1 //===- BasicAliasAnalysis.cpp - Stateless Alias Analysis Impl -------------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This file defines the primary stateless implementation of the
11 // Alias Analysis interface that implements identities (two different
12 // globals cannot alias, etc), but does no stateful analysis.
14 //===----------------------------------------------------------------------===//
16 #include "llvm/Analysis/BasicAliasAnalysis.h"
17 #include "llvm/ADT/SmallVector.h"
18 #include "llvm/ADT/Statistic.h"
19 #include "llvm/Analysis/AliasAnalysis.h"
20 #include "llvm/Analysis/CFG.h"
21 #include "llvm/Analysis/CaptureTracking.h"
22 #include "llvm/Analysis/InstructionSimplify.h"
23 #include "llvm/Analysis/LoopInfo.h"
24 #include "llvm/Analysis/MemoryBuiltins.h"
25 #include "llvm/Analysis/ValueTracking.h"
26 #include "llvm/IR/Constants.h"
27 #include "llvm/IR/DataLayout.h"
28 #include "llvm/IR/DerivedTypes.h"
29 #include "llvm/IR/Dominators.h"
30 #include "llvm/IR/GlobalAlias.h"
31 #include "llvm/IR/GlobalVariable.h"
32 #include "llvm/IR/Instructions.h"
33 #include "llvm/IR/IntrinsicInst.h"
34 #include "llvm/IR/LLVMContext.h"
35 #include "llvm/IR/Operator.h"
36 #include "llvm/Pass.h"
37 #include "llvm/Support/ErrorHandling.h"
41 /// Enable analysis of recursive PHI nodes.
42 static cl::opt<bool> EnableRecPhiAnalysis("basicaa-recphi", cl::Hidden,
45 /// SearchLimitReached / SearchTimes shows how often the limit of
46 /// to decompose GEPs is reached. It will affect the precision
47 /// of basic alias analysis.
48 #define DEBUG_TYPE "basicaa"
49 STATISTIC(SearchLimitReached, "Number of times the limit to "
50 "decompose GEPs is reached");
51 STATISTIC(SearchTimes, "Number of times a GEP is decomposed");
53 /// Cutoff after which to stop analysing a set of phi nodes potentially involved
54 /// in a cycle. Because we are analysing 'through' phi nodes we need to be
55 /// careful with value equivalence. We use reachability to make sure a value
56 /// cannot be involved in a cycle.
57 const unsigned MaxNumPhiBBsValueReachabilityCheck = 20;
59 // The max limit of the search depth in DecomposeGEPExpression() and
60 // GetUnderlyingObject(), both functions need to use the same search
61 // depth otherwise the algorithm in aliasGEP will assert.
62 static const unsigned MaxLookupSearchDepth = 6;
64 //===----------------------------------------------------------------------===//
66 //===----------------------------------------------------------------------===//
68 /// Returns true if the pointer is to a function-local object that never
69 /// escapes from the function.
70 static bool isNonEscapingLocalObject(const Value *V) {
71 // If this is a local allocation, check to see if it escapes.
72 if (isa<AllocaInst>(V) || isNoAliasCall(V))
73 // Set StoreCaptures to True so that we can assume in our callers that the
74 // pointer is not the result of a load instruction. Currently
75 // PointerMayBeCaptured doesn't have any special analysis for the
76 // StoreCaptures=false case; if it did, our callers could be refined to be
78 return !PointerMayBeCaptured(V, false, /*StoreCaptures=*/true);
80 // If this is an argument that corresponds to a byval or noalias argument,
81 // then it has not escaped before entering the function. Check if it escapes
82 // inside the function.
83 if (const Argument *A = dyn_cast<Argument>(V))
84 if (A->hasByValAttr() || A->hasNoAliasAttr())
85 // Note even if the argument is marked nocapture we still need to check
86 // for copies made inside the function. The nocapture attribute only
87 // specifies that there are no copies made that outlive the function.
88 return !PointerMayBeCaptured(V, false, /*StoreCaptures=*/true);
93 /// Returns true if the pointer is one which would have been considered an
94 /// escape by isNonEscapingLocalObject.
95 static bool isEscapeSource(const Value *V) {
96 if (isa<CallInst>(V) || isa<InvokeInst>(V) || isa<Argument>(V))
99 // The load case works because isNonEscapingLocalObject considers all
100 // stores to be escapes (it passes true for the StoreCaptures argument
101 // to PointerMayBeCaptured).
102 if (isa<LoadInst>(V))
108 /// Returns the size of the object specified by V, or UnknownSize if unknown.
109 static uint64_t getObjectSize(const Value *V, const DataLayout &DL,
110 const TargetLibraryInfo &TLI,
111 bool RoundToAlign = false) {
113 if (getObjectSize(V, Size, DL, &TLI, RoundToAlign))
115 return MemoryLocation::UnknownSize;
118 /// Returns true if we can prove that the object specified by V is smaller than
120 static bool isObjectSmallerThan(const Value *V, uint64_t Size,
121 const DataLayout &DL,
122 const TargetLibraryInfo &TLI) {
123 // Note that the meanings of the "object" are slightly different in the
124 // following contexts:
125 // c1: llvm::getObjectSize()
126 // c2: llvm.objectsize() intrinsic
127 // c3: isObjectSmallerThan()
128 // c1 and c2 share the same meaning; however, the meaning of "object" in c3
129 // refers to the "entire object".
131 // Consider this example:
132 // char *p = (char*)malloc(100)
135 // In the context of c1 and c2, the "object" pointed by q refers to the
136 // stretch of memory of q[0:19]. So, getObjectSize(q) should return 20.
138 // However, in the context of c3, the "object" refers to the chunk of memory
139 // being allocated. So, the "object" has 100 bytes, and q points to the middle
140 // the "object". In case q is passed to isObjectSmallerThan() as the 1st
141 // parameter, before the llvm::getObjectSize() is called to get the size of
142 // entire object, we should:
143 // - either rewind the pointer q to the base-address of the object in
144 // question (in this case rewind to p), or
145 // - just give up. It is up to caller to make sure the pointer is pointing
146 // to the base address the object.
148 // We go for 2nd option for simplicity.
149 if (!isIdentifiedObject(V))
152 // This function needs to use the aligned object size because we allow
153 // reads a bit past the end given sufficient alignment.
154 uint64_t ObjectSize = getObjectSize(V, DL, TLI, /*RoundToAlign*/ true);
156 return ObjectSize != MemoryLocation::UnknownSize && ObjectSize < Size;
159 /// Returns true if we can prove that the object specified by V has size Size.
160 static bool isObjectSize(const Value *V, uint64_t Size, const DataLayout &DL,
161 const TargetLibraryInfo &TLI) {
162 uint64_t ObjectSize = getObjectSize(V, DL, TLI);
163 return ObjectSize != MemoryLocation::UnknownSize && ObjectSize == Size;
166 //===----------------------------------------------------------------------===//
167 // GetElementPtr Instruction Decomposition and Analysis
168 //===----------------------------------------------------------------------===//
170 /// Analyzes the specified value as a linear expression: "A*V + B", where A and
171 /// B are constant integers.
173 /// Returns the scale and offset values as APInts and return V as a Value*, and
174 /// return whether we looked through any sign or zero extends. The incoming
175 /// Value is known to have IntegerType and it may already be sign or zero
178 /// Note that this looks through extends, so the high bits may not be
179 /// represented in the result.
