//===- Inliner.cpp - Code common to all inliners --------------------------===// // // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. // See https://llvm.org/LICENSE.txt for license information. // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception // //===----------------------------------------------------------------------===// // // This file implements the mechanics required to implement inlining without // missing any calls and updating the call graph. The decisions of which calls // are profitable to inline are implemented elsewhere. // //===----------------------------------------------------------------------===// #include "llvm/Transforms/IPO/Inliner.h" #include "llvm/ADT/DenseMap.h" #include "llvm/ADT/None.h" #include "llvm/ADT/Optional.h" #include "llvm/ADT/STLExtras.h" #include "llvm/ADT/ScopeExit.h" #include "llvm/ADT/SetVector.h" #include "llvm/ADT/SmallPtrSet.h" #include "llvm/ADT/SmallVector.h" #include "llvm/ADT/Statistic.h" #include "llvm/ADT/StringRef.h" #include "llvm/Analysis/AliasAnalysis.h" #include "llvm/Analysis/AssumptionCache.h" #include "llvm/Analysis/BasicAliasAnalysis.h" #include "llvm/Analysis/BlockFrequencyInfo.h" #include "llvm/Analysis/CGSCCPassManager.h" #include "llvm/Analysis/CallGraph.h" #include "llvm/Analysis/GlobalsModRef.h" #include "llvm/Analysis/InlineAdvisor.h" #include "llvm/Analysis/InlineCost.h" #include "llvm/Analysis/LazyCallGraph.h" #include "llvm/Analysis/OptimizationRemarkEmitter.h" #include "llvm/Analysis/ProfileSummaryInfo.h" #include "llvm/Analysis/TargetLibraryInfo.h" #include "llvm/Analysis/TargetTransformInfo.h" #include "llvm/IR/Attributes.h" #include "llvm/IR/BasicBlock.h" #include "llvm/IR/DataLayout.h" #include "llvm/IR/DebugLoc.h" #include "llvm/IR/DerivedTypes.h" #include "llvm/IR/DiagnosticInfo.h" #include "llvm/IR/Function.h" #include "llvm/IR/InstIterator.h" #include "llvm/IR/Instruction.h" #include "llvm/IR/Instructions.h" #include "llvm/IR/IntrinsicInst.h" #include "llvm/IR/Metadata.h" #include "llvm/IR/Module.h" #include "llvm/IR/PassManager.h" #include "llvm/IR/User.h" #include "llvm/IR/Value.h" #include "llvm/Pass.h" #include "llvm/Support/Casting.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/Debug.h" #include "llvm/Support/raw_ostream.h" #include "llvm/Transforms/Utils/CallPromotionUtils.h" #include "llvm/Transforms/Utils/Cloning.h" #include "llvm/Transforms/Utils/ImportedFunctionsInliningStatistics.h" #include "llvm/Transforms/Utils/Local.h" #include "llvm/Transforms/Utils/ModuleUtils.h" #include #include #include #include #include #include #include using namespace llvm; #define DEBUG_TYPE "inline" STATISTIC(NumInlined, "Number of functions inlined"); STATISTIC(NumCallsDeleted, "Number of call sites deleted, not inlined"); STATISTIC(NumDeleted, "Number of functions deleted because all callers found"); STATISTIC(NumMergedAllocas, "Number of allocas merged together"); /// Flag to disable manual alloca merging. /// /// Merging of allocas was originally done as a stack-size saving technique /// prior to LLVM's code generator having support for stack coloring based on /// lifetime markers. It is now in the process of being removed. To experiment /// with disabling it and relying fully on lifetime marker based stack /// coloring, you can pass this flag to LLVM. static cl::opt DisableInlinedAllocaMerging("disable-inlined-alloca-merging", cl::init(false), cl::Hidden); namespace { enum class InlinerFunctionImportStatsOpts { No = 0, Basic = 1, Verbose = 2, }; } // end anonymous namespace static cl::opt InlinerFunctionImportStats( "inliner-function-import-stats", cl::init(InlinerFunctionImportStatsOpts::No), cl::values(clEnumValN(InlinerFunctionImportStatsOpts::Basic, "basic", "basic statistics"), clEnumValN(InlinerFunctionImportStatsOpts::Verbose, "verbose", "printing of statistics for each inlined function")), cl::Hidden, cl::desc("Enable inliner stats for imported functions")); LegacyInlinerBase::LegacyInlinerBase(char &ID) : CallGraphSCCPass(ID) {} LegacyInlinerBase::LegacyInlinerBase(char &ID, bool InsertLifetime) : CallGraphSCCPass(ID), InsertLifetime(InsertLifetime) {} /// For this class, we declare that we require and preserve the call graph. /// If the derived class implements this method, it should /// always explicitly call the implementation here. void LegacyInlinerBase::getAnalysisUsage(AnalysisUsage &AU) const { AU.addRequired(); AU.addRequired(); AU.addRequired(); getAAResultsAnalysisUsage(AU); CallGraphSCCPass::getAnalysisUsage(AU); } using InlinedArrayAllocasTy = DenseMap>; /// Look at all of the allocas that we inlined through this call site. If we /// have already inlined other allocas through other calls into this function, /// then we know that they have disjoint lifetimes and that we can merge them. /// /// There are many heuristics possible for merging these allocas, and the /// different options have different tradeoffs. One thing