LLVM 17.0.0git
Go to the documentation of this file.
1//===- CodeMetrics.cpp - Code cost measurements ---------------------------===//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
9// This file implements code cost measurement utilities.
18#include "llvm/IR/Function.h"
19#include "llvm/Support/Debug.h"
22#define DEBUG_TYPE "code-metrics"
24using namespace llvm;
26static void
30 const User *U = dyn_cast<User>(V);
31 if (!U)
32 return;
34 for (const Value *Operand : U->operands())
35 if (Visited.insert(Operand).second)
36 if (const auto *I = dyn_cast<Instruction>(Operand))
37 if (!I->mayHaveSideEffects() && !I->isTerminator())
38 Worklist.push_back(I);
44 // Note: We don't speculate PHIs here, so we'll miss instruction chains kept
45 // alive only by ephemeral values.
47 // Walk the worklist using an index but without caching the size so we can
48 // append more entries as we process the worklist. This forms a queue without
49 // quadratic behavior by just leaving processed nodes at the head of the
50 // worklist forever.
51 for (int i = 0; i < (int)Worklist.size(); ++i) {
52 const Value *V = Worklist[i];
54 assert(Visited.count(V) &&
55 "Failed to add a worklist entry to our visited set!");
57 // If all uses of this value are ephemeral, then so is this value.
58 if (!all_of(V->users(), [&](const User *U) { return EphValues.count(U); }))
59 continue;
61 EphValues.insert(V);
62 LLVM_DEBUG(dbgs() << "Ephemeral Value: " << *V << "\n");
64 // Append any more operands to consider.
65 appendSpeculatableOperands(V, Visited, Worklist);
66 }
69// Find all ephemeral values.
71 const Loop *L, AssumptionCache *AC,
76 for (auto &AssumeVH : AC->assumptions()) {
77 if (!AssumeVH)
78 continue;
79 Instruction *I = cast<Instruction>(AssumeVH);
81 // Filter out call sites outside of the loop so we don't do a function's
82 // worth of work for each of its loops (and, in the common case, ephemeral
83 // values in the loop are likely due to @llvm.assume calls in the loop).
84 if (!L->contains(I->getParent()))
85 continue;
87 if (EphValues.insert(I).second)
88 appendSpeculatableOperands(I, Visited, Worklist);
89 }
91 completeEphemeralValues(Visited, Worklist, EphValues);
95 const Function *F, AssumptionCache *AC,
100 for (auto &AssumeVH : AC->assumptions()) {
101 if (!AssumeVH)
102 continue;
103 Instruction *I = cast<Instruction>(AssumeVH);
104 assert(I->getParent()->getParent() == F &&
105 "Found assumption for the wrong function!");
107 if (EphValues.insert(I).second)
108 appendSpeculatableOperands(I, Visited, Worklist);
109 }
111 completeEphemeralValues(Visited, Worklist, EphValues);
114/// Fill in the current structure with information gleaned from the specified
115/// block.
117 const BasicBlock *BB, const TargetTransformInfo &TTI,
118 const SmallPtrSetImpl<const Value *> &EphValues, bool PrepareForLTO) {
119 ++NumBlocks;
120 InstructionCost NumInstsBeforeThisBB = NumInsts;
121 for (const Instruction &I : *BB) {
122 // Skip ephemeral values.
123 if (EphValues.count(&I))
124 continue;
126 // Special handling for calls.
127 if (const auto *Call = dyn_cast<CallBase>(&I)) {
128 if (const Function *F = Call->getCalledFunction()) {
129 bool IsLoweredToCall = TTI.isLoweredToCall(F);
130 // If a function is both internal and has a single use, then it is
131 // extremely likely to get inlined in the future (it was probably
132 // exposed by an interleaved devirtualization pass).
133 // When preparing for LTO, liberally consider calls as inline
134 // candidates.