180 /*static*/ Value *BasicAliasAnalysis::GetLinearExpression(
181 Value *V, APInt &Scale, APInt &Offset, ExtensionKind &Extension,
182 const DataLayout &DL, unsigned Depth, AssumptionCache *AC,
184 assert(V->getType()->isIntegerTy() && "Not an integer value");
186 // Limit our recursion depth.
193 if (ConstantInt *Const = dyn_cast<ConstantInt>(V)) {
194 // if it's a constant, just convert it to an offset
195 // and remove the variable.
196 Offset += Const->getValue();
197 assert(Scale == 0 && "Constant values don't have a scale");
201 if (BinaryOperator *BOp = dyn_cast<BinaryOperator>(V)) {
202 if (ConstantInt *RHSC = dyn_cast<ConstantInt>(BOp->getOperand(1))) {
203 switch (BOp->getOpcode()) {
206 case Instruction::Or:
207 // X|C == X+C if all the bits in C are unset in X. Otherwise we can't
209 if (!MaskedValueIsZero(BOp->getOperand(0), RHSC->getValue(), DL, 0, AC,
213 case Instruction::Add:
214 V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, Extension,
215 DL, Depth + 1, AC, DT);
216 Offset += RHSC->getValue();
218 case Instruction::Mul:
219 V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, Extension,
220 DL, Depth + 1, AC, DT);
221 Offset *= RHSC->getValue();
222 Scale *= RHSC->getValue();
224 case Instruction::Shl:
225 V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, Extension,
226 DL, Depth + 1, AC, DT);
227 Offset <<= RHSC->getValue().getLimitedValue();
228 Scale <<= RHSC->getValue().getLimitedValue();
234 // Since GEP indices are sign extended anyway, we don't care about the high
235 // bits of a sign or zero extended value - just scales and offsets. The
236 // extensions have to be consistent though.
237 if ((isa<SExtInst>(V) && Extension != EK_ZeroExt) ||
238 (isa<ZExtInst>(V) && Extension != EK_SignExt)) {
239 Value *CastOp = cast<CastInst>(V)->getOperand(0);
240 unsigned OldWidth = Scale.getBitWidth();
241 unsigned SmallWidth = CastOp->getType()->getPrimitiveSizeInBits();
242 Scale = Scale.trunc(SmallWidth);
243 Offset = Offset.trunc(SmallWidth);
244 Extension = isa<SExtInst>(V) ? EK_SignExt : EK_ZeroExt;
246 Value *Result = GetLinearExpression(CastOp, Scale, Offset, Extension, DL,
248 Scale = Scale.zext(OldWidth);
250 // We have to sign-extend even if Extension == EK_ZeroExt as we can't
251 // decompose a sign extension (i.e. zext(x - 1) != zext(x) - zext(-1)).
252 Offset = Offset.sext(OldWidth);
262 /// If V is a symbolic pointer expression, decompose it into a base pointer
263 /// with a constant offset and a number of scaled symbolic offsets.
265 /// The scaled symbolic offsets (represented by pairs of a Value* and a scale
266 /// in the VarIndices vector) are Value*'s that are known to be scaled by the
267 /// specified amount, but which may have other unrepresented high bits. As
268 /// such, the gep cannot necessarily be reconstructed from its decomposed form.
270 /// When DataLayout is around, this function is capable of analyzing everything
271 /// that GetUnderlyingObject can look through. To be able to do that
272 /// GetUnderlyingObject and DecomposeGEPExpression must use the same search
273 /// depth (MaxLookupSearchDepth). When DataLayout not is around, it just looks
274 /// through pointer casts.
275 /*static*/ const Value *BasicAliasAnalysis::DecomposeGEPExpression(
276 const Value *V, int64_t &BaseOffs,
277 SmallVectorImpl<VariableGEPIndex> &VarIndices, bool &MaxLookupReached,
278 const DataLayout &DL, AssumptionCache *AC, DominatorTree *DT) {
279 // Limit recursion depth to limit compile time in crazy cases.
280 unsigned MaxLookup = MaxLookupSearchDepth;
281 MaxLookupReached = false;
286 // See if this is a bitcast or GEP.
287 const Operator *Op = dyn_cast<Operator>(V);
289 // The only non-operator case we can handle are GlobalAliases.
290 if (const GlobalAlias *GA = dyn_cast<GlobalAlias>(V)) {
291 if (!GA->mayBeOverridden()) {
292 V = GA->getAliasee();
299 if (Op->getOpcode() == Instruction::BitCast ||
300 Op->getOpcode() == Instruction::AddrSpaceCast) {
301 V = Op->getOperand(0);
305 const GEPOperator *GEPOp = dyn_cast<GEPOperator>(Op);
307 // If it's not a GEP, hand it off to SimplifyInstruction to see if it
308 // can come up with something. This matches what GetUnderlyingObject does.
309 if (const Instruction *I = dyn_cast<Instruction>(V))
310 // TODO: Get a DominatorTree and AssumptionCache and use them here
311 // (these are both now available in this function, but this should be
312 // updated when GetUnderlyingObject is updated). TLI should be
314 if (const Value *Simplified =
315 SimplifyInstruction(const_cast<Instruction *>(I), DL)) {
323 // Don't attempt to analyze GEPs over unsized objects.
324 if (!GEPOp->getOperand(0)->getType()->getPointerElementType()->isSized())
327 unsigned AS = GEPOp->getPointerAddressSpace();
328 // Walk the indices of the GEP, accumulating them into BaseOff/VarIndices.
329 gep_type_iterator GTI = gep_type_begin(GEPOp);
330 for (User::const_op_iterator I = GEPOp->op_begin() + 1, E = GEPOp->op_end();
333 // Compute the (potentially symbolic) offset in bytes for this index.
334 if (StructType *STy = dyn_cast<StructType>(*GTI++)) {
335 // For a struct, add the member offset.
336 unsigned FieldNo = cast<ConstantInt>(Index)->getZExtValue();
340 BaseOffs += DL.getStructLayout(STy)->getElementOffset(FieldNo);
344 // For an array/pointer, add the element offset, explicitly scaled.
345 if (ConstantInt *CIdx = dyn_cast<ConstantInt>(Index)) {
348 BaseOffs += DL.getTypeAllocSize(*GTI) * CIdx->getSExtValue();
352 uint64_t Scale = DL.getTypeAllocSize(*GTI);
353 ExtensionKind Extension = EK_NotExtended;
355 // If the integer type is smaller than the pointer size, it is implicitly
356 // sign extended to pointer size.
357 unsigned Width = Index->getType()->getIntegerBitWidth();
358 if (DL.getPointerSizeInBits(AS) > Width)
359 Extension = EK_SignExt;
361 // Use GetLinearExpression to decompose the index into a C1*V+C2 form.
362 APInt IndexScale(Width, 0), IndexOffset(Width, 0);
363 Index = GetLinearExpression(Index, IndexScale, IndexOffset, Extension, DL,
366 // The GEP index scale ("Scale") scales C1*V+C2, yielding (C1*V+C2)*Scale.
367 // This gives us an aggregate computation of (C1*Scale)*V + C2*Scale.
