISelLowering.cpp -@@ -1653,6 +1653,8 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S +@@ -1682,6 +1682,8 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S const GlobalValue *GV = GA->getGlobal(); EVT PtrVT = getPointerTy(); bool is64bit = Subtarget.isPPC64(); @@ -528,7 +528,7 @@ Index: lib/Target/PowerPC/PPCISelLowering.cpp TLSModel::Model Model = getTargetMachine().getTLSModel(GV); -@@ -1691,7 +1693,10 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S +@@ -1721,7 +1723,10 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S GOTPtr = DAG.getNode(PPCISD::ADDIS_TLSGD_HA, dl, PtrVT, GOTReg, TGA); } else { @@ -540,7 +540,7 @@ Index: lib/Target/PowerPC/PPCISelLowering.cpp } SDValue GOTEntry = DAG.getNode(PPCISD::ADDI_TLSGD_L, dl, PtrVT, GOTPtr, TGA); -@@ -1721,7 +1726,10 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S +@@ -1738,7 +1743,10 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S GOTPtr = DAG.getNode(PPCISD::ADDIS_TLSLD_HA, dl, PtrVT, GOTReg, TGA); } else { @@ -556,7 +556,7 @@ Index: lib/Target/PowerPC/PPCInstrInfo.td =================================================================== --- lib/Target/PowerPC/PPCInstrInfo.td +++ lib/Target/PowerPC/PPCInstrInfo.td -@@ -976,6 +976,9 @@ let isTerminator = 1, isBarrier = 1, PPC970_Unit = +@@ -980,6 +980,9 @@ let isTerminator = 1, isBarrier = 1, PPC970_Unit = let Defs = [LR] in def MovePCtoLR : Pseudo<(outs), (ins), "#MovePCtoLR", []>, PPC970_Unit_BRU; @@ -566,7 +566,7 @@ Index: lib/Target/PowerPC/PPCInstrInfo.td let isBranch = 1, isTerminator = 1, hasCtrlDep = 1, PPC970_Unit = 7 in { let isBarrier = 1 in { -@@ -2444,15 +2447,13 @@ def ADDISdtprelHA32 : Pseudo<(outs gprc:$rD), (ins +@@ -2442,15 +2445,13 @@ def ADDISdtprelHA32 : Pseudo<(outs gprc:$rD), (ins tglobaltlsaddr:$disp))]>; // Support for Position-independent code diff --git a/contrib/llvm/patches/patch-25-llvm-r224415-ppc-local.diff b/contrib/llvm/patches/patch-23-llvm-r224415-ppc-local.diff index 9099743..9099743 100644 --- a/contrib/llvm/patches/patch-25-llvm-r224415-ppc-local.diff +++ b/contrib/llvm/patches/patch-23-llvm-r224415-ppc-local.diff diff --git a/contrib/llvm/patches/patch-26-llvm-r213890-ppc-eh_frame.diff b/contrib/llvm/patches/patch-24-llvm-r213890-ppc-eh_frame.diff index 9a4881f..9a4881f 100644 --- a/contrib/llvm/patches/patch-26-llvm-r213890-ppc-eh_frame.diff +++ b/contrib/llvm/patches/patch-24-llvm-r213890-ppc-eh_frame.diff diff --git a/contrib/llvm/patches/patch-28-llvm-r224890-ppc-ctr-tls-loop.diff b/contrib/llvm/patches/patch-25-llvm-r224890-ppc-ctr-tls-loop.diff index aeb81f2..aeb81f2 100644 --- a/contrib/llvm/patches/patch-28-llvm-r224890-ppc-ctr-tls-loop.diff +++ b/contrib/llvm/patches/patch-25-llvm-r224890-ppc-ctr-tls-loop.diff diff --git a/contrib/llvm/patches/patch-29-clang-r213790-type_traits-crash.diff b/contrib/llvm/patches/patch-26-clang-r213790-type_traits-crash.diff index e721d24..e721d24 100644 --- a/contrib/llvm/patches/patch-29-clang-r213790-type_traits-crash.diff +++ b/contrib/llvm/patches/patch-26-clang-r213790-type_traits-crash.diff diff --git a/contrib/llvm/patches/patch-27-llvm-r221703-ppc-tls_get_addr.diff b/contrib/llvm/patches/patch-27-llvm-r221703-ppc-tls_get_addr.diff deleted file mode 100644 index 6cf2a6a..0000000 --- a/contrib/llvm/patches/patch-27-llvm-r221703-ppc-tls_get_addr.diff +++ /dev/null @@ -1,504 +0,0 @@ -Pull in r221703 from upstream llvm trunk (by Bill Schmidt): - - [PowerPC] Replace foul hackery with real calls to __tls_get_addr - - My original support for the general dynamic and local dynamic TLS - models contained some fairly obtuse hacks to generate calls to - __tls_get_addr when lowering a TargetGlobalAddress. Rather than - generating real calls, special GET_TLS_ADDR nodes were used to wrap - the calls and only reveal them at assembly time. I attempted to - provide correct parameter and return values by chaining CopyToReg and - CopyFromReg nodes onto the GET_TLS_ADDR nodes, but this was also not - fully correct. Problems were seen with two back-to-back stores to TLS - variables, where the call sequences ended up overlapping with unhappy - results. Additionally, since these weren't real calls, the proper - register side effects of a call were not recorded, so clobbered values - were kept live across the calls. - - The proper thing to do is to lower these into calls in the first - place. This is relatively straightforward; see the changes to - PPCTargetLowering::LowerGlobalTLSAddress() in PPCISelLowering.cpp. - The changes here are standard call lowering, except that we need to - track the fact that these calls will require a relocation. This is - done by adding a machine operand flag of MO_TLSLD or MO_TLSGD to the - TargetGlobalAddress operand that appears earlier in the sequence. - - The calls to LowerCallTo() eventually find their way to - LowerCall_64SVR4() or LowerCall_32SVR4(), which call FinishCall(), - which calls PrepareCall(). In PrepareCall(), we detect the calls to - __tls_get_addr and immediately snag the TargetGlobalTLSAddress with - the annotated relocation information. This becomes an extra operand - on the call following the callee, which is expected for nodes of type - tlscall. We change the call opcode to CALL_TLS for this case. Back - in FinishCall(), we change it again to CALL_NOP_TLS for 64-bit only, - since we require a TOC-restore nop following the call for the 64-bit - ABIs. - - During selection, patterns in PPCInstrInfo.td and PPCInstr64Bit.td - convert the CALL_TLS nodes into BL_TLS nodes, and convert the - CALL_NOP_TLS nodes into BL8_NOP_TLS nodes. This replaces the code - removed from PPCAsmPrinter.cpp, as the BL_TLS or BL8_NOP_TLS - nodes can now be emitted normally using their patterns and the - associated printTLSCall print method. - - Finally, as a result of these changes, all references to get-tls-addr - in its various guises are no longer used, so they have been removed. - - There are existing TLS tests to verify the changes haven't messed - anything up). I've added one new test that verifies that the problem - with the original code has been fixed. - -This fixes a fatal "Bad machine code" error when compiling