//===- SampleProfile.cpp - Incorporate sample profiles into the IR --------===// // // The LLVM Compiler Infrastructure // // This file is distributed under the University of Illinois Open Source // License. See LICENSE.TXT for details. // //===----------------------------------------------------------------------===// // // This file implements the SampleProfileLoader transformation. This pass // reads a profile file generated by a sampling profiler (e.g. Linux Perf - // http://perf.wiki.kernel.org/) and generates IR metadata to reflect the // profile information in the given profile. // // This pass generates branch weight annotations on the IR: // // - prof: Represents branch weights. This annotation is added to branches // to indicate the weights of each edge coming out of the branch. // The weight of each edge is the weight of the target block for // that edge. The weight of a block B is computed as the maximum // number of samples found in B. // //===----------------------------------------------------------------------===// #define DEBUG_TYPE "sample-profile" #include "llvm/Transforms/Scalar.h" #include "llvm/ADT/DenseMap.h" #include "llvm/ADT/OwningPtr.h" #include "llvm/ADT/SmallPtrSet.h" #include "llvm/ADT/SmallSet.h" #include "llvm/ADT/StringMap.h" #include "llvm/ADT/StringRef.h" #include "llvm/Analysis/LoopInfo.h" #include "llvm/Analysis/PostDominators.h" #include "llvm/DebugInfo.h" #include "llvm/IR/Constants.h" #include "llvm/IR/Dominators.h" #include "llvm/IR/Function.h" #include "llvm/IR/Instructions.h" #include "llvm/IR/LLVMContext.h" #include "llvm/IR/MDBuilder.h" #include "llvm/IR/Metadata.h" #include "llvm/IR/Module.h" #include "llvm/Pass.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/Debug.h" #include "llvm/Support/InstIterator.h" #include "llvm/Support/LineIterator.h" #include "llvm/Support/MemoryBuffer.h" #include "llvm/Support/Regex.h" #include "llvm/Support/raw_ostream.h" using namespace llvm; // Command line option to specify the file to read samples from. This is // mainly used for debugging. static cl::opt<std::string> SampleProfileFile( "sample-profile-file", cl::init(""), cl::value_desc("filename"), cl::desc("Profile file loaded by -sample-profile"), cl::Hidden); static cl::opt<unsigned> SampleProfileMaxPropagateIterations( "sample-profile-max-propagate-iterations", cl::init(100), cl::desc("Maximum number of iterations to go through when propagating " "sample block/edge weights through the CFG.")); namespace { typedef DenseMap<uint32_t, uint32_t> BodySampleMap; typedef DenseMap<BasicBlock *, uint32_t> BlockWeightMap; typedef DenseMap<BasicBlock *, BasicBlock *> EquivalenceClassMap; typedef std::pair<BasicBlock *, BasicBlock *> Edge; typedef DenseMap<Edge, uint32_t> EdgeWeightMap; typedef DenseMap<BasicBlock *, SmallVector<BasicBlock *, 8> > BlockEdgeMap; /// \brief Representation of the runtime profile for a function. /// /// This data structure contains the runtime profile for a given /// function. It contains the total number of samples collected /// in the function and a map of samples collected in every statement. class SampleFunctionProfile { public: SampleFunctionProfile() : TotalSamples(0), TotalHeadSamples(0), HeaderLineno(0), DT(0), PDT(0), LI(0) {} unsigned getFunctionLoc(Function &F); bool emitAnnotations(Function &F, DominatorTree *DomTree, PostDominatorTree *PostDomTree, LoopInfo *Loops); uint32_t getInstWeight(Instruction &I); uint32_t getBlockWeight(BasicBlock *B); void addTotalSamples(unsigned Num) { TotalSamples += Num; } void addHeadSamples(unsigned Num) { TotalHeadSamples += Num; } void addBodySamples(unsigned LineOffset, unsigned Num) { BodySamples[LineOffset] += Num; } void print(raw_ostream &OS); void printEdgeWeight(raw_ostream &OS, Edge E); void printBlockWeight(raw_ostream &OS, BasicBlock *BB); void printBlockEquivalence(raw_ostream &OS, BasicBlock *BB); bool computeBlockWeights(Function &F); void findEquivalenceClasses(Function &F); void findEquivalencesFor(BasicBlock *BB1, SmallVector<BasicBlock *, 8> Descendants, DominatorTreeBase<BasicBlock> *DomTree); void propagateWeights(Function &F); uint32_t visitEdge(Edge E, unsigned *NumUnknownEdges, Edge *UnknownEdge); void buildEdges(Function &F); bool propagateThroughEdges(Function &F); bool empty() { return BodySamples.empty(); } protected: /// \brief Total number of samples collected inside this function. /// /// Samples are cumulative, they include all the samples collected /// inside this function and all its inlined callees. unsigned TotalSamples; /// \brief Total number of samples collected at the head