that we *really* /// don't want to hurt is SRoA: once inlining happens, often allocas are no /// longer address taken and so they can be promoted. /// /// Our "solution" for that is to only merge allocas whose outermost type is an /// array type. These are usually not promoted because someone is using a /// variable index into them. These are also often the most important ones to /// merge. /// /// A better solution would be to have real memory lifetime markers in the IR /// and not have the inliner do any merging of allocas at all. This would /// allow the backend to do proper stack slot coloring of all allocas that /// *actually make it to the backend*, which is really what we want. /// /// Because we don't have this information, we do this simple and useful hack. static void mergeInlinedArrayAllocas(Function *Caller, InlineFunctionInfo &IFI, InlinedArrayAllocasTy &InlinedArrayAllocas, int InlineHistory) { SmallPtrSet UsedAllocas; // When processing our SCC, check to see if the call site was inlined from // some other call site. For example, if we're processing "A" in this code: // A() { B() } // B() { x = alloca ... C() } // C() { y = alloca ... } // Assume that C was not inlined into B initially, and so we're processing A // and decide to inline B into A. Doing this makes an alloca available for // reuse and makes a callsite (C) available for inlining. When we process // the C call site we don't want to do any alloca merging between X and Y // because their scopes are not disjoint. We could make this smarter by // keeping track of the inline history for each alloca in the // InlinedArrayAllocas but this isn't likely to be a significant win. if (InlineHistory != -1) // Only do merging for top-level call sites in SCC. return; // Loop over all the allocas we have so far and see if they can be merged with // a previously inlined alloca. If not, remember that we had it. for (unsigned AllocaNo = 0, E = IFI.StaticAllocas.size(); AllocaNo != E; ++AllocaNo) { AllocaInst *AI = IFI.StaticAllocas[AllocaNo]; // Don't bother trying to merge array allocations (they will usually be // canonicalized to be an allocation *of* an array), or allocations whose // type is not itself an array (because we're afraid of pessimizing SRoA). ArrayType *ATy = dyn_cast(AI->getAllocatedType()); if (!ATy || AI->isArrayAllocation()) continue; // Get the list of all available allocas for this array type. std::vector &AllocasForType = InlinedArrayAllocas[ATy]; // Loop over the allocas in AllocasForType to see if we can reuse one. Note // that we have to be careful not to reuse the same "available" alloca for // multiple different allocas that we just inlined, we use the 'UsedAllocas' // set to keep track of which "available" allocas are being used by this // function. Also, AllocasForType can be empty of course! bool MergedAwayAlloca = false; for (AllocaInst *AvailableAlloca : AllocasForType) { Align Align1 = AI->getAlign(); Align Align2 = AvailableAlloca->getAlign(); // The available alloca has to be in the right function, not in some other // function in this SCC. if (AvailableAlloca->getParent() != AI->getParent()) continue; // If the inlined function already uses this alloca then we can't reuse // it. if (!UsedAllocas.insert(AvailableAlloca).second) continue; // Otherwise, we *can* reuse it, RAUW AI into AvailableAlloca and declare // success! LLVM_DEBUG(dbgs() << " ***MERGED ALLOCA: " << *AI << "\n\t\tINTO: " << *AvailableAlloca << '\n'); // Move affected dbg.declare calls immediately after the new alloca to // avoid the situation when a dbg.declare precedes its alloca. if (auto *L = LocalAsMetadata::getIfExists(AI)) if (auto *MDV = MetadataAsValue::getIfExists(AI->getContext(), L)) for (User *U : MDV->users()) if (DbgDeclareInst *DDI = dyn_cast(U)) DDI->moveBefore(AvailableAlloca->getNextNode()); AI->replaceAllUsesWith(AvailableAlloca); if (Align1 > Align2) AvailableAlloca->setAlignment(AI->getAlign()); AI->eraseFromParent(); MergedAwayAlloca = true; ++NumMergedAllocas; IFI.StaticAllocas[AllocaNo] = nullptr; break; } // If we already nuked the alloca, we're done with it. if (MergedAwayAlloca) continue; // If we were unable to merge away the alloca either because there are no // allocas of the right type available or because we reused them all // already, remember that this alloca came from an inlined function and mark // it used so we don't reuse it for other allocas from this inline // operation. AllocasForType.push_back(AI); UsedAllocas.insert(AI); } } /// If it is possible to inline the specified call site, /// do so and update the CallGraph for this operation. /// /// This function also does some basic book-keeping to update the IR. The /// InlinedArrayAllocas map keeps track of any allocas that are already /// available from other functions inlined into the caller. If we are able to /// inline this call site we attempt to reuse already available allocas or add /// any