Index: lib/Analysis/InlineCost.cpp =================================================================== --- lib/Analysis/InlineCost.cpp +++ lib/Analysis/InlineCost.cpp @@ -66,6 +66,41 @@ cl::ZeroOrMore, cl::desc("Threshold for hot callsites ")); +// The following options control the heuristic that estimates speedup due to +// inlining and adds a threshold bonus when the speedup is above a certain +// limit. Speedup here refers to savings in weighted cost due to inlining +// relative to the weighted cost of the uninlined callee. +// +/// Threshold bonus to apply when inlining is expected to result in a speedup. +/// The bonus is expressed as a percent value. The unbonused threshold is +/// multiplied by this threshold to arrive at the threshold increase. +static cl::opt + SpeedupBonusPercent("speedup-bonus-percent", cl::Hidden, cl::init(200), + cl::ZeroOrMore, + cl::desc("Bonus for callees showing speedup")); + +/// Minimum speedup percentage required to apply a bonus to the threshold. +/// +/// Speedup is computed as the percentage of weighted cost savings if the callee +/// is inlined. +static cl::opt + MinSpeedupForBonus("min-speedup-for-bonus", cl::Hidden, cl::init(10), + cl::ZeroOrMore, + cl::desc("Minimum weighted cost savings (in percentage) " + "needed to apply the speedup bonus")); + +/// Minimum relative frequency of callsite to apply speedup-based bonus. +/// +/// Speedup bonus is applied if the callsite is hot and not applied if it is +/// cold as determined by profile summary. If the callsite is not known to be +/// hot or cold, the bonus is applied only if the block frequency of the +/// callsite relative to the caller's entry meets this ratio to limit size +/// increase due to speedup bonus by applying the heuristic. +static cl::opt MinBFRatioForSpeedupBonus( + "min-bf-ratio-for-speedup-bonus", cl::Hidden, cl::init(8), cl::ZeroOrMore, + cl::desc("Minimum ratio of callsite's block frequency to caller's entry " + "block frequency to apply the speedup bonus")); + namespace { class CallAnalyzer : public InstVisitor { @@ -95,6 +130,7 @@ /// Tunable parameters that control the analysis. const InlineParams &Params; + BlockFrequencyInfo *CallerBFI, *CalleeBFI; int Threshold; int Cost; @@ -126,22 +162,28 @@ /// allocas on the caller stack which could be simplified through SROA. DenseMap SROAArgValues; + // We track unweighted and weighted(by block frequency) SROA cost savings. + using SROACostTy = std::pair; + /// The mapping of caller Alloca values to their accumulated cost savings. If /// we have to disable SROA for one of the allocas, this tells us how much /// cost must be added. - DenseMap SROAArgCosts; + DenseMap SROAArgCosts; /// Keep track of values which map to a pointer base and constant offset. DenseMap> ConstantOffsetPtrs; // Custom simplification helper routines. bool isAllocaDerivedArg(Value *V); + int getArgPassingCost(CallSite CS, Function *Callee); bool lookupSROAArgAndCost(Value *V, Value *&Arg, - DenseMap::iterator &CostIt); - void disableSROA(DenseMap::iterator CostIt); + DenseMap::iterator &CostIt); + void disableSROA(DenseMap::iterator CostIt); void disableSROA(Value *V); - void accumulateSROACost(DenseMap::iterator CostIt, + void accumulateSROACost(DenseMap::iterator CostIt, int InstructionCost); + void accumulateCost(int InstructionCost); + void accumulateSavings(int Savings); bool isGEPFree(GetElementPtrInst &GEP); bool accumulateGEPOffset(GEPOperator &GEP, APInt &Offset); bool simplifyCallSite(Function *F, CallSite CS); @@ -209,6 +251,14 @@ bool visitCleanupReturnInst(CleanupReturnInst &RI); bool visitCatchReturnInst(CatchReturnInst &RI); bool visitUnreachableInst(UnreachableInst &I); + /// Simplify \p I if its operands are constants and