of the function. /// FIXME: Use head samples to estimate a cold/hot attribute for the function. unsigned TotalHeadSamples; /// \brief Line number for the function header. Used to compute relative /// line numbers from the absolute line LOCs found in instruction locations. /// The relative line numbers are needed to address the samples from the /// profile file. unsigned HeaderLineno; /// \brief Map line offsets to collected samples. /// /// Each entry in this map contains the number of samples /// collected at the corresponding line offset. All line locations /// are an offset from the start of the function. BodySampleMap BodySamples; /// \brief Map basic blocks to their computed weights. /// /// The weight of a basic block is defined to be the maximum /// of all the instruction weights in that block. BlockWeightMap BlockWeights; /// \brief Map edges to their computed weights. /// /// Edge weights are computed by propagating basic block weights in /// SampleProfile::propagateWeights. EdgeWeightMap EdgeWeights; /// \brief Set of visited blocks during propagation. SmallPtrSet<BasicBlock *, 128> VisitedBlocks; /// \brief Set of visited edges during propagation. SmallSet<Edge, 128> VisitedEdges; /// \brief Equivalence classes for block weights. /// /// Two blocks BB1 and BB2 are in the same equivalence class if they /// dominate and post-dominate each other, and they are in the same loop /// nest. When this happens, the two blocks are guaranteed to execute /// the same number of times. EquivalenceClassMap EquivalenceClass; /// \brief Dominance, post-dominance and loop information. DominatorTree *DT; PostDominatorTree *PDT; LoopInfo *LI; /// \brief Predecessors for each basic block in the CFG. BlockEdgeMap Predecessors; /// \brief Successors for each basic block in the CFG. BlockEdgeMap Successors; }; /// \brief Sample-based profile reader. /// /// Each profile contains sample counts for all the functions /// executed. Inside each function, statements are annotated with the /// collected samples on all the instructions associated with that /// statement. /// /// For this to produce meaningful data, the program needs to be /// compiled with some debug information (at minimum, line numbers: /// -gline-tables-only). Otherwise, it will be impossible to match IR /// instructions to the line numbers collected by the profiler. /// /// From the profile file, we are interested in collecting the /// following information: /// /// * A list of functions included in the profile (mangled names). /// /// * For each function F: /// 1. The total number of samples collected in F. /// /// 2. The samples collected at each line in F. To provide some /// protection against source code shuffling, line numbers should /// be relative to the start of the function. class SampleModuleProfile { public: SampleModuleProfile(StringRef F) : Profiles(0), Filename(F) {} void dump(); void loadText(); void loadNative() { llvm_unreachable("not implemented"); } void printFunctionProfile(raw_ostream &OS, StringRef FName); void dumpFunctionProfile(StringRef FName); SampleFunctionProfile &getProfile(const Function &F) { return Profiles[F.getName()]; } /// \brief Report a parse error message and stop compilation. void reportParseError(int64_t LineNumber, Twine Msg) const { report_fatal_error(Filename + ":" + Twine(LineNumber) + ": " + Msg + "\n"); } protected: /// \brief Map every function to its associated profile. /// /// The profile of every function executed at runtime is collected /// in the structure SampleFunctionProfile. This maps function objects /// to their corresponding profiles. StringMap<SampleFunctionProfile> Profiles; /// \brief Path name to the file holding the profile data. /// /// The format of this file is defined by each profiler /// independently. If possible, the profiler should have a text /// version of the profile format to be used in constructing test /// cases and debugging. StringRef Filename; }; /// \brief Sample profile pass. /// /// This pass reads profile data from the file specified by /// -sample-profile-file and annotates every affected function with the /// profile information found in that file. class SampleProfileLoader : public FunctionPass { public: // Class identification, replacement for typeinfo static char ID; SampleProfileLoader(StringRef Name = SampleProfileFile) : FunctionPass(ID), Profiler(0), Filename(Name) { initializeSampleProfileLoaderPass(*PassRegistry::getPassRegistry()); } virtual bool doInitialization(Module &M); void dump() { Profiler->dump(); } virtual const char *getPassName() const { return "Sample profile