new allocas to the set if not possible. static InlineResult inlineCallIfPossible( CallBase &CB, InlineFunctionInfo &IFI, InlinedArrayAllocasTy &InlinedArrayAllocas, int InlineHistory, bool InsertLifetime, function_ref &AARGetter, ImportedFunctionsInliningStatistics &ImportedFunctionsStats) { Function *Callee = CB.getCalledFunction(); Function *Caller = CB.getCaller(); AAResults &AAR = AARGetter(*Callee); // Try to inline the function. Get the list of static allocas that were // inlined. InlineResult IR = InlineFunction(CB, IFI, &AAR, InsertLifetime); if (!IR.isSuccess()) return IR; if (InlinerFunctionImportStats != InlinerFunctionImportStatsOpts::No) ImportedFunctionsStats.recordInline(*Caller, *Callee); AttributeFuncs::mergeAttributesForInlining(*Caller, *Callee); if (!DisableInlinedAllocaMerging) mergeInlinedArrayAllocas(Caller, IFI, InlinedArrayAllocas, InlineHistory); return IR; // success } /// Return true if the specified inline history ID /// indicates an inline history that includes the specified function. static bool inlineHistoryIncludes( Function *F, int InlineHistoryID, const SmallVectorImpl> &InlineHistory) { while (InlineHistoryID != -1) { assert(unsigned(InlineHistoryID) < InlineHistory.size() && "Invalid inline history ID"); if (InlineHistory[InlineHistoryID].first == F) return true; InlineHistoryID = InlineHistory[InlineHistoryID].second; } return false; } bool LegacyInlinerBase::doInitialization(CallGraph &CG) { if (InlinerFunctionImportStats != InlinerFunctionImportStatsOpts::No) ImportedFunctionsStats.setModuleInfo(CG.getModule()); return false; // No changes to CallGraph. } bool LegacyInlinerBase::runOnSCC(CallGraphSCC &SCC) { if (skipSCC(SCC)) return false; return inlineCalls(SCC); } static bool inlineCallsImpl(CallGraphSCC &SCC, CallGraph &CG, std::function GetAssumptionCache, ProfileSummaryInfo *PSI, std::function GetTLI, bool InsertLifetime, function_ref GetInlineCost, function_ref AARGetter, ImportedFunctionsInliningStatistics &ImportedFunctionsStats) { SmallPtrSet SCCFunctions; LLVM_DEBUG(dbgs() << "Inliner visiting SCC:"); for (CallGraphNode *Node : SCC) { Function *F = Node->getFunction(); if (F) SCCFunctions.insert(F); LLVM_DEBUG(dbgs() << " " << (F ? F->getName() : "INDIRECTNODE")); } // Scan through and identify all call sites ahead of time so that we only // inline call sites in the original functions, not call sites that result // from inlining other functions. SmallVector, 16> CallSites; // When inlining a callee produces new call sites, we want to keep track of // the fact that they were inlined from the callee. This allows us to avoid // infinite inlining in some obscure cases. To represent this, we use an // index into the InlineHistory vector. SmallVector, 8> InlineHistory; for (CallGraphNode *Node : SCC) { Function *F = Node->getFunction(); if (!F || F->isDeclaration()) continue; OptimizationRemarkEmitter ORE(F); for (BasicBlock &BB : *F) for (Instruction &I : BB) { auto *CB = dyn_cast(&I); // If this isn't a call, or it is a call to an intrinsic, it can // never be inlined. if (!CB || isa(I)) continue; // If this is a direct call to an external function, we can never inline // it. If it is an indirect call, inlining may resolve it to be a // direct call, so we keep it. if (Function *Callee = CB->getCalledFunction()) if (Callee->isDeclaration()) { using namespace ore; setInlineRemark(*CB, "unavailable definition"); ORE.emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "NoDefinition", &I) << NV("Callee", Callee) << " will not be inlined into " << NV("Caller", CB->getCaller()) << " because its definition is unavailable" << setIsVerbose(); }); continue; } CallSites.push_back(std::make_pair(CB, -1)); } } LLVM_DEBUG(dbgs() << ": " << CallSites.size() << " call sites.\n"); // If there are no calls in this function, exit early. if (CallSites.empty()) return false; // Now that we have all of the call sites, move the ones to functions in the // current SCC to the end of the list. unsigned FirstCallInSCC = CallSites.size(); for (unsigned I = 0; I < FirstCallInSCC; ++I) if (Function *F = CallSites[I].first->getCalledFunction()) if (SCCFunctions.count(F)) std::swap(CallSites[I--], CallSites[--FirstCallInSCC]); InlinedArrayAllocasTy InlinedArrayAllocas; InlineFunctionInfo InlineInfo(&CG, GetAssumptionCache, PSI); // Now that we have all of the call sites, loop over them and inline them if // it looks profitable to do so. bool Changed = false; bool LocalChange; do { LocalChange = false; // Iterate over the outer loop because inlining functions can cause indirect // calls to become direct calls. // CallSites may be modified inside so ranged for loop can not be used. for (unsigned CSi = 0; CSi != CallSites.size(); ++CSi) { auto &P = CallSites[CSi]; CallBase &CB = *P.first; const int InlineHistoryID = P.second; Function *Caller = CB.getCaller(); Function *Callee = CB.getCalledFunction(); // We can only inline direct