blob: a35f5e11f0e0c6ea72b5284bc61320d37c4e821c [file] [log] [blame]
//===- InlineCost.cpp - Cost analysis for inliner -------------------------===//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
// This file implements inline cost analysis.
#include "llvm/Analysis/InlineCost.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/SetVector.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/Analysis/AssumptionCache.h"
#include "llvm/Analysis/BlockFrequencyInfo.h"
#include "llvm/Analysis/CFG.h"
#include "llvm/Analysis/CodeMetrics.h"
#include "llvm/Analysis/ConstantFolding.h"
#include "llvm/Analysis/InstructionSimplify.h"
#include "llvm/Analysis/LoopInfo.h"
#include "llvm/Analysis/ProfileSummaryInfo.h"
#include "llvm/Analysis/TargetLibraryInfo.h"
#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/Analysis/ValueTracking.h"
#include "llvm/Config/llvm-config.h"
#include "llvm/IR/AssemblyAnnotationWriter.h"
#include "llvm/IR/CallingConv.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/GetElementPtrTypeIterator.h"
#include "llvm/IR/GlobalAlias.h"
#include "llvm/IR/InstVisitor.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/Operator.h"
#include "llvm/IR/PatternMatch.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/FormattedStream.h"
#include "llvm/Support/raw_ostream.h"
using namespace llvm;
#define DEBUG_TYPE "inline-cost"
STATISTIC(NumCallsAnalyzed, "Number of call sites analyzed");
static cl::opt<int>
DefaultThreshold("inlinedefault-threshold", cl::Hidden, cl::init(225),
cl::desc("Default amount of inlining to perform"));
static cl::opt<bool> PrintInstructionComments(
"print-instruction-comments", cl::Hidden, cl::init(false),
cl::desc("Prints comments for instruction based on inline cost analysis"));
static cl::opt<int> InlineThreshold(
"inline-threshold", cl::Hidden, cl::init(225), cl::ZeroOrMore,
cl::desc("Control the amount of inlining to perform (default = 225)"));
static cl::opt<int> HintThreshold(
"inlinehint-threshold", cl::Hidden, cl::init(325), cl::ZeroOrMore,
cl::desc("Threshold for inlining functions with inline hint"));
static cl::opt<int>
ColdCallSiteThreshold("inline-cold-callsite-threshold", cl::Hidden,
cl::init(45), cl::ZeroOrMore,
cl::desc("Threshold for inlining cold callsites"));
static cl::opt<bool> InlineEnableCostBenefitAnalysis(
"inline-enable-cost-benefit-analysis", cl::Hidden, cl::init(false),
cl::desc("Enable the cost-benefit analysis for the inliner"));
static cl::opt<int> InlineSavingsMultiplier(
"inline-savings-multiplier", cl::Hidden, cl::init(8), cl::ZeroOrMore,
cl::desc("Multiplier to multiply cycle savings by during inlining"));
static cl::opt<int>
InlineSizeAllowance("inline-size-allowance", cl::Hidden, cl::init(100),
cl::desc("The maximum size of a callee that get's "
"inlined without sufficient cycle savings"));
// We introduce this threshold to help performance of instrumentation based
// PGO before we actually hook up inliner with analysis passes such as BPI and
// BFI.
static cl::opt<int> ColdThreshold(
"inlinecold-threshold", cl::Hidden, cl::init(45), cl::ZeroOrMore,
cl::desc("Threshold for inlining functions with cold attribute"));
static cl::opt<int>
HotCallSiteThreshold("hot-callsite-threshold", cl::Hidden, cl::init(3000),
cl::desc("Threshold for hot callsites "));
static cl::opt<int> LocallyHotCallSiteThreshold(
"locally-hot-callsite-threshold", cl::Hidden, cl::init(525), cl::ZeroOrMore,
cl::desc("Threshold for locally hot callsites "));
static cl::opt<int> ColdCallSiteRelFreq(
"cold-callsite-rel-freq", cl::Hidden, cl::init(2), cl::ZeroOrMore,
cl::desc("Maximum block frequency, expressed as a percentage of caller's "
"entry frequency, for a callsite to be cold in the absence of "
"profile information."));
static cl::opt<int> HotCallSiteRelFreq(
"hot-callsite-rel-freq", cl::Hidden, cl::init(60), cl::ZeroOrMore,
cl::desc("Minimum block frequency, expressed as a multiple of caller's "
"entry frequency, for a callsite to be hot in the absence of "
"profile information."));
static cl::opt<bool> OptComputeFullInlineCost(
"inline-cost-full", cl::Hidden, cl::init(false), cl::ZeroOrMore,
cl::desc("Compute the full inline cost of a call site even when the cost "
"exceeds the threshold."));
static cl::opt<bool> InlineCallerSupersetNoBuiltin(
"inline-caller-superset-nobuiltin", cl::Hidden, cl::init(true),
cl::desc("Allow inlining when caller has a superset of callee's nobuiltin "
static cl::opt<bool> DisableGEPConstOperand(
"disable-gep-const-evaluation", cl::Hidden, cl::init(false),
cl::desc("Disables evaluation of GetElementPtr with constant operands"));
namespace {
class InlineCostCallAnalyzer;
// This struct is used to store information about inline cost of a
// particular instruction
struct InstructionCostDetail {
int CostBefore = 0;
int CostAfter = 0;
int ThresholdBefore = 0;
int ThresholdAfter = 0;
int getThresholdDelta() const { return ThresholdAfter - ThresholdBefore; }
int getCostDelta() const { return CostAfter - CostBefore; }
bool hasThresholdChanged() const { return ThresholdAfter != ThresholdBefore; }
class InlineCostAnnotationWriter : public AssemblyAnnotationWriter {
InlineCostCallAnalyzer *const ICCA;
InlineCostAnnotationWriter(InlineCostCallAnalyzer *ICCA) : ICCA(ICCA) {}
virtual void emitInstructionAnnot(const Instruction *I,
formatted_raw_ostream &OS) override;
/// Carry out call site analysis, in order to evaluate inlinability.
/// NOTE: the type is currently used as implementation detail of functions such
/// as llvm::getInlineCost. Note the function_ref constructor parameters - the
/// expectation is that they come from the outer scope, from the wrapper
/// functions. If we want to support constructing CallAnalyzer objects where
/// lambdas are provided inline at construction, or where the object needs to
/// otherwise survive past the scope of the provided functions, we need to
/// revisit the argument types.
class CallAnalyzer : public InstVisitor<CallAnalyzer, bool> {
typedef InstVisitor<CallAnalyzer, bool> Base;
friend class InstVisitor<CallAnalyzer, bool>;
virtual ~CallAnalyzer() {}
/// The TargetTransformInfo available for this compilation.
const TargetTransformInfo &TTI;
/// Getter for the cache of @llvm.assume intrinsics.
function_ref<AssumptionCache &(Function &)> GetAssumptionCache;
/// Getter for BlockFrequencyInfo
function_ref<BlockFrequencyInfo &(Function &)> GetBFI;
/// Profile summary information.
ProfileSummaryInfo *PSI;
/// The called function.
Function &F;
// Cache the DataLayout since we use it a lot.
const DataLayout &DL;
/// The OptimizationRemarkEmitter available for this compilation.
OptimizationRemarkEmitter *ORE;
/// The candidate callsite being analyzed. Please do not use this to do
/// analysis in the caller function; we want the inline cost query to be
/// easily cacheable. Instead, use the cover function paramHasAttr.
CallBase &CandidateCall;
/// Extension points for handling callsite features.
// Called before a basic block was analyzed.
virtual void onBlockStart(const BasicBlock *BB) {}
/// Called after a basic block was analyzed.
virtual void onBlockAnalyzed(const BasicBlock *BB) {}
/// Called before an instruction was analyzed
virtual void onInstructionAnalysisStart(const Instruction *I) {}
/// Called after an instruction was analyzed
virtual void onInstructionAnalysisFinish(const Instruction *I) {}
/// Called at the end of the analysis of the callsite. Return the outcome of
/// the analysis, i.e. 'InlineResult(true)' if the inlining may happen, or
/// the reason it can't.
virtual InlineResult finalizeAnalysis() { return InlineResult::success(); }
/// Called when we're about to start processing a basic block, and every time
/// we are done processing an instruction. Return true if there is no point in
/// continuing the analysis (e.g. we've determined already the call site is
/// too expensive to inline)
virtual bool shouldStop() { return false; }
/// Called before the analysis of the callee body starts (with callsite
/// contexts propagated). It checks callsite-specific information. Return a
/// reason analysis can't continue if that's the case, or 'true' if it may
/// continue.
virtual InlineResult onAnalysisStart() { return InlineResult::success(); }
/// Called if the analysis engine decides SROA cannot be done for the given
/// alloca.
virtual void onDisableSROA(AllocaInst *Arg) {}
/// Called the analysis engine determines load elimination won't happen.
virtual void onDisableLoadElimination() {}
/// Called to account for a call.
virtual void onCallPenalty() {}
/// Called to account for the expectation the inlining would result in a load
/// elimination.
virtual void onLoadEliminationOpportunity() {}
/// Called to account for the cost of argument setup for the Call in the
/// callee's body (not the callsite currently under analysis).
virtual void onCallArgumentSetup(const CallBase &Call) {}
/// Called to account for a load relative intrinsic.
virtual void onLoadRelativeIntrinsic() {}
/// Called to account for a lowered call.
virtual void onLoweredCall(Function *F, CallBase &Call, bool IsIndirectCall) {
/// Account for a jump table of given size. Return false to stop further
/// processing the switch instruction
virtual bool onJumpTable(unsigned JumpTableSize) { return true; }
/// Account for a case cluster of given size. Return false to stop further
/// processing of the instruction.
virtual bool onCaseCluster(unsigned NumCaseCluster) { return true; }
/// Called at the end of processing a switch instruction, with the given
/// number of case clusters.
virtual void onFinalizeSwitch(unsigned JumpTableSize,
unsigned NumCaseCluster) {}
/// Called to account for any other instruction not specifically accounted
/// for.
virtual void onMissedSimplification() {}
/// Start accounting potential benefits due to SROA for the given alloca.
virtual void onInitializeSROAArg(AllocaInst *Arg) {}
/// Account SROA savings for the AllocaInst value.
virtual void onAggregateSROAUse(AllocaInst *V) {}
bool handleSROA(Value *V, bool DoNotDisable) {
// Check for SROA candidates in comparisons.
if (auto *SROAArg = getSROAArgForValueOrNull(V)) {
if (DoNotDisable) {
return true;
return false;
bool IsCallerRecursive = false;
bool IsRecursiveCall = false;
bool ExposesReturnsTwice = false;
bool HasDynamicAlloca = false;
bool ContainsNoDuplicateCall = false;
bool HasReturn = false;
bool HasIndirectBr = false;
bool HasUninlineableIntrinsic = false;
bool InitsVargArgs = false;
/// Number of bytes allocated statically by the callee.
uint64_t AllocatedSize = 0;
unsigned NumInstructions = 0;
unsigned NumVectorInstructions = 0;
/// While we walk the potentially-inlined instructions, we build up and
/// maintain a mapping of simplified values specific to this callsite. The
/// idea is to propagate any special information we have about arguments to
/// this call through the inlinable section of the function, and account for
/// likely simplifications post-inlining. The most important aspect we track
/// is CFG altering simplifications -- when we prove a basic block dead, that
/// can cause dramatic shifts in the cost of inlining a function.