pass"; } virtual bool runOnFunction(Function &F); virtual void getAnalysisUsage(AnalysisUsage &AU) const { AU.setPreservesCFG(); AU.addRequired<LoopInfo>(); AU.addRequired<DominatorTreeWrapperPass>(); AU.addRequired<PostDominatorTree>(); } protected: /// \brief Profile reader object. OwningPtr<SampleModuleProfile> Profiler; /// \brief Name of the profile file to load. StringRef Filename; }; } /// \brief Print this function profile on stream \p OS. /// /// \param OS Stream to emit the output to. void SampleFunctionProfile::print(raw_ostream &OS) { OS << TotalSamples << ", " << TotalHeadSamples << ", " << BodySamples.size() << " sampled lines\n"; for (BodySampleMap::const_iterator SI = BodySamples.begin(), SE = BodySamples.end(); SI != SE; ++SI) OS << "\tline offset: " << SI->first << ", number of samples: " << SI->second << "\n"; OS << "\n"; } /// \brief Print the weight of edge \p E on stream \p OS. /// /// \param OS Stream to emit the output to. /// \param E Edge to print. void SampleFunctionProfile::printEdgeWeight(raw_ostream &OS, Edge E) { OS << "weight[" << E.first->getName() << "->" << E.second->getName() << "]: " << EdgeWeights[E] << "\n"; } /// \brief Print the equivalence class of block \p BB on stream \p OS. /// /// \param OS Stream to emit the output to. /// \param BB Block to print. void SampleFunctionProfile::printBlockEquivalence(raw_ostream &OS, BasicBlock *BB) { BasicBlock *Equiv = EquivalenceClass[BB]; OS << "equivalence[" << BB->getName() << "]: " << ((Equiv) ? EquivalenceClass[BB]->getName() : "NONE") << "\n"; } /// \brief Print the weight of block \p BB on stream \p OS. /// /// \param OS Stream to emit the output to. /// \param BB Block to print. void SampleFunctionProfile::printBlockWeight(raw_ostream &OS, BasicBlock *BB) { OS << "weight[" << BB->getName() << "]: " << BlockWeights[BB] << "\n"; } /// \brief Print the function profile for \p FName on stream \p OS. /// /// \param OS Stream to emit the output to. /// \param FName Name of the function to print. void SampleModuleProfile::printFunctionProfile(raw_ostream &OS, StringRef FName) { OS << "Function: " << FName << ":\n"; Profiles[FName].print(OS); } /// \brief Dump the function profile for \p FName. /// /// \param FName Name of the function to print. void SampleModuleProfile::dumpFunctionProfile(StringRef FName) { printFunctionProfile(dbgs(), FName); } /// \brief Dump all the function profiles found. void SampleModuleProfile::dump() { for (StringMap<SampleFunctionProfile>::const_iterator I = Profiles.begin(), E = Profiles.end(); I != E; ++I) dumpFunctionProfile(I->getKey()); } /// \brief Load samples from a text file. /// /// The file contains a list of samples for every function executed at /// runtime. Each function profile has the following format: /// /// function1:total_samples:total_head_samples /// offset1[.discriminator]: number_of_samples [fn1:num fn2:num ... ] /// offset2[.discriminator]: number_of_samples [fn3:num fn4:num ... ] /// ... /// offsetN[.discriminator]: number_of_samples [fn5:num fn6:num ... ] /// /// Function names must be mangled in order for the profile loader to /// match them in the current translation unit. The two numbers in the /// function header specify how many total samples were accumulated in /// the function (first number), and the total number of samples accumulated /// at the prologue of the function (second number). This head sample /// count provides an indicator of how frequent is the function invoked. /// /// Each sampled line may contain several items. Some are optional /// (marked below): /// /// a- Source line offset. This number represents the line number /// in the function where the sample was collected. The line number /// is always relative to the line where symbol of the function /// is defined. So, if the function has its header at line 280, /// the offset 13 is at line 293 in the file. /// /// b- [OPTIONAL] Discriminator. This is used if the sampled program /// was compiled with DWARF discriminator support /// (http://wiki.dwarfstd.org/index.php?title=Path_Discriminators) /// This is currently only emitted by GCC and we just ignore it. /// /// FIXME: Handle discriminators, since they are needed to distinguish /// multiple control flow within a single source LOC. /// /// c- Number of samples. This is the number of samples collected by /// the profiler at this source location. /// /// d- [OPTIONAL] Potential call targets and samples. If present, this /// line contains a call instruction. This models both direct and /// indirect calls. Each called target is listed together with the /// number