calls to non-declarations. if (!Callee || Callee->isDeclaration()) continue; bool IsTriviallyDead = isInstructionTriviallyDead(&CB, &GetTLI(*Caller)); if (!IsTriviallyDead) { // If this call site was obtained by inlining another function, verify // that the include path for the function did not include the callee // itself. If so, we'd be recursively inlining the same function, // which would provide the same callsites, which would cause us to // infinitely inline. if (InlineHistoryID != -1 && inlineHistoryIncludes(Callee, InlineHistoryID, InlineHistory)) { setInlineRemark(CB, "recursive"); continue; } } // FIXME for new PM: because of the old PM we currently generate ORE and // in turn BFI on demand. With the new PM, the ORE dependency should // just become a regular analysis dependency. OptimizationRemarkEmitter ORE(Caller); auto OIC = shouldInline(CB, GetInlineCost, ORE); // If the policy determines that we should inline this function, // delete the call instead. if (!OIC) continue; // If this call site is dead and it is to a readonly function, we should // just delete the call instead of trying to inline it, regardless of // size. This happens because IPSCCP propagates the result out of the // call and then we're left with the dead call. if (IsTriviallyDead) { LLVM_DEBUG(dbgs() << " -> Deleting dead call: " << CB << "\n"); // Update the call graph by deleting the edge from Callee to Caller. setInlineRemark(CB, "trivially dead"); CG[Caller]->removeCallEdgeFor(CB); CB.eraseFromParent(); ++NumCallsDeleted; } else { // Get DebugLoc to report. CB will be invalid after Inliner. DebugLoc DLoc = CB.getDebugLoc(); BasicBlock *Block = CB.getParent(); // Attempt to inline the function. using namespace ore; InlineResult IR = inlineCallIfPossible( CB, InlineInfo, InlinedArrayAllocas, InlineHistoryID, InsertLifetime, AARGetter, ImportedFunctionsStats); if (!IR.isSuccess()) { setInlineRemark(CB, std::string(IR.getFailureReason()) + "; " + inlineCostStr(*OIC)); ORE.emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "NotInlined", DLoc, Block) << NV("Callee", Callee) << " will not be inlined into " << NV("Caller", Caller) << ": " << NV("Reason", IR.getFailureReason()); }); continue; } ++NumInlined; emitInlinedInto(ORE, DLoc, Block, *Callee, *Caller, *OIC); // If inlining this function gave us any new call sites, throw them // onto our worklist to process. They are useful inline candidates. if (!InlineInfo.InlinedCalls.empty()) { // Create a new inline history entry for this, so that we remember // that these new callsites came about due to inlining Callee. int NewHistoryID = InlineHistory.size(); InlineHistory.push_back(std::make_pair(Callee, InlineHistoryID)); #ifndef NDEBUG // Make sure no dupplicates in the inline candidates. This could // happen when a callsite is simpilfied to reusing the return value // of another callsite during function cloning, thus the other // callsite will be reconsidered here. DenseSet DbgCallSites; for (auto &II : CallSites) DbgCallSites.insert(II.first); #endif for (Value *Ptr : InlineInfo.InlinedCalls) { #ifndef NDEBUG assert(DbgCallSites.count(dyn_cast(Ptr)) == 0); #endif CallSites.push_back( std::make_pair(dyn_cast(Ptr), NewHistoryID)); } } } // If we inlined or deleted the last possible call site to the function, // delete the function body now. if (Callee && Callee->use_empty() && Callee->hasLocalLinkage() && // TODO: Can remove if in SCC now. !SCCFunctions.count(Callee) && // The function may be apparently dead, but if there are indirect // callgraph references to the node, we cannot delete it yet, this // could invalidate the CGSCC iterator. CG[Callee]->getNumReferences() == 0) { LLVM_DEBUG(dbgs() << " -> Deleting dead function: " << Callee->getName() << "\n"); CallGraphNode *CalleeNode = CG[Callee]; // Remove any call graph edges from the callee to its callees. CalleeNode->removeAllCalledFunctions(); // Removing the node for callee from the call graph and delete it. delete CG.removeFunctionFromModule(CalleeNode); ++NumDeleted; } // Remove this call site from the list. If possible, use // swap/pop_back for efficiency, but do not use it if doing so would // move a call site to a function in this SCC before the // 'FirstCallInSCC' barrier. if (SCC.isSingular()) { CallSites[CSi] = CallSites.back(); CallSites.pop_back(); } else { CallSites.erase(CallSites.begin() + CSi); } --CSi; Changed = true; LocalChange = true; } } while (LocalChange); return Changed; } bool LegacyInlinerBase::inlineCalls(CallGraphSCC &SCC) { CallGraph &CG = getAnalysis().getCallGraph(); ACT = &getAnalysis(); PSI = &getAnalysis().getPSI(); GetTLI = [&](Function &F) -> const TargetLibraryInfo & { return getAnalysis().getTLI(F); }; auto GetAssumptionCache = [&](Function &F) -> AssumptionCache & { return ACT->getAssumptionCache(F); }; return inlineCallsImpl( SCC, CG, GetAssumptionCache, PSI, GetTLI, InsertLifetime, [&](CallBase &CB) { return