DenseMap<Value *, Constant *> SimplifiedValues;
/// Keep track of the values which map back (through function arguments) to
/// allocas on the caller stack which could be simplified through SROA.
DenseMap<Value *, AllocaInst *> SROAArgValues;
/// Keep track of Allocas for which we believe we may get SROA optimization.
DenseSet<AllocaInst *> EnabledSROAAllocas;
/// Keep track of values which map to a pointer base and constant offset.
DenseMap<Value *, std::pair<Value *, APInt>> ConstantOffsetPtrs;
/// Keep track of dead blocks due to the constant arguments.
SetVector<BasicBlock *> DeadBlocks;
/// The mapping of the blocks to their known unique successors due to the
/// constant arguments.
DenseMap<BasicBlock *, BasicBlock *> KnownSuccessors;
/// Model the elimination of repeated loads that is expected to happen
/// whenever we simplify away the stores that would otherwise cause them to be
/// loads.
bool EnableLoadElimination;
SmallPtrSet<Value *, 16> LoadAddrSet;
AllocaInst *getSROAArgForValueOrNull(Value *V) const {
auto It = SROAArgValues.find(V);
if (It == SROAArgValues.end() || EnabledSROAAllocas.count(It->second) == 0)
return nullptr;
return It->second;
// Custom simplification helper routines.
bool isAllocaDerivedArg(Value *V);
void disableSROAForArg(AllocaInst *SROAArg);
void disableSROA(Value *V);
void findDeadBlocks(BasicBlock *CurrBB, BasicBlock *NextBB);
void disableLoadElimination();
bool isGEPFree(GetElementPtrInst &GEP);
bool canFoldInboundsGEP(GetElementPtrInst &I);
bool accumulateGEPOffset(GEPOperator &GEP, APInt &Offset);
bool simplifyCallSite(Function *F, CallBase &Call);
template <typename Callable>
bool simplifyInstruction(Instruction &I, Callable Evaluate);
ConstantInt *stripAndComputeInBoundsConstantOffsets(Value *&V);
/// Return true if the given argument to the function being considered for
/// inlining has the given attribute set either at the call site or the
/// function declaration. Primarily used to inspect call site specific
/// attributes since these can be more precise than the ones on the callee
/// itself.
bool paramHasAttr(Argument *A, Attribute::AttrKind Attr);
/// Return true if the given value is known non null within the callee if
/// inlined through this particular callsite.
bool isKnownNonNullInCallee(Value *V);
/// Return true if size growth is allowed when inlining the callee at \p Call.
bool allowSizeGrowth(CallBase &Call);
// Custom analysis routines.
InlineResult analyzeBlock(BasicBlock *BB,
SmallPtrSetImpl<const Value *> &EphValues);
// Disable several entry points to the visitor so we don't accidentally use
// them by declaring but not defining them here.
void visit(Module *);
void visit(Module &);
void visit(Function *);
void visit(Function &);
void visit(BasicBlock *);
void visit(BasicBlock &);
// Provide base case for our instruction visit.
bool visitInstruction(Instruction &I);
// Our visit overrides.
bool visitAlloca(AllocaInst &I);
bool visitPHI(PHINode &I);
bool visitGetElementPtr(GetElementPtrInst &I);
bool visitBitCast(BitCastInst &I);
bool visitPtrToInt(PtrToIntInst &I);
bool visitIntToPtr(IntToPtrInst &I);
bool visitCastInst(CastInst &I);
bool visitUnaryInstruction(UnaryInstruction &I);
bool visitCmpInst(CmpInst &I);
bool visitSub(BinaryOperator &I);
bool visitBinaryOperator(BinaryOperator &I);
bool visitFNeg(UnaryOperator &I);
bool visitLoad(LoadInst &I);
bool visitStore(StoreInst &I);
bool visitExtractValue(ExtractValueInst &I);
bool visitInsertValue(InsertValueInst &I);
bool visitCallBase(CallBase &Call);
bool visitReturnInst(ReturnInst &RI);
bool visitBranchInst(BranchInst &BI);
bool visitSelectInst(SelectInst &SI);
bool visitSwitchInst(SwitchInst &SI);
bool visitIndirectBrInst(IndirectBrInst &IBI);
bool visitResumeInst(ResumeInst &RI);
bool visitCleanupReturnInst(CleanupReturnInst &RI);
bool visitCatchReturnInst(CatchReturnInst &RI);
bool visitUnreachableInst(UnreachableInst &I);
Function &Callee, CallBase &Call, const TargetTransformInfo &TTI,
function_ref<AssumptionCache &(Function &)> GetAssumptionCache,
function_ref<BlockFrequencyInfo &(Function &)> GetBFI = nullptr,
ProfileSummaryInfo *PSI = nullptr,
OptimizationRemarkEmitter *ORE = nullptr)
: TTI(TTI), GetAssumptionCache(GetAssumptionCache), GetBFI(GetBFI),
PSI(PSI), F(Callee), DL(F.getParent()->getDataLayout()), ORE(ORE),
CandidateCall(Call), EnableLoadElimination(true) {}
InlineResult analyze();
Optional<Constant*> getSimplifiedValue(Instruction *I) {
if (SimplifiedValues.find(I) != SimplifiedValues.end())
return SimplifiedValues[I];
return None;
// Keep a bunch of stats about the cost savings found so we can print them
// out when debugging.
unsigned NumConstantArgs = 0;
unsigned NumConstantOffsetPtrArgs = 0;
unsigned NumAllocaArgs = 0;
unsigned NumConstantPtrCmps = 0;
unsigned NumConstantPtrDiffs = 0;
unsigned NumInstructionsSimplified = 0;
void dump();
/// FIXME: if it is necessary to derive from InlineCostCallAnalyzer, note
/// the FIXME in onLoweredCall, when instantiating an InlineCostCallAnalyzer
class InlineCostCallAnalyzer final : public CallAnalyzer {
const int CostUpperBound = INT_MAX - InlineConstants::InstrCost - 1;
const bool ComputeFullInlineCost;
int LoadEliminationCost = 0;
/// Bonus to be applied when percentage of vector instructions in callee is
/// high (see more details in updateThreshold).
int VectorBonus = 0;
/// Bonus to be applied when the callee has only one reachable basic block.
int SingleBBBonus = 0;
/// Tunable parameters that control the analysis.
const InlineParams &Params;
// This DenseMap stores the delta change in cost and threshold after
// accounting for the given instruction. The map is filled only with the
// flag PrintInstructionComments on.
DenseMap<const Instruction *, InstructionCostDetail> InstructionCostDetailMap;
/// Upper bound for the inlining cost. Bonuses are being applied to account
/// for speculative "expected profit" of the inlining decision.
int Threshold = 0;
/// Attempt to evaluate indirect calls to boost its inline cost.
const bool BoostIndirectCalls;
/// Ignore the threshold when finalizing analysis.
const bool IgnoreThreshold;
// True if the cost-benefit-analysis-based inliner is enabled.
const bool CostBenefitAnalysisEnabled;
/// Inlining cost measured in abstract units, accounts for all the
/// instructions expected to be executed for a given function invocation.
/// Instructions that are statically proven to be dead based on call-site
/// arguments are not counted here.
int Cost = 0;
// The cumulative cost at the beginning of the basic block being analyzed. At
// the end of analyzing each basic block, "Cost - CostAtBBStart" represents
// the size of that basic block.
int CostAtBBStart = 0;
// The static size of live but cold basic blocks. This is "static" in the
// sense that it's not weighted by profile counts at all.
int ColdSize = 0;
bool SingleBB = true;
unsigned SROACostSavings = 0;
unsigned SROACostSavingsLost = 0;
/// 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<AllocaInst *, int> SROAArgCosts;
/// Return true if \p Call is a cold callsite.
bool isColdCallSite(CallBase &Call, BlockFrequencyInfo *CallerBFI);
/// Update Threshold based on callsite properties such as callee
/// attributes and callee hotness for PGO builds. The Callee is explicitly
/// passed to support analyzing indirect calls whose target is inferred by
/// analysis.
void updateThreshold(CallBase &Call, Function &Callee);
/// Return a higher threshold if \p Call is a hot callsite.
Optional<int> getHotCallSiteThreshold(CallBase &Call,
BlockFrequencyInfo *CallerBFI);
/// Handle a capped 'int' increment for Cost.
void addCost(int64_t Inc, int64_t UpperBound = INT_MAX) {
assert(UpperBound > 0 && UpperBound <= INT_MAX && "invalid upper bound");
Cost = (int)std::min(UpperBound, Cost + Inc);
void onDisableSROA(AllocaInst *Arg) override {
auto CostIt = SROAArgCosts.find(Arg);
if (CostIt == SROAArgCosts.end())
SROACostSavings -= CostIt->second;
SROACostSavingsLost += CostIt->second;
void onDisableLoadElimination() override {
LoadEliminationCost = 0;
void onCallPenalty() override { addCost(InlineConstants::CallPenalty); }
void onCallArgumentSetup(const CallBase &Call) override {
// Pay the price of the argument setup. We account for the average 1
// instruction per call argument setup here.
addCost(Call.arg_size() * InlineConstants::InstrCost);
void onLoadRelativeIntrinsic() override {
// This is normally lowered to 4 LLVM instructions.
addCost(3 * InlineConstants::InstrCost);
void onLoweredCall(Function *F, CallBase &Call,
bool IsIndirectCall) override {
// We account for the average 1 instruction per call argument setup here.
addCost(Call.arg_size() * InlineConstants::InstrCost);
// If we have a constant that we are calling as a function, we can peer
// through it and see the function target. This happens not infrequently
// during devirtualization and so we want to give it a hefty bonus for
// inlining, but cap that bonus in the event that inlining wouldn't pan out.
// Pretend to inline the function, with a custom threshold.
if (IsIndirectCall && BoostIndirectCalls) {
auto IndirectCallParams = Params;
IndirectCallParams.DefaultThreshold =
/// FIXME: if InlineCostCallAnalyzer is derived from, this may need
/// to instantiate the derived class.
InlineCostCallAnalyzer CA(*F, Call, IndirectCallParams, TTI,
GetAssumptionCache, GetBFI, PSI, ORE, false);
if (CA.analyze().isSuccess()) {
// 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());
} else
// Otherwise simply add the cost for merely making the call.
void onFinalizeSwitch(unsigned JumpTableSize,
unsigned NumCaseCluster) override {
// If suitable for a jump table, consider the cost for the table size and
// branch to destination.
// Maximum valid cost increased in this function.
if (JumpTableSize) {
int64_t JTCost = (int64_t)JumpTableSize * InlineConstants::InstrCost +
4 * InlineConstants::InstrCost;
addCost(JTCost, (int64_t)CostUpperBound);
// Considering forming a binary search, we should find the number of nodes
// which is same as the number of comparisons when lowered. For a given
// number of clusters, n, we can define a recursive function, f(n), to find
// the number of nodes in the tree. The recursion is :
// f(n) = 1 + f(n/2) + f (n - n/2), when n > 3,
// and f(n) = n, when n <= 3.
// This will lead a binary tree where the leaf should be either f(2) or f(3)
// when n > 3. So, the number of comparisons from leaves should be n, while
// the number of non-leaf should be :
// 2^(log2(n) - 1) - 1
// = 2^log2(n) * 2^-1 - 1
// = n / 2 - 1.