of samples. For example, /// /// 130: 7 foo:3 bar:2 baz:7 /// /// The above means that at relative line offset 130 there is a /// call instruction that calls one of foo(), bar() and baz(). With /// baz() being the relatively more frequent call target. /// /// FIXME: This is currently unhandled, but it has a lot of /// potential for aiding the inliner. /// /// /// Since this is a flat profile, a function that shows up more than /// once gets all its samples aggregated across all its instances. /// /// FIXME: flat profiles are too imprecise to provide good optimization /// opportunities. Convert them to context-sensitive profile. /// /// This textual representation is useful to generate unit tests and /// for debugging purposes, but it should not be used to generate /// profiles for large programs, as the representation is extremely /// inefficient. void SampleModuleProfile::loadText() { OwningPtr<MemoryBuffer> Buffer; error_code EC = MemoryBuffer::getFile(Filename, Buffer); if (EC) report_fatal_error("Could not open file " + Filename + ": " + EC.message()); line_iterator LineIt(*Buffer, '#'); // Read the profile of each function. Since each function may be // mentioned more than once, and we are collecting flat profiles, // accumulate samples as we parse them. Regex HeadRE("^([^:]+):([0-9]+):([0-9]+)$"); Regex LineSample("^([0-9]+)(\\.[0-9]+)?: ([0-9]+)(.*)$"); while (!LineIt.is_at_eof()) { // Read the header of each function. The function header should // have this format: // // function_name:total_samples:total_head_samples // // See above for an explanation of each field. SmallVector<StringRef, 3> Matches; if (!HeadRE.match(*LineIt, &Matches)) reportParseError(LineIt.line_number(), "Expected 'mangled_name:NUM:NUM', found " + *LineIt); assert(Matches.size() == 4); StringRef FName = Matches[1]; unsigned NumSamples, NumHeadSamples; Matches[2].getAsInteger(10, NumSamples); Matches[3].getAsInteger(10, NumHeadSamples); Profiles[FName] = SampleFunctionProfile(); SampleFunctionProfile &FProfile = Profiles[FName]; FProfile.addTotalSamples(NumSamples); FProfile.addHeadSamples(NumHeadSamples); ++LineIt; // Now read the body. The body of the function ends when we reach // EOF or when we see the start of the next function. while (!LineIt.is_at_eof() && isdigit((*LineIt)[0])) { if (!LineSample.match(*LineIt, &Matches)) reportParseError( LineIt.line_number(), "Expected 'NUM[.NUM]: NUM[ mangled_name:NUM]*', found " + *LineIt); assert(Matches.size() == 5); unsigned LineOffset, NumSamples; Matches[1].getAsInteger(10, LineOffset); // FIXME: Handle discriminator information (in Matches[2]). Matches[3].getAsInteger(10, NumSamples); // FIXME: Handle called targets (in Matches[4]). // When dealing with instruction weights, we use the value // zero to indicate the absence of a sample. If we read an // actual zero from the profile file, return it as 1 to // avoid the confusion later on. if (NumSamples == 0) NumSamples = 1; FProfile.addBodySamples(LineOffset, NumSamples); ++LineIt; } } } /// \brief Get the weight for an instruction. /// /// The "weight" of an instruction \p Inst is the number of samples /// collected on that instruction at runtime. To retrieve it, we /// need to compute the line number of \p Inst relative to the start of its /// function. We use HeaderLineno to compute the offset. We then /// look up the samples collected for \p Inst using BodySamples. /// /// \param Inst Instruction to query. /// /// \returns The profiled weight of I. uint32_t SampleFunctionProfile::getInstWeight(Instruction &Inst) { unsigned Lineno = Inst.getDebugLoc().getLine(); if (Lineno < HeaderLineno) return 0; unsigned LOffset = Lineno - HeaderLineno; uint32_t Weight = BodySamples.lookup(LOffset); DEBUG(dbgs() << " " << Lineno << ":" << Inst.getDebugLoc().getCol() << ":" << Inst << " (line offset: " << LOffset << " - weight: " << Weight << ")\n"); return Weight; } /// \brief Compute the weight of a basic block. /// /// The weight of basic block \p B is the maximum weight of all the /// instructions in B. The weight of \p B is computed and cached in /// the BlockWeights map. /// /// \param B The basic block to query. /// /// \returns The computed weight of B. uint32_t SampleFunctionProfile::getBlockWeight(BasicBlock *B) { // If we've computed B's weight before, return it. std::pair<BlockWeightMap::iterator, bool> Entry = BlockWeights.insert(std::make_pair(B, 0)); if (!Entry.second) return Entry.first->second; // Otherwise, compute and cache B's weight. uint32_t Weight = 0; for (BasicBlock::iterator I = B->begin(), E = B->end(); I != E; ++I) { uint32_t