368 BaseOffs += IndexOffset.getSExtValue() * Scale;
369 Scale *= IndexScale.getSExtValue();
371 // If we already had an occurrence of this index variable, merge this
372 // scale into it. For example, we want to handle:
373 // A[x][x] -> x*16 + x*4 -> x*20
374 // This also ensures that 'x' only appears in the index list once.
375 for (unsigned i = 0, e = VarIndices.size(); i != e; ++i) {
376 if (VarIndices[i].V == Index && VarIndices[i].Extension == Extension) {
377 Scale += VarIndices[i].Scale;
378 VarIndices.erase(VarIndices.begin() + i);
383 // Make sure that we have a scale that makes sense for this target's
385 if (unsigned ShiftBits = 64 - DL.getPointerSizeInBits(AS)) {
387 Scale = (int64_t)Scale >> ShiftBits;
391 VariableGEPIndex Entry = {Index, Extension,
392 static_cast<int64_t>(Scale)};
393 VarIndices.push_back(Entry);
397 // Analyze the base pointer next.
398 V = GEPOp->getOperand(0);
399 } while (--MaxLookup);
401 // If the chain of expressions is too deep, just return early.
402 MaxLookupReached = true;
403 SearchLimitReached++;
407 //===----------------------------------------------------------------------===//
408 // BasicAliasAnalysis Pass
409 //===----------------------------------------------------------------------===//
411 // Register the pass...
412 char BasicAliasAnalysis::ID = 0;
413 INITIALIZE_AG_PASS_BEGIN(BasicAliasAnalysis, AliasAnalysis, "basicaa",
414 "Basic Alias Analysis (stateless AA impl)", false,
416 INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)
417 INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
418 INITIALIZE_AG_PASS_END(BasicAliasAnalysis, AliasAnalysis, "basicaa",
419 "Basic Alias Analysis (stateless AA impl)", false, true,
422 ImmutablePass *llvm::createBasicAliasAnalysisPass() {
423 return new BasicAliasAnalysis();
426 /// Returns whether the given pointer value points to memory that is local to
427 /// the function, with global constants being considered local to all
429 bool BasicAliasAnalysis::pointsToConstantMemory(const MemoryLocation &Loc,
431 assert(Visited.empty() && "Visited must be cleared after use!");
433 unsigned MaxLookup = 8;
434 SmallVector<const Value *, 16> Worklist;
435 Worklist.push_back(Loc.Ptr);
437 const Value *V = GetUnderlyingObject(Worklist.pop_back_val(), *DL);
438 if (!Visited.insert(V).second) {
440 return AliasAnalysis::pointsToConstantMemory(Loc, OrLocal);
443 // An alloca instruction defines local memory.
444 if (OrLocal && isa<AllocaInst>(V))
447 // A global constant counts as local memory for our purposes.
448 if (const GlobalVariable *GV = dyn_cast<GlobalVariable>(V)) {
449 // Note: this doesn't require GV to be "ODR" because it isn't legal for a
450 // global to be marked constant in some modules and non-constant in
451 // others. GV may even be a declaration, not a definition.
452 if (!GV->isConstant()) {
454 return AliasAnalysis::pointsToConstantMemory(Loc, OrLocal);
459 // If both select values point to local memory, then so does the select.
460 if (const SelectInst *SI = dyn_cast<SelectInst>(V)) {
461 Worklist.push_back(SI->getTrueValue());
462 Worklist.push_back(SI->getFalseValue());
466 // If all values incoming to a phi node point to local memory, then so does
468 if (const PHINode *PN = dyn_cast<PHINode>(V)) {
469 // Don't bother inspecting phi nodes with many operands.
470 if (PN->getNumIncomingValues() > MaxLookup) {
472 return AliasAnalysis::pointsToConstantMemory(Loc, OrLocal);
474 for (Value *IncValue : PN->incoming_values())
475 Worklist.push_back(IncValue);
479 // Otherwise be conservative.
481 return AliasAnalysis::pointsToConstantMemory(Loc, OrLocal);
483 } while (!Worklist.empty() && --MaxLookup);
486 return Worklist.empty();
489 // FIXME: This code is duplicated with MemoryLocation and should be hoisted to
490 // some common utility location.
491 static bool isMemsetPattern16(const Function *MS,
492 const TargetLibraryInfo &TLI) {
493 if (TLI.has(LibFunc::memset_pattern16) &&
494 MS->getName() == "memset_pattern16") {
495 FunctionType *MemsetType = MS->getFunctionType();
496 if (!MemsetType->isVarArg() && MemsetType->getNumParams() == 3 &&
497 isa<PointerType>(MemsetType->getParamType(0)) &&
498 isa<PointerType>(MemsetType->getParamType(1)) &&
499 isa<IntegerType>(MemsetType->getParamType(2)))
506 /// Returns the behavior when calling the given call site.
507 FunctionModRefBehavior
508 BasicAliasAnalysis::getModRefBehavior(ImmutableCallSite CS) {
509 if (CS.doesNotAccessMemory())
510 // Can't do better than this.
511 return FMRB_DoesNotAccessMemory;
513 FunctionModRefBehavior Min = FMRB_UnknownModRefBehavior;
515 // If the callsite knows it only reads memory, don't return worse
517 if (CS.onlyReadsMemory())
518 Min = FMRB_OnlyReadsMemory;
520 if (CS.onlyAccessesArgMemory())
521 Min = FunctionModRefBehavior(Min & FMRB_OnlyAccessesArgumentPointees);
523 // The AliasAnalysis base class has some smarts, lets use them.
524 return FunctionModRefBehavior(AliasAnalysis::getModRefBehavior(CS) & Min);
527 /// Returns the behavior when calling the given function. For use when the call
528 /// site is not known.
529 FunctionModRefBehavior
530 BasicAliasAnalysis::getModRefBehavior(const Function *F) {
531 // If the function declares it doesn't access memory, we can't do better.
532 if (F->doesNotAccessMemory())
533 return FMRB_DoesNotAccessMemory;
535 // For intrinsics, we can check the table.
536 if (Intrinsic::ID iid = F->getIntrinsicID()) {
537 #define GET_INTRINSIC_MODREF_BEHAVIOR
538 #include "llvm/IR/Intrinsics.gen"
539 #undef GET_INTRINSIC_MODREF_BEHAVIOR
542 FunctionModRefBehavior Min = FMRB_UnknownModRefBehavior;
544 // If the function declares it only reads memory, go with that.
545 if (F->onlyReadsMemory())
546 Min = FMRB_OnlyReadsMemory;
548 if (F->onlyAccessesArgMemory())
549 Min = FunctionModRefBehavior(Min & FMRB_OnlyAccessesArgumentPointees);
551 const TargetLibraryInfo &TLI =
552 getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
553 if (isMemsetPattern16(F, TLI))
554 Min = FMRB_OnlyAccessesArgumentPointees;
556 // Otherwise be conservative.
557 return FunctionModRefBehavior(AliasAnalysis::getModRefBehavior(F) & Min);
560 ModRefInfo BasicAliasAnalysis::getArgModRefInfo(ImmutableCallSite CS,
562 if (const IntrinsicInst *II = dyn_cast<IntrinsicInst>(CS.getInstruction()))
563 switch (II->getIntrinsicID()) {
566 case Intrinsic::memset:
567 case Intrinsic::memcpy:
568 case Intrinsic::memmove:
569 assert((ArgIdx == 0 || ArgIdx == 1) &&
570 "Invalid argument index for memory intrinsic");
571 return ArgIdx ? MRI_Ref : MRI_Mod;
574 // We can bound the aliasing properties of memset_pattern16 just as we can
575 // for memcpy/memset. This is particularly important because the
576 // LoopIdiomRecognizer likes to turn loops into calls to memset_pattern16
577 // whenever possible.