// Considering comparisons from leaf and non-leaf nodes, we can estimate the
// number of comparisons in a simple closed form :
// n + n / 2 - 1 = n * 3 / 2 - 1
if (NumCaseCluster <= 3) {
// Suppose a comparison includes one compare and one conditional branch.
addCost(NumCaseCluster * 2 * InlineConstants::InstrCost);
int64_t ExpectedNumberOfCompare = 3 * (int64_t)NumCaseCluster / 2 - 1;
int64_t SwitchCost =
ExpectedNumberOfCompare * 2 * InlineConstants::InstrCost;
addCost(SwitchCost, (int64_t)CostUpperBound);
void onMissedSimplification() override {
void onInitializeSROAArg(AllocaInst *Arg) override {
assert(Arg != nullptr &&
"Should not initialize SROA costs for null value.");
SROAArgCosts[Arg] = 0;
void onAggregateSROAUse(AllocaInst *SROAArg) override {
auto CostIt = SROAArgCosts.find(SROAArg);
assert(CostIt != SROAArgCosts.end() &&
"expected this argument to have a cost");
CostIt->second += InlineConstants::InstrCost;
SROACostSavings += InlineConstants::InstrCost;
void onBlockStart(const BasicBlock *BB) override { CostAtBBStart = Cost; }
void onBlockAnalyzed(const BasicBlock *BB) override {
if (CostBenefitAnalysisEnabled) {
// Keep track of the static size of live but cold basic blocks. For now,
// we define a cold basic block to be one that's never executed.
assert(GetBFI && "GetBFI must be available");
BlockFrequencyInfo *BFI = &(GetBFI(F));
assert(BFI && "BFI must be available");
auto ProfileCount = BFI->getBlockProfileCount(BB);
if (ProfileCount.getValue() == 0)
ColdSize += Cost - CostAtBBStart;
auto *TI = BB->getTerminator();
// If we had any successors at this point, than post-inlining is likely to
// have them as well. Note that we assume any basic blocks which existed
// due to branches or switches which folded above will also fold after
// inlining.
if (SingleBB && TI->getNumSuccessors() > 1) {
// Take off the bonus we applied to the threshold.
Threshold -= SingleBBBonus;
SingleBB = false;
void onInstructionAnalysisStart(const Instruction *I) override {
// This function is called to store the initial cost of inlining before
// the given instruction was assessed.
if (!PrintInstructionComments)
InstructionCostDetailMap[I].CostBefore = Cost;
InstructionCostDetailMap[I].ThresholdBefore = Threshold;
void onInstructionAnalysisFinish(const Instruction *I) override {
// This function is called to find new values of cost and threshold after
// the instruction has been assessed.
if (!PrintInstructionComments)
InstructionCostDetailMap[I].CostAfter = Cost;
InstructionCostDetailMap[I].ThresholdAfter = Threshold;
bool isCostBenefitAnalysisEnabled() {
if (!InlineEnableCostBenefitAnalysis)
return false;
if (!PSI || !PSI->hasProfileSummary())
return false;
if (!GetBFI)
return false;
auto *Caller = CandidateCall.getParent()->getParent();
if (!Caller->getEntryCount())
return false;
BlockFrequencyInfo *CallerBFI = &(GetBFI(*Caller));
if (!CallerBFI)
return false;
// For now, limit to hot call site.
if (!PSI->isHotCallSite(CandidateCall, CallerBFI))
return false;
if (!F.getEntryCount())
return false;
BlockFrequencyInfo *CalleeBFI = &(GetBFI(F));
if (!CalleeBFI)
return false;
return true;
// Determine whether we should inline the given call site, taking into account
// both the size cost and the cycle savings. Return None if we don't have
// suficient profiling information to determine.
Optional<bool> costBenefitAnalysis() {
if (!CostBenefitAnalysisEnabled)
return None;
// buildInlinerPipeline in the pass builder sets HotCallSiteThreshold to 0
// for the prelink phase of the AutoFDO + ThinLTO build. Honor the logic by
// falling back to the cost-based metric.
// TODO: Improve this hacky condition.
if (Threshold == 0)
return None;
BlockFrequencyInfo *CalleeBFI = &(GetBFI(F));
// The cycle savings expressed as the sum of InlineConstants::InstrCost
// multiplied by the estimated dynamic count of each instruction we can
// avoid. Savings come from the call site cost, such as argument setup and
// the call instruction, as well as the instructions that are folded.
// We use 128-bit APInt here to avoid potential overflow. This variable
// should stay well below 10^^24 (or 2^^80) in practice. This "worst" case
// assumes that we can avoid or fold a billion instructions, each with a
// profile count of 10^^15 -- roughly the number of cycles for a 24-hour
// period on a 4GHz machine.
APInt CycleSavings(128, 0);
for (auto &BB : F) {
APInt CurrentSavings(128, 0);
for (auto &I : BB) {
if (BranchInst *BI = dyn_cast<BranchInst>(&I)) {
// Count a conditional branch as savings if it becomes unconditional.
if (BI->isConditional() &&
SimplifiedValues.lookup(BI->getCondition()))) {
CurrentSavings += InlineConstants::InstrCost;
} else if (Value *V = dyn_cast<Value>(&I)) {
// Count an instruction as savings if we can fold it.
if (SimplifiedValues.count(V)) {
CurrentSavings += InlineConstants::InstrCost;
// TODO: Consider other forms of savings like switch statements,
// indirect calls becoming direct, SROACostSavings, LoadEliminationCost,
// etc.
auto ProfileCount = CalleeBFI->getBlockProfileCount(&BB);
CurrentSavings *= ProfileCount.getValue();
CycleSavings += CurrentSavings;
// Compute the cycle savings per call.
auto EntryProfileCount = F.getEntryCount();
auto EntryCount = EntryProfileCount.getCount();
CycleSavings += EntryCount / 2;
CycleSavings = CycleSavings.udiv(EntryCount);
// Compute the total savings for the call site.
auto *CallerBB = CandidateCall.getParent();
BlockFrequencyInfo *CallerBFI = &(GetBFI(*(CallerBB->getParent())));
CycleSavings += getCallsiteCost(this->CandidateCall, DL);
CycleSavings *= CallerBFI->getBlockProfileCount(CallerBB).getValue();
// Remove the cost of the cold basic blocks.
int Size = Cost - ColdSize;
// Allow tiny callees to be inlined regardless of whether they meet the
// savings threshold.
Size = Size > InlineSizeAllowance ? Size - InlineSizeAllowance : 1;
// Return true if the savings justify the cost of inlining. Specifically,
// we evaluate the following inequality:
// CycleSavings PSI->getOrCompHotCountThreshold()
// -------------- >= -----------------------------------
// Size InlineSavingsMultiplier
// Note that the left hand side is specific to a call site. The right hand
// side is a constant for the entire executable.
APInt LHS = CycleSavings;
LHS *= InlineSavingsMultiplier;
APInt RHS(128, PSI->getOrCompHotCountThreshold());
RHS *= Size;
return LHS.uge(RHS);
InlineResult finalizeAnalysis() override {
// Loops generally act a lot like calls in that they act like barriers to
// movement, require a certain amount of setup, etc. So when optimising for
// size, we penalise any call sites that perform loops. We do this after all
// other costs here, so will likely only be dealing with relatively small
// functions (and hence DT and LI will hopefully be cheap).
auto *Caller = CandidateCall.getFunction();
if (Caller->hasMinSize()) {
DominatorTree DT(F);
LoopInfo LI(DT);
int NumLoops = 0;
for (Loop *L : LI) {
// Ignore loops that will not be executed
if (DeadBlocks.count(L->getHeader()))
addCost(NumLoops * InlineConstants::CallPenalty);
// We applied the maximum possible vector bonus at the beginning. Now,
// subtract the excess bonus, if any, from the Threshold before
// comparing against Cost.
if (NumVectorInstructions <= NumInstructions / 10)
Threshold -= VectorBonus;
else if (NumVectorInstructions <= NumInstructions / 2)
Threshold -= VectorBonus / 2;
if (auto Result = costBenefitAnalysis()) {
if (Result.getValue())
return InlineResult::success();
return InlineResult::failure("Cost over threshold.");
if (IgnoreThreshold || Cost < std::max(1, Threshold))
return InlineResult::success();
return InlineResult::failure("Cost over threshold.");
bool shouldStop() override {
// Bail out the moment we cross the threshold. This means we'll under-count
// the cost, but only when undercounting doesn't matter.
return !IgnoreThreshold && Cost >= Threshold && !ComputeFullInlineCost;
void onLoadEliminationOpportunity() override {
LoadEliminationCost += InlineConstants::InstrCost;
InlineResult onAnalysisStart() override {
// Perform some tweaks to the cost and threshold based on the direct
// callsite information.
// We want to more aggressively inline vector-dense kernels, so up the
// threshold, and we'll lower it if the % of vector instructions gets too
// low. Note that these bonuses are some what arbitrary and evolved over
// time by accident as much as because they are principled bonuses.
// FIXME: It would be nice to remove all such bonuses. At least it would be
// nice to base the bonus values on something more scientific.
assert(NumInstructions == 0);
assert(NumVectorInstructions == 0);
// Update the threshold based on callsite properties
updateThreshold(CandidateCall, F);
// While Threshold depends on commandline options that can take negative
// values, we want to enforce the invariant that the computed threshold and
// bonuses are non-negative.
assert(Threshold >= 0);
assert(SingleBBBonus >= 0);
assert(VectorBonus >= 0);
// 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 + VectorBonus);
// Give out bonuses for the callsite, as the instructions setting them up
// will be gone after inlining.
addCost(-getCallsiteCost(this->CandidateCall, DL));
// If this function uses the coldcc calling convention, prefer not to inline
// it.
if (F.getCallingConv() == CallingConv::Cold)
Cost += InlineConstants::ColdccPenalty;
// Check if we're done. This can happen due to bonuses and penalties.
if (Cost >= Threshold && !ComputeFullInlineCost)
return InlineResult::failure("high cost");
return InlineResult::success();
Function &Callee, CallBase &Call, const InlineParams &Params,
const TargetTransformInfo &TTI,
function_ref<AssumptionCache &(Function &)> GetAssumptionCache,
function_ref<BlockFrequencyInfo &(Function &)> GetBFI = nullptr,
ProfileSummaryInfo *PSI = nullptr,
OptimizationRemarkEmitter *ORE = nullptr, bool BoostIndirect = true,
bool IgnoreThreshold = false)
: CallAnalyzer(Callee, Call, TTI, GetAssumptionCache, GetBFI, PSI, ORE),
ComputeFullInlineCost(OptComputeFullInlineCost ||
Params.ComputeFullInlineCost || ORE ||
Params(Params), Threshold(Params.DefaultThreshold),
BoostIndirectCalls(BoostIndirect), IgnoreThreshold(IgnoreThreshold),
Writer(this) {}
/// Annotation Writer for instruction details
InlineCostAnnotationWriter Writer;
void dump();
// Prints the same analysis as dump(), but its definition is not dependent
// on the build.
void print();
Optional<InstructionCostDetail> getCostDetails(const Instruction *I) {
if (InstructionCostDetailMap.find(I) != InstructionCostDetailMap.end())
return InstructionCostDetailMap[I];
return None;
virtual ~InlineCostCallAnalyzer() {}
int getThreshold() { return Threshold; }
int getCost() { return Cost; }
} // namespace
/// Test whether the given value is an Alloca-derived function argument.
bool CallAnalyzer::isAllocaDerivedArg(Value *V) {
return SROAArgValues.count(V);
void CallAnalyzer::disableSROAForArg(AllocaInst *SROAArg) {
void InlineCostAnnotationWriter::emitInstructionAnnot(const Instruction *I,
formatted_raw_ostream &OS) {
// The cost of inlining of the given instruction is printed always.