InstWeight = getInstWeight(*I); if (InstWeight > Weight) Weight = InstWeight; } Entry.first->second = Weight; return Weight; } /// \brief Compute and store the weights of every basic block. /// /// This populates the BlockWeights map by computing /// the weights of every basic block in the CFG. /// /// \param F The function to query. bool SampleFunctionProfile::computeBlockWeights(Function &F) { bool Changed = false; DEBUG(dbgs() << "Block weights\n"); for (Function::iterator B = F.begin(), E = F.end(); B != E; ++B) { uint32_t Weight = getBlockWeight(B); Changed |= (Weight > 0); DEBUG(printBlockWeight(dbgs(), B)); } return Changed; } /// \brief Find equivalence classes for the given block. /// /// This finds all the blocks that are guaranteed to execute the same /// number of times as \p BB1. To do this, it traverses all the the /// descendants of \p BB1 in the dominator or post-dominator tree. /// /// A block BB2 will be in the same equivalence class as \p BB1 if /// the following holds: /// /// 1- \p BB1 is a descendant of BB2 in the opposite tree. So, if BB2 /// is a descendant of \p BB1 in the dominator tree, then BB2 should /// dominate BB1 in the post-dominator tree. /// /// 2- Both BB2 and \p BB1 must be in the same loop. /// /// For every block BB2 that meets those two requirements, we set BB2's /// equivalence class to \p BB1. /// /// \param BB1 Block to check. /// \param Descendants Descendants of \p BB1 in either the dom or pdom tree. /// \param DomTree Opposite dominator tree. If \p Descendants is filled /// with blocks from \p BB1's dominator tree, then /// this is the post-dominator tree, and vice versa. void SampleFunctionProfile::findEquivalencesFor( BasicBlock *BB1, SmallVector<BasicBlock *, 8> Descendants, DominatorTreeBase<BasicBlock> *DomTree) { for (SmallVectorImpl<BasicBlock *>::iterator I = Descendants.begin(), E = Descendants.end(); I != E; ++I) { BasicBlock *BB2 = *I; bool IsDomParent = DomTree->dominates(BB2, BB1); bool IsInSameLoop = LI->getLoopFor(BB1) == LI->getLoopFor(BB2); if (BB1 != BB2 && VisitedBlocks.insert(BB2) && IsDomParent && IsInSameLoop) { EquivalenceClass[BB2] = BB1; // If BB2 is heavier than BB1, make BB2 have the same weight // as BB1. // // Note that we don't worry about the opposite situation here // (when BB2 is lighter than BB1). We will deal with this // during the propagation phase. Right now, we just want to // make sure that BB1 has the largest weight of all the // members of its equivalence set. uint32_t &BB1Weight = BlockWeights[BB1]; uint32_t &BB2Weight = BlockWeights[BB2]; BB1Weight = std::max(BB1Weight, BB2Weight); } } } /// \brief Find equivalence classes. /// /// Since samples may be missing from blocks, we can fill in the gaps by setting /// the weights of all the blocks in the same equivalence class to the same /// weight. To compute the concept of equivalence, we use dominance and loop /// information. Two blocks B1 and B2 are in the same equivalence class if B1 /// dominates B2, B2 post-dominates B1 and both are in the same loop. /// /// \param F The function to query. void SampleFunctionProfile::findEquivalenceClasses(Function &F) { SmallVector<BasicBlock *, 8> DominatedBBs; DEBUG(dbgs() << "\nBlock equivalence classes\n"); // Find equivalence sets based on dominance and post-dominance information. for (Function::iterator B = F.begin(), E = F.end(); B != E; ++B) { BasicBlock *BB1 = B; // Compute BB1's equivalence class once. if (EquivalenceClass.count(BB1)) { DEBUG(printBlockEquivalence(dbgs(), BB1)); continue; } // By default, blocks are in their own equivalence class. EquivalenceClass[BB1] = BB1; // Traverse all the blocks dominated by BB1. We are looking for // every basic block BB2 such that: // // 1- BB1 dominates BB2. // 2- BB2 post-dominates BB1. // 3- BB1 and BB2 are in the same loop nest. // // If all those conditions hold, it means that BB2 is executed // as many times as BB1, so they are placed in the same equivalence // class by making BB2's equivalence class be BB1. DominatedBBs.clear(); DT->getDescendants(BB1, DominatedBBs); findEquivalencesFor(BB1, DominatedBBs, PDT->DT); // Repeat the same logic for all the blocks post-dominated by BB1. // We are looking for every basic block BB2 such that: // // 1- BB1 post-dominates BB2. // 2- BB2 dominates BB1. // 3- BB1 and BB2 are in the same loop nest. // // If all those conditions hold, BB2's equivalence class is BB1. DominatedBBs.clear(); PDT->getDescendants(BB1, DominatedBBs); findEquivalencesFor(BB1, DominatedBBs, DT); DEBUG(printBlockEquivalence(dbgs(), BB1)); } // Assign weights to