update SimplifiedValues. + /// Evaluate is a lambda specific to instruction type that evaluates the + /// instruction when all the operands are constants. + bool + simplifyInstruction(Instruction &I, + function_ref)> Evaluate); + int getSpeedupBonus(CallSite &CS, int Threshold); + bool hasLargeSpeedup(); public: CallAnalyzer(const TargetTransformInfo &TTI, @@ -226,8 +276,8 @@ FiftyPercentVectorBonus(0), TenPercentVectorBonus(0), VectorBonus(0), NumConstantArgs(0), NumConstantOffsetPtrArgs(0), NumAllocaArgs(0), NumConstantPtrCmps(0), NumConstantPtrDiffs(0), - NumInstructionsSimplified(0), SROACostSavings(0), - SROACostSavingsLost(0) {} + NumInstructionsSimplified(0), SROACostSavings(0), WeightedSavings(0), + WeightedCost(0), SROACostSavingsLost(0), CurrBBFreq(1) {} bool analyzeCall(CallSite CS); @@ -243,7 +293,10 @@ unsigned NumConstantPtrDiffs; unsigned NumInstructionsSimplified; unsigned SROACostSavings; + int WeightedSavings, WeightedCost; unsigned SROACostSavingsLost; + /// The block frequency of the current block being analyzed. + uint64_t CurrBBFreq; void dump(); }; @@ -255,10 +308,47 @@ return SROAArgValues.count(V); } +/// Return the cost of passing arguments to the callee at callsite \param CS. +/// +/// This cost is negative since inlining would eliminate the instructions needed +/// for setting up the arguments +int CallAnalyzer::getArgPassingCost(CallSite CS, Function *Callee) { + // Give out bonuses per argument, as the instructions setting them up will + // be gone after inlining. + int Cost = 0; + const DataLayout &DL = Callee->getParent()->getDataLayout(); + for (unsigned I = 0, E = CS.arg_size(); I != E; ++I) { + if (CS.isByValArgument(I)) { + // We approximate the number of loads and stores needed by dividing the + // size of the byval type by the target's pointer size. + PointerType *PTy = cast(CS.getArgument(I)->getType()); + unsigned TypeSize = DL.getTypeSizeInBits(PTy->getElementType()); + unsigned PointerSize = DL.getPointerSizeInBits(); + // Ceiling division. + unsigned NumStores = (TypeSize + PointerSize - 1) / PointerSize; + + // If it generates more than 8 stores it is likely to be expanded as an + // inline memcpy so we take that as an upper bound. Otherwise we assume + // one load and one store per word copied. + // FIXME: The maxStoresPerMemcpy setting from the target should be used + // here instead of a magic number of 8, but it's not available via + // DataLayout. + NumStores = std::min(NumStores, 8U); + + Cost -= 2 * NumStores * InlineConstants::InstrCost; + } else { + // For non-byval arguments subtract off one instruction per call + // argument. + Cost -= InlineConstants::InstrCost; + } + } + return Cost; +} + /// \brief Lookup the SROA-candidate argument and cost iterator which V maps to. /// Returns false if V does not map to a SROA-candidate. bool CallAnalyzer::lookupSROAArgAndCost( - Value *V, Value *&Arg, DenseMap::iterator &CostIt) { + Value *V, Value *&Arg, DenseMap::iterator &CostIt) { if (SROAArgValues.empty() || SROAArgCosts.empty()) return false; @@ -275,28 +365,37 @@ /// /// This marks the candidate as no longer viable for SROA, and adds the cost /// savings associated with it back into the inline cost measurement. -void CallAnalyzer::disableSROA(DenseMap::iterator CostIt) { +void CallAnalyzer::disableSROA(DenseMap::iterator CostIt) { // If we're no longer able to perform SROA we need to undo its cost savings // and prevent subsequent analysis. - Cost += CostIt->second; - SROACostSavings -= CostIt->second; - SROACostSavingsLost += CostIt->second; + Cost += CostIt->second.first; + WeightedCost += CostIt->second.second; + SROACostSavings -= CostIt->second.first; + WeightedSavings -= CostIt->second.second; + SROACostSavingsLost += CostIt->second.first; SROAArgCosts.erase(CostIt); } /// \brief If 'V' maps to a SROA candidate, disable SROA for it. void CallAnalyzer::disableSROA(Value *V) { Value *SROAArg; - DenseMap::iterator CostIt; + DenseMap::iterator CostIt; if (lookupSROAArgAndCost(V, SROAArg, CostIt)) disableSROA(CostIt); } /// \brief Accumulate the given cost for a particular SROA candidate. -void CallAnalyzer::accumulateSROACost(DenseMap::iterator CostIt, - int InstructionCost) { - CostIt->second += InstructionCost; +void CallAnalyzer::accumulateSROACost( + DenseMap::iterator CostIt, int InstructionCost) { + CostIt->second.first += InstructionCost; SROACostSavings += InstructionCost; + // FIXME: We use saturating multiply on uint64_t here and below where we + // compute weighted cost/savings. If this proves to be less precise, consider + // using 128 bit APInt and also use relative block frequency (scale block + // frequencies relative to entry block). + auto WeightedCost = SaturatingMultiply(CurrBBFreq, (uint64_t)InstructionCost); + CostIt->second.second += WeightedCost; + WeightedSavings += WeightedCost; } /// \brief Accumulate a constant GEP offset into an APInt if possible. @@ -394,9 +493,14 @@ return true; } +void CallAnalyzer::accumulateCost(int InstructionCost) { + Cost += InstructionCost; + WeightedCost += SaturatingMultiply(CurrBBFreq, (uint64_t)InstructionCost); +} + bool CallAnalyzer::visitGetElementPtr(GetElementPtrInst &I) { Value *SROAArg; - DenseMap::iterator CostIt; + DenseMap::iterator CostIt; bool SROACandidate = lookupSROAArgAndCost(I.getPointerOperand(), SROAArg, CostIt); @@ -413,6 +517,10 @@ // Non-constant GEPs aren't folded, and disable SROA. if (SROACandidate) disableSROA(CostIt); + // isGEPFree looks up SimplifiedValues and so we should ideally be + // tracking savings here, but that would require calling the TTI hook + // with and without simplified values. So we take the conservative route + // and assume no savings due to the lookup. return isGEPFree(I); } @@ -428,11 +536,21 @@ } } + // Lambda to check whether a GEP's indices are all constant. auto IsGEPOffsetConstant = [&](GetElementPtrInst &GEP) { - for (User::op_iterator I = GEP.idx_begin(), E = GEP.idx_end(); I != E; ++I) - if (!isa(*I) && !SimplifiedValues.lookup(*I)) - return false; + bool SVLookup = false; + for (User::op_iterator I = GEP.idx_begin(), E = GEP.idx_end(); I != E; + ++I) { + if (!isa(*I)) { + if (SimplifiedValues.lookup(*I)) + SVLookup = true; + else + return false; + } + } + if (SVLookup) + accumulateSavings(InlineConstants::InstrCost); return true; }; @@ -450,16 +568,23 @@ return isGEPFree(I); } +void CallAnalyzer::accumulateSavings(int Savings) { + WeightedSavings += SaturatingMultiply(CurrBBFreq, (uint64_t)Savings); +} + /// Simplify \p I if its operands are constants and update SimplifiedValues. /// \p Evaluate is a callable specific to instruction type that evaluates the /// instruction when all the operands are constants. template bool CallAnalyzer::simplifyInstruction(Instruction &I, Callable Evaluate) { SmallVector COps; + bool SVLookup = false; for (Value *Op : I.operands()) { Constant *COp = dyn_cast(Op); - if (!COp) + if (!COp) { COp = SimplifiedValues.lookup(Op); + SVLookup = true; + } if (!COp) return false; COps.push_back(COp); @@ -468,9 +593,47 @@ if (!C) return false; SimplifiedValues[&I] = C; + // If SVLookup is true, then the constness of at least one of the operands is + // inferred only after looking up the SimplifiedValues map. In other words, + // the constness of the operands (and consequently the result) cannot be + // inferred without inlining. Thus we account the savings associated with the + // elimination of this instruction to inlining + if (SVLookup) + accumulateSavings(InlineConstants::InstrCost); return true; } +/// Get the bonus to be applied for callsites with significant speedup. +int CallAnalyzer::getSpeedupBonus(CallSite &CS, int Threshold) { + if (!CallerBFI || !CalleeBFI) + return 0; + auto EntryFreq = CallerBFI->getEntryFreq(); + auto *BB = CS.getInstruction()->getParent(); + auto CallSiteFreq = CallerBFI->getBlockFreq(BB).getFrequency(); + int Bonus = SpeedupBonusPercent * Threshold / 100; + // Always apply the bonus for hot callsites and never apply this for cold + // callsites. + if (PSI) { + if (PSI->isHotCallSite(CS, CallerBFI)) + return Bonus; + if (PSI->isColdCallSite(CS, CallerBFI)) + return 0; + } + // In the absence of profile summary or when the callsite is neither hot nor + // cold, apply the bonus if callsite's frequency exceeds + // MinBFRatioForSpeedupBonus. + if (CallSiteFreq / EntryFreq >= MinBFRatioForSpeedupBonus) + return Bonus; + return 0; +} + +/// Return true if the estimated speedup is large enough to apply bonus. +bool CallAnalyzer::hasLargeSpeedup() { + int D = std::max(1, WeightedCost + WeightedSavings); + int Speedup = WeightedSavings * 100 / D; + return Speedup > MinSpeedupForBonus; +} + bool CallAnalyzer::visitBitCast(BitCastInst &I) { // Propagate constants through bitcasts. if (simplifyInstruction(I, [&](SmallVectorImpl &COps) { @@ -487,7 +650,7 @@ // Also look for SROA candidates here. Value *SROAArg; - DenseMap::iterator CostIt; + DenseMap::iterator CostIt; if (lookupSROAArgAndCost(I.getOperand(0), SROAArg, CostIt)) SROAArgValues[&I] = SROAArg; @@ -521,7 +684,7 @@ // preserved either cannot fire on an integer, or won't in-and-of themselves // disable SROA (ext) w/o some later use that we would see and disable. Value *SROAArg; - DenseMap::iterator CostIt; + DenseMap::iterator CostIt; if (lookupSROAArgAndCost(I.getOperand(0), SROAArg, CostIt)) SROAArgValues[&I] = SROAArg; @@ -548,7 +711,7 @@ // "Propagate" SROA here in the same manner as we do for ptrtoint above. Value *SROAArg; - DenseMap::iterator CostIt; + DenseMap::iterator CostIt; if (lookupSROAArgAndCost(Op, SROAArg, CostIt)) SROAArgValues[&I] = SROAArg; @@ -667,7 +830,6 @@ if (Callee.hasFnAttribute(Attribute::InlineHint)) Threshold = MaxIfValid(Threshold, Params.HintThreshold); if (PSI) { - BlockFrequencyInfo *CallerBFI = GetBFI ? &((*GetBFI)(*Caller)) : nullptr; if (PSI->isHotCallSite(CS, CallerBFI)) { DEBUG(dbgs() << "Hot callsite.\n"); Threshold = MaxIfValid(Threshold, Params.HotCallSiteThreshold); @@ -716,6 +878,7 @@ Constant *CLHS = ConstantInt::get(LHS->getContext(), LHSOffset); Constant *CRHS = ConstantInt::get(RHS->getContext(), RHSOffset); if (Constant *C = ConstantExpr::getICmp(I.getPredicate(), CLHS, CRHS)) { + accumulateSavings(InlineConstants::InstrCost); SimplifiedValues[&I] = C; ++NumConstantPtrCmps; return true; @@ -727,6 +890,8 @@ // if we know the value (argument) can't be null if (I.isEquality() && isa(I.getOperand(1)) && isKnownNonNullInCallee(I.getOperand(0))) { + if (isAllocaDerivedArg(I.getOperand(0))) + accumulateSavings(InlineConstants::InstrCost); bool IsNotEqual = I.getPredicate() == CmpInst::ICMP_NE; SimplifiedValues[&I] = IsNotEqual ? ConstantInt::getTrue(I.getType()) : ConstantInt::getFalse(I.getType()); @@ -734,7 +899,7 @@ } // Finally check for SROA candidates in comparisons. Value *SROAArg; - DenseMap::iterator CostIt; + DenseMap::iterator CostIt; if (lookupSROAArgAndCost(I.getOperand(0), SROAArg, CostIt)) { if (isa(I.getOperand(1))) { accumulateSROACost(CostIt, InlineConstants::InstrCost); @@ -762,6 +927,7 @@ Constant *CLHS = ConstantInt::get(LHS->getContext(), LHSOffset); Constant *CRHS = ConstantInt::get(RHS->getContext(), RHSOffset); if (Constant *C = ConstantExpr::getSub(CLHS, CRHS)) { + accumulateSavings(InlineConstants::InstrCost); SimplifiedValues[&I] = C; ++NumConstantPtrDiffs; return true; @@ -799,7 +965,7 @@ bool CallAnalyzer::visitLoad(LoadInst &I) { Value *SROAArg; - DenseMap::iterator CostIt; + DenseMap::iterator CostIt; if (lookupSROAArgAndCost(I.getPointerOperand(), SROAArg, CostIt)) { if (I.isSimple()) { accumulateSROACost(CostIt, InlineConstants::InstrCost); @@ -814,7 +980,7 @@ bool CallAnalyzer::visitStore(StoreInst &I) { Value *SROAArg; - DenseMap::iterator CostIt; + DenseMap::iterator CostIt; if (lookupSROAArgAndCost(I.getPointerOperand(), SROAArg, CostIt)) { if (I.isSimple()) { accumulateSROACost(CostIt, InlineConstants::InstrCost); @@ -864,15 +1030,17 @@ // inside of instsimplify, directly constant fold calls here. if (!canConstantFoldCallTo(F)) return false; - + bool SVLookup = false; // Try to re-map the arguments to constants. SmallVector ConstantArgs; ConstantArgs.reserve(CS.arg_size()); for (CallSite::arg_iterator I = CS.arg_begin(), E = CS.arg_end(); I != E; ++I) { Constant *C = dyn_cast(*I); - if (!C) + if (!C) { C = dyn_cast_or_null(SimplifiedValues.lookup(*I)); + SVLookup = true; + } if (!C) return false; // This argument doesn't map to a constant. @@ -880,6 +1048,19 @@ } if (Constant *C = ConstantFoldCall(F, ConstantArgs)) { SimplifiedValues[CS.getInstruction()] = C; + + // If SVLookup is true, then the constness of at least one argument is + // inferred only after looking up the SimplifiedValues map. In other words, + // it is not possible to infer the the constness of all the arguments (and + // hence the result of the call due to constant folding) without inlining. + // Thus we account the savings associated with the elimination of this + // callsite to inlining. + // + // FIXME: Increase the savings associated with simplifying a callsite. + if (SVLookup) + accumulateSavings(InlineConstants::InstrCost + + InlineConstants::CallPenalty); + return true; } @@ -910,7 +1091,7 @@ case Intrinsic::load_relative: // This is normally lowered to 4 LLVM instructions. - Cost += 3 * InlineConstants::InstrCost; + accumulateCost(3 * InlineConstants::InstrCost); return false; case Intrinsic::memset: @@ -934,12 +1115,12 @@ if (TTI.isLoweredToCall(F)) { // We account for the average 1 instruction per call argument setup // here. - Cost += CS.arg_size() * InlineConstants::InstrCost; + accumulateCost(CS.arg_size() * InlineConstants::InstrCost); // Everything other than inline ASM will also have a significant cost // merely from making the call. if (!isa(CS.getCalledValue())) - Cost += InlineConstants::CallPenalty; + accumulateCost(InlineConstants::CallPenalty); } return Base::visitCallSite(CS); @@ -951,7 +1132,7 @@ // First, pay the price of the argument setup. We account for the average // 1 instruction per call argument setup here. - Cost += CS.arg_size() * InlineConstants::InstrCost; + accumulateCost(CS.arg_size() * InlineConstants::InstrCost); // Next, check if this happens to be an indirect function call to a known // function in this inline context. If not, we've done all we can. @@ -972,6 +1153,13 @@ // We were able to inline the indirect call! Subtract the cost from the // threshold to get the bonus we want to apply, but don't go below zero. Cost -= std::max(0, CA.getThreshold() - CA.getCost()); + // FIXME: This underestimates the savings due to removing a call. Perhaps we + // should get the weighted cost and savings of inlining the indirect call, + // scale it based on callsite and callee's enrtry frequencies and add them + // up to the current callee's weighted cost and savings. + int ArgPassingCost = getArgPassingCost(CS, F); + accumulateSavings(InlineConstants::InstrCost + + InlineConstants::CallPenalty - ArgPassingCost); } return