// The threshold delta is printed only when it is non-zero. It happens
// when we decided to give a bonus at a particular instruction.
Optional<InstructionCostDetail> Record = ICCA->getCostDetails(I);
if (!Record)
OS << "; No analysis for the instruction";
else {
OS << "; cost before = " << Record->CostBefore
<< ", cost after = " << Record->CostAfter
<< ", threshold before = " << Record->ThresholdBefore
<< ", threshold after = " << Record->ThresholdAfter << ", ";
OS << "cost delta = " << Record->getCostDelta();
if (Record->hasThresholdChanged())
OS << ", threshold delta = " << Record->getThresholdDelta();
auto C = ICCA->getSimplifiedValue(const_cast<Instruction *>(I));
if (C) {
OS << ", simplified to ";
C.getValue()->print(OS, true);
OS << "\n";
/// If 'V' maps to a SROA candidate, disable SROA for it.
void CallAnalyzer::disableSROA(Value *V) {
if (auto *SROAArg = getSROAArgForValueOrNull(V)) {
void CallAnalyzer::disableLoadElimination() {
if (EnableLoadElimination) {
EnableLoadElimination = false;
/// Accumulate a constant GEP offset into an APInt if possible.
/// Returns false if unable to compute the offset for any reason. Respects any
/// simplified values known during the analysis of this callsite.
bool CallAnalyzer::accumulateGEPOffset(GEPOperator &GEP, APInt &Offset) {
unsigned IntPtrWidth = DL.getIndexTypeSizeInBits(GEP.getType());
assert(IntPtrWidth == Offset.getBitWidth());
for (gep_type_iterator GTI = gep_type_begin(GEP), GTE = gep_type_end(GEP);
GTI != GTE; ++GTI) {
ConstantInt *OpC = dyn_cast<ConstantInt>(GTI.getOperand());
if (!OpC)
if (Constant *SimpleOp = SimplifiedValues.lookup(GTI.getOperand()))
OpC = dyn_cast<ConstantInt>(SimpleOp);
if (!OpC)
return false;
if (OpC->isZero())
// Handle a struct index, which adds its field offset to the pointer.
if (StructType *STy = GTI.getStructTypeOrNull()) {
unsigned ElementIdx = OpC->getZExtValue();
const StructLayout *SL = DL.getStructLayout(STy);
Offset += APInt(IntPtrWidth, SL->getElementOffset(ElementIdx));
APInt TypeSize(IntPtrWidth, DL.getTypeAllocSize(GTI.getIndexedType()));
Offset += OpC->getValue().sextOrTrunc(IntPtrWidth) * TypeSize;
return true;
/// Use TTI to check whether a GEP is free.
/// Respects any simplified values known during the analysis of this callsite.
bool CallAnalyzer::isGEPFree(GetElementPtrInst &GEP) {
SmallVector<Value *, 4> Operands;
for (const Use &Op : GEP.indices())
if (Constant *SimpleOp = SimplifiedValues.lookup(Op))
return TargetTransformInfo::TCC_Free ==
TTI.getUserCost(&GEP, Operands,
bool CallAnalyzer::visitAlloca(AllocaInst &I) {
// Check whether inlining will turn a dynamic alloca into a static
// alloca and handle that case.
if (I.isArrayAllocation()) {
Constant *Size = SimplifiedValues.lookup(I.getArraySize());
if (auto *AllocSize = dyn_cast_or_null<ConstantInt>(Size)) {
// Sometimes a dynamic alloca could be converted into a static alloca
// after this constant prop, and become a huge static alloca on an
// unconditional CFG path. Avoid inlining if this is going to happen above
// a threshold.
// FIXME: If the threshold is removed or lowered too much, we could end up
// being too pessimistic and prevent inlining non-problematic code. This
// could result in unintended perf regressions. A better overall strategy
// is needed to track stack usage during inlining.
Type *Ty = I.getAllocatedType();
AllocatedSize = SaturatingMultiplyAdd(
AllocSize->getLimitedValue(), DL.getTypeAllocSize(Ty).getKnownMinSize(),
if (AllocatedSize > InlineConstants::MaxSimplifiedDynamicAllocaToInline) {
HasDynamicAlloca = true;
return false;
return Base::visitAlloca(I);
// Accumulate the allocated size.
if (I.isStaticAlloca()) {
Type *Ty = I.getAllocatedType();
AllocatedSize =
SaturatingAdd(DL.getTypeAllocSize(Ty).getKnownMinSize(), AllocatedSize);
// We will happily inline static alloca instructions.
if (I.isStaticAlloca())
return Base::visitAlloca(I);
// FIXME: This is overly conservative. Dynamic allocas are inefficient for
// a variety of reasons, and so we would like to not inline them into
// functions which don't currently have a dynamic alloca. This simply
// disables inlining altogether in the presence of a dynamic alloca.
HasDynamicAlloca = true;
return false;
bool CallAnalyzer::visitPHI(PHINode &I) {
// FIXME: We need to propagate SROA *disabling* through phi nodes, even
// though we don't want to propagate it's bonuses. The idea is to disable
// SROA if it *might* be used in an inappropriate manner.
// Phi nodes are always zero-cost.
// FIXME: Pointer sizes may differ between different address spaces, so do we
// need to use correct address space in the call to getPointerSizeInBits here?
// Or could we skip the getPointerSizeInBits call completely? As far as I can
// see the ZeroOffset is used as a dummy value, so we can probably use any
// bit width for the ZeroOffset?
APInt ZeroOffset = APInt::getNullValue(DL.getPointerSizeInBits(0));
bool CheckSROA = I.getType()->isPointerTy();
// Track the constant or pointer with constant offset we've seen so far.
Constant *FirstC = nullptr;
std::pair<Value *, APInt> FirstBaseAndOffset = {nullptr, ZeroOffset};
Value *FirstV = nullptr;
for (unsigned i = 0, e = I.getNumIncomingValues(); i != e; ++i) {
BasicBlock *Pred = I.getIncomingBlock(i);
// If the incoming block is dead, skip the incoming block.
if (DeadBlocks.count(Pred))
// If the parent block of phi is not the known successor of the incoming
// block, skip the incoming block.
BasicBlock *KnownSuccessor = KnownSuccessors[Pred];
if (KnownSuccessor && KnownSuccessor != I.getParent())
Value *V = I.getIncomingValue(i);
// If the incoming value is this phi itself, skip the incoming value.
if (&I == V)
Constant *C = dyn_cast<Constant>(V);
if (!C)
C = SimplifiedValues.lookup(V);
std::pair<Value *, APInt> BaseAndOffset = {nullptr, ZeroOffset};
if (!C && CheckSROA)
BaseAndOffset = ConstantOffsetPtrs.lookup(V);
if (!C && !BaseAndOffset.first)
// The incoming value is neither a constant nor a pointer with constant
// offset, exit early.
return true;
if (FirstC) {
if (FirstC == C)
// If we've seen a constant incoming value before and it is the same
// constant we see this time, continue checking the next incoming value.
// Otherwise early exit because we either see a different constant or saw
// a constant before but we have a pointer with constant offset this time.
return true;
if (FirstV) {
// The same logic as above, but check pointer with constant offset here.
if (FirstBaseAndOffset == BaseAndOffset)
return true;
if (C) {
// This is the 1st time we've seen a constant, record it.
FirstC = C;
// The remaining case is that this is the 1st time we've seen a pointer with
// constant offset, record it.
FirstV = V;
FirstBaseAndOffset = BaseAndOffset;
// Check if we can map phi to a constant.
if (FirstC) {
SimplifiedValues[&I] = FirstC;
return true;
// Check if we can map phi to a pointer with constant offset.
if (FirstBaseAndOffset.first) {
ConstantOffsetPtrs[&I] = FirstBaseAndOffset;
if (auto *SROAArg = getSROAArgForValueOrNull(FirstV))
SROAArgValues[&I] = SROAArg;
return true;
/// Check we can fold GEPs of constant-offset call site argument pointers.
/// This requires target data and inbounds GEPs.
/// \return true if the specified GEP can be folded.
bool CallAnalyzer::canFoldInboundsGEP(GetElementPtrInst &I) {
// Check if we have a base + offset for the pointer.
std::pair<Value *, APInt> BaseAndOffset =
if (!BaseAndOffset.first)
return false;
// Check if the offset of this GEP is constant, and if so accumulate it
// into Offset.
if (!accumulateGEPOffset(cast<GEPOperator>(I), BaseAndOffset.second))
return false;
// Add the result as a new mapping to Base + Offset.
ConstantOffsetPtrs[&I] = BaseAndOffset;
return true;
bool CallAnalyzer::visitGetElementPtr(GetElementPtrInst &I) {
auto *SROAArg = getSROAArgForValueOrNull(I.getPointerOperand());
// Lambda to check whether a GEP's indices are all constant.
auto IsGEPOffsetConstant = [&](GetElementPtrInst &GEP) {
for (const Use &Op : GEP.indices())
if (!isa<Constant>(Op) && !SimplifiedValues.lookup(Op))
return false;
return true;
if (!DisableGEPConstOperand)
if (simplifyInstruction(I, [&](SmallVectorImpl<Constant *> &COps) {
SmallVector<Constant *, 2> Indices;
for (unsigned int Index = 1 ; Index < COps.size() ; ++Index)
return ConstantExpr::getGetElementPtr(I.getSourceElementType(), COps[0],
Indices, I.isInBounds());
return true;
if ((I.isInBounds() && canFoldInboundsGEP(I)) || IsGEPOffsetConstant(I)) {
if (SROAArg)
SROAArgValues[&I] = SROAArg;
// Constant GEPs are modeled as free.
return true;
// Variable GEPs will require math and will disable SROA.
if (SROAArg)
return isGEPFree(I);
/// 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 <typename Callable>
bool CallAnalyzer::simplifyInstruction(Instruction &I, Callable Evaluate) {
SmallVector<Constant *, 2> COps;
for (Value *Op : I.operands()) {
Constant *COp = dyn_cast<Constant>(Op);
if (!COp)
COp = SimplifiedValues.lookup(Op);
if (!COp)
return false;
auto *C = Evaluate(COps);
if (!C)
return false;
SimplifiedValues[&I] = C;
return true;
bool CallAnalyzer::visitBitCast(BitCastInst &I) {
// Propagate constants through bitcasts.
if (simplifyInstruction(I, [&](SmallVectorImpl<Constant *> &COps) {
return ConstantExpr::getBitCast(COps[0], I.getType());
return true;
// Track base/offsets through casts
std::pair<Value *, APInt> BaseAndOffset =
// Casts don't change the offset, just wrap it up.
if (BaseAndOffset.first)
ConstantOffsetPtrs[&I] = BaseAndOffset;
// Also look for SROA candidates here.