equivalence classes. // // All the basic blocks in the same equivalence class will execute // the same number of times. Since we know that the head block in // each equivalence class has the largest weight, assign that weight // to all the blocks in that equivalence class. DEBUG(dbgs() << "\nAssign the same weight to all blocks in the same class\n"); for (Function::iterator B = F.begin(), E = F.end(); B != E; ++B) { BasicBlock *BB = B; BasicBlock *EquivBB = EquivalenceClass[BB]; if (BB != EquivBB) BlockWeights[BB] = BlockWeights[EquivBB]; DEBUG(printBlockWeight(dbgs(), BB)); } } /// \brief Visit the given edge to decide if it has a valid weight. /// /// If \p E has not been visited before, we copy to \p UnknownEdge /// and increment the count of unknown edges. /// /// \param E Edge to visit. /// \param NumUnknownEdges Current number of unknown edges. /// \param UnknownEdge Set if E has not been visited before. /// /// \returns E's weight, if known. Otherwise, return 0. uint32_t SampleFunctionProfile::visitEdge(Edge E, unsigned *NumUnknownEdges, Edge *UnknownEdge) { if (!VisitedEdges.count(E)) { (*NumUnknownEdges)++; *UnknownEdge = E; return 0; } return EdgeWeights[E]; } /// \brief Propagate weights through incoming/outgoing edges. /// /// If the weight of a basic block is known, and there is only one edge /// with an unknown weight, we can calculate the weight of that edge. /// /// Similarly, if all the edges have a known count, we can calculate the /// count of the basic block, if needed. /// /// \param F Function to process. /// /// \returns True if new weights were assigned to edges or blocks. bool SampleFunctionProfile::propagateThroughEdges(Function &F) { bool Changed = false; DEBUG(dbgs() << "\nPropagation through edges\n"); for (Function::iterator BI = F.begin(), EI = F.end(); BI != EI; ++BI) { BasicBlock *BB = BI; // Visit all the predecessor and successor edges to determine // which ones have a weight assigned already. Note that it doesn't // matter that we only keep track of a single unknown edge. The // only case we are interested in handling is when only a single // edge is unknown (see setEdgeOrBlockWeight). for (unsigned i = 0; i < 2; i++) { uint32_t TotalWeight = 0; unsigned NumUnknownEdges = 0; Edge UnknownEdge, SelfReferentialEdge; if (i == 0) { // First, visit all predecessor edges. for (size_t I = 0; I < Predecessors[BB].size(); I++) { Edge E = std::make_pair(Predecessors[BB][I], BB); TotalWeight += visitEdge(E, &NumUnknownEdges, &UnknownEdge); if (E.first == E.second) SelfReferentialEdge = E; } } else { // On the second round, visit all successor edges. for (size_t I = 0; I < Successors[BB].size(); I++) { Edge E = std::make_pair(BB, Successors[BB][I]); TotalWeight += visitEdge(E, &NumUnknownEdges, &UnknownEdge); } } // After visiting all the edges, there are three cases that we // can handle immediately: // // - All the edge weights are known (i.e., NumUnknownEdges == 0). // In this case, we simply check that the sum of all the edges // is the same as BB's weight. If not, we change BB's weight // to match. Additionally, if BB had not been visited before, // we mark it visited. // // - Only one edge is unknown and BB has already been visited. // In this case, we can compute the weight of the edge by // subtracting the total block weight from all the known // edge weights. If the edges weight more than BB, then the // edge of the last remaining edge is set to zero. // // - There exists a self-referential edge and the weight of BB is // known. In this case, this edge can be based on BB's weight. // We add up all the other known edges and set the weight on // the self-referential edge as we did in the previous case. // // In any other case, we must continue iterating. Eventually, // all edges will get a weight, or iteration will stop when // it reaches SampleProfileMaxPropagateIterations. if (NumUnknownEdges <= 1) { uint32_t &BBWeight = BlockWeights[BB]; if (NumUnknownEdges == 0) { // If we already know the weight of all edges, the weight of the // basic block can be computed. It should be no larger than the sum // of all edge weights. if (TotalWeight > BBWeight) { BBWeight = TotalWeight; Changed = true; DEBUG(dbgs() << "All edge weights for " << BB->getName() << " known. Set weight for block: "; printBlockWeight(dbgs(), BB);); } if (VisitedBlocks.insert(BB)) Changed = true; } else if (NumUnknownEdges == 1 && VisitedBlocks.count(BB)) { // If there is a single unknown edge and the block has been // visited, then we can compute E's weight. if (BBWeight >= TotalWeight) EdgeWeights[UnknownEdge] = BBWeight - TotalWeight; else EdgeWeights[UnknownEdge] = 0; VisitedEdges.insert(UnknownEdge); Changed = true; DEBUG(dbgs() << "Set weight for edge: "; printEdgeWeight(dbgs(), UnknownEdge)); } } else if (SelfReferentialEdge.first && VisitedBlocks.count(BB)) { uint32_t &BBWeight = BlockWeights[BB]; // We have a self-referential edge and the weight of BB is known. if (BBWeight >= TotalWeight) EdgeWeights[SelfReferentialEdge] = BBWeight - TotalWeight; else EdgeWeights[SelfReferentialEdge] = 0; VisitedEdges.insert(SelfReferentialEdge); Changed = true; DEBUG(dbgs() << "Set self-referential edge weight to: "; printEdgeWeight(dbgs(), SelfReferentialEdge)); } } } return Changed; } /// \brief Build in/out edge lists for each basic block in the CFG. /// /// We are interested in unique edges. If a block B1 has multiple /// edges to another block B2, we only add a single B1->B2 edge. void SampleFunctionProfile::buildEdges(Function &F) { for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) { BasicBlock *B1 = I; // Add predecessors for B1. SmallPtrSet<BasicBlock *, 16> Visited; if (!Predecessors[B1].empty()) llvm_unreachable("Found a stale predecessors list in a basic block."); for (pred_iterator PI = pred_begin(B1), PE = pred_end(B1); PI != PE; ++PI) { BasicBlock *B2 = *PI; if (Visited.insert(B2)) Predecessors[B1].push_back(B2); } // Add successors for B1. Visited.clear(); if (!Successors[B1].empty()) llvm_unreachable("Found a stale successors list in a basic block."); for (succ_iterator SI = succ_begin(B1), SE = succ_end(B1); SI != SE; ++SI) { BasicBlock *B2 = *SI; if (Visited.insert(B2)) Successors[B1].push_back(B2); } } } /// \brief Propagate weights into edges /// /// The following rules are applied to every block B in the CFG: /// /// - If B has a single predecessor/successor, then the weight /// of that edge is the weight of the block. /// /// - If all incoming or outgoing edges are known except one, and the /// weight of the block is already known, the weight of the unknown /// edge will be the weight of the block minus the sum of all the known /// edges. If the sum of all the known edges is larger than B's weight, /// we set the unknown edge weight to zero. /// /// - If there is a self-referential edge, and the weight of the block is /// known, the weight for that edge is set to the weight of the block /// minus the weight of the other incoming edges to that block (if /// known). void SampleFunctionProfile::propagateWeights(Function &F) { bool Changed = true; unsigned i = 0; // Before propagation starts, build, for each block, a list of // unique predecessors and successors. This is necessary to handle // identical edges in multiway branches. Since we visit all blocks and all // edges of the CFG, it is cleaner to build these lists once at the start // of the pass. buildEdges(F); // Propagate until we converge or we go past the iteration limit. while (Changed && i++ < SampleProfileMaxPropagateIterations) { Changed = propagateThroughEdges(F); } // Generate MD_prof metadata for every branch instruction using the // edge weights computed during propagation. DEBUG(dbgs() << "\nPropagation complete. Setting branch weights\n"); MDBuilder MDB(F.getContext()); for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) { BasicBlock *B = I; TerminatorInst *TI = B->getTerminator(); if (TI->getNumSuccessors() == 1) continue; if (!isa<BranchInst>(TI) && !isa<SwitchInst>(TI)) continue; DEBUG(dbgs() << "\nGetting weights for branch at line " << TI->getDebugLoc().getLine() << ":" << TI->getDebugLoc().getCol() << ".\n"); SmallVector<uint32_t, 4> Weights; bool AllWeightsZero = true; for (unsigned I = 0; I < TI->getNumSuccessors(); ++I) { BasicBlock *Succ = TI->getSuccessor(I); Edge E = std::make_pair(B, Succ); uint32_t Weight = EdgeWeights[E]; DEBUG(dbgs() << "\t"; printEdgeWeight(dbgs(), E)); Weights.push_back(Weight); if (Weight != 0) AllWeightsZero = false; } // Only set weights if there is at least one non-zero weight. // In any other case, let the analyzer set weights. if (!AllWeightsZero) { DEBUG(dbgs() << "SUCCESS. Found non-zero weights.\n"); TI->setMetadata(llvm::LLVMContext::MD_prof, MDB.createBranchWeights(Weights)); } else { DEBUG(dbgs() << "SKIPPED. All branch weights are zero.