diff options
Diffstat (limited to 'lib/Transforms/Instrumentation/PGOInstrumentation.cpp')
-rw-r--r-- | lib/Transforms/Instrumentation/PGOInstrumentation.cpp | 220 |
1 files changed, 77 insertions, 143 deletions
diff --git a/lib/Transforms/Instrumentation/PGOInstrumentation.cpp b/lib/Transforms/Instrumentation/PGOInstrumentation.cpp index 6fec3c9c79ee..ca1bb62389e9 100644 --- a/lib/Transforms/Instrumentation/PGOInstrumentation.cpp +++ b/lib/Transforms/Instrumentation/PGOInstrumentation.cpp @@ -48,6 +48,7 @@ //===----------------------------------------------------------------------===// #include "CFGMST.h" +#include "ValueProfileCollector.h" #include "llvm/ADT/APInt.h" #include "llvm/ADT/ArrayRef.h" #include "llvm/ADT/STLExtras.h" @@ -61,7 +62,6 @@ #include "llvm/Analysis/BlockFrequencyInfo.h" #include "llvm/Analysis/BranchProbabilityInfo.h" #include "llvm/Analysis/CFG.h" -#include "llvm/Analysis/IndirectCallVisitor.h" #include "llvm/Analysis/LoopInfo.h" #include "llvm/Analysis/OptimizationRemarkEmitter.h" #include "llvm/Analysis/ProfileSummaryInfo.h" @@ -96,6 +96,7 @@ #include "llvm/ProfileData/InstrProf.h" #include "llvm/ProfileData/InstrProfReader.h" #include "llvm/Support/BranchProbability.h" +#include "llvm/Support/CRC.h" #include "llvm/Support/Casting.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/DOTGraphTraits.h" @@ -103,11 +104,11 @@ #include "llvm/Support/Error.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Support/GraphWriter.h" -#include "llvm/Support/JamCRC.h" #include "llvm/Support/raw_ostream.h" #include "llvm/Transforms/Instrumentation.h" #include "llvm/Transforms/Instrumentation/PGOInstrumentation.h" #include "llvm/Transforms/Utils/BasicBlockUtils.h" +#include "llvm/Transforms/Utils/MisExpect.h" #include <algorithm> #include <cassert> #include <cstdint> @@ -120,6 +121,7 @@ using namespace llvm; using ProfileCount = Function::ProfileCount; +using VPCandidateInfo = ValueProfileCollector::CandidateInfo; #define DEBUG_TYPE "pgo-instrumentation" @@ -286,6 +288,11 @@ static std::string getBranchCondString(Instruction *TI) { return result; } +static const char *ValueProfKindDescr[] = { +#define VALUE_PROF_KIND(Enumerator, Value, Descr) Descr, +#include "llvm/ProfileData/InstrProfData.inc" +}; + namespace { /// The select instruction visitor plays three roles specified @@ -348,50 +355,6 @@ struct SelectInstVisitor : public InstVisitor<SelectInstVisitor> { unsigned getNumOfSelectInsts() const { return NSIs; } }; -/// Instruction Visitor class to visit memory intrinsic calls. -struct MemIntrinsicVisitor : public InstVisitor<MemIntrinsicVisitor> { - Function &F; - unsigned NMemIs = 0; // Number of memIntrinsics instrumented. - VisitMode Mode = VM_counting; // Visiting mode. - unsigned CurCtrId = 0; // Current counter index. - unsigned TotalNumCtrs = 0; // Total number of counters - GlobalVariable *FuncNameVar = nullptr; - uint64_t FuncHash = 0; - PGOUseFunc *UseFunc = nullptr; - std::vector<Instruction *> Candidates; - - MemIntrinsicVisitor(Function &Func) : F(Func) {} - - void countMemIntrinsics(Function &Func) { - NMemIs = 0; - Mode = VM_counting; - visit(Func); - } - - void