parts of -libgomp for 32-bit PowerPC. - -Introduced here: http://svnweb.freebsd.org/changeset/base/276301 - -Index: lib/Target/PowerPC/PPC.h -=================================================================== ---- lib/Target/PowerPC/PPC.h -+++ lib/Target/PowerPC/PPC.h -@@ -96,7 +96,12 @@ namespace llvm { - MO_TOC_LO = 7 << 4, - - // Symbol for VK_PPC_TLS fixup attached to an ADD instruction -- MO_TLS = 8 << 4 -+ MO_TLS = 8 << 4, -+ -+ // Symbols for VK_PPC_TLSGD and VK_PPC_TLSLD in __tls_get_addr -+ // call sequences. -+ MO_TLSLD = 9 << 4, -+ MO_TLSGD = 10 << 4 - }; - } // end namespace PPCII - -Index: lib/Target/PowerPC/PPCAsmPrinter.cpp -=================================================================== ---- lib/Target/PowerPC/PPCAsmPrinter.cpp -+++ lib/Target/PowerPC/PPCAsmPrinter.cpp -@@ -689,35 +689,6 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - .addExpr(SymGotTlsGD)); - return; - } -- case PPC::GETtlsADDR: -- // Transform: %X3 = GETtlsADDR %X3, <ga:@sym> -- // Into: BL8_NOP_TLS __tls_get_addr(sym@tlsgd) -- case PPC::GETtlsADDR32: { -- // Transform: %R3 = GETtlsADDR32 %R3, <ga:@sym> -- // Into: BL_TLS __tls_get_addr(sym@tlsgd)@PLT -- -- StringRef Name = "__tls_get_addr"; -- MCSymbol *TlsGetAddr = OutContext.GetOrCreateSymbol(Name); -- MCSymbolRefExpr::VariantKind Kind = MCSymbolRefExpr::VK_None; -- -- if (!Subtarget.isPPC64() && !Subtarget.isDarwin() && -- TM.getRelocationModel() == Reloc::PIC_) -- Kind = MCSymbolRefExpr::VK_PLT; -- const MCSymbolRefExpr *TlsRef = -- MCSymbolRefExpr::Create(TlsGetAddr, Kind, OutContext); -- const MachineOperand &MO = MI->getOperand(2); -- const GlobalValue *GValue = MO.getGlobal(); -- MCSymbol *MOSymbol = getSymbol(GValue); -- const MCExpr *SymVar = -- MCSymbolRefExpr::Create(MOSymbol, MCSymbolRefExpr::VK_PPC_TLSGD, -- OutContext); -- EmitToStreamer(OutStreamer, -- MCInstBuilder(Subtarget.isPPC64() ? -- PPC::BL8_NOP_TLS : PPC::BL_TLS) -- .addExpr(TlsRef) -- .addExpr(SymVar)); -- return; -- } - case PPC::ADDIStlsldHA: { - // Transform: %Xd = ADDIStlsldHA %X2, <ga:@sym> - // Into: %Xd = ADDIS8 %X2, sym@got@tlsld@ha -@@ -755,36 +726,6 @@ void PPCAsmPrinter::EmitInstruction(const MachineI - .addExpr(SymGotTlsLD)); - return; - } -- case PPC::GETtlsldADDR: -- // Transform: %X3 = GETtlsldADDR %X3, <ga:@sym> -- // Into: BL8_NOP_TLS __tls_get_addr(sym@tlsld) -- case PPC::GETtlsldADDR32: { -- // Transform: %R3 = GETtlsldADDR32 %R3, <ga:@sym> -- // Into: BL_TLS __tls_get_addr(sym@tlsld)@PLT -- -- StringRef Name = "__tls_get_addr"; -- MCSymbol *TlsGetAddr = OutContext.GetOrCreateSymbol(Name); -- MCSymbolRefExpr::VariantKind Kind = MCSymbolRefExpr::VK_None; -- -- if (!Subtarget.isPPC64() && !Subtarget.isDarwin() && -- TM.getRelocationModel() == Reloc::PIC_) -- Kind = MCSymbolRefExpr::VK_PLT; -- -- const MCSymbolRefExpr *TlsRef = -- MCSymbolRefExpr::Create(TlsGetAddr, Kind, OutContext); -- const MachineOperand &MO = MI->getOperand(2); -- const GlobalValue *GValue = MO.getGlobal(); -- MCSymbol *MOSymbol = getSymbol(GValue); -- const MCExpr *SymVar = -- MCSymbolRefExpr::Create(MOSymbol, MCSymbolRefExpr::VK_PPC_TLSLD, -- OutContext); -- EmitToStreamer(OutStreamer, -- MCInstBuilder(Subtarget.isPPC64() ? -- PPC::BL8_NOP_TLS : PPC::BL_TLS) -- .addExpr(TlsRef) -- .addExpr(SymVar)); -- return; -- } - case PPC::ADDISdtprelHA: - // Transform: %Xd = ADDISdtprelHA %X3, <ga:@sym> - // Into: %Xd = ADDIS8 %X3, sym@dtprel@ha -Index: lib/Target/PowerPC/PPCISelLowering.cpp -=================================================================== ---- lib/Target/PowerPC/PPCISelLowering.cpp -+++ lib/Target/PowerPC/PPCISelLowering.cpp -@@ -781,6 +781,8 @@ const char *PPCTargetLowering::getTargetNodeName(u - case PPCISD::SHL: return "PPCISD::SHL"; - case PPCISD::CALL: return "PPCISD::CALL"; - case PPCISD::CALL_NOP: return "PPCISD::CALL_NOP"; -+ case PPCISD::CALL_TLS: return "PPCISD::CALL_TLS"; -+ case PPCISD::CALL_NOP_TLS: return "PPCISD::CALL_NOP_TLS"; - case PPCISD::MTCTR: return "PPCISD::MTCTR"; - case PPCISD::BCTRL: return "PPCISD::BCTRL"; - case PPCISD::RET_FLAG: return "PPCISD::RET_FLAG"; -@@ -810,10 +812,8 @@ const char *PPCTargetLowering::getTargetNodeName(u - case PPCISD::ADD_TLS: return "PPCISD::ADD_TLS"; - case PPCISD::ADDIS_TLSGD_HA: return "PPCISD::ADDIS_TLSGD_HA"; - case PPCISD::ADDI_TLSGD_L: return "PPCISD::ADDI_TLSGD_L"; -- case PPCISD::GET_TLS_ADDR: return "PPCISD::GET_TLS_ADDR"; - case PPCISD::ADDIS_TLSLD_HA: return "PPCISD::ADDIS_TLSLD_HA"; - case PPCISD::ADDI_TLSLD_L: return "PPCISD::ADDI_TLSLD_L"; -- case PPCISD::GET_TLSLD_ADDR: return "PPCISD::GET_TLSLD_ADDR"; - case PPCISD::ADDIS_DTPREL_HA: return "PPCISD::ADDIS_DTPREL_HA"; - case PPCISD::ADDI_DTPREL_L: return "PPCISD::ADDI_DTPREL_L"; - case PPCISD::VADD_SPLAT: return "PPCISD::VADD_SPLAT"; -@@ -1641,6 +1641,27 @@ SDValue PPCTargetLowering::LowerBlockAddress(SDVal - return LowerLabelRef(TgtBAHi, TgtBALo, isPIC, DAG); - } - -+// Generate a call to __tls_get_addr for the given GOT entry Op. -+std::pair<SDValue,SDValue> -+PPCTargetLowering::lowerTLSCall(SDValue Op, SDLoc dl, -+ SelectionDAG &DAG) const { -+ -+ Type *IntPtrTy = getDataLayout()->getIntPtrType(*DAG.getContext()); -+ TargetLowering::ArgListTy Args; -+ TargetLowering::ArgListEntry Entry; -+ Entry.Node = Op; -+ Entry.Ty = IntPtrTy; -+ Args.push_back(Entry); -+ -+ TargetLowering::CallLoweringInfo CLI(DAG); -+ CLI.setDebugLoc(dl).setChain(DAG.getEntryNode()) -+ .setCallee(CallingConv::C, IntPtrTy, -+ DAG.getTargetExternalSymbol("__tls_get_addr", getPointerTy()), -+ std::move(Args), 0); -+ -+ return LowerCallTo(CLI); -+} -+ - SDValue PPCTargetLowering::LowerGlobalTLSAddress(SDValue Op, - SelectionDAG &DAG) const { - -@@ -1686,7 +1707,8 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S - } - - if (Model == TLSModel::GeneralDynamic) { -- SDValue TGA = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, 0); -+ SDValue TGA = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, -+ PPCII::MO_TLSGD); - SDValue GOTPtr; - if (is64bit) { - SDValue GOTReg = DAG.getRegister(PPC::X2, MVT::i64); -@@ -1700,26 +1722,13 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S - } - SDValue GOTEntry = DAG.getNode(PPCISD::ADDI_TLSGD_L, dl, PtrVT, - GOTPtr, TGA); -- -- // We need a chain node, and don't have one handy. The underlying -- // call has no side effects, so using the function entry node -- // suffices. -- SDValue Chain = DAG.getEntryNode(); -- Chain = DAG.getCopyToReg(Chain, dl, -- is64bit ? PPC::X3 : PPC::R3, GOTEntry); -- SDValue ParmReg = DAG.getRegister(is64bit ? PPC::X3 : PPC::R3, -- is64bit ? MVT::i64 : MVT::i32); -- SDValue TLSAddr = DAG.getNode(PPCISD::GET_TLS_ADDR, dl, -- PtrVT, ParmReg, TGA); -- // The return value from GET_TLS_ADDR really is in X3 already, but -- // some hacks are needed here to tie everything together. The extra -- // copies dissolve during subsequent transforms. -- Chain = DAG.getCopyToReg(Chain, dl, is64bit ? PPC::X3 : PPC::R3, TLSAddr); -- return DAG.getCopyFromReg(Chain, dl, is64bit ? PPC::X3 : PPC::R3, PtrVT); -+ std::pair<SDValue, SDValue> CallResult = lowerTLSCall(GOTEntry, dl, DAG); -+ return CallResult.first; - } - - if (Model == TLSModel::LocalDynamic) { -- SDValue TGA = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, 0); -+ SDValue TGA = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, -+ PPCII::MO_TLSLD); - SDValue GOTPtr; - if (is64bit) { - SDValue GOTReg = DAG.getRegister(PPC::X2, MVT::i64); -@@ -1733,23 +1742,11 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(S - } - SDValue GOTEntry = DAG.getNode(PPCISD::ADDI_TLSLD_L, dl, PtrVT, - GOTPtr, TGA); -- -- // We need a chain node, and don't have one handy. The underlying -- // call has no side effects, so using the function entry node -- // suffices. -- SDValue Chain = DAG.getEntryNode(); -- Chain = DAG.getCopyToReg(Chain, dl, -- is64bit ? PPC::X3 : PPC::R3, GOTEntry); -- SDValue ParmReg = DAG.getRegister(is64bit ? PPC::X3 : PPC::R3, -- is64bit ? MVT::i64 : MVT::i32); -- SDValue TLSAddr = DAG.getNode(PPCISD::GET_TLSLD_ADDR, dl, -- PtrVT, ParmReg, TGA); -- // The return value from GET_TLSLD_ADDR really is in X3 already, but -- // some hacks are needed here to tie everything together. The extra -- // copies dissolve during subsequent transforms. -- Chain = DAG.getCopyToReg(Chain, dl, is64bit ? PPC::X3 : PPC::R3, TLSAddr); -+ std::pair<SDValue, SDValue> CallResult = lowerTLSCall(GOTEntry, dl, DAG); -+ SDValue TLSAddr = CallResult.first; -+ SDValue Chain = CallResult.second; - SDValue DtvOffsetHi = DAG.getNode(PPCISD::ADDIS_DTPREL_HA, dl, PtrVT, -- Chain, ParmReg, TGA); -+ Chain, TLSAddr, TGA); - return DAG.getNode(PPCISD::ADDI_DTPREL_L, dl, PtrVT, DtvOffsetHi, TGA); - } - -@@ -3712,6 +3709,23 @@ unsigned PrepareCall(SelectionDAG &DAG, SDValue &C - if (Callee.getNode()) { - Ops.push_back(Chain); - Ops.push_back(Callee); -+ -+ // If this is a call to __tls_get_addr, find the symbol whose address -+ // is to be taken and add it to the list. This will be used to -+ // generate __tls_get_addr(<sym>@tlsgd) or __tls_get_addr(<sym>@tlsld). -+ // We find the symbol by walking the chain to the CopyFromReg, walking -+ // back from the CopyFromReg to the ADDI_TLSGD_L or ADDI_TLSLD_L, and -+ // pulling the symbol from that node. -+ if (ExternalSymbolSDNode *S = dyn_cast<ExternalSymbolSDNode>(Callee)) -+ if (!strcmp(S->getSymbol(), "__tls_get_addr")) { -+ assert(!needIndirectCall && "Indirect call to __tls_get_addr???"); -+ SDNode *AddI = Chain.getNode()->getOperand(2).getNode(); -+ SDValue TGTAddr = AddI->getOperand(1); -+ assert(TGTAddr.getNode()->getOpcode() == ISD::TargetGlobalTLSAddress && -+ "Didn't find target global TLS address where we expected one"); -+ Ops.push_back(TGTAddr); -+ CallOpc = PPCISD::CALL_TLS; -+ } - } - // If this is a tail call add stack pointer delta. - if (isTailCall) -@@ -3863,7 +3877,9 @@ PPCTargetLowering::FinishCall(CallingConv::ID Call - DAG.getTarget().getRelocationModel() == Reloc::PIC_)) { - // Otherwise insert NOP for non-local calls. - CallOpc = PPCISD::CALL_NOP; -- } -+ } else if (CallOpc == PPCISD::CALL_TLS) -+ // For 64-bit SVR4, TLS calls are always non-local. -+ CallOpc = PPCISD::CALL_NOP_TLS; - } - - Chain = DAG.getNode(CallOpc, dl, NodeTys, Ops); -Index: lib/Target/PowerPC/PPCISelLowering.h -=================================================================== ---- lib/Target/PowerPC/PPCISelLowering.h -+++ lib/Target/PowerPC/PPCISelLowering.h -@@ -99,6 +99,10 @@ namespace llvm { - /// SVR4 calls. - CALL, CALL_NOP, - -+ /// CALL_TLS and CALL_NOP_TLS - Versions of CALL and CALL_NOP used -+ /// to access TLS variables. -+ CALL_TLS, CALL_NOP_TLS, -+ - /// CHAIN,FLAG = MTCTR(VAL, CHAIN[, INFLAG]) - Directly corresponds to a - /// MTCTR instruction. - MTCTR, -@@ -214,10 +218,6 @@ namespace llvm { - /// sym\@got\@tlsgd\@l. - ADDI_TLSGD_L, - -- /// G8RC = GET_TLS_ADDR %X3, Symbol - For the general-dynamic TLS -- /// model, produces a call to __tls_get_addr(sym\@tlsgd). -- GET_TLS_ADDR, -- - /// G8RC = ADDIS_TLSLD_HA %X2, Symbol - For the local-dynamic TLS - /// model, produces an ADDIS8 instruction that adds the GOT base - /// register to sym\@got\@tlsld\@ha. -@@ -228,10 +228,6 @@ namespace llvm { - /// sym\@got\@tlsld\@l. - ADDI_TLSLD_L, - -- /// G8RC = GET_TLSLD_ADDR %X3, Symbol - For the local-dynamic TLS -- /// model, produces a call to __tls_get_addr(sym\@tlsld). -- GET_TLSLD_ADDR, -- - /// G8RC = ADDIS_DTPREL_HA %X3, Symbol, Chain - For the - /// local-dynamic TLS model, produces an ADDIS8 instruction - /// that adds X3 to sym\@dtprel\@ha. The Chain operand is needed -@@ -552,6 +548,8 @@ namespace llvm { - SDValue LowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const; - SDValue LowerConstantPool(SDValue Op, SelectionDAG &DAG) const; - SDValue LowerBlockAddress(SDValue Op, SelectionDAG &DAG) const; -+ std::pair<SDValue,SDValue> lowerTLSCall(SDValue Op, SDLoc dl, -+ SelectionDAG &DAG) const; - SDValue LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const; - SDValue LowerGlobalAddress(SDValue Op, SelectionDAG &DAG) const; - SDValue LowerJumpTable(SDValue Op, SelectionDAG &DAG) const; -Index: lib/Target/PowerPC/PPCInstr64Bit.td -=================================================================== ---- lib/Target/PowerPC/PPCInstr64Bit.td -+++ lib/Target/PowerPC/PPCInstr64Bit.td -@@ -188,6 +188,9 @@ def : Pat<(PPCcall (i64 texternalsym:$dst)), - def : Pat<(PPCcall_nop (i64 texternalsym:$dst)), - (BL8_NOP texternalsym:$dst)>; - -+def : Pat<(PPCcall_nop_tls texternalsym:$func, tglobaltlsaddr:$sym), -+ (BL8_NOP_TLS texternalsym:$func, tglobaltlsaddr:$sym)>; -+ - // Atomic operations - let usesCustomInserter = 1 in { - let Defs = [CR0] in { -@@ -872,11 +875,6 @@ def ADDItlsgdL : Pseudo<(outs g8rc:$rD), (ins g8rc - [(set i64:$rD, - (PPCaddiTlsgdL i64:$reg, tglobaltlsaddr:$disp))]>, - isPPC64; --def GETtlsADDR : Pseudo<(outs g8rc:$rD), (ins g8rc:$reg, tlsgd:$sym), -- "#GETtlsADDR", -- [(set i64:$rD, -- (PPCgetTlsAddr i64:$reg, tglobaltlsaddr:$sym))]>, -- isPPC64; - def ADDIStlsldHA: Pseudo<(outs g8rc:$rD), (ins g8rc_nox0:$reg, s16imm64:$disp), - "#ADDIStlsldHA", - [(set i64:$rD, -@@ -887,11 +885,6 @@ def ADDItlsldL : Pseudo<(outs g8rc:$rD), (ins g8rc - [(set i64:$rD, - (PPCaddiTlsldL i64:$reg, tglobaltlsaddr:$disp))]>, - isPPC64; --def GETtlsldADDR : Pseudo<(outs g8rc:$rD), (ins g8rc:$reg, tlsgd:$sym), -- "#GETtlsldADDR", -- [(set i64:$rD, -- (PPCgetTlsldAddr i64:$reg, tglobaltlsaddr:$sym))]>, -- isPPC64; - def ADDISdtprelHA: Pseudo<(outs g8rc:$rD), (ins g8rc_nox0:$reg, s16imm64:$disp), - "#ADDISdtprelHA", - [(set i64:$rD, -Index: lib/Target/PowerPC/PPCInstrInfo.td -=================================================================== ---- lib/Target/PowerPC/PPCInstrInfo.td -+++ lib/Target/PowerPC/PPCInstrInfo.td -@@ -110,10 +110,8 @@ def PPCldGotTprelL : SDNode<"PPCISD::LD_GOT_TPREL_ - def PPCaddTls : SDNode<"PPCISD::ADD_TLS", SDTIntBinOp, []>; - def PPCaddisTlsgdHA : SDNode<"PPCISD::ADDIS_TLSGD_HA", SDTIntBinOp>; - def PPCaddiTlsgdL : SDNode<"PPCISD::ADDI_TLSGD_L", SDTIntBinOp>; --def PPCgetTlsAddr : SDNode<"PPCISD::GET_TLS_ADDR", SDTIntBinOp>; - def PPCaddisTlsldHA : SDNode<"PPCISD::ADDIS_TLSLD_HA", SDTIntBinOp>; - def PPCaddiTlsldL : SDNode<"PPCISD::ADDI_TLSLD_L", SDTIntBinOp>; --def PPCgetTlsldAddr : SDNode<"PPCISD::GET_TLSLD_ADDR", SDTIntBinOp>; - def PPCaddisDtprelHA : SDNode<"PPCISD::ADDIS_DTPREL_HA", SDTIntBinOp, - [SDNPHasChain]>; - def PPCaddiDtprelL : SDNode<"PPCISD::ADDI_DTPREL_L", SDTIntBinOp>; -@@ -136,9 +134,15 @@ def SDT_PPCCall : SDTypeProfile<0, -1, [SDTCisIn - def PPCcall : SDNode<"PPCISD::CALL", SDT_PPCCall, - [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue, - SDNPVariadic]>; -+def PPCcall_tls : SDNode<"PPCISD::CALL_TLS", SDT_PPCCall, -+ [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue, -+ SDNPVariadic]>; - def PPCcall_nop : SDNode<"PPCISD::CALL_NOP", SDT_PPCCall, - [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue, - SDNPVariadic]>; -+def PPCcall_nop_tls : SDNode<"PPCISD::CALL_NOP_TLS", SDT_PPCCall, -+ [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue, -+ SDNPVariadic]>; - def PPCload : SDNode<"PPCISD::LOAD", SDTypeProfile<1, 1, []>, - [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue]>; - def PPCload_toc : SDNode<"PPCISD::LOAD_TOC", SDTypeProfile<0, 1, []>, -@@ -2369,6 +2373,8 @@ def : Pat<(PPCcall (i32 tglobaladdr:$dst)), - def : Pat<(PPCcall (i32 texternalsym:$dst)), - (BL texternalsym:$dst)>; - -+def : Pat<(PPCcall_tls texternalsym:$func, tglobaltlsaddr:$sym), -+ (BL_TLS texternalsym:$func, tglobaltlsaddr:$sym)>; - - def : Pat<(PPCtc_return (i32 tglobaladdr:$dst), imm:$imm), - (TCRETURNdi tglobaladdr:$dst, imm:$imm)>; -@@ -2424,18 +2430,10 @@ def ADDItlsgdL32 : Pseudo<(outs gprc:$rD), (ins gp - "#ADDItlsgdL32", - [(set i32:$rD, - (PPCaddiTlsgdL i32:$reg, tglobaltlsaddr:$disp))]>; --def GETtlsADDR32 : Pseudo<(outs gprc:$rD), (ins gprc:$reg, tlsgd32:$sym), -- "#GETtlsADDR32", -- [(set i32:$rD, -- (PPCgetTlsAddr i32:$reg, tglobaltlsaddr:$sym))]>; - def ADDItlsldL32 : Pseudo<(outs gprc:$rD), (ins gprc_nor0:$reg, s16imm:$disp), - "#ADDItlsldL32", - [(set i32:$rD, - (PPCaddiTlsldL i32:$reg, tglobaltlsaddr:$disp))]>; --def GETtlsldADDR32 : Pseudo<(outs gprc:$rD), (ins gprc:$reg, tlsgd32:$sym), -- "#GETtlsldADDR32", -- [(set i32:$rD, -- (PPCgetTlsldAddr i32:$reg, tglobaltlsaddr:$sym))]>; - def ADDIdtprelL32 : Pseudo<(outs gprc:$rD), (ins gprc_nor0:$reg, s16imm:$disp), - "#ADDIdtprelL32", - [(set i32:$rD, -Index: lib/Target/PowerPC/PPCMCInstLower.cpp -=================================================================== ---- lib/Target/PowerPC/PPCMCInstLower.cpp -+++ lib/Target/PowerPC/PPCMCInstLower.cpp -@@ -137,6 +137,12 @@ static MCOperand GetSymbolRef(const MachineOperand - case PPCII::MO_TLS: - RefKind = MCSymbolRefExpr::VK_PPC_TLS; - break; -+ case PPCII::MO_TLSGD: -+ RefKind = MCSymbolRefExpr::VK_PPC_TLSGD; -+ break; -+ case PPCII::MO_TLSLD: -+ RefKind = MCSymbolRefExpr::VK_PPC_TLSLD; -+ break; - } - - if (MO.getTargetFlags() == PPCII::MO_PLT_OR_STUB && !isDarwin) -Index: test/CodeGen/PowerPC/tls-store2.ll -=================================================================== ---- test/CodeGen/PowerPC/tls-store2.ll -+++ test/CodeGen/PowerPC/tls-store2.ll -@@ -0,0 +1,33 @@ -+; RUN: llc -march=ppc64 -mcpu=pwr7 -O2 -relocation-model=pic < %s | FileCheck %s -+ -+target datalayout = "e-m:e-i64:64-n32:64" -+target triple = "powerpc64le-unknown-linux-gnu" -+ -+; Test back-to-back stores of TLS variables to ensure call sequences no -+; longer overlap. -+ -+@__once_callable = external thread_local global i8** -+@__once_call = external thread_local global void ()* -+ -+define i64 @call_once(i64 %flag, i8* %ptr) { -+entry: -+ %var = alloca i8*, align 8 -+ store i8* %ptr, i8** %var, align 8 -+ store i8** %var, i8*** @__once_callable, align 8 -+ store void ()* @__once_call_impl, void ()** @__once_call, align 8 -+ ret i64 %flag -+} -+ -+; CHECK-LABEL: call_once: -+; CHECK: addis 3, 2, __once_callable@got@tlsgd@ha -+; CHECK: addi 3, 3, __once_callable@got@tlsgd@l -+; CHECK: bl __tls_get_addr(__once_callable@tlsgd) -+; CHECK-NEXT: nop -+; CHECK: std {{[0-9]+}}, 0(3) -+; CHECK: addis 3, 2, __once_call@got@tlsgd@ha -+; CHECK: addi 3, 3, __once_call@got@tlsgd@l -+; CHECK: bl __tls_get_addr(__once_call@tlsgd) -+; CHECK-NEXT: nop -+; CHECK: std {{[0-9]+}}, 0(3) -+ -+declare void @__once_call_impl() diff --git a/contrib/llvm/patches/patch-30-llvm-r222587-arm-add-pc.diff b/contrib/llvm/patches/patch-27-llvm-r222587-arm-add-pc.diff