getInlineCost(CB); }, LegacyAARGetter(*this), ImportedFunctionsStats); } /// Remove now-dead linkonce functions at the end of /// processing to avoid breaking the SCC traversal. bool LegacyInlinerBase::doFinalization(CallGraph &CG) { if (InlinerFunctionImportStats != InlinerFunctionImportStatsOpts::No) ImportedFunctionsStats.dump(InlinerFunctionImportStats == InlinerFunctionImportStatsOpts::Verbose); return removeDeadFunctions(CG); } /// Remove dead functions that are not included in DNR (Do Not Remove) list. bool LegacyInlinerBase::removeDeadFunctions(CallGraph &CG, bool AlwaysInlineOnly) { SmallVector FunctionsToRemove; SmallVector DeadFunctionsInComdats; auto RemoveCGN = [&](CallGraphNode *CGN) { // Remove any call graph edges from the function to its callees. CGN->removeAllCalledFunctions(); // Remove any edges from the external node to the function's call graph // node. These edges might have been made irrelegant due to // optimization of the program. CG.getExternalCallingNode()->removeAnyCallEdgeTo(CGN); // Removing the node for callee from the call graph and delete it. FunctionsToRemove.push_back(CGN); }; // Scan for all of the functions, looking for ones that should now be removed // from the program. Insert the dead ones in the FunctionsToRemove set. for (const auto &I : CG) { CallGraphNode *CGN = I.second.get(); Function *F = CGN->getFunction(); if (!F || F->isDeclaration()) continue; // Handle the case when this function is called and we only want to care // about always-inline functions. This is a bit of a hack to share code // between here and the InlineAlways pass. if (AlwaysInlineOnly && !F->hasFnAttribute(Attribute::AlwaysInline)) continue; // If the only remaining users of the function are dead constants, remove // them. F->removeDeadConstantUsers(); if (!F->isDefTriviallyDead()) continue; // It is unsafe to drop a function with discardable linkage from a COMDAT // without also dropping the other members of the COMDAT. // The inliner doesn't visit non-function entities which are in COMDAT // groups so it is unsafe to do so *unless* the linkage is local. if (!F->hasLocalLinkage()) { if (F->hasComdat()) { DeadFunctionsInComdats.push_back(F); continue; } } RemoveCGN(CGN); } if (!DeadFunctionsInComdats.empty()) { // Filter out the functions whose comdats remain alive. filterDeadComdatFunctions(CG.getModule(), DeadFunctionsInComdats); // Remove the rest. for (Function *F : DeadFunctionsInComdats) RemoveCGN(CG[F]); } if (FunctionsToRemove.empty()) return false; // Now that we know which functions to delete, do so. We didn't want to do // this inline, because that would invalidate our CallGraph::iterator // objects. :( // // Note that it doesn't matter that we are iterating over a non-stable order // here to do this, it doesn't matter which order the functions are deleted // in. array_pod_sort(FunctionsToRemove.begin(), FunctionsToRemove.end()); FunctionsToRemove.erase( std::unique(FunctionsToRemove.begin(), FunctionsToRemove.end()), FunctionsToRemove.end()); for (CallGraphNode *CGN : FunctionsToRemove) { delete CG.removeFunctionFromModule(CGN); ++NumDeleted; } return true; } InlinerPass::~InlinerPass() { if (ImportedFunctionsStats) { assert(InlinerFunctionImportStats != InlinerFunctionImportStatsOpts::No); ImportedFunctionsStats->dump(InlinerFunctionImportStats == InlinerFunctionImportStatsOpts::Verbose); } } InlineAdvisor & InlinerPass::getAdvisor(const ModuleAnalysisManagerCGSCCProxy::Result &MAM, FunctionAnalysisManager &FAM, Module &M) { auto *IAA = MAM.getCachedResult(M); if (!IAA) { // It should still be possible to run the inliner as a stand-alone SCC pass, // for test scenarios. In that case, we default to the // DefaultInlineAdvisor, which doesn't need to keep state between SCC pass // runs. It also uses just the default InlineParams. // In this case, we need to use the provided FAM, which is valid for the // duration of the inliner pass, and thus the lifetime of the owned advisor. // The one we would get from the MAM can be invalidated as a result of the // inliner's activity. OwnedDefaultAdvisor.emplace(FAM, getInlineParams()); return *OwnedDefaultAdvisor; } assert(IAA->getAdvisor() && "Expected a present InlineAdvisorAnalysis also have an " "InlineAdvisor initialized"); return *IAA->getAdvisor(); } PreservedAnalyses InlinerPass::run(LazyCallGraph::SCC &InitialC, CGSCCAnalysisManager &AM, LazyCallGraph &CG, CGSCCUpdateResult &UR) { const auto &MAMProxy = AM.getResult(InitialC, CG); bool Changed = false; assert(InitialC.size() > 0 && "Cannot handle an empty SCC!"); Module &M = *InitialC.begin()->getFunction().getParent(); ProfileSummaryInfo *PSI = MAMProxy.getCachedResult(M); FunctionAnalysisManager &FAM = AM.getResult(InitialC, CG) .getManager(); InlineAdvisor &Advisor = getAdvisor(MAMProxy, FAM, M); Advisor.onPassEntry(); auto AdvisorOnExit = make_scope_exit([&] { Advisor.onPassExit(); }); if (!ImportedFunctionsStats && InlinerFunctionImportStats != InlinerFunctionImportStatsOpts::No) { ImportedFunctionsStats = std::make_unique(); ImportedFunctionsStats->setModuleInfo(M); } // We use a single common worklist for calls across the entire SCC. We // process these in-order and append new calls introduced during inlining to // the end. // // Note that this particular order of processing is actually critical to // avoid very bad behaviors. Consider *highly connected* call graphs where // each function contains a small amonut of code and a couple of calls to // other functions. Because the LLVM inliner is fundamentally a bottom-up // inliner, it can handle gracefully the fact that these all appear to be // reasonable inlining candidates as it will flatten things until they become // too big to inline, and then move on and flatten another batch. // // However, when processing call edges *within* an SCC we cannot rely on this // bottom-up behavior. As a consequence, with heavily connected *SCCs* of // functions we can end up incrementally inlining N calls into each of // N functions because each incremental inlining decision looks good and we // don't have a topological ordering to prevent explosions. // // To compensate for this, we don't process transitive edges made immediate // by inlining until we've done one pass of inlining across the entire SCC. // Large, highly connected SCCs still lead to some amount of code bloat in // this model, but it is uniformly spread across all the functions in the SCC // and eventually they all become too large to inline, rather than // incrementally maknig a single function grow in a super linear fashion. SmallVector, 16> Calls; // Populate the initial list of calls in this SCC. for (auto &N : InitialC) { auto &ORE = FAM.getResult(N.getFunction()); // We want to generally process call sites top-down in order for // simplifications stemming from replacing the call with the returned value // after inlining to be visible to subsequent inlining decisions. // FIXME: Using instructions sequence is a really bad way to do this. // Instead we should do an actual RPO walk of the function body. for (Instruction &I : instructions(N.getFunction())) if (auto *CB = dyn_cast(&I)) if (Function *Callee = CB->getCalledFunction()) { if (!Callee->isDeclaration()) Calls.push_back({CB, -1}); else if (!isa(I)) { using namespace ore; setInlineRemark(*CB, "unavailable definition"); ORE.emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "NoDefinition", &I) << NV("Callee", Callee) << " will not be inlined into " << NV("Caller", CB->getCaller()) << " because its definition is unavailable" << setIsVerbose(); }); } } } if (Calls.empty()) return PreservedAnalyses::all(); // Capture updatable variables for the current SCC and RefSCC. auto *C = &InitialC; auto *RC = &C->getOuterRefSCC(); // When inlining a callee produces new call sites, we want to keep track of // the fact that they were inlined from the callee. This allows us to avoid // infinite inlining in some obscure cases. To represent this, we use an // index into the InlineHistory vector. SmallVector, 16> InlineHistory; // Track a set vector of inlined callees so that we can augment the caller // with all of their edges in the call graph before pruning out the ones that // got simplified away. SmallSetVector InlinedCallees; // Track the dead functions to delete once finished with inlining calls. We // defer deleting these to make it easier to handle the call graph updates. SmallVector DeadFunctions; // Loop forward over all of the calls. Note that we cannot cache the size as // inlining can introduce new calls that need to be processed. for (int I = 0; I < (int)Calls.size(); ++I) { // We expect the calls to typically be batched with sequences of calls that // have the same caller, so we first set up some shared infrastructure for // this caller. We also do any pruning we can at this layer on the caller // alone. Function &F = *Calls[I].first->getCaller(); LazyCallGraph::Node &N = *CG.lookup(F); if (CG.lookupSCC(N) != C) continue; if (!Calls[I].first->getCalledFunction()->hasFnAttribute( Attribute::AlwaysInline) && F.hasOptNone()) { setInlineRemark(*Calls[I].first, "optnone attribute"); continue; } LLVM_DEBUG(dbgs() << "Inlining calls in: " << F.getName() << "\n"); auto GetAssumptionCache = [&](Function &F) -> AssumptionCache & { return FAM.getResult(F); }; // Now process as many calls as we have within this caller in the sequence. // We bail out as soon as the caller has to change so we can update the // call graph and prepare the context of that new caller. bool DidInline = false; for (; I < (int)Calls.size() && Calls[I].first->getCaller() == &F; ++I) { auto &P = Calls[I]; CallBase *CB = P.first; const int InlineHistoryID = P.second; Function &Callee = *CB->getCalledFunction(); if (InlineHistoryID != -1 && inlineHistoryIncludes(&Callee, InlineHistoryID, InlineHistory)) { setInlineRemark(*CB, "recursive"); continue; } // Check