578 if (CS.getCalledFunction() &&
579 isMemsetPattern16(CS.getCalledFunction(), *TLI)) {
580 assert((ArgIdx == 0 || ArgIdx == 1) &&
581 "Invalid argument index for memset_pattern16");
582 return ArgIdx ? MRI_Ref : MRI_Mod;
584 // FIXME: Handle memset_pattern4 and memset_pattern8 also.
586 return AliasAnalysis::getArgModRefInfo(CS, ArgIdx);
589 static bool isAssumeIntrinsic(ImmutableCallSite CS) {
590 const IntrinsicInst *II = dyn_cast<IntrinsicInst>(CS.getInstruction());
591 if (II && II->getIntrinsicID() == Intrinsic::assume)
597 bool BasicAliasAnalysis::doInitialization(Module &M) {
598 InitializeAliasAnalysis(this, &M.getDataLayout());
602 /// Checks to see if the specified callsite can clobber the specified memory
605 /// Since we only look at local properties of this function, we really can't
606 /// say much about this query. We do, however, use simple "address taken"
607 /// analysis on local objects.
608 ModRefInfo BasicAliasAnalysis::getModRefInfo(ImmutableCallSite CS,
609 const MemoryLocation &Loc) {
610 assert(notDifferentParent(CS.getInstruction(), Loc.Ptr) &&
611 "AliasAnalysis query involving multiple functions!");
613 const Value *Object = GetUnderlyingObject(Loc.Ptr, *DL);
615 // If this is a tail call and Loc.Ptr points to a stack location, we know that
616 // the tail call cannot access or modify the local stack.
617 // We cannot exclude byval arguments here; these belong to the caller of
618 // the current function not to the current function, and a tail callee
619 // may reference them.
620 if (isa<AllocaInst>(Object))
621 if (const CallInst *CI = dyn_cast<CallInst>(CS.getInstruction()))
622 if (CI->isTailCall())
625 // If the pointer is to a locally allocated object that does not escape,
626 // then the call can not mod/ref the pointer unless the call takes the pointer
627 // as an argument, and itself doesn't capture it.
628 if (!isa<Constant>(Object) && CS.getInstruction() != Object &&
629 isNonEscapingLocalObject(Object)) {
630 bool PassedAsArg = false;
632 for (ImmutableCallSite::arg_iterator CI = CS.arg_begin(), CE = CS.arg_end();
633 CI != CE; ++CI, ++ArgNo) {
634 // Only look at the no-capture or byval pointer arguments. If this
635 // pointer were passed to arguments that were neither of these, then it
636 // couldn't be no-capture.
637 if (!(*CI)->getType()->isPointerTy() ||
638 (!CS.doesNotCapture(ArgNo) && !CS.isByValArgument(ArgNo)))
641 // If this is a no-capture pointer argument, see if we can tell that it
642 // is impossible to alias the pointer we're checking. If not, we have to
643 // assume that the call could touch the pointer, even though it doesn't
645 if (!isNoAlias(MemoryLocation(*CI), MemoryLocation(Object))) {
655 // While the assume intrinsic is marked as arbitrarily writing so that
656 // proper control dependencies will be maintained, it never aliases any
657 // particular memory location.
658 if (isAssumeIntrinsic(CS))
661 // The AliasAnalysis base class has some smarts, lets use them.
662 return AliasAnalysis::getModRefInfo(CS, Loc);
665 ModRefInfo BasicAliasAnalysis::getModRefInfo(ImmutableCallSite CS1,
666 ImmutableCallSite CS2) {
667 // While the assume intrinsic is marked as arbitrarily writing so that
668 // proper control dependencies will be maintained, it never aliases any
669 // particular memory location.
670 if (isAssumeIntrinsic(CS1) || isAssumeIntrinsic(CS2))
673 // The AliasAnalysis base class has some smarts, lets use them.
674 return AliasAnalysis::getModRefInfo(CS1, CS2);
677 /// Provide ad-hoc rules to disambiguate accesses through two GEP operators,
678 /// both having the exact same pointer operand.
679 static AliasResult aliasSameBasePointerGEPs(const GEPOperator *GEP1,
681 const GEPOperator *GEP2,
683 const DataLayout &DL) {
685 assert(GEP1->getPointerOperand() == GEP2->getPointerOperand() &&
686 "Expected GEPs with the same pointer operand");
688 // Try to determine whether GEP1 and GEP2 index through arrays, into structs,
689 // such that the struct field accesses provably cannot alias.
690 // We also need at least two indices (the pointer, and the struct field).
691 if (GEP1->getNumIndices() != GEP2->getNumIndices() ||
692 GEP1->getNumIndices() < 2)
695 // If we don't know the size of the accesses through both GEPs, we can't
696 // determine whether the struct fields accessed can't alias.
697 if (V1Size == MemoryLocation::UnknownSize ||
698 V2Size == MemoryLocation::UnknownSize)
702 dyn_cast<ConstantInt>(GEP1->getOperand(GEP1->getNumOperands() - 1));
704 dyn_cast<ConstantInt>(GEP2->getOperand(GEP2->getNumOperands() - 1));
706 // If the last (struct) indices aren't constants, we can't say anything.
707 // If they're identical, the other indices might be also be dynamically
708 // equal, so the GEPs can alias.
709 if (!C1 || !C2 || C1 == C2)
712 // Find the last-indexed type of the GEP, i.e., the type you'd get if
713 // you stripped the last index.
714 // On the way, look at each indexed type. If there's something other
715 // than an array, different indices can lead to different final types.
716 SmallVector<Value *, 8> IntermediateIndices;
718 // Insert the first index; we don't need to check the type indexed
719 // through it as it only drops the pointer indirection.
720 assert(GEP1->getNumIndices() > 1 && "Not enough GEP indices to examine");
721 IntermediateIndices.push_back(GEP1->getOperand(1));
723 // Insert all the remaining indices but the last one.
724 // Also, check that they all index through arrays.
725 for (unsigned i = 1, e = GEP1->getNumIndices() - 1; i != e; ++i) {
726 if (!isa<ArrayType>(GetElementPtrInst::getIndexedType(
727 GEP1->getSourceElementType(), IntermediateIndices)))
729 IntermediateIndices.push_back(GEP1->getOperand(i + 1));
732 StructType *LastIndexedStruct =
733 dyn_cast<StructType>(GetElementPtrInst::getIndexedType(
734 GEP1->getSourceElementType(), IntermediateIndices));
736 if (!LastIndexedStruct)
740 // - both GEPs begin indexing from the exact same pointer;
741 // - the last indices in both GEPs are constants, indexing into a struct;
742 // - said indices are different, hence, the pointed-to fields are different;
743 // - both GEPs only index through arrays prior to that.
745 // This lets us determine that the struct that GEP1 indexes into and the
746 // struct that GEP2 indexes into must either precisely overlap or be
747 // completely disjoint. Because they cannot partially overlap, indexing into
748 // different non-overlapping fields of the struct will never alias.