index 46b2a04..46b2a04 100644 --- a/contrib/llvm/patches/patch-30-llvm-r222587-arm-add-pc.diff +++ b/contrib/llvm/patches/patch-27-llvm-r222587-arm-add-pc.diff diff --git a/contrib/llvm/patches/patch-31-llvm-r222292-aarch64-no-neon.diff b/contrib/llvm/patches/patch-28-llvm-r222292-aarch64-no-neon.diff index c1120bb..c1120bb 100644 --- a/contrib/llvm/patches/patch-31-llvm-r222292-aarch64-no-neon.diff +++ b/contrib/llvm/patches/patch-28-llvm-r222292-aarch64-no-neon.diff diff --git a/contrib/llvm/tools/clang/include/clang/Basic/DiagnosticSemaKinds.td b/contrib/llvm/tools/clang/include/clang/Basic/DiagnosticSemaKinds.td index 89749ed..1665a45 100644 --- a/contrib/llvm/tools/clang/include/clang/Basic/DiagnosticSemaKinds.td +++ b/contrib/llvm/tools/clang/include/clang/Basic/DiagnosticSemaKinds.td @@ -2115,6 +2115,9 @@ def warn_attribute_invalid_on_definition : Warning< InGroup<IgnoredAttributes>; def err_attribute_dll_redeclaration : Error< "redeclaration of %q0 cannot add %q1 attribute">; +def warn_attribute_dll_redeclaration : Warning< + "redeclaration of %q0 should not add %q1 attribute">, + InGroup<DiagGroup<"dll-attribute-on-redeclaration">>; def err_attribute_dllimport_function_definition : Error< "dllimport cannot be applied to non-inline function definition">; def err_attribute_dll_deleted : Error< diff --git a/contrib/llvm/tools/clang/lib/Basic/Version.cpp b/contrib/llvm/tools/clang/lib/Basic/Version.cpp index 2b577d6..2d3875d 100644 --- a/contrib/llvm/tools/clang/lib/Basic/Version.cpp +++ b/contrib/llvm/tools/clang/lib/Basic/Version.cpp @@ -36,7 +36,7 @@ std::string getClangRepositoryPath() { // If the SVN_REPOSITORY is empty, try to use the SVN keyword. This helps us // pick up a tag in an SVN export, for example. - StringRef SVNRepository("$URL: https://llvm.org/svn/llvm-project/cfe/tags/RELEASE_350/final/lib/Basic/Version.cpp $"); + StringRef SVNRepository("$URL: https://llvm.org/svn/llvm-project/cfe/tags/RELEASE_351/final/lib/Basic/Version.cpp $"); if (URL.empty()) { URL = SVNRepository.slice(SVNRepository.find(':'), SVNRepository.find("/lib/Basic")); diff --git a/contrib/llvm/tools/clang/lib/CodeGen/CGExprScalar.cpp b/contrib/llvm/tools/clang/lib/CodeGen/CGExprScalar.cpp index 9e0fbcf..140e9aa 100644 --- a/contrib/llvm/tools/clang/lib/CodeGen/CGExprScalar.cpp +++ b/contrib/llvm/tools/clang/lib/CodeGen/CGExprScalar.cpp @@ -3216,18 +3216,26 @@ Value *ScalarExprEmitter::VisitChooseExpr(ChooseExpr *E) { Value *ScalarExprEmitter::VisitVAArgExpr(VAArgExpr *VE) { QualType Ty = VE->getType(); + if (Ty->isVariablyModifiedType()) CGF.EmitVariablyModifiedType(Ty); llvm::Value *ArgValue = CGF.EmitVAListRef(VE->getSubExpr()); llvm::Value *ArgPtr = CGF.EmitVAArg(ArgValue, VE->getType()); + llvm::Type *ArgTy = ConvertType(VE->getType()); // If EmitVAArg fails, we fall back to the LLVM instruction. if (!ArgPtr) - return Builder.CreateVAArg(ArgValue, ConvertType(VE->getType())); + return Builder.CreateVAArg(ArgValue, ArgTy); // FIXME Volatility. - return Builder.CreateLoad(ArgPtr); + llvm::Value *Val = Builder.CreateLoad(ArgPtr); + + // If EmitVAArg promoted the type, we must truncate it. + if (ArgTy != Val->getType()) + Val = Builder.CreateTrunc(Val, ArgTy); + + return Val; } Value *ScalarExprEmitter::VisitBlockExpr(const BlockExpr *block) { diff --git a/contrib/llvm/tools/clang/lib/CodeGen/TargetInfo.cpp b/contrib/llvm/tools/clang/lib/CodeGen/TargetInfo.cpp index ccb8c7e..05bbc4f 100644 --- a/contrib/llvm/tools/clang/lib/CodeGen/TargetInfo.cpp +++ b/contrib/llvm/tools/clang/lib/CodeGen/TargetInfo.cpp @@ -5544,15 +5544,19 @@ MipsABIInfo::classifyArgumentType(QualType Ty, uint64_t &Offset) const { // If we have reached here, aggregates are passed directly by coercing to // another structure type. Padding is inserted if the offset of the // aggregate is unaligned. - return ABIArgInfo::getDirect(HandleAggregates(Ty, TySize), 0, - getPaddingType(OrigOffset, CurrOffset)); + ABIArgInfo ArgInfo = + ABIArgInfo::getDirect(HandleAggregates(Ty, TySize), 0, + getPaddingType(OrigOffset, CurrOffset)); + ArgInfo.setInReg(true); + return ArgInfo; } // Treat an enum type as its underlying type. if (const EnumType *EnumTy = Ty->getAs<EnumType>()) Ty = EnumTy->getDecl()->getIntegerType(); - if (Ty->isPromotableIntegerType()) + // All integral types are promoted to the GPR width. + if (Ty->isIntegralOrEnumerationType()) return ABIArgInfo::getExtend(); return ABIArgInfo::getDirect( @@ -5604,7 +5608,12 @@ MipsABIInfo::returnAggregateInRegs(QualType RetTy, uint64_t Size) const { ABIArgInfo MipsABIInfo::classifyReturnType(QualType RetTy) const { uint64_t Size = getContext().getTypeSize(RetTy); - if (RetTy->isVoidType() || Size == 0) + if (RetTy->isVoidType()) + return ABIArgInfo::getIgnore(); + + // O32 doesn't treat zero-sized structs differently from other structs. + // However, N32/N64 ignores zero sized return values. + if (!IsO32 && Size == 0) return ABIArgInfo::getIgnore(); if (isAggregateTypeForABI(RetTy) || RetTy->isVectorType()) { @@ -5612,12 +5621,15 @@ ABIArgInfo MipsABIInfo::classifyReturnType(QualType RetTy) const { if (RetTy->isAnyComplexType()) return ABIArgInfo::getDirect(); - // O32 returns integer vectors in registers. - if (IsO32 && RetTy->isVectorType() && !RetTy->hasFloatingRepresentation()) - return ABIArgInfo::getDirect(returnAggregateInRegs(RetTy, Size)); - - if (!IsO32) - return ABIArgInfo::getDirect(returnAggregateInRegs(RetTy, Size)); + // O32 returns integer vectors in registers and N32/N64 returns all small + // aggregates in registers.. + if (!IsO32 || + (RetTy->isVectorType() && !RetTy->hasFloatingRepresentation())) { + ABIArgInfo ArgInfo = + ABIArgInfo::getDirect(returnAggregateInRegs(RetTy, Size)); + ArgInfo.setInReg(true); + return ArgInfo; + } } return ABIArgInfo::getIndirect(0); @@ -5647,11 +5659,20 @@ llvm::Value* MipsABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty, CodeGenFunction &CGF) const { llvm::Type *BP = CGF.Int8PtrTy; llvm::Type *BPP = CGF.Int8PtrPtrTy; + + // Integer arguments are promoted 32-bit on O32 and 64-bit on N32/N64. + unsigned SlotSizeInBits = IsO32 ? 