if this inlining may repeat breaking an SCC apart that has // already been split once before. In that case, inlining here may // trigger infinite inlining, much like is prevented within the inliner // itself by the InlineHistory above, but spread across CGSCC iterations // and thus hidden from the full inline history. if (CG.lookupSCC(*CG.lookup(Callee)) == C && UR.InlinedInternalEdges.count({&N, C})) { LLVM_DEBUG(dbgs() << "Skipping inlining internal SCC edge from a node " "previously split out of this SCC by inlining: " << F.getName() << " -> " << Callee.getName() << "\n"); setInlineRemark(*CB, "recursive SCC split"); continue; } auto Advice = Advisor.getAdvice(*CB); // Check whether we want to inline this callsite. if (!Advice->isInliningRecommended()) { Advice->recordUnattemptedInlining(); continue; } // Setup the data structure used to plumb customization into the // `InlineFunction` routine. InlineFunctionInfo IFI( /*cg=*/nullptr, GetAssumptionCache, PSI, &FAM.getResult(*(CB->getCaller())), &FAM.getResult(Callee)); InlineResult IR = InlineFunction(*CB, IFI); if (!IR.isSuccess()) { Advice->recordUnsuccessfulInlining(IR); continue; } DidInline = true; InlinedCallees.insert(&Callee); ++NumInlined; // Add any new callsites to defined functions to the worklist. if (!IFI.InlinedCallSites.empty()) { int NewHistoryID = InlineHistory.size(); InlineHistory.push_back({&Callee, InlineHistoryID}); for (CallBase *ICB : reverse(IFI.InlinedCallSites)) { Function *NewCallee = ICB->getCalledFunction(); if (!NewCallee) { // Try to promote an indirect (virtual) call without waiting for // the post-inline cleanup and the next DevirtSCCRepeatedPass // iteration because the next iteration may not happen and we may // miss inlining it. if (tryPromoteCall(*ICB)) NewCallee = ICB->getCalledFunction(); } if (NewCallee) if (!NewCallee->isDeclaration()) Calls.push_back({ICB, NewHistoryID}); } } if (InlinerFunctionImportStats != InlinerFunctionImportStatsOpts::No) ImportedFunctionsStats->recordInline(F, Callee); // Merge the attributes based on the inlining. AttributeFuncs::mergeAttributesForInlining(F, Callee); // For local functions, check whether this makes the callee trivially // dead. In that case, we can drop the body of the function eagerly // which may reduce the number of callers of other functions to one, // changing inline cost thresholds. bool CalleeWasDeleted = false; if (Callee.hasLocalLinkage()) { // To check this we also need to nuke any dead constant uses (perhaps // made dead by this operation on other functions). Callee.removeDeadConstantUsers(); if (Callee.use_empty() && !CG.isLibFunction(Callee)) { Calls.erase( std::remove_if(Calls.begin() + I + 1, Calls.end(), [&](const std::pair &Call) { return Call.first->getCaller() == &Callee; }), Calls.end()); // Clear the body and queue the function itself for deletion when we // finish inlining and call graph updates. // Note that after this point, it is an error to do anything other // than use the callee's address or delete it. Callee.dropAllReferences(); assert(find(DeadFunctions, &Callee) == DeadFunctions.end() && "Cannot put cause a function to become dead twice!"); DeadFunctions.push_back(&Callee); CalleeWasDeleted = true; } } if (CalleeWasDeleted) Advice->recordInliningWithCalleeDeleted(); else Advice->recordInlining(); } // Back the call index up by one to put us in a good position to go around // the outer loop. --I; if (!DidInline) continue; Changed = true; // Add all the inlined callees' edges as ref edges to the caller. These are // by definition trivial edges as we always have *some* transitive ref edge // chain. While in some cases these edges are direct calls inside the // callee, they have to be modeled in the inliner as reference edges as // there may be a reference edge anywhere along the chain from the current // caller to the callee that causes the whole thing to appear like // a (transitive) reference edge that will require promotion to a call edge // below. for (Function *InlinedCallee : InlinedCallees) { LazyCallGraph::Node &CalleeN = *CG.lookup(*InlinedCallee); for (LazyCallGraph::Edge &E : *CalleeN) RC->insertTrivialRefEdge(N, E.getNode()); } // At this point, since we have made changes we have at least removed // a call instruction. However, in the process we do some incremental // simplification of the surrounding code. This simplification can // essentially do all of the same things as a function pass and we can // re-use the exact same logic for updating the call graph to reflect the // change. // Inside the update, we also update the FunctionAnalysisManager in the // proxy for this particular SCC. We do this as the SCC may have changed and // as we're going to mutate this particular function we want to make sure // the proxy is in place to forward any invalidation events. LazyCallGraph::SCC *OldC = C; C = &updateCGAndAnalysisManagerForFunctionPass(CG, *C, N, AM, UR, FAM); LLVM_DEBUG(dbgs() << "Updated