X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTarget%2FARM%2FARMTargetMachine.cpp;h=0778ec372a8ee1bd0d1f43897df162b895887996;hb=d44dd1789ad34be8e9537f544fd3e815d7a4b721;hp=1af622cfdd7e115f3d0e821d574bf0de4653b56f;hpb=71f56c4aac3a6062738f95eeff42ba54f549d0cc;p=oota-llvm.git diff --git a/lib/Target/ARM/ARMTargetMachine.cpp b/lib/Target/ARM/ARMTargetMachine.cpp index 1af622cfdd7..0778ec372a8 100644 --- a/lib/Target/ARM/ARMTargetMachine.cpp +++ b/lib/Target/ARM/ARMTargetMachine.cpp @@ -11,13 +11,14 @@ //===----------------------------------------------------------------------===// #include "ARM.h" -#include "ARMTargetMachine.h" #include "ARMFrameLowering.h" +#include "ARMTargetMachine.h" #include "ARMTargetObjectFile.h" +#include "ARMTargetTransformInfo.h" #include "llvm/CodeGen/Passes.h" #include "llvm/IR/Function.h" +#include "llvm/IR/LegacyPassManager.h" #include "llvm/MC/MCAsmInfo.h" -#include "llvm/PassManager.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/FormattedStream.h" #include "llvm/Support/TargetRegistry.h" @@ -36,6 +37,16 @@ EnableAtomicTidy("arm-atomic-cfg-tidy", cl::Hidden, " to make use of cmpxchg flow-based information"), cl::init(true)); +static cl::opt +EnableARMLoadStoreOpt("arm-load-store-opt", cl::Hidden, + cl::desc("Enable ARM load/store optimization pass"), + cl::init(true)); + +// FIXME: Unify control over GlobalMerge. +static cl::opt +EnableGlobalMerge("arm-global-merge", cl::Hidden, + cl::desc("Enable the global merge pass")); + extern "C" void LLVMInitializeARMTarget() { // Register the target. RegisterTargetMachine X(TheARMLETarget); @@ -52,15 +63,121 @@ static std::unique_ptr createTLOF(const Triple &TT) { return make_unique(); } +static ARMBaseTargetMachine::ARMABI +computeTargetABI(const Triple &TT, StringRef CPU, + const TargetOptions &Options) { + if (Options.MCOptions.getABIName().startswith("aapcs")) + return ARMBaseTargetMachine::ARM_ABI_AAPCS; + else if (Options.MCOptions.getABIName().startswith("apcs")) + return ARMBaseTargetMachine::ARM_ABI_APCS; + + assert(Options.MCOptions.getABIName().empty() && + "Unknown target-abi option!"); + + ARMBaseTargetMachine::ARMABI TargetABI = + ARMBaseTargetMachine::ARM_ABI_UNKNOWN; + + // FIXME: This is duplicated code from the front end and should be unified. + if (TT.isOSBinFormatMachO()) { + if (TT.getEnvironment() == llvm::Triple::EABI || + (TT.getOS() == llvm::Triple::UnknownOS && TT.isOSBinFormatMachO()) || + CPU.startswith("cortex-m")) { + TargetABI = ARMBaseTargetMachine::ARM_ABI_AAPCS; + } else { + TargetABI = ARMBaseTargetMachine::ARM_ABI_APCS; + } + } else if (TT.isOSWindows()) { + // FIXME: this is invalid for WindowsCE + TargetABI = ARMBaseTargetMachine::ARM_ABI_AAPCS; + } else { + // Select the default based on the platform. + switch (TT.getEnvironment()) { + case llvm::Triple::Android: + case llvm::Triple::GNUEABI: + case llvm::Triple::GNUEABIHF: + case llvm::Triple::EABIHF: + case llvm::Triple::EABI: + TargetABI = ARMBaseTargetMachine::ARM_ABI_AAPCS; + break; + case llvm::Triple::GNU: + TargetABI = ARMBaseTargetMachine::ARM_ABI_APCS; + break; + default: + if (TT.isOSNetBSD()) + TargetABI = ARMBaseTargetMachine::ARM_ABI_APCS; + else + TargetABI = ARMBaseTargetMachine::ARM_ABI_AAPCS; + break; + } + } + + return TargetABI; +} + +static std::string computeDataLayout(const Triple &TT, StringRef CPU, + const TargetOptions &Options, + bool isLittle) { + auto ABI = computeTargetABI(TT, CPU, Options); + std::string Ret = ""; + + if (isLittle) + // Little endian. + Ret += "e"; + else + // Big endian. + Ret += "E"; + + Ret += DataLayout::getManglingComponent(TT); + + // Pointers are 32 bits and aligned to 32 bits. + Ret += "-p:32:32"; + + // ABIs other than APCS have 64 bit integers with natural alignment. + if (ABI != ARMBaseTargetMachine::ARM_ABI_APCS) + Ret += "-i64:64"; + + // We have 64 bits floats. The APCS ABI requires them to be aligned to 32 + // bits, others to 64 bits. We always try