if (auto *SROAArg = getSROAArgForValueOrNull(I.getOperand(0)))
SROAArgValues[&I] = SROAArg;
// Bitcasts are always zero cost.
return true;
bool CallAnalyzer::visitPtrToInt(PtrToIntInst &I) {
// Propagate constants through ptrtoint.
if (simplifyInstruction(I, [&](SmallVectorImpl<Constant *> &COps) {
return ConstantExpr::getPtrToInt(COps[0], I.getType());
return true;
// Track base/offset pairs when converted to a plain integer provided the
// integer is large enough to represent the pointer.
unsigned IntegerSize = I.getType()->getScalarSizeInBits();
unsigned AS = I.getOperand(0)->getType()->getPointerAddressSpace();
if (IntegerSize == DL.getPointerSizeInBits(AS)) {
std::pair<Value *, APInt> BaseAndOffset =
if (BaseAndOffset.first)
ConstantOffsetPtrs[&I] = BaseAndOffset;
// This is really weird. Technically, ptrtoint will disable SROA. However,
// unless that ptrtoint is *used* somewhere in the live basic blocks after
// inlining, it will be nuked, and SROA should proceed. All of the uses which
// would block SROA would also block SROA if applied directly to a pointer,
// and so we can just add the integer in here. The only places where SROA is
// 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.
if (auto *SROAArg = getSROAArgForValueOrNull(I.getOperand(0)))
SROAArgValues[&I] = SROAArg;
return TargetTransformInfo::TCC_Free ==
TTI.getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency);
bool CallAnalyzer::visitIntToPtr(IntToPtrInst &I) {
// Propagate constants through ptrtoint.
if (simplifyInstruction(I, [&](SmallVectorImpl<Constant *> &COps) {
return ConstantExpr::getIntToPtr(COps[0], I.getType());
return true;
// Track base/offset pairs when round-tripped through a pointer without
// modifications provided the integer is not too large.
Value *Op = I.getOperand(0);
unsigned IntegerSize = Op->getType()->getScalarSizeInBits();
if (IntegerSize <= DL.getPointerTypeSizeInBits(I.getType())) {
std::pair<Value *, APInt> BaseAndOffset = ConstantOffsetPtrs.lookup(Op);
if (BaseAndOffset.first)
ConstantOffsetPtrs[&I] = BaseAndOffset;
// "Propagate" SROA here in the same manner as we do for ptrtoint above.
if (auto *SROAArg = getSROAArgForValueOrNull(Op))
SROAArgValues[&I] = SROAArg;
return TargetTransformInfo::TCC_Free ==
TTI.getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency);
bool CallAnalyzer::visitCastInst(CastInst &I) {
// Propagate constants through casts.
if (simplifyInstruction(I, [&](SmallVectorImpl<Constant *> &COps) {
return ConstantExpr::getCast(I.getOpcode(), COps[0], I.getType());
return true;
// Disable SROA in the face of arbitrary casts we don't explicitly list
// elsewhere.
// If this is a floating-point cast, and the target says this operation
// is expensive, this may eventually become a library call. Treat the cost
// as such.
switch (I.getOpcode()) {
case Instruction::FPTrunc:
case Instruction::FPExt:
case Instruction::UIToFP:
case Instruction::SIToFP:
case Instruction::FPToUI:
case Instruction::FPToSI:
if (TTI.getFPOpCost(I.getType()) == TargetTransformInfo::TCC_Expensive)
return TargetTransformInfo::TCC_Free ==
TTI.getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency);
bool CallAnalyzer::visitUnaryInstruction(UnaryInstruction &I) {
Value *Operand = I.getOperand(0);
if (simplifyInstruction(I, [&](SmallVectorImpl<Constant *> &COps) {
return ConstantFoldInstOperands(&I, COps[0], DL);
return true;
// Disable any SROA on the argument to arbitrary unary instructions.
return false;
bool CallAnalyzer::paramHasAttr(Argument *A, Attribute::AttrKind Attr) {
return CandidateCall.paramHasAttr(A->getArgNo(), Attr);
bool CallAnalyzer::isKnownNonNullInCallee(Value *V) {
// Does the *call site* have the NonNull attribute set on an argument? We
// use the attribute on the call site to memoize any analysis done in the
// caller. This will also trip if the callee function has a non-null
// parameter attribute, but that's a less interesting case because hopefully
// the callee would already have been simplified based on that.
if (Argument *A = dyn_cast<Argument>(V))
if (paramHasAttr(A, Attribute::NonNull))
return true;
// Is this an alloca in the caller? This is distinct from the attribute case
// above because attributes aren't updated within the inliner itself and we
// always want to catch the alloca derived case.
if (isAllocaDerivedArg(V))
// We can actually predict the result of comparisons between an
// alloca-derived value and null. Note that this fires regardless of
// SROA firing.
return true;
return false;
bool CallAnalyzer::allowSizeGrowth(CallBase &Call) {
// If the normal destination of the invoke or the parent block of the call
// site is unreachable-terminated, there is little point in inlining this
// unless there is literally zero cost.
// FIXME: Note that it is possible that an unreachable-terminated block has a
// hot entry. For example, in below scenario inlining hot_call_X() may be
// beneficial :
// main() {
// hot_call_1();
// ...
// hot_call_N()
// exit(0);
// }
// For now, we are not handling this corner case here as it is rare in real
// code. In future, we should elaborate this based on BPI and BFI in more
// general threshold adjusting heuristics in updateThreshold().
if (InvokeInst *II = dyn_cast<InvokeInst>(&Call)) {
if (isa<UnreachableInst>(II->getNormalDest()->getTerminator()))
return false;
} else if (isa<UnreachableInst>(Call.getParent()->getTerminator()))
return false;
return true;
bool InlineCostCallAnalyzer::isColdCallSite(CallBase &Call,
BlockFrequencyInfo *CallerBFI) {
// If global profile summary is available, then callsite's coldness is
// determined based on that.
if (PSI && PSI->hasProfileSummary())
return PSI->isColdCallSite(Call, CallerBFI);
// Otherwise we need BFI to be available.
if (!CallerBFI)
return false;
// Determine if the callsite is cold relative to caller's entry. We could
// potentially cache the computation of scaled entry frequency, but the added
// complexity is not worth it unless this scaling shows up high in the
// profiles.
const BranchProbability ColdProb(ColdCallSiteRelFreq, 100);
auto CallSiteBB = Call.getParent();
auto CallSiteFreq = CallerBFI->getBlockFreq(CallSiteBB);
auto CallerEntryFreq =
return CallSiteFreq < CallerEntryFreq * ColdProb;
InlineCostCallAnalyzer::getHotCallSiteThreshold(CallBase &Call,
BlockFrequencyInfo *CallerBFI) {
// If global profile summary is available, then callsite's hotness is
// determined based on that.
if (PSI && PSI->hasProfileSummary() && PSI->isHotCallSite(Call, CallerBFI))
return Params.HotCallSiteThreshold;
// Otherwise we need BFI to be available and to have a locally hot callsite
// threshold.
if (!CallerBFI || !Params.LocallyHotCallSiteThreshold)
return None;
// Determine if the callsite is hot relative to caller's entry. We could
// potentially cache the computation of scaled entry frequency, but the added
// complexity is not worth it unless this scaling shows up high in the
// profiles.
auto CallSiteBB = Call.getParent();
auto CallSiteFreq = CallerBFI->getBlockFreq(CallSiteBB).getFrequency();
auto CallerEntryFreq = CallerBFI->getEntryFreq();
if (CallSiteFreq >= CallerEntryFreq * HotCallSiteRelFreq)
return Params.LocallyHotCallSiteThreshold;
// Otherwise treat it normally.
return None;
void InlineCostCallAnalyzer::updateThreshold(CallBase &Call, Function &Callee) {
// If no size growth is allowed for this inlining, set Threshold to 0.
if (!allowSizeGrowth(Call)) {
Threshold = 0;
Function *Caller = Call.getCaller();
// return min(A, B) if B is valid.
auto MinIfValid = [](int A, Optional<int> B) {
return B ? std::min(A, B.getValue()) : A;
// return max(A, B) if B is valid.
auto MaxIfValid = [](int A, Optional<int> B) {
return B ? std::max(A, B.getValue()) : A;
// Various bonus percentages. These are multiplied by Threshold to get the
// bonus values.
// SingleBBBonus: This bonus is applied if the callee has a single reachable
// basic block at the given callsite context. This is speculatively applied
// and withdrawn if more than one basic block is seen.
// LstCallToStaticBonus: This large bonus is applied to ensure the inlining
// of the last call to a static function as inlining such functions is
// guaranteed to reduce code size.
// These bonus percentages may be set to 0 based on properties of the caller
// and the callsite.
int SingleBBBonusPercent = 50;
int VectorBonusPercent = TTI.getInlinerVectorBonusPercent();
int LastCallToStaticBonus = InlineConstants::LastCallToStaticBonus;
// Lambda to set all the above bonus and bonus percentages to 0.
auto DisallowAllBonuses = [&]() {
SingleBBBonusPercent = 0;
VectorBonusPercent = 0;
LastCallToStaticBonus = 0;
// Use the OptMinSizeThreshold or OptSizeThreshold knob if they are available
// and reduce the threshold if the caller has the necessary attribute.
if (Caller->hasMinSize()) {
Threshold = MinIfValid(Threshold, Params.OptMinSizeThreshold);
// For minsize, we want to disable the single BB bonus and the vector
// bonuses, but not the last-call-to-static bonus. Inlining the last call to
// a static function will, at the minimum, eliminate the parameter setup and
// call/return instructions.
SingleBBBonusPercent = 0;
VectorBonusPercent = 0;
} else if (Caller->hasOptSize())
Threshold = MinIfValid(Threshold, Params.OptSizeThreshold);
// Adjust the threshold based on inlinehint attribute and profile based
// hotness information if the caller does not have MinSize attribute.
if (!Caller->hasMinSize()) {
if (Callee.hasFnAttribute(Attribute::InlineHint))
Threshold = MaxIfValid(Threshold, Params.HintThreshold);
// FIXME: After switching to the new passmanager, simplify the logic below
// by checking only the callsite hotness/coldness as we will reliably
// have local profile information.
// Callsite hotness and coldness can be determined if sample profile is
// used (which adds hotness metadata to calls) or if caller's
// BlockFrequencyInfo is available.
BlockFrequencyInfo *CallerBFI = GetBFI ? &(GetBFI(*Caller)) : nullptr;
auto HotCallSiteThreshold = getHotCallSiteThreshold(Call, CallerBFI);
if (!Caller->hasOptSize() && HotCallSiteThreshold) {
LLVM_DEBUG(dbgs() << "Hot callsite.\n");
// FIXME: This should update the threshold only if it exceeds the
// current threshold, but AutoFDO + ThinLTO currently relies on this
// behavior to prevent inlining of hot callsites during ThinLTO
// compile phase.
Threshold = HotCallSiteThreshold.getValue();
} else if (isColdCallSite(Call, CallerBFI)) {
LLVM_DEBUG(dbgs() << "Cold callsite.\n");
// Do not apply bonuses for a cold callsite including the
// LastCallToStatic bonus. While this bonus might result in code size
// reduction, it can cause the size of a non-cold caller to increase
// preventing it from being inlined.