750 // Therefore, the only remaining thing needed to show that both GEPs can't
751 // alias is that the fields are not overlapping.
752 const StructLayout *SL = DL.getStructLayout(LastIndexedStruct);
753 const uint64_t StructSize = SL->getSizeInBytes();
754 const uint64_t V1Off = SL->getElementOffset(C1->getZExtValue());
755 const uint64_t V2Off = SL->getElementOffset(C2->getZExtValue());
757 auto EltsDontOverlap = [StructSize](uint64_t V1Off, uint64_t V1Size,
758 uint64_t V2Off, uint64_t V2Size) {
759 return V1Off < V2Off && V1Off + V1Size <= V2Off &&
760 ((V2Off + V2Size <= StructSize) ||
761 (V2Off + V2Size - StructSize <= V1Off));
764 if (EltsDontOverlap(V1Off, V1Size, V2Off, V2Size) ||
765 EltsDontOverlap(V2Off, V2Size, V1Off, V1Size))
771 /// Provides a bunch of ad-hoc rules to disambiguate a GEP instruction against
774 /// We know that V1 is a GEP, but we don't know anything about V2.
775 /// UnderlyingV1 is GetUnderlyingObject(GEP1, DL), UnderlyingV2 is the same for
777 AliasResult BasicAliasAnalysis::aliasGEP(
778 const GEPOperator *GEP1, uint64_t V1Size, const AAMDNodes &V1AAInfo,
779 const Value *V2, uint64_t V2Size, const AAMDNodes &V2AAInfo,
780 const Value *UnderlyingV1, const Value *UnderlyingV2) {
781 int64_t GEP1BaseOffset;
782 bool GEP1MaxLookupReached;
783 SmallVector<VariableGEPIndex, 4> GEP1VariableIndices;
785 // We have to get two AssumptionCaches here because GEP1 and V2 may be from
786 // different functions.
787 // FIXME: This really doesn't make any sense. We get a dominator tree below
788 // that can only refer to a single function. But this function (aliasGEP) is
789 // a method on an immutable pass that can be called when there *isn't*
790 // a single function. The old pass management layer makes this "work", but
791 // this isn't really a clean solution.
792 AssumptionCacheTracker &ACT = getAnalysis<AssumptionCacheTracker>();
793 AssumptionCache *AC1 = nullptr, *AC2 = nullptr;
794 if (auto *GEP1I = dyn_cast<Instruction>(GEP1))
795 AC1 = &ACT.getAssumptionCache(
796 const_cast<Function &>(*GEP1I->getParent()->getParent()));
797 if (auto *I2 = dyn_cast<Instruction>(V2))
798 AC2 = &ACT.getAssumptionCache(
799 const_cast<Function &>(*I2->getParent()->getParent()));
801 DominatorTreeWrapperPass *DTWP =
802 getAnalysisIfAvailable<DominatorTreeWrapperPass>();
803 DominatorTree *DT = DTWP ? &DTWP->getDomTree() : nullptr;
805 // If we have two gep instructions with must-alias or not-alias'ing base
806 // pointers, figure out if the indexes to the GEP tell us anything about the
808 if (const GEPOperator *GEP2 = dyn_cast<GEPOperator>(V2)) {
809 // Do the base pointers alias?
810 AliasResult BaseAlias =
811 aliasCheck(UnderlyingV1, MemoryLocation::UnknownSize, AAMDNodes(),
812 UnderlyingV2, MemoryLocation::UnknownSize, AAMDNodes());
814 // Check for geps of non-aliasing underlying pointers where the offsets are
816 if ((BaseAlias == MayAlias) && V1Size == V2Size) {
817 // Do the base pointers alias assuming type and size.
818 AliasResult PreciseBaseAlias = aliasCheck(UnderlyingV1, V1Size, V1AAInfo,
819 UnderlyingV2, V2Size, V2AAInfo);
820 if (PreciseBaseAlias == NoAlias) {
821 // See if the computed offset from the common pointer tells us about the
822 // relation of the resulting pointer.
823 int64_t GEP2BaseOffset;
824 bool GEP2MaxLookupReached;
825 SmallVector<VariableGEPIndex, 4> GEP2VariableIndices;
826 const Value *GEP2BasePtr =
827 DecomposeGEPExpression(GEP2, GEP2BaseOffset, GEP2VariableIndices,
828 GEP2MaxLookupReached, *DL, AC2, DT);
829 const Value *GEP1BasePtr =
830 DecomposeGEPExpression(GEP1, GEP1BaseOffset, GEP1VariableIndices,
831 GEP1MaxLookupReached, *DL, AC1, DT);
832 // DecomposeGEPExpression and GetUnderlyingObject should return the
833 // same result except when DecomposeGEPExpression has no DataLayout.
834 if (GEP1BasePtr != UnderlyingV1 || GEP2BasePtr != UnderlyingV2) {
836 "DecomposeGEPExpression and GetUnderlyingObject disagree!");
839 // If the max search depth is reached the result is undefined
840 if (GEP2MaxLookupReached || GEP1MaxLookupReached)
844 if (GEP1BaseOffset == GEP2BaseOffset &&
845 GEP1VariableIndices == GEP2VariableIndices)
847 GEP1VariableIndices.clear();
851 // If we get a No or May, then return it immediately, no amount of analysis
852 // will improve this situation.
853 if (BaseAlias != MustAlias)
856 // Otherwise, we have a MustAlias. Since the base pointers alias each other
857 // exactly, see if the computed offset from the common pointer tells us
858 // about the relation of the resulting pointer.
859 const Value *GEP1BasePtr =
860 DecomposeGEPExpression(GEP1, GEP1BaseOffset, GEP1VariableIndices,
861 GEP1MaxLookupReached, *DL, AC1, DT);
863 int64_t GEP2BaseOffset;
864 bool GEP2MaxLookupReached;
865 SmallVector<VariableGEPIndex, 4> GEP2VariableIndices;
866 const Value *GEP2BasePtr =
867 DecomposeGEPExpression(GEP2, GEP2BaseOffset, GEP2VariableIndices,
868 GEP2MaxLookupReached, *DL, AC2, DT);
870 // DecomposeGEPExpression and GetUnderlyingObject should return the
871 // same result except when DecomposeGEPExpression has no DataLayout.
872 if (GEP1BasePtr != UnderlyingV1 || GEP2BasePtr != UnderlyingV2) {
873 assert(!DL && "DecomposeGEPExpression and GetUnderlyingObject disagree!");
877 // If we know the two GEPs are based off of the exact same pointer (and not
878 // just the same underlying object), see if that tells us anything about
879 // the resulting pointers.
880 if (DL && GEP1->getPointerOperand() == GEP2->getPointerOperand()) {
881 AliasResult R = aliasSameBasePointerGEPs(GEP1, V1Size, GEP2, V2Size, *DL);
882 // If we couldn't find anything interesting, don't abandon just yet.
887 // If the max search depth is reached the result is undefined
888 if (GEP2MaxLookupReached || GEP1MaxLookupReached)
891 // Subtract the GEP2 pointer from the GEP1 pointer to find out their
892 // symbolic difference.