135 if (!Call->isNoInline() && IsLoweredToCall &&
136 ((F->hasInternalLinkage() && F->hasOneLiveUse()) ||
137 PrepareForLTO)) {
139 }
141 // If this call is to function itself, then the function is recursive.
142 // Inlining it into other functions is a bad idea, because this is
143 // basically just a form of loop peeling, and our metrics aren't useful
144 // for that case.
145 if (F == BB->getParent())
146 isRecursive = true;
148 if (IsLoweredToCall)
149 ++NumCalls;
150 } else {
151 // We don't want inline asm to count as a call - that would prevent loop
152 // unrolling. The argument setup cost is still real, though.
153 if (!Call->isInlineAsm())
154 ++NumCalls;
155 }
156 }
158 if (const AllocaInst *AI = dyn_cast<AllocaInst>(&I)) {
159 if (!AI->isStaticAlloca())
160 this->usesDynamicAlloca = true;
161 }
163 if (isa<ExtractElementInst>(I) || I.getType()->isVectorTy())
166 if (I.getType()->isTokenTy() && I.isUsedOutsideOfBlock(BB))
167 notDuplicatable = true;
169 if (const CallInst *CI = dyn_cast<CallInst>(&I)) {
170 if (CI->cannotDuplicate())
171 notDuplicatable = true;
172 if (CI->isConvergent())
173 convergent = true;
174 }
176 if (const InvokeInst *InvI = dyn_cast<InvokeInst>(&I))
177 if (InvI->cannotDuplicate())
178 notDuplicatable = true;
181 }
183 if (isa<ReturnInst>(BB->getTerminator()))
184 ++NumRets;
186 // We never want to inline functions that contain an indirectbr. This is
187 // incorrect because all the blockaddress's (in static global initializers
188 // for example) would be referring to the original function, and this indirect
189 // jump would jump from the inlined copy of the function into the original
190 // function which is extremely undefined behavior.
191 // FIXME: This logic isn't really right; we can safely inline functions
192 // with indirectbr's as long as no other function or global references the
193 // blockaddress of a block within the current function. And as a QOI issue,
194 // if someone is using a blockaddress without an indirectbr, and that
195 // reference somehow ends up in another function or global, we probably
196 // don't want to inline this function.
197 notDuplicatable |= isa<IndirectBrInst>(BB->getTerminator());
199 // Remember NumInsts for this BB.
200 InstructionCost NumInstsThisBB = NumInsts - NumInstsBeforeThisBB;
201 NumBBInsts[BB] = NumInstsThisBB;
static void appendSpeculatableOperands(const Value *V, SmallPtrSetImpl< const Value * > &Visited, SmallVectorImpl< const Value * > &Worklist)
Definition: CodeMetrics.cpp:27
static void completeEphemeralValues(SmallPtrSetImpl< const Value * > &Visited, SmallVectorImpl< const Value * > &Worklist, SmallPtrSetImpl< const Value * > &EphValues)
Definition: CodeMetrics.cpp:41
#define LLVM_DEBUG(X)
Definition: Debug.h:101
This file defines an InstructionCost class that is used when calculating the cost of an instruction,...
#define F(x, y, z)
Definition: MD5.cpp:55
#define I(x, y, z)
Definition: MD5.cpp:58
assert(ImpDefSCC.getReg()==AMDGPU::SCC &&ImpDefSCC.isDef())
This file defines the SmallPtrSet class.
This pass exposes codegen information to IR-level passes.
an instruction to allocate memory on the stack
Definition: Instructions.h:58
A cache of @llvm.assume calls within a function.
MutableArrayRef< ResultElem > assumptions()
Access the list of assumption handles currently tracked for this function.
LLVM Basic Block Representation.
Definition: BasicBlock.h:56
const Function * getParent() const
Return the enclosing method, or null if none.
Definition: BasicBlock.h:112
const Instruction * getTerminator() const LLVM_READONLY
Returns the terminator instruction if the block is well formed or null if the block is not well forme...