to align to 64 bits. + if (ABI == ARMBaseTargetMachine::ARM_ABI_APCS) + Ret += "-f64:32:64"; + + // We have 128 and 64 bit vectors. The APCS ABI aligns them to 32 bits, others + // to 64. We always ty to give them natural alignment. + if (ABI == ARMBaseTargetMachine::ARM_ABI_APCS) + Ret += "-v64:32:64-v128:32:128"; + else + Ret += "-v128:64:128"; + + // Try to align aggregates to 32 bits (the default is 64 bits, which has no + // particular hardware support on 32-bit ARM). + Ret += "-a:0:32"; + + // Integer registers are 32 bits. + Ret += "-n32"; + + // The stack is 128 bit aligned on NaCl, 64 bit aligned on AAPCS and 32 bit + // aligned everywhere else. + if (TT.isOSNaCl()) + Ret += "-S128"; + else if (ABI == ARMBaseTargetMachine::ARM_ABI_AAPCS) + Ret += "-S64"; + else + Ret += "-S32"; + + return Ret; +} + /// TargetMachine ctor - Create an ARM architecture model. /// -ARMBaseTargetMachine::ARMBaseTargetMachine(const Target &T, StringRef TT, +ARMBaseTargetMachine::ARMBaseTargetMachine(const Target &T, const Triple &TT, StringRef CPU, StringRef FS, const TargetOptions &Options, Reloc::Model RM, CodeModel::Model CM, CodeGenOpt::Level OL, bool isLittle) - : LLVMTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL), - TLOF(createTLOF(Triple(getTargetTriple()))), + : LLVMTargetMachine(T, computeDataLayout(TT, CPU, Options, isLittle), TT, + CPU, FS, Options, RM, CM, OL), + TargetABI(computeTargetABI(TT, CPU, Options)), + TLOF(createTLOF(getTargetTriple())), Subtarget(TT, CPU, FS, *this, isLittle), isLittle(isLittle) { // Default to triple-appropriate float ABI @@ -73,11 +190,8 @@ ARMBaseTargetMachine::~ARMBaseTargetMachine() {} const ARMSubtarget * ARMBaseTargetMachine::getSubtargetImpl(const Function &F) const { - AttributeSet FnAttrs = F.getAttributes(); - Attribute CPUAttr = - FnAttrs.getAttribute(AttributeSet::FunctionIndex, "target-cpu"); - Attribute FSAttr = - FnAttrs.getAttribute(AttributeSet::FunctionIndex, "target-features"); + Attribute CPUAttr = F.getFnAttribute("target-cpu"); + Attribute FSAttr = F.getFnAttribute("target-features"); std::string CPU = !CPUAttr.hasAttribute(Attribute::None) ? CPUAttr.getValueAsString().str() @@ -91,14 +205,15 @@ ARMBaseTargetMachine::getSubtargetImpl(const Function &F) const { // function before we can generate a subtarget. We also need to use // it as a key for the subtarget since that can be the only difference // between two functions. - Attribute SFAttr = - FnAttrs.getAttribute(AttributeSet::FunctionIndex, "use-soft-float"); - bool SoftFloat = !SFAttr.hasAttribute(Attribute::None) - ? SFAttr.getValueAsString() == "true" - : Options.UseSoftFloat; - - auto &I = SubtargetMap[CPU + FS + (SoftFloat ? "use-soft-float=true" - : "use-soft-float=false")]; + bool SoftFloat = + F.hasFnAttribute("use-soft-float") && + F.getFnAttribute("use-soft-float").getValueAsString() == "true"; + // If the soft float attribute is set on the function turn on the soft float + // subtarget feature. + if (SoftFloat) + FS += FS.empty() ? "+soft-float" : ",+soft-float"; + + auto &I = SubtargetMap[CPU + FS]; if (!I) { // This needs to be done before we create a new subtarget since any // creation will depend on the TM and the code generation flags on the @@ -109,19 +224,17 @@ ARMBaseTargetMachine::getSubtargetImpl(const Function &F) const { return I.get(); } -void ARMBaseTargetMachine::addAnalysisPasses(PassManagerBase &PM) { - // Add first the target-independent BasicTTI pass, then our ARM pass. This - // allows the ARM pass to delegate to the target independent layer when - // appropriate. - PM.add(createBasicTargetTransformInfoPass(this)); - PM.add(createARMTargetTransformInfoPass(this)); +TargetIRAnalysis ARMBaseTargetMachine::getTargetIRAnalysis() { + return TargetIRAnalysis( + [this](Function &F) { return TargetTransformInfo(ARMTTIImpl(this, F)); }); } void ARMTargetMachine::anchor() { } -ARMTargetMachine::ARMTargetMachine(const Target &T, StringRef TT, StringRef