893 GEP1BaseOffset -= GEP2BaseOffset;
894 GetIndexDifference(GEP1VariableIndices, GEP2VariableIndices);
897 // Check to see if these two pointers are related by the getelementptr
898 // instruction. If one pointer is a GEP with a non-zero index of the other
899 // pointer, we know they cannot alias.
901 // If both accesses are unknown size, we can't do anything useful here.
902 if (V1Size == MemoryLocation::UnknownSize &&
903 V2Size == MemoryLocation::UnknownSize)
906 AliasResult R = aliasCheck(UnderlyingV1, MemoryLocation::UnknownSize,
907 AAMDNodes(), V2, V2Size, V2AAInfo);
909 // If V2 may alias GEP base pointer, conservatively returns MayAlias.
910 // If V2 is known not to alias GEP base pointer, then the two values
911 // cannot alias per GEP semantics: "A pointer value formed from a
912 // getelementptr instruction is associated with the addresses associated
913 // with the first operand of the getelementptr".
916 const Value *GEP1BasePtr =
917 DecomposeGEPExpression(GEP1, GEP1BaseOffset, GEP1VariableIndices,
918 GEP1MaxLookupReached, *DL, AC1, DT);
920 // DecomposeGEPExpression and GetUnderlyingObject should return the
921 // same result except when DecomposeGEPExpression has no DataLayout.
922 if (GEP1BasePtr != UnderlyingV1) {
923 assert(!DL && "DecomposeGEPExpression and GetUnderlyingObject disagree!");
926 // If the max search depth is reached the result is undefined
927 if (GEP1MaxLookupReached)
931 // In the two GEP Case, if there is no difference in the offsets of the
932 // computed pointers, the resultant pointers are a must alias. This
933 // hapens when we have two lexically identical GEP's (for example).
935 // In the other case, if we have getelementptr <ptr>, 0, 0, 0, 0, ... and V2
936 // must aliases the GEP, the end result is a must alias also.
937 if (GEP1BaseOffset == 0 && GEP1VariableIndices.empty())
940 // If there is a constant difference between the pointers, but the difference
941 // is less than the size of the associated memory object, then we know
942 // that the objects are partially overlapping. If the difference is
943 // greater, we know they do not overlap.
944 if (GEP1BaseOffset != 0 && GEP1VariableIndices.empty()) {
945 if (GEP1BaseOffset >= 0) {
946 if (V2Size != MemoryLocation::UnknownSize) {
947 if ((uint64_t)GEP1BaseOffset < V2Size)
952 // We have the situation where:
955 // ---------------->|
956 // |-->V1Size |-------> V2Size
958 // We need to know that V2Size is not unknown, otherwise we might have
959 // stripped a gep with negative index ('gep <ptr>, -1, ...).
960 if (V1Size != MemoryLocation::UnknownSize &&
961 V2Size != MemoryLocation::UnknownSize) {
962 if (-(uint64_t)GEP1BaseOffset < V1Size)
969 if (!GEP1VariableIndices.empty()) {
971 bool AllPositive = true;
972 for (unsigned i = 0, e = GEP1VariableIndices.size(); i != e; ++i) {
974 // Try to distinguish something like &A[i][1] against &A[42][0].
975 // Grab the least significant bit set in any of the scales. We
976 // don't need std::abs here (even if the scale's negative) as we'll
977 // be ^'ing Modulo with itself later.
978 Modulo |= (uint64_t)GEP1VariableIndices[i].Scale;
981 // If the Value could change between cycles, then any reasoning about
982 // the Value this cycle may not hold in the next cycle. We'll just
983 // give up if we can't determine conditions that hold for every cycle:
984 const Value *V = GEP1VariableIndices[i].V;
986 bool SignKnownZero, SignKnownOne;
987 ComputeSignBit(const_cast<Value *>(V), SignKnownZero, SignKnownOne, *DL,
988 0, AC1, nullptr, DT);
990 // Zero-extension widens the variable, and so forces the sign
992 bool IsZExt = GEP1VariableIndices[i].Extension == EK_ZeroExt;
993 SignKnownZero |= IsZExt;
994 SignKnownOne &= !IsZExt;
996 // If the variable begins with a zero then we know it's
997 // positive, regardless of whether the value is signed or
999 int64_t Scale = GEP1VariableIndices[i].Scale;
1001 (SignKnownZero && Scale >= 0) || (SignKnownOne && Scale < 0);
1005 Modulo = Modulo ^ (Modulo & (Modulo - 1));
1007 // We can compute the difference between the two addresses
1008 // mod Modulo. Check whether that difference guarantees that the
1009 // two locations do not alias.
1010 uint64_t ModOffset = (uint64_t)GEP1BaseOffset & (Modulo - 1);
1011 if (V1Size != MemoryLocation::UnknownSize &&
1012 V2Size != MemoryLocation::UnknownSize && ModOffset >= V2Size &&
1013 V1Size <= Modulo - ModOffset)
1016 // If we know all the variables are positive, then GEP1 >= GEP1BasePtr.
1017 // If GEP1BasePtr > V2 (GEP1BaseOffset > 0) then we know the pointers
1018 // don't alias if V2Size can fit in the gap between V2 and GEP1BasePtr.
1019 if (AllPositive && GEP1BaseOffset > 0 && V2Size <= (uint64_t)GEP1BaseOffset)
1023 // Statically, we can see that the base objects are the same, but the
1024 // pointers have dynamic offsets which we can't resolve. And none of our
1025 // little tricks above worked.
1027 // TODO: Returning PartialAlias instead of MayAlias is a mild hack; the
1028 // practical effect of this is protecting TBAA in the case of dynamic
1029 // indices into arrays of unions or malloc'd memory.
1030 return PartialAlias;
1033 static AliasResult MergeAliasResults(AliasResult A, AliasResult B) {
1034 // If the results agree, take it.
1037 // A mix of PartialAlias and MustAlias is PartialAlias.
1038 if ((A == PartialAlias && B == MustAlias) ||
1039 (B == PartialAlias && A == MustAlias))
1040 return PartialAlias;
1041 // Otherwise, we don't know anything.
1045 /// Provides a bunch of ad-hoc rules to disambiguate a Select instruction
1046 /// against another.
1047 AliasResult BasicAliasAnalysis::aliasSelect(const SelectInst *SI,
1049 const AAMDNodes &SIAAInfo,
1050 const Value *V2, uint64_t V2Size,
1051 const AAMDNodes &V2AAInfo) {
1052 // If the values are Selects with the same condition, we can do a more precise
1053 // check: just check for aliases between the values on corresponding arms.
1054 if (const SelectInst *SI2 = dyn_cast<SelectInst>(V2))
1055 if (SI->getCondition() == SI2->getCondition()) {
1056 AliasResult Alias = aliasCheck(SI->getTrueValue(), SISize, SIAAInfo,
1057 SI2->getTrueValue(), V2Size, V2AAInfo);
1058 if (Alias == MayAlias)
1060 AliasResult ThisAlias =
1061 aliasCheck(SI->getFalseValue(), SISize, SIAAInfo,
1062 SI2->getFalseValue(), V2Size, V2AAInfo);
1063 return MergeAliasResults(ThisAlias, Alias);
1066 // If both arms of the Select node NoAlias or MustAlias V2, then returns
1067 // NoAlias / MustAlias. Otherwise, returns MayAlias.