Threshold = MinIfValid(Threshold, Params.ColdCallSiteThreshold);
} else if (PSI) {
// Use callee's global profile information only if we have no way of
// determining this via callsite information.
if (PSI->isFunctionEntryHot(&Callee)) {
LLVM_DEBUG(dbgs() << "Hot callee.\n");
// If callsite hotness can not be determined, we may still know
// that the callee is hot and treat it as a weaker hint for threshold
// increase.
Threshold = MaxIfValid(Threshold, Params.HintThreshold);
} else if (PSI->isFunctionEntryCold(&Callee)) {
LLVM_DEBUG(dbgs() << "Cold callee.\n");
// Do not apply bonuses for a cold callee including the
// LastCallToStatic bonus. While this bonus might result in code size
// reduction, it can cause the size of a non-cold caller to increase
// preventing it from being inlined.
Threshold = MinIfValid(Threshold, Params.ColdThreshold);
// Finally, take the target-specific inlining threshold multiplier into
// account.
Threshold *= TTI.getInliningThresholdMultiplier();
Threshold += TTI.adjustInliningThreshold(&Call);
SingleBBBonus = Threshold * SingleBBBonusPercent / 100;
VectorBonus = Threshold * VectorBonusPercent / 100;
bool OnlyOneCallAndLocalLinkage =
F.hasLocalLinkage() && F.hasOneUse() && &F == Call.getCalledFunction();
// If there is only one call of the function, and it has internal linkage,
// the cost of inlining it drops dramatically. It may seem odd to update
// Cost in updateThreshold, but the bonus depends on the logic in this method.
if (OnlyOneCallAndLocalLinkage)
Cost -= LastCallToStaticBonus;
bool CallAnalyzer::visitCmpInst(CmpInst &I) {
Value *LHS = I.getOperand(0), *RHS = I.getOperand(1);
// First try to handle simplified comparisons.
if (simplifyInstruction(I, [&](SmallVectorImpl<Constant *> &COps) {
return ConstantExpr::getCompare(I.getPredicate(), COps[0], COps[1]);
return true;
if (I.getOpcode() == Instruction::FCmp)
return false;
// Otherwise look for a comparison between constant offset pointers with
// a common base.
Value *LHSBase, *RHSBase;
APInt LHSOffset, RHSOffset;
std::tie(LHSBase, LHSOffset) = ConstantOffsetPtrs.lookup(LHS);
if (LHSBase) {
std::tie(RHSBase, RHSOffset) = ConstantOffsetPtrs.lookup(RHS);
if (RHSBase && LHSBase == RHSBase) {
// We have common bases, fold the icmp to a constant based on the
// offsets.
Constant *CLHS = ConstantInt::get(LHS->getContext(), LHSOffset);
Constant *CRHS = ConstantInt::get(RHS->getContext(), RHSOffset);
if (Constant *C = ConstantExpr::getICmp(I.getPredicate(), CLHS, CRHS)) {
SimplifiedValues[&I] = C;
return true;
// If the comparison is an equality comparison with null, we can simplify it
// if we know the value (argument) can't be null
if (I.isEquality() && isa<ConstantPointerNull>(I.getOperand(1)) &&
isKnownNonNullInCallee(I.getOperand(0))) {
bool IsNotEqual = I.getPredicate() == CmpInst::ICMP_NE;
SimplifiedValues[&I] = IsNotEqual ? ConstantInt::getTrue(I.getType())
: ConstantInt::getFalse(I.getType());
return true;
return handleSROA(I.getOperand(0), isa<ConstantPointerNull>(I.getOperand(1)));
bool CallAnalyzer::visitSub(BinaryOperator &I) {
// Try to handle a special case: we can fold computing the difference of two
// constant-related pointers.
Value *LHS = I.getOperand(0), *RHS = I.getOperand(1);
Value *LHSBase, *RHSBase;
APInt LHSOffset, RHSOffset;
std::tie(LHSBase, LHSOffset) = ConstantOffsetPtrs.lookup(LHS);
if (LHSBase) {
std::tie(RHSBase, RHSOffset) = ConstantOffsetPtrs.lookup(RHS);
if (RHSBase && LHSBase == RHSBase) {
// We have common bases, fold the subtract to a constant based on the
// offsets.
Constant *CLHS = ConstantInt::get(LHS->getContext(), LHSOffset);
Constant *CRHS = ConstantInt::get(RHS->getContext(), RHSOffset);
if (Constant *C = ConstantExpr::getSub(CLHS, CRHS)) {
SimplifiedValues[&I] = C;
return true;
// Otherwise, fall back to the generic logic for simplifying and handling
// instructions.
return Base::visitSub(I);
bool CallAnalyzer::visitBinaryOperator(BinaryOperator &I) {
Value *LHS = I.getOperand(0), *RHS = I.getOperand(1);
Constant *CLHS = dyn_cast<Constant>(LHS);
if (!CLHS)
CLHS = SimplifiedValues.lookup(LHS);
Constant *CRHS = dyn_cast<Constant>(RHS);
if (!CRHS)
CRHS = SimplifiedValues.lookup(RHS);
Value *SimpleV = nullptr;
if (auto FI = dyn_cast<FPMathOperator>(&I))
SimpleV = SimplifyBinOp(I.getOpcode(), CLHS ? CLHS : LHS, CRHS ? CRHS : RHS,
FI->getFastMathFlags(), DL);
SimpleV =
SimplifyBinOp(I.getOpcode(), CLHS ? CLHS : LHS, CRHS ? CRHS : RHS, DL);
if (Constant *C = dyn_cast_or_null<Constant>(SimpleV))
SimplifiedValues[&I] = C;
if (SimpleV)
return true;
// Disable any SROA on arguments to arbitrary, unsimplified binary operators.
// If the instruction is floating point, and the target says this operation
// is expensive, this may eventually become a library call. Treat the cost
// as such. Unless it's fneg which can be implemented with an xor.
using namespace llvm::PatternMatch;
if (I.getType()->isFloatingPointTy() &&
TTI.getFPOpCost(I.getType()) == TargetTransformInfo::TCC_Expensive &&
!match(&I, m_FNeg(m_Value())))
return false;
bool CallAnalyzer::visitFNeg(UnaryOperator &I) {
Value *Op = I.getOperand(0);
Constant *COp = dyn_cast<Constant>(Op);
if (!COp)
COp = SimplifiedValues.lookup(Op);
Value *SimpleV = SimplifyFNegInst(
COp ? COp : Op, cast<FPMathOperator>(I).getFastMathFlags(), DL);
if (Constant *C = dyn_cast_or_null<Constant>(SimpleV))
SimplifiedValues[&I] = C;
if (SimpleV)
return true;
// Disable any SROA on arguments to arbitrary, unsimplified fneg.
return false;
bool CallAnalyzer::visitLoad(LoadInst &I) {
if (handleSROA(I.getPointerOperand(), I.isSimple()))
return true;
// If the data is already loaded from this address and hasn't been clobbered
// by any stores or calls, this load is likely to be redundant and can be
// eliminated.
if (EnableLoadElimination &&
!LoadAddrSet.insert(I.getPointerOperand()).second && I.isUnordered()) {
return true;
return false;
bool CallAnalyzer::visitStore(StoreInst &I) {
if (handleSROA(I.getPointerOperand(), I.isSimple()))
return true;
// The store can potentially clobber loads and prevent repeated loads from
// being eliminated.
// 1. We can probably keep an initial set of eliminatable loads substracted
// from the cost even when we finally see a store. We just need to disable
// *further* accumulation of elimination savings.
// 2. We should probably at some point thread MemorySSA for the callee into
// this and then use that to actually compute *really* precise savings.
return false;
bool CallAnalyzer::visitExtractValue(ExtractValueInst &I) {
// Constant folding for extract value is trivial.
if (simplifyInstruction(I, [&](SmallVectorImpl<Constant *> &COps) {
return ConstantExpr::getExtractValue(COps[0], I.getIndices());
return true;
// SROA can look through these but give them a cost.
return false;
bool CallAnalyzer::visitInsertValue(InsertValueInst &I) {
// Constant folding for insert value is trivial.
if (simplifyInstruction(I, [&](SmallVectorImpl<Constant *> &COps) {
return ConstantExpr::getInsertValue(/*AggregateOperand*/ COps[0],
/*InsertedValueOperand*/ COps[1],
return true;
// SROA can look through these but give them a cost.
return false;
/// Try to simplify a call site.
/// Takes a concrete function and callsite and tries to actually simplify it by
/// analyzing the arguments and call itself with instsimplify. Returns true if
/// it has simplified the callsite to some other entity (a constant), making it
/// free.
bool CallAnalyzer::simplifyCallSite(Function *F, CallBase &Call) {
// FIXME: Using the instsimplify logic directly for this is inefficient
// because we have to continually rebuild the argument list even when no
// simplifications can be performed. Until that is fixed with remapping
// inside of instsimplify, directly constant fold calls here.
if (!canConstantFoldCallTo(&Call, F))
return false;
// Try to re-map the arguments to constants.
SmallVector<Constant *, 4> ConstantArgs;
for (Value *I : Call.args()) {
Constant *C = dyn_cast<Constant>(I);
if (!C)
C = dyn_cast_or_null<Constant>(SimplifiedValues.lookup(I));
if (!C)
return false; // This argument doesn't map to a constant.
if (Constant *C = ConstantFoldCall(&Call, F, ConstantArgs)) {
SimplifiedValues[&Call] = C;
return true;
return false;
bool CallAnalyzer::visitCallBase(CallBase &Call) {
if (Call.hasFnAttr(Attribute::ReturnsTwice) &&
!F.hasFnAttribute(Attribute::ReturnsTwice)) {
// This aborts the entire analysis.
ExposesReturnsTwice = true;
return false;
if (isa<CallInst>(Call) && cast<CallInst>(Call).cannotDuplicate())
ContainsNoDuplicateCall = true;
Value *Callee = Call.getCalledOperand();
Function *F = dyn_cast_or_null<Function>(Callee);
bool IsIndirectCall = !F;
if (IsIndirectCall) {
// 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.
F = dyn_cast_or_null<Function>(SimplifiedValues.lookup(Callee));
if (!F) {
if (!Call.onlyReadsMemory())
return Base::visitCallBase(Call);
assert(F && "Expected a call to a known function");
// When we have a concrete function, first try to simplify it directly.
if (simplifyCallSite(F, Call))
return true;
// Next check if it is an intrinsic we know about.
// FIXME: Lift this into part of the InstVisitor.
if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(&Call)) {
switch (II->getIntrinsicID()) {
if (!Call.onlyReadsMemory() && !isAssumeLikeIntrinsic(II))
return Base::visitCallBase(Call);
case Intrinsic::load_relative:
return false;
case Intrinsic::memset:
case Intrinsic::memcpy:
case Intrinsic::memmove:
// SROA can usually chew through these intrinsics, but they aren't free.
return false;
case Intrinsic::icall_branch_funnel:
case Intrinsic::localescape:
HasUninlineableIntrinsic = true;
return false;
case Intrinsic::vastart:
InitsVargArgs = true;
return false;
if (F == Call.getFunction()) {
// This flag will fully abort the analysis, so don't bother with anything
// else.