\n"); } } } /// \brief Get the line number for the function header. /// /// This looks up function \p F in the current compilation unit and /// retrieves the line number where the function is defined. This is /// line 0 for all the samples read from the profile file. Every line /// number is relative to this line. /// /// \param F Function object to query. /// /// \returns the line number where \p F is defined. unsigned SampleFunctionProfile::getFunctionLoc(Function &F) { NamedMDNode *CUNodes = F.getParent()->getNamedMetadata("llvm.dbg.cu"); if (CUNodes) { for (unsigned I = 0, E1 = CUNodes->getNumOperands(); I != E1; ++I) { DICompileUnit CU(CUNodes->getOperand(I)); DIArray Subprograms = CU.getSubprograms(); for (unsigned J = 0, E2 = Subprograms.getNumElements(); J != E2; ++J) { DISubprogram Subprogram(Subprograms.getElement(J)); if (Subprogram.describes(&F)) return Subprogram.getLineNumber(); } } } report_fatal_error("No debug information found in function " + F.getName() + "\n"); } /// \brief Generate branch weight metadata for all branches in \p F. /// /// Branch weights are computed out of instruction samples using a /// propagation heuristic. Propagation proceeds in 3 phases: /// /// 1- Assignment of block weights. All the basic blocks in the function /// are initial assigned the same weight as their most frequently /// executed instruction. /// /// 2- Creation of equivalence classes. Since samples may be missing from /// blocks, we can fill in the gaps by setting the weights of all the /// blocks in the same equivalence class to the same weight. To compute /// the concept of equivalence, we use dominance and loop information. /// Two blocks B1 and B2 are in the same equivalence class if B1 /// dominates B2, B2 post-dominates B1 and both are in the same loop. /// /// 3- Propagation of block weights into edges. This uses a simple /// propagation heuristic. The following rules are applied to every /// block B in the CFG: /// /// - If B has a single predecessor/successor, then the weight /// of that edge is the weight of the block. /// /// - If all the edges are known except one, and the weight of the /// block is already known, the weight of the unknown edge will /// be the weight of the block minus the sum of all the known /// edges. If the sum of all the known edges is larger than B's weight, /// we set the unknown edge weight to zero. /// /// - If there is a self-referential edge, and the weight of the block is /// known, the weight for that edge is set to the weight of the block /// minus the weight of the other incoming edges to that block (if /// known). /// /// Since this propagation is not guaranteed to finalize for every CFG, we /// only allow it to proceed for a limited number of iterations (controlled /// by -sample-profile-max-propagate-iterations). /// /// FIXME: Try to replace this propagation heuristic with a scheme /// that is guaranteed to finalize. A work-list approach similar to /// the standard value propagation algorithm used by SSA-CCP might /// work here. /// /// Once all the branch weights are computed, we emit the MD_prof /// metadata on B using the computed values for each of its branches. /// /// \param F The function to query. bool SampleFunctionProfile::emitAnnotations(Function &F, DominatorTree *DomTree, PostDominatorTree *PostDomTree, LoopInfo *Loops) { bool Changed = false; // Initialize invariants used during computation and propagation. HeaderLineno = getFunctionLoc(F); DEBUG(dbgs() << "Line number for the first instruction in " << F.getName() << ": " << HeaderLineno << "\n"); DT = DomTree; PDT = PostDomTree; LI = Loops; // Compute basic block weights. Changed |= computeBlockWeights(F); if (Changed) { // Find equivalence classes. findEquivalenceClasses(F); // Propagate weights to all edges. propagateWeights(F); } return Changed; } char SampleProfileLoader::ID = 0; INITIALIZE_PASS_BEGIN(SampleProfileLoader, "sample-profile", "Sample Profile loader", false, false) INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass) INITIALIZE_PASS_DEPENDENCY(PostDominatorTree) INITIALIZE_PASS_DEPENDENCY(LoopInfo) INITIALIZE_PASS_END(SampleProfileLoader, "sample-profile", "Sample Profile loader", false, false) bool SampleProfileLoader::doInitialization(Module &M) { Profiler.reset(new SampleModuleProfile(Filename)); Profiler->loadText(); return true; } FunctionPass *llvm::createSampleProfileLoaderPass() { return new SampleProfileLoader(SampleProfileFile); } FunctionPass *llvm::createSampleProfileLoaderPass(StringRef Name) { return new SampleProfileLoader(Name); } bool SampleProfileLoader::runOnFunction(Function &F) { DominatorTree *DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree(); PostDominatorTree *PDT = &getAnalysis<PostDominatorTree>(); LoopInfo *LI = &getAnalysis<LoopInfo>(); SampleFunctionProfile &FunctionProfile = Profiler->getProfile(F); if (!FunctionProfile.empty()) return FunctionProfile.emitAnnotations(F, DT, PDT, LI); return false; }