X-Git-Url: http://plrg.eecs.uci.edu/git/?p=oota-llvm.git;a=blobdiff_plain;f=lib%2FAnalysis%2FBasicAliasAnalysis.cpp;h=a6a25f9326a015f23fa39c95b1b75f29c288d5c9;hp=2da64c279f901333ef4cc27e26f54403b9bb08df;hb=37d8ade35ce1bb8feb46f8731340d0b0b5f7ab6c;hpb=4d7ed3960cbe2ec937ed42fb3e80237f0c26453c diff --git a/lib/Analysis/BasicAliasAnalysis.cpp b/lib/Analysis/BasicAliasAnalysis.cpp index 2da64c279f9..a6a25f9326a 100644 --- a/lib/Analysis/BasicAliasAnalysis.cpp +++ b/lib/Analysis/BasicAliasAnalysis.cpp @@ -13,24 +13,20 @@ // //===----------------------------------------------------------------------===// -#include "llvm/Analysis/Passes.h" -#include "llvm/ADT/SmallPtrSet.h" +#include "llvm/Analysis/BasicAliasAnalysis.h" #include "llvm/ADT/SmallVector.h" +#include "llvm/ADT/Statistic.h" #include "llvm/Analysis/AliasAnalysis.h" -#include "llvm/Analysis/AssumptionCache.h" #include "llvm/Analysis/CFG.h" #include "llvm/Analysis/CaptureTracking.h" #include "llvm/Analysis/InstructionSimplify.h" #include "llvm/Analysis/LoopInfo.h" #include "llvm/Analysis/MemoryBuiltins.h" -#include "llvm/Analysis/TargetLibraryInfo.h" #include "llvm/Analysis/ValueTracking.h" #include "llvm/IR/Constants.h" #include "llvm/IR/DataLayout.h" #include "llvm/IR/DerivedTypes.h" #include "llvm/IR/Dominators.h" -#include "llvm/IR/Function.h" -#include "llvm/IR/GetElementPtrTypeIterator.h" #include "llvm/IR/GlobalAlias.h" #include "llvm/IR/GlobalVariable.h" #include "llvm/IR/Instructions.h" @@ -42,6 +38,18 @@ #include using namespace llvm; +/// Enable analysis of recursive PHI nodes. +static cl::opt EnableRecPhiAnalysis("basicaa-recphi", cl::Hidden, + cl::init(false)); + +/// SearchLimitReached / SearchTimes shows how often the limit of +/// to decompose GEPs is reached. It will affect the precision +/// of basic alias analysis. +#define DEBUG_TYPE "basicaa" +STATISTIC(SearchLimitReached, "Number of times the limit to " + "decompose GEPs is reached"); +STATISTIC(SearchTimes, "Number of times a GEP is decomposed"); + /// Cutoff after which to stop analysing a set of phi nodes potentially involved /// in a cycle. Because we are analysing 'through' phi nodes we need to be /// careful with value equivalence. We use reachability to make sure a value @@ -57,8 +65,8 @@ static const unsigned MaxLookupSearchDepth = 6; // Useful predicates //===----------------------------------------------------------------------===// -/// isNonEscapingLocalObject - Return true if the pointer is to a function-local -/// object that never escapes from the function. +/// Returns true if the pointer is to a function-local object that never +/// escapes from the function. static bool isNonEscapingLocalObject(const Value *V) { // If this is a local allocation, check to see if it escapes. if (isa(V) || isNoAliasCall(V)) @@ -82,8 +90,8 @@ static bool isNonEscapingLocalObject(const Value *V) { return false; } -/// isEscapeSource - Return true if the pointer is one which would have -/// been considered an escape by isNonEscapingLocalObject. +/// Returns true if the pointer is one which would have been considered an +/// escape by isNonEscapingLocalObject. static bool isEscapeSource(const Value *V) { if (isa(V) || isa(V) || isa(V)) return true; @@ -97,19 +105,18 @@ static bool isEscapeSource(const Value *V) { return false; } -/// getObjectSize - Return the size of the object specified by V, or -/// UnknownSize if unknown. +/// Returns the size of the object specified by V, or UnknownSize if unknown. static uint64_t getObjectSize(const Value *V, const DataLayout &DL, const TargetLibraryInfo &TLI, bool RoundToAlign = false) { uint64_t Size; if (getObjectSize(V, Size, DL, &TLI, RoundToAlign)) return Size; - return AliasAnalysis::UnknownSize; + return MemoryLocation::UnknownSize; } -/// isObjectSmallerThan - Return true if we can prove that the object specified -/// by V is smaller than Size. +/// Returns true if we can prove that the object specified by V is smaller than +/// Size. static bool isObjectSmallerThan(const Value *V, uint64_t Size, const DataLayout &DL, const TargetLibraryInfo &TLI) { @@ -144,59 +151,36 @@ static bool isObjectSmallerThan(const Value *V, uint64_t Size, // This function needs to use the aligned object size because we allow // reads a bit past