instrumentMemIntrinsics(Function &Func, unsigned TotalNC, - GlobalVariable *FNV, uint64_t FHash) { - Mode = VM_instrument; - TotalNumCtrs = TotalNC; - FuncHash = FHash; - FuncNameVar = FNV; - visit(Func); - } - - std::vector<Instruction *> findMemIntrinsics(Function &Func) { - Candidates.clear(); - Mode = VM_annotate; - visit(Func); - return Candidates; - } - - // Visit the IR stream and annotate all mem intrinsic call instructions. - void instrumentOneMemIntrinsic(MemIntrinsic &MI); - - // Visit \p MI instruction and perform tasks according to visit mode. - void visitMemIntrinsic(MemIntrinsic &SI); - - unsigned getNumOfMemIntrinsics() const { return NMemIs; } -}; class PGOInstrumentationGenLegacyPass : public ModulePass { public: @@ -563,13 +526,14 @@ private: // A map that stores the Comdat group in function F. std::unordered_multimap<Comdat *, GlobalValue *> &ComdatMembers; + ValueProfileCollector VPC; + void computeCFGHash(); void renameComdatFunction(); public: - std::vector<std::vector<Instruction *>> ValueSites; + std::vector<std::vector<VPCandidateInfo>> ValueSites; SelectInstVisitor SIVisitor; - MemIntrinsicVisitor MIVisitor; std::string FuncName; GlobalVariable *FuncNameVar; @@ -604,23 +568,21 @@ public: std::unordered_multimap<Comdat *, GlobalValue *> &ComdatMembers, bool CreateGlobalVar = false, BranchProbabilityInfo *BPI = nullptr, BlockFrequencyInfo *BFI = nullptr, bool IsCS = false) - : F(Func), IsCS(IsCS), ComdatMembers(ComdatMembers), - ValueSites(IPVK_Last + 1), SIVisitor(Func), MIVisitor(Func), - MST(F, BPI, BFI) { + : F(Func), IsCS(IsCS), ComdatMembers(ComdatMembers), VPC(Func), + ValueSites(IPVK_Last + 1), SIVisitor(Func), MST(F, BPI, BFI) { // This should be done before CFG hash computation. SIVisitor.countSelects(Func); - MIVisitor.countMemIntrinsics(Func); + ValueSites[IPVK_MemOPSize] = VPC.get(IPVK_MemOPSize); if (!IsCS) { NumOfPGOSelectInsts += SIVisitor.getNumOfSelectInsts(); - NumOfPGOMemIntrinsics += MIVisitor.getNumOfMemIntrinsics(); + NumOfPGOMemIntrinsics += ValueSites[IPVK_MemOPSize].size(); NumOfPGOBB += MST.BBInfos.size(); - ValueSites[IPVK_IndirectCallTarget] = findIndirectCalls(Func); + ValueSites[IPVK_IndirectCallTarget] = VPC.get(IPVK_IndirectCallTarget); } else { NumOfCSPGOSelectInsts += SIVisitor.getNumOfSelectInsts(); - NumOfCSPGOMemIntrinsics += MIVisitor.getNumOfMemIntrinsics(); + NumOfCSPGOMemIntrinsics += ValueSites[IPVK_MemOPSize].size(); NumOfCSPGOBB += MST.BBInfos.size(); } - ValueSites[IPVK_MemOPSize] = MIVisitor.findMemIntrinsics(Func); FuncName = getPGOFuncName(F); computeCFGHash(); @@ -647,7 +609,7 @@ public: // value of each BB in the CFG. The higher 32 bits record the number of edges. template <class Edge, class BBInfo> void FuncPGOInstrumentation<Edge, BBInfo>::computeCFGHash() { - std::vector<char> Indexes; + std::vector<uint8_t> Indexes; JamCRC JC; for (auto &BB : F) { const Instruction *TI = BB.getTerminator(); @@ -658,7 +620,7 @@ void FuncPGOInstrumentation<Edge, BBInfo>::computeCFGHash() { continue; uint32_t Index = BI->Index; for (int J = 0; J < 4; J++) - Indexes.push_back((char)(Index >> (J * 8))); + Indexes.push_back((uint8_t)(Index >> (J * 8))); } } JC.update(Indexes); @@ -874,28 +836,36 @@ static void instrumentOneFunc( if (DisableValueProfiling) return; - unsigned NumIndirectCalls = 0; - for (auto &I : FuncInfo.ValueSites[IPVK_IndirectCallTarget]) { - CallSite CS(I); - Value *Callee = CS.getCalledValue(); - LLVM_DEBUG(dbgs() << "Instrument one indirect call: CallSite Index = " - << NumIndirectCalls << "\n"); - IRBuilder<> Builder(I); - assert(Builder.GetInsertPoint() != I->getParent()->end() && - "Cannot get the Instrumentation point"); - Builder.CreateCall( - Intrinsic::getDeclaration(M, Intrinsic::instrprof_value_profile), - {ConstantExpr::getBitCast(FuncInfo.FuncNameVar, I8PtrTy), - Builder.getInt64(FuncInfo.FunctionHash), - Builder.CreatePtrToInt(Callee, Builder.getInt64Ty()), - Builder.getInt32(IPVK_IndirectCallTarget), - Builder.getInt32(NumIndirectCalls++)}); - } - NumOfPGOICall += NumIndirectCalls; + NumOfPGOICall += FuncInfo.ValueSites[IPVK_IndirectCallTarget].size(); - // Now instrument memop intrinsic calls. - FuncInfo.MIVisitor.instrumentMemIntrinsics( - F, NumCounters, FuncInfo.FuncNameVar, FuncInfo.FunctionHash); + // For each VP Kind, walk the VP candidates and instrument each one. + for (uint32_t Kind = IPVK_First; Kind <= IPVK_Last; ++Kind) { + unsigned SiteIndex = 0; + if (Kind == IPVK_MemOPSize && !PGOInstrMemOP) + continue; + + for (VPCandidateInfo Cand : FuncInfo.ValueSites[Kind]) { + LLVM_DEBUG(dbgs() << "Instrument one VP " << ValueProfKindDescr[Kind] + << " site: CallSite Index = " << SiteIndex << "\n"); + + IRBuilder<> Builder(Cand.InsertPt); + assert(Builder.GetInsertPoint() != Cand.InsertPt->getParent()->end() && + "Cannot get the Instrumentation point"); + + Value *ToProfile = nullptr; + if (Cand.V->getType()->isIntegerTy()) + ToProfile = Builder.CreateZExtOrTrunc(Cand.V, Builder.getInt64Ty()); + else if (Cand.V->getType()->isPointerTy()) + ToProfile = Builder.CreatePtrToInt(Cand.V, Builder.getInt64Ty()); + assert(ToProfile && "value profiling Value is of unexpected type"); + + Builder.CreateCall( + Intrinsic::getDeclaration(M, Intrinsic::instrprof_value_profile), + {ConstantExpr::getBitCast(FuncInfo.FuncNameVar, I8PtrTy), + Builder.getInt64(FuncInfo.FunctionHash), ToProfile, + Builder.getInt32(Kind), Builder.getInt32(SiteIndex++)}); + } + } // IPVK_First <= Kind <= IPVK_Last } namespace { @@ -984,9 +954,9 @@ class PGOUseFunc { public: PGOUseFunc(Function &Func, Module *Modu, std::unordered_multimap<Comdat *, GlobalValue *> &ComdatMembers, - BranchProbabilityInfo *BPI = nullptr, - BlockFrequencyInfo *BFIin = nullptr, bool IsCS = false) - : F(Func), M(Modu), BFI(BFIin), + BranchProbabilityInfo *BPI, BlockFrequencyInfo *BFIin, + ProfileSummaryInfo *PSI, bool IsCS) + : F(Func), M(Modu), BFI(BFIin), PSI(PSI), FuncInfo(Func, ComdatMembers, false, BPI, BFIin, IsCS), FreqAttr(FFA_Normal), IsCS(IsCS) {} @@ -1041,6 +1011,7 @@ private: Function &F; Module *M; BlockFrequencyInfo *BFI; + ProfileSummaryInfo *PSI; // This member stores the shared information with class PGOGenFunc. FuncPGOInstrumentation<PGOUseEdge, UseBBInfo> FuncInfo; @@ -1078,15 +1049,9 @@ private: // FIXME: This function should be removed once the functionality in // the inliner is