Definition: BasicBlock.h:127
This class represents a function call, abstracting a target machine's calling convention.
Invoke instruction.
bool contains(const LoopT *L) const
Return true if the specified loop is contained within in this loop.
Definition: LoopInfo.h:139
Represents a single loop in the control flow graph.
Definition: LoopInfo.h:547
A templated base class for SmallPtrSet which provides the typesafe interface that is common across al...
Definition: SmallPtrSet.h:344
size_type count(ConstPtrType Ptr) const
count - Return 1 if the specified pointer is in the set, 0 otherwise.
Definition: SmallPtrSet.h:383
std::pair< iterator, bool > insert(PtrType Ptr)
Inserts Ptr if and only if there is no element in the container equal to Ptr.
Definition: SmallPtrSet.h:365
SmallPtrSet - This class implements a set which is optimized for holding SmallSize or less elements.
Definition: SmallPtrSet.h:450
size_t size() const
Definition: SmallVector.h:91
This class consists of common code factored out of the SmallVector class to reduce code duplication b...
Definition: SmallVector.h:577
void push_back(const T &Elt)
Definition: SmallVector.h:416
This is a 'vector' (really, a variable-sized array), optimized for the case when the array is small.
Definition: SmallVector.h:1200
This pass provides access to the codegen interfaces that are needed for IR-level transformations.
@ TCK_CodeSize
Instruction code size.
bool isLoweredToCall(const Function *F) const
Test whether calls to a function lower to actual program function calls.
InstructionCost getInstructionCost(const User *U, ArrayRef< const Value * > Operands, TargetCostKind CostKind) const
Estimate the cost of a given IR user when lowered.
op_range operands()
Definition: User.h:242
LLVM Value Representation.
Definition: Value.h:74
iterator_range< user_iterator > users()
Definition: Value.h:421
This is an optimization pass for GlobalISel generic memory operations.
Definition: AddressRanges.h:18
bool all_of(R &&range, UnaryPredicate P)
Provide wrappers to std::all_of which take ranges instead of having to pass begin/end explicitly.
Definition: STLExtras.h:1735
raw_ostream & dbgs()
dbgs() - This returns a reference to a raw_ostream for debugging messages.
Definition: Debug.cpp:163
bool usesDynamicAlloca
True if this function calls alloca (in the C sense).
Definition: CodeMetrics.h:49
unsigned NumBlocks
Number of analyzed blocks.
Definition: CodeMetrics.h:55
bool notDuplicatable
True if this function cannot be duplicated.
Definition: CodeMetrics.h:43
unsigned NumInlineCandidates
The number of calls to internal functions with a single caller.
Definition: CodeMetrics.h:67
bool isRecursive
True if this function calls itself.
Definition: CodeMetrics.h:37
static void collectEphemeralValues(const Loop *L, AssumptionCache *AC, SmallPtrSetImpl< const Value * > &EphValues)
Collect a loop's ephemeral values (those used only by an assume or similar intrinsics in the loop).
Definition: CodeMetrics.cpp:70
unsigned NumRets
How many 'ret' instructions the blocks contain.
Definition: CodeMetrics.h:75
void analyzeBasicBlock(const BasicBlock *BB, const TargetTransformInfo &TTI, const SmallPtrSetImpl< const Value * > &EphValues, bool PrepareForLTO=false)
Add information about a block to the current state.
DenseMap< const BasicBlock *, InstructionCost > NumBBInsts
Keeps track of basic block code size estimates.
Definition: CodeMetrics.h:58
bool convergent
True if this function contains a call to a convergent function.
Definition: CodeMetrics.h:46
unsigned NumCalls
Keep track of the number of calls to 'big' functions.
Definition: CodeMetrics.h:61
unsigned NumVectorInsts
How many instructions produce vector values.
Definition: CodeMetrics.h:72
InstructionCost NumInsts
Code size cost of the analyzed blocks.
Definition: CodeMetrics.h:52