the end given sufficient alignment. - uint64_t ObjectSize = getObjectSize(V, DL, TLI, /*RoundToAlign*/true); + uint64_t ObjectSize = getObjectSize(V, DL, TLI, /*RoundToAlign*/ true); - return ObjectSize != AliasAnalysis::UnknownSize && ObjectSize < Size; + return ObjectSize != MemoryLocation::UnknownSize && ObjectSize < Size; } -/// isObjectSize - Return true if we can prove that the object specified -/// by V has size Size. -static bool isObjectSize(const Value *V, uint64_t Size, - const DataLayout &DL, const TargetLibraryInfo &TLI) { +/// Returns true if we can prove that the object specified by V has size Size. +static bool isObjectSize(const Value *V, uint64_t Size, const DataLayout &DL, + const TargetLibraryInfo &TLI) { uint64_t ObjectSize = getObjectSize(V, DL, TLI); - return ObjectSize != AliasAnalysis::UnknownSize && ObjectSize == Size; + return ObjectSize != MemoryLocation::UnknownSize && ObjectSize == Size; } //===----------------------------------------------------------------------===// // GetElementPtr Instruction Decomposition and Analysis //===----------------------------------------------------------------------===// -namespace { - enum ExtensionKind { - EK_NotExtended, - EK_SignExt, - EK_ZeroExt - }; - - struct VariableGEPIndex { - const Value *V; - ExtensionKind Extension; - int64_t Scale; - - bool operator==(const VariableGEPIndex &Other) const { - return V == Other.V && Extension == Other.Extension && - Scale == Other.Scale; - } - - bool operator!=(const VariableGEPIndex &Other) const { - return !operator==(Other); - } - }; -} - - -/// GetLinearExpression - Analyze the specified value as a linear expression: -/// "A*V + B", where A and B are constant integers. Return the scale and offset -/// values as APInts and return V as a Value*, and return whether we looked -/// through any sign or zero extends. The incoming Value is known to have -/// IntegerType and it may already be sign or zero extended. +/// Analyzes the specified value as a linear expression: "A*V + B", where A and +/// B are constant integers. +/// +/// Returns the scale and offset values as APInts and return V as a Value*, and +/// return whether we looked through any sign or zero extends. The incoming +/// Value is known to have IntegerType and it may already be sign or zero +/// extended. /// /// Note that this looks through extends, so the high bits may not be /// represented in the result. -static Value *GetLinearExpression(Value *V, APInt &Scale, APInt &Offset, - ExtensionKind &Extension, - const DataLayout &DL, unsigned Depth, - AssumptionCache *AC, DominatorTree *DT) { +/*static*/ const Value *BasicAliasAnalysis::GetLinearExpression( + const Value *V, APInt &Scale, APInt &Offset, unsigned &ZExtBits, + unsigned &SExtBits, const DataLayout &DL, unsigned Depth, + AssumptionCache *AC, DominatorTree *DT, bool &NSW, bool &NUW) { assert(V->getType()->isIntegerTy() && "Not an integer value"); // Limit our recursion depth. @@ -206,65 +190,125 @@ static Value *GetLinearExpression(Value *V, APInt &Scale, APInt &Offset, return V; } - if (ConstantInt *Const = dyn_cast(V)) { - // if it's a constant, just convert it to an offset - // and remove the variable. - Offset += Const->getValue(); + if (const ConstantInt *Const = dyn_cast(V)) { + // if it's a constant, just convert it to an offset and remove the variable. + // If we've been called recursively the Offset bit width will be greater + // than the constant's (the Offset's always as wide as the outermost call), + // so we'll zext here and process any extension in the isa & + // isa cases below. + Offset += Const->getValue().zextOrSelf(Offset.getBitWidth()); assert(Scale == 0 && "Constant values don't have a scale"); return V; } - if (BinaryOperator *BOp = dyn_cast(V)) { + if (const BinaryOperator *BOp = dyn_cast(V)) { if (ConstantInt *RHSC = dyn_cast(BOp->getOperand(1))) { + + // If we've been called recursively then Offset and Scale will be wider + // that the BOp operands. We'll always zext it here as we'll process sign + // extensions below (see the isa / isa cases). + APInt RHS = RHSC->getValue().zextOrSelf(Offset.getBitWidth()); + switch (BOp->getOpcode()) { - default: break; + default: + // We don't understand this instruction, so we can't decompose it any + // further. + Scale = 1; + Offset = 0; + return V; case Instruction::Or: // X|C == X+C if all the bits in C are unset in X. Otherwise we can't // analyze it. if (!MaskedValueIsZero(BOp->getOperand(0), RHSC->getValue(), DL, 0, AC, - BOp, DT)) - break; - // FALL THROUGH. + BOp, DT)) { + Scale = 1; + Offset = 0; + return V; + } + // FALL THROUGH. case Instruction::Add: - V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, Extension, - DL, Depth + 1, AC, DT); - Offset += RHSC->getValue(); - return V; + V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, ZExtBits, + SExtBits, DL, Depth + 1, AC, DT, NSW, NUW); + Offset += RHS; + break; + case Instruction::Sub: + V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, ZExtBits, + SExtBits, DL, Depth + 1, AC, DT, NSW, NUW); + Offset -= RHS; + break; case Instruction::Mul: - V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, Extension, - DL, Depth + 1, AC, DT); - Offset *= RHSC->getValue(); - Scale *= RHSC->getValue(); - return V; + V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, ZExtBits, + SExtBits, DL, Depth + 1, AC, DT, NSW, NUW); + Offset *= RHS; + Scale *= RHS; + break; case Instruction::Shl: - V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, Extension, - DL, Depth + 1, AC, DT); - Offset <<= RHSC->getValue().getLimitedValue(); - Scale <<= RHSC->getValue().getLimitedValue(); + V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, ZExtBits, + SExtBits, DL, Depth + 1, AC, DT, NSW, NUW); + Offset <<= RHS.getLimitedValue(); + Scale <<= RHS.getLimitedValue(); + // the semantics of nsw and nuw for left shifts don't match those of + // multiplications, so we won't propagate them. + NSW = NUW = false; return V; } + + if (isa(BOp)) { + NUW &= BOp->hasNoUnsignedWrap(); + NSW &= BOp->hasNoSignedWrap(); + } + return V; } } // Since GEP indices are sign extended anyway, we don't care about the high // bits of a sign or zero extended value - just scales and offsets. The // extensions have to be consistent though. - if ((isa(V) && Extension != EK_ZeroExt) || - (isa(V) && Extension != EK_SignExt)) { + if (isa(V) || isa(V)) { Value *CastOp = cast(V)->getOperand(0); - unsigned OldWidth = Scale.getBitWidth(); + unsigned NewWidth = V->getType()->getPrimitiveSizeInBits(); unsigned SmallWidth = CastOp->getType()->getPrimitiveSizeInBits(); - Scale = Scale.trunc(SmallWidth); - Offset = Offset.trunc(SmallWidth); - Extension = isa(V) ? EK_SignExt : EK_ZeroExt; - - Value *Result = GetLinearExpression(CastOp, Scale, Offset, Extension, DL, - Depth + 1, AC, DT); - Scale = Scale.zext(OldWidth); - - // We have to sign-extend even if Extension == EK_ZeroExt as we can't - // decompose a sign extension (i.e. zext(x - 1) != zext(x) - zext(-1)). - Offset = Offset.sext(OldWidth); + unsigned OldZExtBits = ZExtBits, OldSExtBits = SExtBits; + const Value *Result = + GetLinearExpression(CastOp, Scale, Offset, ZExtBits, SExtBits, DL, + Depth + 1, AC, DT, NSW, NUW); + + // zext(zext(%x)) == zext(%x), and similiarly for sext; we'll handle this + // by just incrementing the number of bits we've extended by. + unsigned ExtendedBy = NewWidth - SmallWidth; + + if (isa(V) && ZExtBits == 0) { + // sext(sext(%x, a), b) == sext(%x, a + b) + + if (NSW) { + // We haven't sign-wrapped, so it's valid to decompose sext(%x + c) + // into sext(%x) + sext(c). We'll sext the Offset ourselves: + unsigned OldWidth = Offset.getBitWidth(); + Offset = Offset.trunc(SmallWidth).sext(NewWidth).zextOrSelf(OldWidth); + } else { + // We may have signed-wrapped, so don't decompose sext(%x + c) into + // sext(%x) + sext(c) + Scale = 1; + Offset = 0; + Result = CastOp; + ZExtBits = OldZExtBits; + SExtBits = OldSExtBits; + } + SExtBits += ExtendedBy; + } else { + // sext(zext(%x, a), b) = zext(zext(%x, a), b) = zext(%x, a + b) + + if (!NUW) { + // We may have unsigned-wrapped, so don't decompose zext(%x + c) into + // zext(%x) + zext(c) + Scale = 1; + Offset = 0; + Result = CastOp; + ZExtBits = OldZExtBits; + SExtBits = OldSExtBits; + } + ZExtBits += ExtendedBy; + } return Result; } @@ -274,29 +318,27 @@ static Value *GetLinearExpression(Value *V, APInt &Scale, APInt &Offset, return V; } -/// DecomposeGEPExpression - If V is a symbolic pointer expression, decompose it -/// into a base