implemented. void markFunctionAttributes(uint64_t EntryCount, uint64_t MaxCount) { - if (ProgramMaxCount == 0) - return; - // Threshold of the hot functions. - const BranchProbability HotFunctionThreshold(1, 100); - // Threshold of the cold functions. - const BranchProbability ColdFunctionThreshold(2, 10000); - if (EntryCount >= HotFunctionThreshold.scale(ProgramMaxCount)) + if (PSI->isHotCount(EntryCount)) FreqAttr = FFA_Hot; - else if (MaxCount <= ColdFunctionThreshold.scale(ProgramMaxCount)) + else if (PSI->isColdCount(MaxCount)) FreqAttr = FFA_Cold; } }; @@ -1433,43 +1398,6 @@ void SelectInstVisitor::visitSelectInst(SelectInst &SI) { llvm_unreachable("Unknown visiting mode"); } -void MemIntrinsicVisitor::instrumentOneMemIntrinsic(MemIntrinsic &MI) { - Module *M = F.getParent(); - IRBuilder<> Builder(&MI); - Type *Int64Ty = Builder.getInt64Ty(); - Type *I8PtrTy = Builder.getInt8PtrTy(); - Value *Length = MI.getLength(); - assert(!isa<ConstantInt>(Length)); - Builder.CreateCall( - Intrinsic::getDeclaration(M, Intrinsic::instrprof_value_profile), - {ConstantExpr::getBitCast(FuncNameVar, I8PtrTy), - Builder.getInt64(FuncHash), Builder.CreateZExtOrTrunc(Length, Int64Ty), - Builder.getInt32(IPVK_MemOPSize), Builder.getInt32(CurCtrId)}); - ++CurCtrId; -} - -void MemIntrinsicVisitor::visitMemIntrinsic(MemIntrinsic &MI) { - if (!PGOInstrMemOP) - return; - Value *Length = MI.getLength(); - // Not instrument constant length calls. - if (dyn_cast<ConstantInt>(Length)) - return; - - switch (Mode) { - case VM_counting: - NMemIs++; - return; - case VM_instrument: - instrumentOneMemIntrinsic(MI); - return; - case VM_annotate: - Candidates.push_back(&MI); - return; - } - llvm_unreachable("Unknown visiting mode"); -} - // Traverse all valuesites and annotate the instructions for all value kind. void PGOUseFunc::annotateValueSites() { if (DisableValueProfiling) @@ -1482,11 +1410,6 @@ void PGOUseFunc::annotateValueSites() { annotateValueSites(Kind); } -static const char *ValueProfKindDescr[] = { -#define VALUE_PROF_KIND(Enumerator, Value, Descr) Descr, -#include "llvm/ProfileData/InstrProfData.inc" -}; - // Annotate the instructions for a specific value kind. void PGOUseFunc::annotateValueSites(uint32_t Kind) { assert(Kind <= IPVK_Last); @@ -1505,11 +1428,11 @@ void PGOUseFunc::annotateValueSites(uint32_t Kind) { return; } - for (auto &I : ValueSites) { + for (VPCandidateInfo &I : ValueSites) { LLVM_DEBUG(dbgs() << "Read one value site profile (kind = " << Kind << "): Index = " << ValueSiteIndex << " out of " << NumValueSites << "\n"); - annotateValueSite(*M, *I, ProfileRecord, + annotateValueSite(*M, *I.AnnotatedInst, ProfileRecord, static_cast<InstrProfValueKind>(Kind), ValueSiteIndex, Kind == IPVK_MemOPSize ? MaxNumMemOPAnnotations : MaxNumAnnotations); @@ -1595,7 +1518,8 @@ PreservedAnalyses PGOInstrumentationGen::run(Module &M, static bool annotateAllFunctions( Module &M, StringRef ProfileFileName, StringRef ProfileRemappingFileName, function_ref<BranchProbabilityInfo *(Function &)> LookupBPI, - function_ref<BlockFrequencyInfo *(Function &)> LookupBFI, bool IsCS) { + function_ref<BlockFrequencyInfo *(Function &)> LookupBFI, + ProfileSummaryInfo *PSI, bool IsCS) { LLVM_DEBUG(dbgs() << "Read in profile counters: "); auto &Ctx = M.getContext(); // Read the counter array from file. @@ -1626,6 +1550,13 @@ static bool annotateAllFunctions( return false; } + // Add the profile summary (read from the header of the indexed summary) here + // so that we can use it below when reading counters (which checks if the + // function should be marked with a cold or inlinehint attribute). + M.setProfileSummary(PGOReader->getSummary(IsCS).getMD(M.getContext()), + IsCS ? ProfileSummary::PSK_CSInstr + : ProfileSummary::PSK_Instr); + std::unordered_multimap<Comdat *, GlobalValue *> ComdatMembers; collectComdatMembers(M, ComdatMembers); std::vector<Function *> HotFunctions; @@ -1638,7 +1569,7 @@ static bool annotateAllFunctions( // Split indirectbr critical edges here before computing the MST rather than // later in getInstrBB() to avoid invalidating it. SplitIndirectBrCriticalEdges(F, BPI, BFI); - PGOUseFunc Func(F, &M, ComdatMembers, BPI, BFI, IsCS); + PGOUseFunc Func(F, &M, ComdatMembers, BPI, BFI, PSI, IsCS); bool AllZeros = false; if (!Func.readCounters(PGOReader.get(), AllZeros)) continue; @@ -1662,9 +1593,9 @@ static bool annotateAllFunctions( F.getName().equals(ViewBlockFreqFuncName))) { LoopInfo LI{DominatorTree(F)}; std::unique_ptr<BranchProbabilityInfo> NewBPI = - llvm::make_unique<BranchProbabilityInfo>(F, LI); + std::make_unique<BranchProbabilityInfo>(F, LI); std::unique_ptr<BlockFrequencyInfo> NewBFI = - llvm::make_unique<BlockFrequencyInfo>(F, *NewBPI, LI); + std::make_unique<BlockFrequencyInfo>(F, *NewBPI, LI); if (PGOViewCounts == PGOVCT_Graph) NewBFI->view(); else if (PGOViewCounts == PGOVCT_Text) { @@ -1686,9 +1617,6 @@ static bool annotateAllFunctions( } } } - M.setProfileSummary(PGOReader->getSummary(IsCS).getMD(M.getContext()), - IsCS ? ProfileSummary::PSK_CSInstr - : ProfileSummary::PSK_Instr); // Set function hotness attribute from the profile. // We have to apply these attributes at the end because their presence @@ -1730,8 +1658,10 @@ PreservedAnalyses PGOInstrumentationUse::run(Module &M, return &FAM.getResult<BlockFrequencyAnalysis>(F); }; + auto *PSI = &AM.getResult<ProfileSummaryAnalysis>(M); + if (!annotateAllFunctions(M, ProfileFileName, ProfileRemappingFileName, - LookupBPI, LookupBFI, IsCS)) + LookupBPI, LookupBFI, PSI, IsCS)) return PreservedAnalyses::all(); return PreservedAnalyses::none(); @@ -1748,7 +1678,8 @@ bool PGOInstrumentationUseLegacyPass::runOnModule(Module &M) { return &this->getAnalysis<BlockFrequencyInfoWrapperPass>(F).getBFI(); }; - return annotateAllFunctions(M, ProfileFileName, "", LookupBPI, LookupBFI, + auto *PSI = &getAnalysis<ProfileSummaryInfoWrapperPass>().getPSI(); + return annotateAllFunctions(M, ProfileFileName, "", LookupBPI, LookupBFI, PSI, IsCS); } @@ -1776,6 +1707,9 @@ void llvm::setProfMetadata(Module *M, Instruction *TI, : Weights) { dbgs() << W << " "; } dbgs() << "\n";); + + misexpect::verifyMisExpect(TI, Weights, TI->getContext()); + TI->setMetadata(LLVMContext::MD_prof, MDB.createBranchWeights(Weights)); if (EmitBranchProbability) { std::string BrCondStr = getBranchCondString(TI); |