IsRecursiveCall = true;
return false;
if (TTI.isLoweredToCall(F)) {
onLoweredCall(F, Call, IsIndirectCall);
if (!(Call.onlyReadsMemory() || (IsIndirectCall && F->onlyReadsMemory())))
return Base::visitCallBase(Call);
bool CallAnalyzer::visitReturnInst(ReturnInst &RI) {
// At least one return instruction will be free after inlining.
bool Free = !HasReturn;
HasReturn = true;
return Free;
bool CallAnalyzer::visitBranchInst(BranchInst &BI) {
// We model unconditional branches as essentially free -- they really
// 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<ConstantInt>(BI.getCondition()) ||
bool CallAnalyzer::visitSelectInst(SelectInst &SI) {
bool CheckSROA = SI.getType()->isPointerTy();
Value *TrueVal = SI.getTrueValue();
Value *FalseVal = SI.getFalseValue();
Constant *TrueC = dyn_cast<Constant>(TrueVal);
if (!TrueC)
TrueC = SimplifiedValues.lookup(TrueVal);
Constant *FalseC = dyn_cast<Constant>(FalseVal);
if (!FalseC)
FalseC = SimplifiedValues.lookup(FalseVal);
Constant *CondC =
if (!CondC) {
// Select C, X, X => X
if (TrueC == FalseC && TrueC) {
SimplifiedValues[&SI] = TrueC;
return true;
if (!CheckSROA)
return Base::visitSelectInst(SI);
std::pair<Value *, APInt> TrueBaseAndOffset =
std::pair<Value *, APInt> FalseBaseAndOffset =
if (TrueBaseAndOffset == FalseBaseAndOffset && TrueBaseAndOffset.first) {
ConstantOffsetPtrs[&SI] = TrueBaseAndOffset;
if (auto *SROAArg = getSROAArgForValueOrNull(TrueVal))
SROAArgValues[&SI] = SROAArg;
return true;
return Base::visitSelectInst(SI);
// Select condition is a constant.
Value *SelectedV = CondC->isAllOnesValue()
? TrueVal
: (CondC->isNullValue()) ? FalseVal : nullptr;
if (!SelectedV) {
// Condition is a vector constant that is not all 1s or all 0s. If all
// operands are constants, ConstantExpr::getSelect() can handle the cases
// such as select vectors.
if (TrueC && FalseC) {
if (auto *C = ConstantExpr::getSelect(CondC, TrueC, FalseC)) {
SimplifiedValues[&SI] = C;
return true;
return Base::visitSelectInst(SI);
// Condition is either all 1s or all 0s. SI can be simplified.
if (Constant *SelectedC = dyn_cast<Constant>(SelectedV)) {
SimplifiedValues[&SI] = SelectedC;
return true;
if (!CheckSROA)
return true;
std::pair<Value *, APInt> BaseAndOffset =
if (BaseAndOffset.first) {
ConstantOffsetPtrs[&SI] = BaseAndOffset;
if (auto *SROAArg = getSROAArgForValueOrNull(SelectedV))
SROAArgValues[&SI] = SROAArg;
return true;
bool CallAnalyzer::visitSwitchInst(SwitchInst &SI) {
// We model unconditional switches as free, see the comments on handling
// branches.
if (isa<ConstantInt>(SI.getCondition()))
return true;
if (Value *V = SimplifiedValues.lookup(SI.getCondition()))
if (isa<ConstantInt>(V))
return true;
// Assume the most general case where the switch is lowered into
// either a jump table, bit test, or a balanced binary tree consisting of
// case clusters without merging adjacent clusters with the same
// destination. We do not consider the switches that are lowered with a mix
// of jump table/bit test/binary search tree. The cost of the switch is
// proportional to the size of the tree or the size of jump table range.
// 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.
unsigned JumpTableSize = 0;
BlockFrequencyInfo *BFI = GetBFI ? &(GetBFI(F)) : nullptr;
unsigned NumCaseCluster =
TTI.getEstimatedNumberOfCaseClusters(SI, JumpTableSize, PSI, BFI);
onFinalizeSwitch(JumpTableSize, NumCaseCluster);
return false;
bool CallAnalyzer::visitIndirectBrInst(IndirectBrInst &IBI) {
// We never want to inline functions that contain an indirectbr. This is
// incorrect because all the blockaddress's (in static global initializers
// for example) would be referring to the original function, and this
// indirect jump would jump from the inlined copy of the function into the
// original function which is extremely undefined behavior.
// FIXME: This logic isn't really right; we can safely inline functions with
// indirectbr's as long as no other function or global references the
// blockaddress of a block within the current function.
HasIndirectBr = true;
return false;
bool CallAnalyzer::visitResumeInst(ResumeInst &RI) {
// FIXME: It's not clear that a single instruction is an accurate model for
// the inline cost of a resume instruction.
return false;
bool CallAnalyzer::visitCleanupReturnInst(CleanupReturnInst &CRI) {
// FIXME: It's not clear that a single instruction is an accurate model for
// the inline cost of a cleanupret instruction.
return false;
bool CallAnalyzer::visitCatchReturnInst(CatchReturnInst &CRI) {
// FIXME: It's not clear that a single instruction is an accurate model for
// the inline cost of a catchret instruction.
return false;
bool CallAnalyzer::visitUnreachableInst(UnreachableInst &I) {
// FIXME: It might be reasonably to discount the cost of instructions leading
// to unreachable as they have the lowest possible impact on both runtime and
// code size.
return true; // No actual code is needed for unreachable.
bool CallAnalyzer::visitInstruction(Instruction &I) {
// Some instructions are free. All of the free intrinsics can also be
// handled by SROA, etc.
if (TargetTransformInfo::TCC_Free ==
TTI.getUserCost(&I, TargetTransformInfo::TCK_SizeAndLatency))
return true;
// We found something we don't understand or can't handle. Mark any SROA-able
// values in the operand list as no longer viable.
for (const Use &Op : I.operands())
return false;
/// Analyze a basic block for its contribution to the inline cost.
/// This method walks the analyzer over every instruction in the given basic
/// block and accounts for their cost during inlining at this callsite. It
/// aborts early if the threshold has been exceeded or an impossible to inline
/// construct has been detected. It returns false if inlining is no longer
/// viable, and true if inlining remains viable.
CallAnalyzer::analyzeBlock(BasicBlock *BB,
SmallPtrSetImpl<const Value *> &EphValues) {
for (Instruction &I : *BB) {
// FIXME: Currently, the number of instructions in a function regardless of
// our ability to simplify them during inline to constants or dead code,
// are actually used by the vector bonus heuristic. As long as that's true,
// we have to special case debug intrinsics here to prevent differences in
// inlining due to debug symbols. Eventually, the number of unsimplified
// instructions shouldn't factor into the cost computation, but until then,
// hack around it here.
if (isa<DbgInfoIntrinsic>(I))
// Skip pseudo-probes.
if (isa<PseudoProbeInst>(I))
// Skip ephemeral values.
if (EphValues.count(&I))
if (isa<ExtractElementInst>(I) || I.getType()->isVectorTy())
// If the instruction simplified to a constant, there is no cost to this
// instruction. Visit the instructions using our InstVisitor to account for
// all of the per-instruction logic. The visit tree returns true if we
// consumed the instruction in any way, and false if the instruction's base
// cost should count against inlining.
if (Base::visit(&I))
using namespace ore;
// If the visit this instruction detected an uninlinable pattern, abort.
InlineResult IR = InlineResult::success();
if (IsRecursiveCall)
IR = InlineResult::failure("recursive");
else if (ExposesReturnsTwice)
IR = InlineResult::failure("exposes returns twice");
else if (HasDynamicAlloca)
IR = InlineResult::failure("dynamic alloca");
else if (HasIndirectBr)
IR = InlineResult::failure("indirect branch");
else if (HasUninlineableIntrinsic)
IR = InlineResult::failure("uninlinable intrinsic");
else if (InitsVargArgs)
IR = InlineResult::failure("varargs");
if (!IR.isSuccess()) {
if (ORE)
ORE->emit([&]() {
return OptimizationRemarkMissed(DEBUG_TYPE, "NeverInline",
<< NV("Callee", &F) << " has uninlinable pattern ("
<< NV("InlineResult", IR.getFailureReason())
<< ") and cost is not fully computed";
return IR;
// If the caller is a recursive function then we don't want to inline
// functions which allocate a lot of stack space because it would increase
// the caller stack usage dramatically.
if (IsCallerRecursive &&
AllocatedSize > InlineConstants::TotalAllocaSizeRecursiveCaller) {
auto IR =
InlineResult::failure("recursive and allocates too much stack space");
if (ORE)
ORE->emit([&]() {
return OptimizationRemarkMissed(DEBUG_TYPE, "NeverInline",
<< NV("Callee", &F) << " is "
<< NV("InlineResult", IR.getFailureReason())
<< ". Cost is not fully computed";
return IR;
if (shouldStop())
return InlineResult::failure(
"Call site analysis is not favorable to inlining.");
return InlineResult::success();
/// Compute the base pointer and cumulative constant offsets for V.
/// This strips all constant offsets off of V, leaving it the base pointer, and
/// accumulates the total constant offset applied in the returned constant. It
/// returns 0 if V is not a pointer, and returns the constant '0' if there are
/// no constant offsets applied.
ConstantInt *CallAnalyzer::stripAndComputeInBoundsConstantOffsets(Value *&V) {
if (!V->getType()->isPointerTy())
return nullptr;
unsigned AS = V->getType()->getPointerAddressSpace();
unsigned IntPtrWidth = DL.getIndexSizeInBits(AS);
APInt Offset = APInt::getNullValue(IntPtrWidth);
// Even though we don't look through PHI nodes, we could be called on an
// instruction in an unreachable block, which may be on a cycle.
SmallPtrSet<Value *, 4> Visited;
do {
if (GEPOperator *GEP = dyn_cast<GEPOperator>(V)) {
if (!GEP->isInBounds() || !accumulateGEPOffset(*GEP, Offset))
return nullptr;
V = GEP->getPointerOperand();
} else if (Operator::getOpcode(V) == Instruction::BitCast) {
V = cast<Operator>(V)->getOperand(0);
} else if (GlobalAlias *GA = dyn_cast<GlobalAlias>(V)) {
if (GA->isInterposable())
V = GA->getAliasee();
} else {
assert(V->getType()->isPointerTy() && "Unexpected operand type!");
} while (Visited.insert(V).second);
Type *IdxPtrTy = DL.getIndexType(V->getType());
return cast<ConstantInt>(ConstantInt::get(IdxPtrTy, Offset));
/// Find dead blocks due to deleted CFG edges during inlining.
/// If we know the successor of the current block, \p CurrBB, has to be \p
/// NextBB, the other successors of \p CurrBB are dead if these successors have
/// no live incoming CFG edges. If one block is found to be dead, we can
/// continue growing the dead block list by checking the successors of the dead
/// blocks to see if all their incoming edges are dead or not.
void CallAnalyzer::findDeadBlocks(BasicBlock *CurrBB, BasicBlock *NextBB) {
auto IsEdgeDead = [&](BasicBlock *Pred, BasicBlock *Succ) {
// A CFG edge is dead if the predecessor is dead or the predecessor has a
// known successor which is not the one under exam.
return (DeadBlocks.count(Pred) ||
(KnownSuccessors[Pred] && KnownSuccessors[Pred] != Succ));
auto IsNewlyDead = [&](BasicBlock *BB) {
// If all the edges to a block are dead, the block is also dead.
return (!DeadBlocks.count(BB) &&
[&](BasicBlock *P) { return IsEdgeDead(P, BB); }));
for (BasicBlock *Succ : successors(CurrBB)) {
if (Succ == NextBB || !IsNewlyDead(Succ))
SmallVector<BasicBlock *, 4> NewDead;
while (!NewDead.empty()) {
BasicBlock *Dead = NewDead.pop_back_val();
if (DeadBlocks.insert(Dead))
// Continue growing the dead block lists.
for (BasicBlock *S : successors(Dead))
if (IsNewlyDead(S))
/// Analyze a call site for potential inlining.