Base::visitCallSite(CS); @@ -989,9 +1177,14 @@ // shouldn't exist at all, but handling them makes the behavior of the // inliner more regular and predictable. Interestingly, conditional branches // which will fold away are also free. - return BI.isUnconditional() || isa(BI.getCondition()) || - dyn_cast_or_null( - SimplifiedValues.lookup(BI.getCondition())); + if (BI.isUnconditional() || isa(BI.getCondition())) + return true; + if (dyn_cast_or_null( + SimplifiedValues.lookup(BI.getCondition()))) { + accumulateSavings(InlineConstants::InstrCost); + return true; + } + return false; } bool CallAnalyzer::visitSwitchInst(SwitchInst &SI) { @@ -999,26 +1192,35 @@ // branches. if (isa(SI.getCondition())) return true; + + // Lambda to compute the cost of switch instruction. + auto SwitchInstCost = [](SwitchInst &SI) { + // We need to compute a cost proportional to the number of + // distinct successor blocks. This fan-out in the CFG cannot be represented + // for free even if we can represent the core switch as a jumptable that + // takes a single instruction. + // + // NB: We convert large switches which are just used to initialize large phi + // nodes to lookup tables instead in simplify-cfg, so this shouldn't prevent + // inlining those. It will prevent inlining in cases where the optimization + // does not (yet) fire. + SmallPtrSet SuccessorBlocks; + SuccessorBlocks.insert(SI.getDefaultDest()); + for (auto I = SI.case_begin(), E = SI.case_end(); I != E; ++I) + SuccessorBlocks.insert(I.getCaseSuccessor()); + + // Add cost corresponding to the number of distinct destinations. The first + // we model as free because of fallthrough. + return (SuccessorBlocks.size() - 1) * InlineConstants::InstrCost; + }; + if (Value *V = SimplifiedValues.lookup(SI.getCondition())) - if (isa(V)) + if (isa(V)) { + accumulateSavings(SwitchInstCost(SI)); return true; + } - // Otherwise, we need to accumulate a cost proportional to the number of - // distinct successor blocks. This fan-out in the CFG cannot be represented - // for free even if we can represent the core switch as a jumptable that - // takes a single instruction. - // - // NB: We convert large switches which are just used to initialize large phi - // nodes to lookup tables instead in simplify-cfg, so this shouldn't prevent - // inlining those. It will prevent inlining in cases where the optimization - // does not (yet) fire. - SmallPtrSet SuccessorBlocks; - SuccessorBlocks.insert(SI.getDefaultDest()); - for (auto I = SI.case_begin(), E = SI.case_end(); I != E; ++I) - SuccessorBlocks.insert(I.getCaseSuccessor()); - // Add cost corresponding to the number of distinct destinations. The first - // we model as free because of fallthrough. - Cost += (SuccessorBlocks.size() - 1) * InlineConstants::InstrCost; + accumulateCost(SwitchInstCost(SI)); return false; } @@ -1083,6 +1285,9 @@ /// viable, and true if inlining remains viable. bool CallAnalyzer::analyzeBlock(BasicBlock *BB, SmallPtrSetImpl &EphValues) { + if (CalleeBFI) + CurrBBFreq = CalleeBFI->getBlockFreq(BB).getFrequency(); + for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; ++I) { // FIXME: Currently, the number of instructions in a function regardless of // our ability to simplify them during inline to constants or dead code, @@ -1119,7 +1324,7 @@ if (TTI.getFPOpCost(I->getType()) == TargetTransformInfo::TCC_Expensive || hasSoftFloatAttr) - Cost += InlineConstants::CallPenalty; + accumulateCost(InlineConstants::CallPenalty); } // If the instruction simplified to a constant, there is no cost to this @@ -1130,7 +1335,7 @@ if (Base::visit(&*I)) ++NumInstructionsSimplified; else - Cost += InlineConstants::InstrCost; + accumulateCost(InlineConstants::InstrCost); // If the visit this