1069 aliasCheck(V2, V2Size, V2AAInfo, SI->getTrueValue(), SISize, SIAAInfo);
1070 if (Alias == MayAlias)
1073 AliasResult ThisAlias =
1074 aliasCheck(V2, V2Size, V2AAInfo, SI->getFalseValue(), SISize, SIAAInfo);
1075 return MergeAliasResults(ThisAlias, Alias);
1078 /// Provide a bunch of ad-hoc rules to disambiguate a PHI instruction against
1080 AliasResult BasicAliasAnalysis::aliasPHI(const PHINode *PN, uint64_t PNSize,
1081 const AAMDNodes &PNAAInfo,
1082 const Value *V2, uint64_t V2Size,
1083 const AAMDNodes &V2AAInfo) {
1084 // Track phi nodes we have visited. We use this information when we determine
1085 // value equivalence.
1086 VisitedPhiBBs.insert(PN->getParent());
1088 // If the values are PHIs in the same block, we can do a more precise
1089 // as well as efficient check: just check for aliases between the values
1090 // on corresponding edges.
1091 if (const PHINode *PN2 = dyn_cast<PHINode>(V2))
1092 if (PN2->getParent() == PN->getParent()) {
1093 LocPair Locs(MemoryLocation(PN, PNSize, PNAAInfo),
1094 MemoryLocation(V2, V2Size, V2AAInfo));
1096 std::swap(Locs.first, Locs.second);
1097 // Analyse the PHIs' inputs under the assumption that the PHIs are
1099 // If the PHIs are May/MustAlias there must be (recursively) an input
1100 // operand from outside the PHIs' cycle that is MayAlias/MustAlias or
1101 // there must be an operation on the PHIs within the PHIs' value cycle
1102 // that causes a MayAlias.
1103 // Pretend the phis do not alias.
1104 AliasResult Alias = NoAlias;
1105 assert(AliasCache.count(Locs) &&
1106 "There must exist an entry for the phi node");
1107 AliasResult OrigAliasResult = AliasCache[Locs];
1108 AliasCache[Locs] = NoAlias;
1110 for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
1111 AliasResult ThisAlias =
1112 aliasCheck(PN->getIncomingValue(i), PNSize, PNAAInfo,
1113 PN2->getIncomingValueForBlock(PN->getIncomingBlock(i)),
1115 Alias = MergeAliasResults(ThisAlias, Alias);
1116 if (Alias == MayAlias)
1120 // Reset if speculation failed.
1121 if (Alias != NoAlias)
1122 AliasCache[Locs] = OrigAliasResult;
1127 SmallPtrSet<Value *, 4> UniqueSrc;
1128 SmallVector<Value *, 4> V1Srcs;
1129 bool isRecursive = false;
1130 for (Value *PV1 : PN->incoming_values()) {
1131 if (isa<PHINode>(PV1))
1132 // If any of the source itself is a PHI, return MayAlias conservatively
1133 // to avoid compile time explosion. The worst possible case is if both
1134 // sides are PHI nodes. In which case, this is O(m x n) time where 'm'
1135 // and 'n' are the number of PHI sources.
1138 if (EnableRecPhiAnalysis)
1139 if (GEPOperator *PV1GEP = dyn_cast<GEPOperator>(PV1)) {
1140 // Check whether the incoming value is a GEP that advances the pointer
1141 // result of this PHI node (e.g. in a loop). If this is the case, we
1142 // would recurse and always get a MayAlias. Handle this case specially
1144 if (PV1GEP->getPointerOperand() == PN && PV1GEP->getNumIndices() == 1 &&
1145 isa<ConstantInt>(PV1GEP->idx_begin())) {
1151 if (UniqueSrc.insert(PV1).second)
1152 V1Srcs.push_back(PV1);
1155 // If this PHI node is recursive, set the size of the accessed memory to
1156 // unknown to represent all the possible values the GEP could advance the
1159 PNSize = MemoryLocation::UnknownSize;
1162 aliasCheck(V2, V2Size, V2AAInfo, V1Srcs[0], PNSize, PNAAInfo);
1164 // Early exit if the check of the first PHI source against V2 is MayAlias.
1165 // Other results are not possible.
1166 if (Alias == MayAlias)
1169 // If all sources of the PHI node NoAlias or MustAlias V2, then returns
1170 // NoAlias / MustAlias. Otherwise, returns MayAlias.
1171 for (unsigned i = 1, e = V1Srcs.size(); i != e; ++i) {
1172 Value *V = V1Srcs[i];
1174 AliasResult ThisAlias =
1175 aliasCheck(V2, V2Size, V2AAInfo, V, PNSize, PNAAInfo);
1176 Alias = MergeAliasResults(ThisAlias, Alias);
1177 if (Alias == MayAlias)
1184 /// Provideis a bunch of ad-hoc rules to disambiguate in common cases, such as
1185 /// array references.
1186 AliasResult BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size,
1187 AAMDNodes V1AAInfo, const Value *V2,
1189 AAMDNodes V2AAInfo) {
1190 // If either of the memory references is empty, it doesn't matter what the
1191 // pointer values are.
1192 if (V1Size == 0 || V2Size == 0)
1195 // Strip off any casts if they exist.
1196 V1 = V1->stripPointerCasts();
1197 V2 = V2->stripPointerCasts();
1199 // If V1 or V2 is undef, the result is NoAlias because we can always pick a
1200 // value for undef that aliases nothing in the program.
1201 if (isa<UndefValue>(V1) || isa<UndefValue>(V2))
1204 // Are we checking for alias of the same value?
1205 // Because we look 'through' phi nodes we could look at "Value" pointers from
1206 // different iterations. We must therefore make sure that this is not the
1207 // case. The function isValueEqualInPotentialCycles ensures that this cannot
1208 // happen by looking at the visited phi nodes and making sure they cannot
1210 if (isValueEqualInPotentialCycles(V1, V2))
1213 if (!V1->getType()->isPointerTy() || !V2->getType()->isPointerTy())
1214 return NoAlias; // Scalars cannot alias each other
1216 // Figure out what objects these things are pointing to if we can.
1217 const Value *O1 = GetUnderlyingObject(V1, *DL, MaxLookupSearchDepth);
1218 const Value *O2 = GetUnderlyingObject(V2, *DL, MaxLookupSearchDepth);
1220 // Null values in the default address space don't point to any object, so they
1221 // don't alias any other pointer.
1222 if (const ConstantPointerNull *CPN = dyn_cast<ConstantPointerNull>(O1))
1223 if (CPN->getType()->getAddressSpace() == 0)
1225 if (const ConstantPointerNull *CPN = dyn_cast<ConstantPointerNull>(O2))
1226 if (CPN->getType()->getAddressSpace() == 0)
1230 // If V1/V2 point to two different objects we know that we have no alias.
1231 if (isIdentifiedObject(O1) && isIdentifiedObject(O2))
1234 // Constant pointers can't alias with non-const isIdentifiedObject objects.
1235 if ((isa<Constant>(O1) && isIdentifiedObject(O2) && !isa<Constant>(O2)) ||
1236 (isa<Constant>(O2) && isIdentifiedObject(O1) && !isa<Constant>(O1)))
1239 // Function arguments can't alias with things that are known to be
1240 // unambigously identified at the function level.