32 : 64; + if (Ty->isIntegerType() && + CGF.getContext().getIntWidth(Ty) < SlotSizeInBits) { + Ty = CGF.getContext().getIntTypeForBitwidth(SlotSizeInBits, + Ty->isSignedIntegerType()); + } CGBuilderTy &Builder = CGF.Builder; llvm::Value *VAListAddrAsBPP = Builder.CreateBitCast(VAListAddr, BPP, "ap"); llvm::Value *Addr = Builder.CreateLoad(VAListAddrAsBPP, "ap.cur"); - int64_t TypeAlign = getContext().getTypeAlign(Ty) / 8; + int64_t TypeAlign = + std::min(getContext().getTypeAlign(Ty) / 8, StackAlignInBytes); llvm::Type *PTy = llvm::PointerType::getUnqual(CGF.ConvertType(Ty)); llvm::Value *AddrTyped; unsigned PtrWidth = getTarget().getPointerWidth(0); @@ -5670,8 +5691,8 @@ llvm::Value* MipsABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty, llvm::Value *AlignedAddr = Builder.CreateBitCast(AddrTyped, BP); TypeAlign = std::max((unsigned)TypeAlign, MinABIStackAlignInBytes); - uint64_t Offset = - llvm::RoundUpToAlignment(CGF.getContext().getTypeSize(Ty) / 8, TypeAlign); + unsigned ArgSizeInBits = CGF.getContext().getTypeSize(Ty); + uint64_t Offset = llvm::RoundUpToAlignment(ArgSizeInBits / 8, TypeAlign); llvm::Value *NextAddr = Builder.CreateGEP(AlignedAddr, llvm::ConstantInt::get(IntTy, Offset), "ap.next"); diff --git a/contrib/llvm/tools/clang/lib/Sema/SemaDecl.cpp b/contrib/llvm/tools/clang/lib/Sema/SemaDecl.cpp index bbe6930..8716227 100644 --- a/contrib/llvm/tools/clang/lib/Sema/SemaDecl.cpp +++ b/contrib/llvm/tools/clang/lib/Sema/SemaDecl.cpp @@ -5020,7 +5020,7 @@ static void checkDLLAttributeRedeclaration(Sema &S, NamedDecl *OldDecl, NewDecl = NewTD->getTemplatedDecl(); if (!OldDecl || !NewDecl) - return; + return; const DLLImportAttr *OldImportAttr = OldDecl->getAttr<DLLImportAttr>(); const DLLExportAttr *OldExportAttr = OldDecl->getAttr<DLLExportAttr>(); @@ -5037,13 +5037,30 @@ static void checkDLLAttributeRedeclaration(Sema &S, NamedDecl *OldDecl, // Implicitly generated declarations are also excluded for now because there // is no other way to switch these to use dllimport or dllexport. bool AddsAttr = !(OldImportAttr || OldExportAttr) && HasNewAttr; + if (AddsAttr && !IsSpecialization && !OldDecl->isImplicit()) { - S.Diag(NewDecl->getLocation(), diag::err_attribute_dll_redeclaration) - << NewDecl - << (NewImportAttr ? (const Attr *)NewImportAttr : NewExportAttr); + // If the declaration hasn't been used yet, allow with a warning for + // free functions and global variables. + bool JustWarn = false; + if (!OldDecl->isUsed() && !OldDecl->isCXXClassMember()) { + auto *VD = dyn_cast<VarDecl>(OldDecl); + if (VD && !VD->getDescribedVarTemplate()) + JustWarn = true; + auto *FD = dyn_cast<FunctionDecl>(OldDecl); + if (FD && FD->getTemplatedKind() == FunctionDecl::TK_NonTemplate) + JustWarn = true; + } + + unsigned DiagID = JustWarn ? diag::warn_attribute_dll_redeclaration + : diag::err_attribute_dll_redeclaration; + S.Diag(NewDecl->getLocation(), DiagID) + << NewDecl + << (NewImportAttr ? (const Attr *)NewImportAttr : NewExportAttr); S.Diag(OldDecl->getLocation(), diag::note_previous_declaration); - NewDecl->setInvalidDecl(); - return; + if (!JustWarn) { + NewDecl->setInvalidDecl(); + return; + } } // A redeclaration is not allowed to drop a dllimport attribute, the only diff --git a/contrib/llvm/tools/clang/lib/Sema/SemaTemplate.cpp b/contrib/llvm/tools/clang/lib/Sema/SemaTemplate.cpp index 63581a4..7e8f0b7 100644 --- a/contrib/llvm/tools/clang/lib/Sema/SemaTemplate.cpp +++ b/contrib/llvm/tools/clang/lib/Sema/SemaTemplate.cpp @@ -3692,12 +3692,12 @@ bool Sema::CheckTemplateArgumentList(TemplateDecl *Template, ArgumentPack.size(), Converted)) return true; - if (TemplateArgs[ArgIdx].getArgument().isPackExpansion() && - isa<TypeAliasTemplateDecl>(Template) && - !(Param + 1 == ParamEnd && (*Param)->isTemplateParameterPack() && - !getExpandedPackSize(*Param))) { + bool PackExpansionIntoNonPack = + TemplateArgs[ArgIdx].getArgument().isPackExpansion() && + (!(*Param)->isTemplateParameterPack() || getExpandedPackSize(*Param)); + if (PackExpansionIntoNonPack && isa<TypeAliasTemplateDecl>(Template)) { // Core issue 1430: we have a pack expansion as an argument to an - // alias template, and it's not part of a final parameter pack. This + // alias template, and it's not part of a parameter pack. This // can't be canonicalized, so reject it now. Diag(TemplateArgs[ArgIdx].getLocation(), diag::err_alias_template_expansion_into_fixed_list) @@ -3720,16 +3720,11 @@ bool Sema::CheckTemplateArgumentList(TemplateDecl *Template, ++Param; } - // If we just saw a pack expansion, then directly convert the remaining - // arguments, because we don't know what parameters they'll match up - // with. - if (TemplateArgs[ArgIdx-1].getArgument().isPackExpansion()) { - bool InFinalParameterPack = Param != ParamEnd && - Param + 1 == ParamEnd && - (*Param)->isTemplateParameterPack() && - !getExpandedPackSize(*Param); - - if (!InFinalParameterPack && !ArgumentPack.empty()) { + // If we just saw a pack expansion into a non-pack, then directly convert + // the remaining arguments, because we don't know what parameters they'll + // match up with. + if (PackExpansionIntoNonPack) { + if (!ArgumentPack.empty()) { // If we were part way through filling in an expanded parameter pack, // fall back to just producing individual arguments. Converted.insert(Converted.end(), @@ -3738,22 +3733,10 @@ bool Sema::CheckTemplateArgumentList(TemplateDecl *Template, } while (ArgIdx < NumArgs) { - if (InFinalParameterPack) - ArgumentPack.push_back(TemplateArgs[ArgIdx].getArgument()); - else - Converted.push_back(TemplateArgs[ArgIdx].getArgument()); + Converted.push_back(TemplateArgs[ArgIdx].getArgument()); ++ArgIdx; } - // Push the argument pack onto the list of converted arguments. - if (InFinalParameterPack) { - Converted.push_back( - TemplateArgument::CreatePackCopy(Context, - ArgumentPack.data(), - ArgumentPack.size())); - ArgumentPack.clear(); - } - return false; } diff --git a/contrib/llvm/utils/TableGen/CallingConvEmitter.cpp b/contrib/llvm/utils/TableGen/CallingConvEmitter.cpp index 6d43e8e..ec2251d 100644 --- a/contrib/llvm/utils/TableGen/CallingConvEmitter.cpp +++ b/contrib/llvm/utils/TableGen/CallingConvEmitter.cpp @@ -35,23 +35,26 @@ private: } // End anonymous namespace void CallingConvEmitter::run(raw_ostream &O) { - std::vector<Record*> CCs = Records.getAllDerivedDefinitions("CallingConv"); - - // Emit prototypes for all of the CC's so that they can forward ref each - // other. + + // Emit prototypes for all of the non-custom CC's so that they can forward ref + // each other. + for (unsigned i = 0, e = CCs.size(); i != e; ++i) { + if (!CCs[i]->getValueAsBit("Custom")) { + O << "static bool " << CCs[i]->getName() + << "(unsigned ValNo, MVT ValVT,\n" + << std::string(CCs[i]->getName().size() + 13, ' ') + << "MVT LocVT, CCValAssign::LocInfo LocInfo,\n" + << std::string(CCs[i]->getName().size() + 13, ' ') + << "ISD::ArgFlagsTy ArgFlags, CCState &State);\n"; + } + } + + // Emit each non-custom calling convention description in full. for (unsigned i = 0, e = CCs.size(); i != e; ++i) { - O << "static bool " << CCs[i]->getName() - << "(unsigned ValNo, MVT ValVT,\n" - << std::string(CCs[i]->getName().size()+13, ' ') - << "MVT LocVT, CCValAssign::LocInfo LocInfo,\n" - << std::string(CCs[i]->getName().size()+13, ' ') - << "ISD::ArgFlagsTy ArgFlags, CCState &State);\n"; + if (!CCs[i]->getValueAsBit("Custom")) + EmitCallingConv(CCs[i], O); } - - // Emit each calling convention description in full. - for (unsigned i = 0, e = CCs.size(); i != e; ++i) - EmitCallingConv(CCs[i], O); } @@ -227,6 +230,21 @@ void CallingConvEmitter::EmitAction(Record *Action, << IndentStr << "else\n" << IndentStr << IndentStr << "LocInfo = CCValAssign::AExt;\n"; } + } else if (Action->isSubClassOf("CCPromoteToUpperBitsInType")) { + Record *DestTy = Action->getValueAsDef("DestTy"); + MVT::SimpleValueType DestVT = getValueType(DestTy); + O << IndentStr << "LocVT = " << getEnumName(DestVT) << ";\n"; + if (MVT(DestVT).isFloatingPoint()) { + PrintFatalError("CCPromoteToUpperBitsInType does not handle floating " + "point"); + } else { + O << IndentStr << "if (ArgFlags.isSExt())\n" + << IndentStr << IndentStr << "LocInfo = CCValAssign::SExtUpper;\n" + << IndentStr << "else if (ArgFlags.isZExt())\n" + << IndentStr << IndentStr << "LocInfo = CCValAssign::ZExtUpper;\n" + << IndentStr << "else\n" + << IndentStr << IndentStr << "LocInfo = CCValAssign::AExtUpper;\n"; + } } else if (Action->isSubClassOf("CCBitConvertToType")) { Record *DestTy = Action->getValueAsDef("DestTy"); O << IndentStr << "LocVT = " << getEnumName(getValueType(DestTy)) <<";\n"; diff --git a/etc/mtree/BSD.debug.dist b/etc/mtree/BSD.debug.dist index 62d33e6..bf53896 100644 --- a/etc/mtree/BSD.debug.dist +++ b/etc/mtree/BSD.debug.dist @@ -25,7 +25,7 @@ .. lib clang - 3.5.0 + 3.5.1 lib freebsd .. diff --git a/etc/mtree/BSD.include.dist b/etc/mtree/BSD.include.dist index 9332ca7..24c896d 100644 --- a/etc/mtree/BSD.include.dist +++ b/etc/mtree/BSD.include.dist @@ -96,7 +96,7 @@ .. .. clang - 3.5.0 + 3.5.1 .. .. crypto diff --git a/etc/mtree/BSD.usr.dist b/etc/mtree/BSD.usr.dist index 022db7a..7a74ab3 100644 --- a/etc/mtree/BSD.usr.dist +++ b/etc/mtree/BSD.usr.dist @@ -15,7 +15,7 @@ aout .. clang - 3.5.0 + 3.5.1 lib freebsd .. diff --git a/lib/clang/include/Makefile b/lib/clang/include/Makefile index 4b44bfd..d8f1864 100644 --- a/lib/clang/include/Makefile +++ b/lib/clang/include/Makefile @@ -8,7 +8,7 @@ LLVM_SRCS= ${.CURDIR}/../../../contrib/llvm .PATH: ${LLVM_SRCS}/tools/clang/lib/Headers -INCSDIR=${INCLUDEDIR}/clang/3.5.0 +INCSDIR=${INCLUDEDIR}/clang/3.5.1 INCS= __wmmintrin_aes.h \ __wmmintrin_pclmul.h \ diff --git a/lib/clang/include/clang/Basic/Version.inc b/lib/clang/include/clang/Basic/Version.inc index 08ba886..2f66e03 100644 --- a/lib/clang/include/clang/Basic/Version.inc +++ b/lib/clang/include/clang/Basic/Version.inc @@ -1,11 +1,11 @@ /* $FreeBSD$ */ -#define CLANG_VERSION 3.5.0 +#define CLANG_VERSION 3.5.1 #define CLANG_VERSION_MAJOR 3 #define CLANG_VERSION_MINOR 5 -#define CLANG_VERSION_PATCHLEVEL 0 +#define CLANG_VERSION_PATCHLEVEL 1 #define CLANG_VENDOR "FreeBSD " -#define CLANG_VENDOR_SUFFIX " 20141124" +#define CLANG_VENDOR_SUFFIX " 20150115" -#define SVN_REVISION "216957" +#define SVN_REVISION "225668" diff --git a/lib/clang/include/clang/Config/config.h b/lib/clang/include/clang/Config/config.h index f7a5c99..b37098b 100644 --- a/lib/clang/include/clang/Config/config.h +++ b/lib/clang/include/clang/Config/config.h @@ -28,7 +28,7 @@ /* Define if we have libxml2 */ /* #undef CLANG_HAVE_LIBXML */ -#define PACKAGE_STRING "LLVM 3.5.0" +#define PACKAGE_STRING "LLVM 3.5.1" /* The LLVM product name and version */ #define BACKEND_PACKAGE_STRING PACKAGE_STRING diff --git a/lib/clang/include/llvm/Config/config.h b/lib/clang/include/llvm/Config/config.h index 4092d3e..79c2825 100644 --- a/lib/clang/include/llvm/Config/config.h +++ b/lib/clang/include/llvm/Config/config.h @@ -564,13 +564,13 @@ #define PACKAGE_NAME "LLVM" /* Define to the full name and version of this package. */ -#define PACKAGE_STRING "LLVM 3.5.0" +#define PACKAGE_STRING "LLVM 3.5.1" /* Define to the one symbol short name of this package. */ #define PACKAGE_TARNAME "llvm" /* Define to the version of this package. */ -#define PACKAGE_VERSION "3.5.0" +#define PACKAGE_VERSION "3.5.1" /* Define as the return type of signal handlers (`int' or `void'). */ #define RETSIGTYPE void diff --git a/lib/clang/libllvmmipscodegen/Makefile b/lib/clang/libllvmmipscodegen/Makefile index ae68d93..1c91920 100644 --- a/lib/clang/libllvmmipscodegen/Makefile +++ b/lib/clang/libllvmmipscodegen/Makefile @@ -12,8 +12,10 @@ SRCS= Mips16FrameLowering.cpp \ Mips16ISelLowering.cpp \ Mips16InstrInfo.cpp \ Mips16RegisterInfo.cpp \ + MipsABIInfo.cpp \ MipsAnalyzeImmediate.cpp \ MipsAsmPrinter.cpp \ + MipsCCState.cpp \ MipsCodeEmitter.cpp \ MipsConstantIslandPass.cpp \ MipsDelaySlotFiller.cpp \ diff --git a/lib/libclang_rt/Makefile.inc b/lib/libclang_rt/Makefile.inc index b0ddd28..ec35e9e 100644 --- a/lib/libclang_rt/Makefile.inc +++ b/lib/libclang_rt/Makefile.inc @@ -5,7 +5,7 @@ CRTARCH=${MACHINE_CPUARCH:C/amd64/x86_64/} CRTSRC=${.CURDIR}/../../../contrib/compiler-rt -LIBDIR=/usr/lib/clang/3.5.0/lib/freebsd +LIBDIR=/usr/lib/clang/3.5.1/lib/freebsd NO_PIC= MK_PROFILE=no diff --git