CPU, - StringRef FS, const TargetOptions &Options, +ARMTargetMachine::ARMTargetMachine(const Target &T, const Triple &TT, + StringRef CPU, StringRef FS, + const TargetOptions &Options, Reloc::Model RM, CodeModel::Model CM, CodeGenOpt::Level OL, bool isLittle) : ARMBaseTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, isLittle) { @@ -133,7 +246,7 @@ ARMTargetMachine::ARMTargetMachine(const Target &T, StringRef TT, StringRef CPU, void ARMLETargetMachine::anchor() { } -ARMLETargetMachine::ARMLETargetMachine(const Target &T, StringRef TT, +ARMLETargetMachine::ARMLETargetMachine(const Target &T, const Triple &TT, StringRef CPU, StringRef FS, const TargetOptions &Options, Reloc::Model RM, CodeModel::Model CM, @@ -142,7 +255,7 @@ ARMLETargetMachine::ARMLETargetMachine(const Target &T, StringRef TT, void ARMBETargetMachine::anchor() { } -ARMBETargetMachine::ARMBETargetMachine(const Target &T, StringRef TT, +ARMBETargetMachine::ARMBETargetMachine(const Target &T, const Triple &TT, StringRef CPU, StringRef FS, const TargetOptions &Options, Reloc::Model RM, CodeModel::Model CM, @@ -151,19 +264,18 @@ ARMBETargetMachine::ARMBETargetMachine(const Target &T, StringRef TT, void ThumbTargetMachine::anchor() { } -ThumbTargetMachine::ThumbTargetMachine(const Target &T, StringRef TT, +ThumbTargetMachine::ThumbTargetMachine(const Target &T, const Triple &TT, StringRef CPU, StringRef FS, const TargetOptions &Options, Reloc::Model RM, CodeModel::Model CM, CodeGenOpt::Level OL, bool isLittle) - : ARMBaseTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, - isLittle) { + : ARMBaseTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, isLittle) { initAsmInfo(); } void ThumbLETargetMachine::anchor() { } -ThumbLETargetMachine::ThumbLETargetMachine(const Target &T, StringRef TT, +ThumbLETargetMachine::ThumbLETargetMachine(const Target &T, const Triple &TT, StringRef CPU, StringRef FS, const TargetOptions &Options, Reloc::Model RM, CodeModel::Model CM, @@ -172,7 +284,7 @@ ThumbLETargetMachine::ThumbLETargetMachine(const Target &T, StringRef TT, void ThumbBETargetMachine::anchor() { } -ThumbBETargetMachine::ThumbBETargetMachine(const Target &T, StringRef TT, +ThumbBETargetMachine::ThumbBETargetMachine(const Target &T, const Triple &TT, StringRef CPU, StringRef FS, const TargetOptions &Options, Reloc::Model RM, CodeModel::Model CM, @@ -190,10 +302,6 @@ public: return getTM(); } - const ARMSubtarget &getARMSubtarget() const { - return *getARMTargetMachine().getSubtargetImpl(); - } - void addIRPasses() override; bool addPreISel() override; bool addInstSelector() override; @@ -216,17 +324,38 @@ void ARMPassConfig::addIRPasses() { // Cmpxchg instructions are often used with a subsequent comparison to // determine whether it succeeded. We can exploit existing control-flow in // ldrex/strex loops to simplify this, but it needs tidying up. - const ARMSubtarget *Subtarget = &getARMSubtarget(); - if (Subtarget->hasAnyDataBarrier() && !Subtarget->isThumb1Only()) - if (TM->getOptLevel() != CodeGenOpt::None && EnableAtomicTidy) - addPass(createCFGSimplificationPass()); + if (TM->getOptLevel() != CodeGenOpt::None && EnableAtomicTidy) + addPass(createCFGSimplificationPass(-1, [this](const Function &F) { + const auto &ST = this->TM->getSubtarget(F); + return ST.hasAnyDataBarrier() && !ST.isThumb1Only(); + })); TargetPassConfig::addIRPasses(); + + // Match interleaved memory accesses to ldN/stN intrinsics. + if (TM->getOptLevel() != CodeGenOpt::None) + addPass(createInterleavedAccessPass(TM)); } bool ARMPassConfig::addPreISel() { - if (TM->getOptLevel() != CodeGenOpt::None) - addPass(createGlobalMergePass(TM)); + if ((TM->getOptLevel() != CodeGenOpt::None && + EnableGlobalMerge == cl::BOU_UNSET) || + EnableGlobalMerge == cl::BOU_TRUE) { + // FIXME: This is using the thumb1 only constant value for + // maximal global offset for merging globals. We may want + // to look into using the old value for non-thumb1 code of + // 4095 