inlining SCC: " << *C << "\n"); RC = &C->getOuterRefSCC(); // If this causes an SCC to split apart into multiple smaller SCCs, there // is a subtle risk we need to prepare for. Other transformations may // expose an "infinite inlining" opportunity later, and because of the SCC // mutation, we will revisit this function and potentially re-inline. If we // do, and that re-inlining also has the potentially to mutate the SCC // structure, the infinite inlining problem can manifest through infinite // SCC splits and merges. To avoid this, we capture the originating caller // node and the SCC containing the call edge. This is a slight over // approximation of the possible inlining decisions that must be avoided, // but is relatively efficient to store. We use C != OldC to know when // a new SCC is generated and the original SCC may be generated via merge // in later iterations. // // It is also possible that even if no new SCC is generated // (i.e., C == OldC), the original SCC could be split and then merged // into the same one as itself. and the original SCC will be added into // UR.CWorklist again, we want to catch such cases too. // // FIXME: This seems like a very heavyweight way of retaining the inline // history, we should look for a more efficient way of tracking it. if ((C != OldC || UR.CWorklist.count(OldC)) && llvm::any_of(InlinedCallees, [&](Function *Callee) { return CG.lookupSCC(*CG.lookup(*Callee)) == OldC; })) { LLVM_DEBUG(dbgs() << "Inlined an internal call edge and split an SCC, " "retaining this to avoid infinite inlining.\n"); UR.InlinedInternalEdges.insert({&N, OldC}); } InlinedCallees.clear(); } // Now that we've finished inlining all of the calls across this SCC, delete // all of the trivially dead functions, updating the call graph and the CGSCC // pass manager in the process. // // Note that this walks a pointer set which has non-deterministic order but // that is OK as all we do is delete things and add pointers to unordered // sets. for (Function *DeadF : DeadFunctions) { // Get the necessary information out of the call graph and nuke the // function there. Also, clear out any cached analyses. auto &DeadC = *CG.lookupSCC(*CG.lookup(*DeadF)); FAM.clear(*DeadF, DeadF->getName()); AM.clear(DeadC, DeadC.getName()); auto &DeadRC = DeadC.getOuterRefSCC(); CG.removeDeadFunction(*DeadF); // Mark the relevant parts of the call graph as invalid so we don't visit // them. UR.InvalidatedSCCs.insert(&DeadC); UR.InvalidatedRefSCCs.insert(&DeadRC); // And delete the actual function from the module. // The Advisor may use Function pointers to efficiently index various // internal maps, e.g. for memoization. Function cleanup passes like // argument promotion create new functions. It is possible for a new // function to be allocated at the address of a deleted function. We could // index using names, but that's inefficient. Alternatively, we let the // Advisor free the functions when it sees fit. DeadF->getBasicBlockList().clear(); M.getFunctionList().remove(DeadF); ++NumDeleted; } if (!Changed) return PreservedAnalyses::all(); // Even if we change the IR, we update the core CGSCC data structures and so // can preserve the proxy to the function analysis manager. PreservedAnalyses PA; PA.preserve(); return PA; } ModuleInlinerWrapperPass::ModuleInlinerWrapperPass(InlineParams Params, bool Debugging, InliningAdvisorMode Mode, unsigned MaxDevirtIterations) : Params(Params), Mode(Mode), MaxDevirtIterations(MaxDevirtIterations), PM(Debugging), MPM(Debugging) { // Run the inliner first. The theory is that we are walking bottom-up and so // the callees have already been fully optimized, and we want to inline them // into the callers so that our optimizations can reflect that. // For PreLinkThinLTO pass, we disable hot-caller heuristic for sample PGO // because it makes profile annotation in the backend inaccurate. PM.addPass(InlinerPass()); } PreservedAnalyses ModuleInlinerWrapperPass::run(Module &M, ModuleAnalysisManager &MAM) { auto &IAA = MAM.getResult(M); if (!IAA.tryCreate(Params, Mode)) { M.getContext().emitError( "Could not setup Inlining Advisor for the requested " "mode and/or options"); return PreservedAnalyses::all(); } // We wrap the CGSCC pipeline in a devirtualization repeater. This will try // to detect when we devirtualize indirect calls and iterate the SCC passes // in that case to try and catch knock-on inlining or function attrs // opportunities. Then we add it to the module pipeline by walking the SCCs // in postorder (or bottom-up). // If MaxDevirtIterations is 0, we just don't use the devirtualization // wrapper. if (MaxDevirtIterations == 0) MPM.addPass(createModuleToPostOrderCGSCCPassAdaptor(std::move(PM))); else MPM.addPass(createModuleToPostOrderCGSCCPassAdaptor( createDevirtSCCRepeatedPass(std::move(PM), MaxDevirtIterations))); auto Ret = MPM.run(M, MAM); IAA.clear(); return Ret; }