1241 if ((isa<Argument>(O1) && isIdentifiedFunctionLocal(O2)) ||
1242 (isa<Argument>(O2) && isIdentifiedFunctionLocal(O1)))
1245 // Most objects can't alias null.
1246 if ((isa<ConstantPointerNull>(O2) && isKnownNonNull(O1)) ||
1247 (isa<ConstantPointerNull>(O1) && isKnownNonNull(O2)))
1250 // If one pointer is the result of a call/invoke or load and the other is a
1251 // non-escaping local object within the same function, then we know the
1252 // object couldn't escape to a point where the call could return it.
1254 // Note that if the pointers are in different functions, there are a
1255 // variety of complications. A call with a nocapture argument may still
1256 // temporary store the nocapture argument's value in a temporary memory
1257 // location if that memory location doesn't escape. Or it may pass a
1258 // nocapture value to other functions as long as they don't capture it.
1259 if (isEscapeSource(O1) && isNonEscapingLocalObject(O2))
1261 if (isEscapeSource(O2) && isNonEscapingLocalObject(O1))
1265 // If the size of one access is larger than the entire object on the other
1266 // side, then we know such behavior is undefined and can assume no alias.
1268 if ((V1Size != MemoryLocation::UnknownSize &&
1269 isObjectSmallerThan(O2, V1Size, *DL, *TLI)) ||
1270 (V2Size != MemoryLocation::UnknownSize &&
1271 isObjectSmallerThan(O1, V2Size, *DL, *TLI)))
1274 // Check the cache before climbing up use-def chains. This also terminates
1275 // otherwise infinitely recursive queries.
1276 LocPair Locs(MemoryLocation(V1, V1Size, V1AAInfo),
1277 MemoryLocation(V2, V2Size, V2AAInfo));
1279 std::swap(Locs.first, Locs.second);
1280 std::pair<AliasCacheTy::iterator, bool> Pair =
1281 AliasCache.insert(std::make_pair(Locs, MayAlias));
1283 return Pair.first->second;
1285 // FIXME: This isn't aggressively handling alias(GEP, PHI) for example: if the
1286 // GEP can't simplify, we don't even look at the PHI cases.
1287 if (!isa<GEPOperator>(V1) && isa<GEPOperator>(V2)) {
1289 std::swap(V1Size, V2Size);
1291 std::swap(V1AAInfo, V2AAInfo);
1293 if (const GEPOperator *GV1 = dyn_cast<GEPOperator>(V1)) {
1294 AliasResult Result =
1295 aliasGEP(GV1, V1Size, V1AAInfo, V2, V2Size, V2AAInfo, O1, O2);
1296 if (Result != MayAlias)
1297 return AliasCache[Locs] = Result;
1300 if (isa<PHINode>(V2) && !isa<PHINode>(V1)) {
1302 std::swap(V1Size, V2Size);
1303 std::swap(V1AAInfo, V2AAInfo);
1305 if (const PHINode *PN = dyn_cast<PHINode>(V1)) {
1306 AliasResult Result = aliasPHI(PN, V1Size, V1AAInfo, V2, V2Size, V2AAInfo);
1307 if (Result != MayAlias)
1308 return AliasCache[Locs] = Result;
1311 if (isa<SelectInst>(V2) && !isa<SelectInst>(V1)) {
1313 std::swap(V1Size, V2Size);
1314 std::swap(V1AAInfo, V2AAInfo);
1316 if (const SelectInst *S1 = dyn_cast<SelectInst>(V1)) {
1317 AliasResult Result =
1318 aliasSelect(S1, V1Size, V1AAInfo, V2, V2Size, V2AAInfo);
1319 if (Result != MayAlias)
1320 return AliasCache[Locs] = Result;
1323 // If both pointers are pointing into the same object and one of them
1324 // accesses is accessing the entire object, then the accesses must
1325 // overlap in some way.
1327 if ((V1Size != MemoryLocation::UnknownSize &&
1328 isObjectSize(O1, V1Size, *DL, *TLI)) ||
1329 (V2Size != MemoryLocation::UnknownSize &&
1330 isObjectSize(O2, V2Size, *DL, *TLI)))
1331 return AliasCache[Locs] = PartialAlias;
1333 AliasResult Result =
1334 AliasAnalysis::alias(MemoryLocation(V1, V1Size, V1AAInfo),
1335 MemoryLocation(V2, V2Size, V2AAInfo));
1336 return AliasCache[Locs] = Result;
1339 /// Check whether two Values can be considered equivalent.
1341 /// In addition to pointer equivalence of \p V1 and \p V2 this checks whether
1342 /// they can not be part of a cycle in the value graph by looking at all
1343 /// visited phi nodes an making sure that the phis cannot reach the value. We
1344 /// have to do this because we are looking through phi nodes (That is we say
1345 /// noalias(V, phi(VA, VB)) if noalias(V, VA) and noalias(V, VB).
1346 bool BasicAliasAnalysis::isValueEqualInPotentialCycles(const Value *V,
1351 const Instruction *Inst = dyn_cast<Instruction>(V);
1355 if (VisitedPhiBBs.empty())
1358 if (VisitedPhiBBs.size() > MaxNumPhiBBsValueReachabilityCheck)
1361 // Use dominance or loop info if available.
1362 DominatorTreeWrapperPass *DTWP =
1363 getAnalysisIfAvailable<DominatorTreeWrapperPass>();
1364 DominatorTree *DT = DTWP ? &DTWP->getDomTree() : nullptr;
1365 auto *LIWP = getAnalysisIfAvailable<LoopInfoWrapperPass>();
1366 LoopInfo *LI = LIWP ? &LIWP->getLoopInfo() : nullptr;
1368 // Make sure that the visited phis cannot reach the Value. This ensures that
1369 // the Values cannot come from different iterations of a potential cycle the
1370 // phi nodes could be involved in.
1371 for (auto *P : VisitedPhiBBs)
1372 if (isPotentiallyReachable(P->begin(), Inst, DT, LI))
1378 /// Computes the symbolic difference between two de-composed GEPs.
1380 /// Dest and Src are the variable indices from two decomposed GetElementPtr
1381 /// instructions GEP1 and GEP2 which have common base pointers.
1382 void BasicAliasAnalysis::GetIndexDifference(
1383 SmallVectorImpl<VariableGEPIndex> &Dest,
1384 const SmallVectorImpl<VariableGEPIndex> &Src) {
1388 for (unsigned i = 0, e = Src.size(); i != e; ++i) {
1389 const Value *V = Src[i].V;
1390 ExtensionKind Extension = Src[i].Extension;
1391 int64_t Scale = Src[i].Scale;
1393 // Find V in Dest. This is N^2, but pointer indices almost never have more
1394 // than a few variable indexes.
1395 for (unsigned j = 0, e = Dest.size(); j != e; ++j) {
1396 if (!isValueEqualInPotentialCycles(Dest[j].V, V) ||
1397 Dest[j].Extension != Extension)
1400 // If we found it, subtract off Scale V's from the entry in Dest. If it
1401 // goes to zero, remove the entry.
1402 if (Dest[j].Scale != Scale)
1403 Dest[j].Scale -= Scale;
1405 Dest.erase(Dest.begin() + j);
1410 // If we didn't consume this entry, add it to the end of the Dest list.
1412 VariableGEPIndex Entry = {V, Extension, -Scale};
1413 Dest.push_back(Entry);