a/tools/build/mk/OptionalObsoleteFiles.inc b/tools/build/mk/OptionalObsoleteFiles.inc index af7784d..4f3b6e1 100644 --- a/tools/build/mk/OptionalObsoleteFiles.inc +++ b/tools/build/mk/OptionalObsoleteFiles.inc @@ -489,60 +489,60 @@ OLD_FILES+=usr/bin/clang++ OLD_FILES+=usr/bin/clang-cpp OLD_FILES+=usr/bin/clang-tblgen OLD_FILES+=usr/bin/tblgen -OLD_FILES+=usr/include/clang/3.5.0/__wmmintrin_aes.h -OLD_FILES+=usr/include/clang/3.5.0/__wmmintrin_pclmul.h -OLD_FILES+=usr/include/clang/3.5.0/altivec.h -OLD_FILES+=usr/include/clang/3.5.0/ammintrin.h -OLD_FILES+=usr/include/clang/3.5.0/arm_acle.h -OLD_FILES+=usr/include/clang/3.5.0/arm_neon.h -OLD_FILES+=usr/include/clang/3.5.0/avx2intrin.h -OLD_FILES+=usr/include/clang/3.5.0/avxintrin.h -OLD_FILES+=usr/include/clang/3.5.0/bmi2intrin.h -OLD_FILES+=usr/include/clang/3.5.0/bmiintrin.h -OLD_FILES+=usr/include/clang/3.5.0/cpuid.h -OLD_FILES+=usr/include/clang/3.5.0/emmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/f16cintrin.h -OLD_FILES+=usr/include/clang/3.5.0/fma4intrin.h -OLD_FILES+=usr/include/clang/3.5.0/fmaintrin.h -OLD_FILES+=usr/include/clang/3.5.0/ia32intrin.h -OLD_FILES+=usr/include/clang/3.5.0/immintrin.h -OLD_FILES+=usr/include/clang/3.5.0/lzcntintrin.h -OLD_FILES+=usr/include/clang/3.5.0/mm3dnow.h -OLD_FILES+=usr/include/clang/3.5.0/mm_malloc.h -OLD_FILES+=usr/include/clang/3.5.0/mmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/module.modulemap -OLD_FILES+=usr/include/clang/3.5.0/nmmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/pmmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/popcntintrin.h -OLD_FILES+=usr/include/clang/3.5.0/prfchwintrin.h -OLD_FILES+=usr/include/clang/3.5.0/rdseedintrin.h -OLD_FILES+=usr/include/clang/3.5.0/rtmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/shaintrin.h -OLD_FILES+=usr/include/clang/3.5.0/smmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/tbmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/tmmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/wmmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/x86intrin.h -OLD_FILES+=usr/include/clang/3.5.0/xmmintrin.h -OLD_FILES+=usr/include/clang/3.5.0/xopintrin.h -OLD_DIRS+=usr/include/clang/3.5.0 +OLD_FILES+=usr/include/clang/3.5.1/__wmmintrin_aes.h +OLD_FILES+=usr/include/clang/3.5.1/__wmmintrin_pclmul.h +OLD_FILES+=usr/include/clang/3.5.1/altivec.h +OLD_FILES+=usr/include/clang/3.5.1/ammintrin.h +OLD_FILES+=usr/include/clang/3.5.1/arm_acle.h +OLD_FILES+=usr/include/clang/3.5.1/arm_neon.h +OLD_FILES+=usr/include/clang/3.5.1/avx2intrin.h +OLD_FILES+=usr/include/clang/3.5.1/avxintrin.h +OLD_FILES+=usr/include/clang/3.5.1/bmi2intrin.h +OLD_FILES+=usr/include/clang/3.5.1/bmiintrin.h +OLD_FILES+=usr/include/clang/3.5.1/cpuid.h +OLD_FILES+=usr/include/clang/3.5.1/emmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/f16cintrin.h +OLD_FILES+=usr/include/clang/3.5.1/fma4intrin.h +OLD_FILES+=usr/include/clang/3.5.1/fmaintrin.h +OLD_FILES+=usr/include/clang/3.5.1/ia32intrin.h +OLD_FILES+=usr/include/clang/3.5.1/immintrin.h +OLD_FILES+=usr/include/clang/3.5.1/lzcntintrin.h +OLD_FILES+=usr/include/clang/3.5.1/mm3dnow.h +OLD_FILES+=usr/include/clang/3.5.1/mm_malloc.h +OLD_FILES+=usr/include/clang/3.5.1/mmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/module.modulemap +OLD_FILES+=usr/include/clang/3.5.1/nmmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/pmmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/popcntintrin.h +OLD_FILES+=usr/include/clang/3.5.1/prfchwintrin.h +OLD_FILES+=usr/include/clang/3.5.1/rdseedintrin.h +OLD_FILES+=usr/include/clang/3.5.1/rtmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/shaintrin.h +OLD_FILES+=usr/include/clang/3.5.1/smmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/tbmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/tmmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/wmmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/x86intrin.h +OLD_FILES+=usr/include/clang/3.5.1/xmmintrin.h +OLD_FILES+=usr/include/clang/3.5.1/xopintrin.h +OLD_DIRS+=usr/include/clang/3.5.1 OLD_DIRS+=usr/include/clang -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.asan-i386.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.asan-x86_64.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.asan_cxx-i386.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.asan_cxx-x86_64.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.profile-arm.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.profile-i386.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.profile-x86_64.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.san-i386.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.san-x86_64.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.ubsan-i386.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.ubsan-x86_64.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.ubsan_cxx-i386.a -OLD_FILES+=usr/lib/clang/3.5.0/lib/freebsd/libclang_rt.ubsan_cxx-x86_64.a -OLD_DIRS+=usr/lib/clang/3.5.0/lib/freebsd -OLD_DIRS+=usr/lib/clang/3.5.0/lib -OLD_DIRS+=usr/lib/clang/3.5.0 +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.asan-i386.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.asan-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.asan_cxx-i386.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.asan_cxx-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.profile-arm.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.profile-i386.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.profile-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.san-i386.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.san-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.ubsan-i386.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.ubsan-x86_64.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.ubsan_cxx-i386.a +OLD_FILES+=usr/lib/clang/3.5.1/lib/freebsd/libclang_rt.ubsan_cxx-x86_64.a +OLD_DIRS+=usr/lib/clang/3.5.1/lib/freebsd +OLD_DIRS+=usr/lib/clang/3.5.1/lib +OLD_DIRS+=usr/lib/clang/3.5.1 OLD_DIRS+=usr/lib/clang OLD_FILES+=usr/share/doc/llvm/clang/LICENSE.TXT OLD_DIRS+=usr/share/doc/llvm/clang |