/// Returns true if inlining this call is viable, and false if it is not
/// viable. It computes the cost and adjusts the threshold based on numerous
/// factors and heuristics. If this method returns false but the computed cost
/// is below the computed threshold, then inlining was forcibly disabled by
/// some artifact of the routine.
InlineResult CallAnalyzer::analyze() {
auto Result = onAnalysisStart();
if (!Result.isSuccess())
return Result;
if (F.empty())
return InlineResult::success();
Function *Caller = CandidateCall.getFunction();
// Check if the caller function is recursive itself.
for (User *U : Caller->users()) {
CallBase *Call = dyn_cast<CallBase>(U);
if (Call && Call->getFunction() == Caller) {
IsCallerRecursive = true;
// Populate our simplified values by mapping from function arguments to call
// arguments with known important simplifications.
auto CAI = CandidateCall.arg_begin();
for (Argument &FAI : F.args()) {
assert(CAI != CandidateCall.arg_end());
if (Constant *C = dyn_cast<Constant>(CAI))
SimplifiedValues[&FAI] = C;
Value *PtrArg = *CAI;
if (ConstantInt *C = stripAndComputeInBoundsConstantOffsets(PtrArg)) {
ConstantOffsetPtrs[&FAI] = std::make_pair(PtrArg, C->getValue());
// We can SROA any pointer arguments derived from alloca instructions.
if (auto *SROAArg = dyn_cast<AllocaInst>(PtrArg)) {
SROAArgValues[&FAI] = SROAArg;
NumConstantArgs = SimplifiedValues.size();
NumConstantOffsetPtrArgs = ConstantOffsetPtrs.size();
NumAllocaArgs = SROAArgValues.size();
// FIXME: If a caller has multiple calls to a callee, we end up recomputing
// the ephemeral values multiple times (and they're completely determined by
// the callee, so this is purely duplicate work).
SmallPtrSet<const Value *, 32> EphValues;
CodeMetrics::collectEphemeralValues(&F, &GetAssumptionCache(F), EphValues);
// The worklist of live basic blocks in the callee *after* inlining. We avoid
// adding basic blocks of the callee which can be proven to be dead for this
// particular call site in order to get more accurate cost estimates. This
// requires a somewhat heavyweight iteration pattern: we need to walk the
// basic blocks in a breadth-first order as we insert live successors. To
// accomplish this, prioritizing for small iterations because we exit after
// crossing our threshold, we use a small-size optimized SetVector.
typedef SetVector<BasicBlock *, SmallVector<BasicBlock *, 16>,
SmallPtrSet<BasicBlock *, 16>>
BBSetVector BBWorklist;
// Note that we *must not* cache the size, this loop grows the worklist.
for (unsigned Idx = 0; Idx != BBWorklist.size(); ++Idx) {
if (shouldStop())
BasicBlock *BB = BBWorklist[Idx];
if (BB->empty())
// Disallow inlining a blockaddress with uses other than strictly callbr.
// A blockaddress only has defined behavior for an indirect branch in the
// same function, and we do not currently support inlining indirect
// branches. But, the inliner may not see an indirect branch that ends up
// being dead code at a particular call site. If the blockaddress escapes
// the function, e.g., via a global variable, inlining may lead to an
// invalid cross-function reference.
// FIXME: pr/39560: continue relaxing this overt restriction.
if (BB->hasAddressTaken())
for (User *U : BlockAddress::get(&*BB)->users())
if (!isa<CallBrInst>(*U))
return InlineResult::failure("blockaddress used outside of callbr");
// Analyze the cost of this block. If we blow through the threshold, this
// returns false, and we can bail on out.
InlineResult IR = analyzeBlock(BB, EphValues);
if (!IR.isSuccess())
return IR;
Instruction *TI = BB->getTerminator();
// Add in the live successors by first checking whether we have terminator
// that may be simplified based on the values simplified by this call.
if (BranchInst *BI = dyn_cast<BranchInst>(TI)) {
if (BI->isConditional()) {
Value *Cond = BI->getCondition();
if (ConstantInt *SimpleCond =
dyn_cast_or_null<ConstantInt>(SimplifiedValues.lookup(Cond))) {
BasicBlock *NextBB = BI->getSuccessor(SimpleCond->isZero() ? 1 : 0);
KnownSuccessors[BB] = NextBB;
findDeadBlocks(BB, NextBB);
} else if (SwitchInst *SI = dyn_cast<SwitchInst>(TI)) {
Value *Cond = SI->getCondition();
if (ConstantInt *SimpleCond =
dyn_cast_or_null<ConstantInt>(SimplifiedValues.lookup(Cond))) {
BasicBlock *NextBB = SI->findCaseValue(SimpleCond)->getCaseSuccessor();
KnownSuccessors[BB] = NextBB;
findDeadBlocks(BB, NextBB);
// If we're unable to select a particular successor, just count all of
// them.
for (unsigned TIdx = 0, TSize = TI->getNumSuccessors(); TIdx != TSize;
bool OnlyOneCallAndLocalLinkage = F.hasLocalLinkage() && F.hasOneUse() &&
&F == CandidateCall.getCalledFunction();
// If this is a noduplicate call, we can still inline as long as
// inlining this would cause the removal of the caller (so the instruction
// is not actually duplicated, just moved).
if (!OnlyOneCallAndLocalLinkage && ContainsNoDuplicateCall)
return InlineResult::failure("noduplicate");
return finalizeAnalysis();
void InlineCostCallAnalyzer::print() {
#define DEBUG_PRINT_STAT(x) dbgs() << " " #x ": " << x << "\n"
if (PrintInstructionComments)
F.print(dbgs(), &Writer);
#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
/// Dump stats about this call's analysis.
LLVM_DUMP_METHOD void InlineCostCallAnalyzer::dump() {
/// Test that there are no attribute conflicts between Caller and Callee
/// that prevent inlining.
static bool functionsHaveCompatibleAttributes(
Function *Caller, Function *Callee, TargetTransformInfo &TTI,
function_ref<const TargetLibraryInfo &(Function &)> &GetTLI) {
// Note that CalleeTLI must be a copy not a reference. The legacy pass manager
// caches the most recently created TLI in the TargetLibraryInfoWrapperPass
// object, and always returns the same object (which is overwritten on each
// GetTLI call). Therefore we copy the first result.
auto CalleeTLI = GetTLI(*Callee);
return TTI.areInlineCompatible(Caller, Callee) &&
InlineCallerSupersetNoBuiltin) &&
AttributeFuncs::areInlineCompatible(*Caller, *Callee);
int llvm::getCallsiteCost(CallBase &Call, const DataLayout &DL) {
int Cost = 0;
for (unsigned I = 0, E = Call.arg_size(); I != E; ++I) {
if (Call.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<PointerType>(Call.getArgOperand(I)->getType());
unsigned TypeSize = DL.getTypeSizeInBits(PTy->getElementType());
unsigned AS = PTy->getAddressSpace();
unsigned PointerSize = DL.getPointerSizeInBits(AS);
// 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;
// The call instruction also disappears after inlining.
Cost += InlineConstants::InstrCost + InlineConstants::CallPenalty;
return Cost;
InlineCost llvm::getInlineCost(
CallBase &Call, const InlineParams &Params, TargetTransformInfo &CalleeTTI,
function_ref<AssumptionCache &(Function &)> GetAssumptionCache,
function_ref<const TargetLibraryInfo &(Function &)> GetTLI,
function_ref<BlockFrequencyInfo &(Function &)> GetBFI,
ProfileSummaryInfo *PSI, OptimizationRemarkEmitter *ORE) {
return getInlineCost(Call, Call.getCalledFunction(), Params, CalleeTTI,
GetAssumptionCache, GetTLI, GetBFI, PSI, ORE);
Optional<int> llvm::getInliningCostEstimate(
CallBase &Call, TargetTransformInfo &CalleeTTI,
function_ref<AssumptionCache &(Function &)> GetAssumptionCache,
function_ref<BlockFrequencyInfo &(Function &)> GetBFI,
ProfileSummaryInfo *PSI, OptimizationRemarkEmitter *ORE) {
const InlineParams Params = {/* DefaultThreshold*/ 0,
/*HintThreshold*/ {},
/*ColdThreshold*/ {},
/*OptSizeThreshold*/ {},
/*OptMinSizeThreshold*/ {},
/*HotCallSiteThreshold*/ {},
/*LocallyHotCallSiteThreshold*/ {},
/*ColdCallSiteThreshold*/ {},
/*ComputeFullInlineCost*/ true,
/*EnableDeferral*/ true};
InlineCostCallAnalyzer CA(*Call.getCalledFunction(), Call, Params, CalleeTTI,
GetAssumptionCache, GetBFI, PSI, ORE, true,
/*IgnoreThreshold*/ true);
auto R = CA.analyze();
if (!R.isSuccess())
return None;
return CA.getCost();
Optional<InlineResult> llvm::getAttributeBasedInliningDecision(
CallBase &Call, Function *Callee, TargetTransformInfo &CalleeTTI,
function_ref<const TargetLibraryInfo &(Function &)> GetTLI) {
// Cannot inline indirect calls.
if (!Callee)
return InlineResult::failure("indirect call");
// When callee coroutine function is inlined into caller coroutine function
// before coro-split pass,
// coro-early pass can not handle this quiet well.
// So we won't inline the coroutine function if it have not been unsplited
if (Callee->isPresplitCoroutine())
return InlineResult::failure("unsplited coroutine call");
// Never inline calls with byval arguments that does not have the alloca
// address space. Since byval arguments can be replaced with a copy to an
// alloca, the inlined code would need to be adjusted to handle that the
// argument is in the alloca address space (so it is a little bit complicated
// to solve).
unsigned AllocaAS = Callee->getParent()->getDataLayout().getAllocaAddrSpace();
for (unsigned I = 0, E = Call.arg_size(); I != E; ++I)
if (Call.isByValArgument(I)) {
PointerType *PTy = cast<PointerType>(Call.getArgOperand(I)->getType());
if (PTy->getAddressSpace() != AllocaAS)
return InlineResult::failure("byval arguments without alloca"
" address space");
// Calls to functions with always-inline attributes should be inlined
// whenever possible.
if (Call.hasFnAttr(Attribute::AlwaysInline)) {
auto IsViable = isInlineViable(*Callee);
if (IsViable.isSuccess())
return InlineResult::success();
return InlineResult::failure(IsViable.getFailureReason());
// Never inline functions with conflicting attributes (unless callee has
// always-inline attribute).
Function *Caller = Call.getCaller();
if (!functionsHaveCompatibleAttributes(Caller, Callee, CalleeTTI, GetTLI))