based on the TargetMachine, but this starts to become + // tricky when doing code gen per function. + bool OnlyOptimizeForSize = (TM->getOptLevel() < CodeGenOpt::Aggressive) && + (EnableGlobalMerge == cl::BOU_UNSET); + // Merging of extern globals is enabled by default on non-Mach-O as we + // expect it to be generally either beneficial or harmless. On Mach-O it + // is disabled as we emit the .subsections_via_symbols directive which + // means that merging extern globals is not safe. + bool MergeExternalByDefault = !TM->getTargetTriple().isOSBinFormatMachO(); + addPass(createGlobalMergePass(TM, 127, OnlyOptimizeForSize, + MergeExternalByDefault)); + } return false; } @@ -234,60 +363,59 @@ bool ARMPassConfig::addPreISel() { bool ARMPassConfig::addInstSelector() { addPass(createARMISelDag(getARMTargetMachine(), getOptLevel())); - const ARMSubtarget *Subtarget = &getARMSubtarget(); - if (Subtarget->isTargetELF() && !Subtarget->isThumb1Only() && - TM->Options.EnableFastISel) + if (TM->getTargetTriple().isOSBinFormatELF() && TM->Options.EnableFastISel) addPass(createARMGlobalBaseRegPass()); return false; } void ARMPassConfig::addPreRegAlloc() { - if (getOptLevel() != CodeGenOpt::None) - addPass(createARMLoadStoreOptimizationPass(true), false); - if (getOptLevel() != CodeGenOpt::None && getARMSubtarget().isCortexA9()) - addPass(createMLxExpansionPass(), false); - // Since the A15SDOptimizer pass can insert VDUP instructions, it can only be - // enabled when NEON is available. - if (getOptLevel() != CodeGenOpt::None && getARMSubtarget().isCortexA15() && - getARMSubtarget().hasNEON() && !DisableA15SDOptimization) { - addPass(createA15SDOptimizerPass()); + if (getOptLevel() != CodeGenOpt::None) { + addPass(createMLxExpansionPass()); + + if (EnableARMLoadStoreOpt) + addPass(createARMLoadStoreOptimizationPass(/* pre-register alloc */ true)); + + if (!DisableA15SDOptimization) + addPass(createA15SDOptimizerPass()); } } void ARMPassConfig::addPreSched2() { if (getOptLevel() != CodeGenOpt::None) { - addPass(createARMLoadStoreOptimizationPass(), false); + if (EnableARMLoadStoreOpt) + addPass(createARMLoadStoreOptimizationPass()); - if (getARMSubtarget().hasNEON()) - addPass(createExecutionDependencyFixPass(&ARM::DPRRegClass), false); + addPass(createExecutionDependencyFixPass(&ARM::DPRRegClass)); } // Expand some pseudo instructions into multiple instructions to allow // proper scheduling. - addPass(createARMExpandPseudoPass(), false); + addPass(createARMExpandPseudoPass()); if (getOptLevel() != CodeGenOpt::None) { - if (!getARMSubtarget().isThumb1Only()) { - // in v8, IfConversion depends on Thumb instruction widths - if (getARMSubtarget().restrictIT() && - !getARMSubtarget().prefers32BitThumb()) - addPass(createThumb2SizeReductionPass(), false); - addPass(&IfConverterID, false); - } + // in v8, IfConversion depends on Thumb instruction widths + addPass(createThumb2SizeReductionPass([this](const Function &F) { + return this->TM->getSubtarget(F).restrictIT(); + })); + + addPass(createIfConverter([this](const Function &F) { + return !this->TM->getSubtarget(F).isThumb1Only(); + })); } - if (getARMSubtarget().isThumb2()) - addPass(createThumb2ITBlockPass()); + addPass(createThumb2ITBlockPass()); } void ARMPassConfig::addPreEmitPass() { - if (getARMSubtarget().isThumb2()) { - if (!getARMSubtarget().prefers32BitThumb()) - addPass(createThumb2SizeReductionPass(), false); + addPass(createThumb2SizeReductionPass()); - // Constant island pass work on unbundled instructions. - addPass(&UnpackMachineBundlesID, false); - } + // Constant island pass work on unbundled instructions. + addPass(createUnpackMachineBundles([this](const Function &F) { + return this->TM->getSubtarget(F).isThumb2(); + })); + + // Don't optimize barriers at -O0. + if (getOptLevel() != CodeGenOpt::None) + addPass(createARMOptimizeBarriersPass()); - addPass(createARMOptimizeBarriersPass(), false); addPass(createARMConstantIslandPass()); }