instruction detected an uninlinable pattern, abort. if (IsRecursiveCall || ExposesReturnsTwice || HasDynamicAlloca || @@ -1215,54 +1420,45 @@ assert(NumInstructions == 0); assert(NumVectorInstructions == 0); + Function *Caller = CS.getInstruction()->getParent()->getParent(); + CallerBFI = nullptr; + CalleeBFI = nullptr; + if (GetBFI) { + CallerBFI = &((*GetBFI)(*Caller)); + if (!F.isDeclaration()) + CalleeBFI = &((*GetBFI)(F)); + // While evaluating the weighted savings due to removal of argument setup + // and the call overhead, we want to use the entry block's frequency. + CurrBBFreq = CalleeBFI->getEntryFreq(); + } // Update the threshold based on callsite properties updateThreshold(CS, F); FiftyPercentVectorBonus = 3 * Threshold / 2; TenPercentVectorBonus = 3 * Threshold / 4; - const DataLayout &DL = F.getParent()->getDataLayout(); // Track whether the post-inlining function would have more than one basic // block. A single basic block is often intended for inlining. Balloon the // threshold by 50% until we pass the single-BB phase. bool SingleBB = true; int SingleBBBonus = Threshold / 2; + int SpeedupBonus = getSpeedupBonus(CS, Threshold); // Speculatively apply all possible bonuses to Threshold. If cost exceeds // this Threshold any time, and cost cannot decrease, we can stop processing // the rest of the function body. - Threshold += (SingleBBBonus + FiftyPercentVectorBonus); - - // Give out bonuses per argument, as the instructions setting them up will - // be gone after inlining. - for (unsigned I = 0, E = CS.arg_size(); I != E; ++I) { - if (CS.isByValArgument(I)) { - // We approximate the number of loads and stores needed by dividing the - // size of the byval type by the target's pointer size. - PointerType *PTy = cast(CS.getArgument(I)->getType()); - unsigned TypeSize = DL.getTypeSizeInBits(PTy->getElementType()); - unsigned PointerSize = DL.getPointerSizeInBits(); - // Ceiling division. - unsigned NumStores = (TypeSize + PointerSize - 1) / PointerSize; + Threshold += (SingleBBBonus + FiftyPercentVectorBonus + SpeedupBonus); - // If it generates more than 8 stores it is likely to be expanded as an - // inline memcpy so we take that as an upper bound. Otherwise we assume - // one load and one store per word copied. - // FIXME: The maxStoresPerMemcpy setting from the target should be used - // here instead of a magic number of 8, but it's not available via - // DataLayout. - NumStores = std::min(NumStores, 8U); + int ArgPassingCost = getArgPassingCost(CS, &F); + accumulateCost(ArgPassingCost); + // Argument passing cost is negative. We negate that to get savings due to + // inlining. + accumulateSavings(-ArgPassingCost); - Cost -= 2 * NumStores * InlineConstants::InstrCost; - } else { - // For non-byval arguments subtract off one instruction per call - // argument. - Cost -= InlineConstants::InstrCost; - } - } // The call instruction also disappears after inlining. - Cost -= InlineConstants::InstrCost + InlineConstants::CallPenalty; - + accumulateCost(-InlineConstants::InstrCost - InlineConstants::CallPenalty); + accumulateSavings(InlineConstants::InstrCost + InlineConstants::CallPenalty); + // If there is only one call of the function, and it has internal linkage, // the cost of inlining it drops dramatically. bool OnlyOneCallAndLocalLinkage = @@ -1273,7 +1469,7 @@ // If this function uses the coldcc calling convention, prefer not to inline // it. if (F.getCallingConv() == CallingConv::Cold) - Cost += InlineConstants::ColdccPenalty; + accumulateCost(InlineConstants::ColdccPenalty); // Check if we're done. This can happen due to bonuses and penalties. if (Cost > Threshold) @@ -1282,7 +1478,6 @@ if (F.empty()) return true; - Function *Caller = CS.getInstruction()->getParent()->getParent(); // Check if the caller function is recursive itself. for (User *U : Caller->users()) { CallSite Site(U); @@ -1311,7 +1506,7 @@ // We can SROA any pointer arguments derived from alloca instructions. if (isa(PtrArg)) { SROAArgValues[&*FAI] = PtrArg; - SROAArgCosts[PtrArg] = 0; + SROAArgCosts[PtrArg] = {0, 0}; } } } @@ -1415,6 +1610,9 @@ else if (NumVectorInstructions <= NumInstructions / 2) Threshold -= (FiftyPercentVectorBonus - TenPercentVectorBonus); + if (!hasLargeSpeedup()) + Threshold -= SpeedupBonus; + return Cost < std::max(1, Threshold); } @@ -1433,6 +1631,8 @@ DEBUG_PRINT_STAT(SROACostSavingsLost); DEBUG_PRINT_STAT(ContainsNoDuplicateCall); DEBUG_PRINT_STAT(Cost); + DEBUG_PRINT_STAT(WeightedCost); + DEBUG_PRINT_STAT(WeightedSavings); DEBUG_PRINT_STAT(Threshold); #undef DEBUG_PRINT_STAT } Index: test/Transforms/Inline/speedup-analysis.ll =================================================================== --- /dev/null +++ test/Transforms/Inline/speedup-analysis.ll @@ -0,0 +1,35 @@ +; RUN: opt < %s -passes='require,cgscc(inline)' -inline-threshold=10 -speedup-bonus-percent=1000 +; Test that a callee that does not fit within the threshold gets inlined +; because of the estimated speedup heuristic. +define i32 @caller(i32 %n) { +; CHECK-LABEL: define i32 @caller +entry: + br label %loop +loop: + %r = phi i32 [%n, %entry], [%result, %loop] +; CHECK-NOT: call i32 @callee + %result = call i32 @callee(i32 %r) + %cond = icmp sle i32 %result, 100 + br i1 %cond, label %loop, label %exit +exit: +; CHECK: ret + ret i32 %result +} + +define i32 @callee(i32 %n) { + %cond = icmp sle i32 %n, 100 + br i1 %cond, label %cond_true, label %cond_false, !prof !0 + +cond_true: + %n1 = add i32 %n, 1 + %n2 = add i32 %n1, 1 + ret i32 %n2 +cond_false: + call void @extern() + call void @extern() + call void @extern() + ret i32 0 +} +declare void @extern() + +!0 = !{!"branch_weights", i32 1, i32 0} Index: test/Transforms/Inline/speedup-analysis2.ll =================================================================== --- /dev/null +++ test/Transforms/Inline/speedup-analysis2.ll @@ -0,0 +1,69 @@ +; RUN: opt < %s -passes='require,cgscc(inline)' -inline-threshold=10 -speedup-bonus-percent=1000 +; Test that a callee that does not fit within the threshold gets inlined +; because of the estimated speedup heuristic. The callee has a switch statement. +; Since only one of the cases is executed per invocation of the callee, the +; weighted cost of the callee is low and results in a big relative speedup as the +; benefits of removing the function call is accounted for. + +define i32 @caller(i32 %n) { +; CHECK-LABEL: define i32 @caller +entry: + br label %loop +loop: + %r = phi i32 [%n, %entry], [%result, %loop] +; CHECK-NOT: call i32 @callee + %result = call i32 @callee(i32 %r) + %cond = icmp sle i32 %result, 100 + br i1 %cond, label %loop, label %exit +exit: +; CHECK: ret + ret i32 %result +} + +define i32 @callee(i32 %n) { +entry: + switch i32 %n, label %return [ + i32 1, label %sw.bb1 + i32 2, label %sw.bb2 + i32 3, label %sw.bb3 + i32 4, label %sw.bb4 + i32 5, label %sw.bb5 + i32 6, label %sw.bb6 + i32 7, label %sw.bb7 + ] + +sw.bb1: + %r1 = add i32 %n, 1 + br label %return + +sw.bb2: + %r2 = add i32 %n, 2 + br label %return + +sw.bb3: + %r3 = add i32 %n, 3 + br label %return + +sw.bb4: + %r4 = add i32 %n, 4 + br label %return + +sw.bb5: + %r5 = add i32 %n, 5 + br label %return + +sw.bb6: + %r6 = add i32 %n, 6 + br label %return + +sw.bb7: + %r7 = add i32 %n, 7 + br label %return + +return: + %res = phi i32 [%n, %entry], [%r1, %sw.bb1], [%r2, %sw.bb2], [%r3, %sw.bb3], [%r4, %sw.bb4], [%r5, %sw.bb5], [%r6, %sw.bb6], [%r7, %sw.bb7] + ret i32 %res + +} +declare void @extern() +