pointer with a constant offset and a number of scaled symbolic -/// offsets. +/// If V is a symbolic pointer expression, decompose it into a base pointer +/// with a constant offset and a number of scaled symbolic offsets. /// -/// The scaled symbolic offsets (represented by pairs of a Value* and a scale in -/// the VarIndices vector) are Value*'s that are known to be scaled by the -/// specified amount, but which may have other unrepresented high bits. As such, -/// the gep cannot necessarily be reconstructed from its decomposed form. +/// The scaled symbolic offsets (represented by pairs of a Value* and a scale +/// in the VarIndices vector) are Value*'s that are known to be scaled by the +/// specified amount, but which may have other unrepresented high bits. As +/// such, the gep cannot necessarily be reconstructed from its decomposed form. /// /// When DataLayout is around, this function is capable of analyzing everything /// that GetUnderlyingObject can look through. To be able to do that /// GetUnderlyingObject and DecomposeGEPExpression must use the same search -/// depth (MaxLookupSearchDepth). -/// When DataLayout not is around, it just looks through pointer casts. -/// -static const Value * -DecomposeGEPExpression(const Value *V, int64_t &BaseOffs, - SmallVectorImpl &VarIndices, - bool &MaxLookupReached, const DataLayout &DL, - AssumptionCache *AC, DominatorTree *DT) { +/// depth (MaxLookupSearchDepth). When DataLayout not is around, it just looks +/// through pointer casts. +/*static*/ const Value *BasicAliasAnalysis::DecomposeGEPExpression( + const Value *V, int64_t &BaseOffs, + SmallVectorImpl &VarIndices, bool &MaxLookupReached, + const DataLayout &DL, AssumptionCache *AC, DominatorTree *DT) { // Limit recursion depth to limit compile time in crazy cases. unsigned MaxLookup = MaxLookupSearchDepth; MaxLookupReached = false; + SearchTimes++; BaseOffs = 0; do { @@ -329,7 +371,7 @@ DecomposeGEPExpression(const Value *V, int64_t &BaseOffs, // updated when GetUnderlyingObject is updated). TLI should be // provided also. if (const Value *Simplified = - SimplifyInstruction(const_cast(I), DL)) { + SimplifyInstruction(const_cast(I), DL)) { V = Simplified; continue; } @@ -344,43 +386,47 @@ DecomposeGEPExpression(const Value *V, int64_t &BaseOffs, unsigned AS = GEPOp->getPointerAddressSpace(); // Walk the indices of the GEP, accumulating them into BaseOff/VarIndices. gep_type_iterator GTI = gep_type_begin(GEPOp); - for (User::const_op_iterator I = GEPOp->op_begin()+1, - E = GEPOp->op_end(); I != E; ++I) { - Value *Index = *I; + for (User::const_op_iterator I = GEPOp->op_begin() + 1, E = GEPOp->op_end(); + I != E; ++I) { + const Value *Index = *I; // Compute the (potentially symbolic) offset in bytes for this index. if (StructType *STy = dyn_cast(*GTI++)) { // For a struct, add the member offset. unsigned FieldNo = cast(Index)->getZExtValue(); - if (FieldNo == 0) continue; + if (FieldNo == 0) + continue; BaseOffs += DL.getStructLayout(STy)->getElementOffset(FieldNo); continue; } // For an array/pointer, add the element offset, explicitly scaled. - if (ConstantInt *CIdx = dyn_cast(Index)) { - if (CIdx->isZero()) continue; + if (const ConstantInt *CIdx = dyn_cast(Index)) { + if (CIdx->isZero()) + continue; BaseOffs += DL.getTypeAllocSize(*GTI) * CIdx->getSExtValue(); continue; } uint64_t Scale = DL.getTypeAllocSize(*GTI); - ExtensionKind Extension = EK_NotExtended; + unsigned ZExtBits = 0, SExtBits = 0; // If the integer type is smaller than the pointer size, it is implicitly // sign extended to pointer size. unsigned Width = Index->getType()->getIntegerBitWidth(); - if (DL.getPointerSizeInBits(AS) > Width) - Extension = EK_SignExt; + unsigned PointerSize = DL.getPointerSizeInBits(AS); + if (PointerSize > Width) + SExtBits += PointerSize - Width; // Use GetLinearExpression to decompose the index into a C1*V+C2 form. APInt IndexScale(Width, 0), IndexOffset(Width, 0); - Index = GetLinearExpression(Index, IndexScale, IndexOffset, Extension, DL, - 0, AC, DT); + bool NSW = true, NUW = true; + Index = GetLinearExpression(Index, IndexScale, IndexOffset, ZExtBits, + SExtBits, DL, 0, AC, DT, NSW, NUW); // The GEP index scale ("Scale") scales C1*V+C2, yielding (C1*V+C2)*Scale. // This gives us an aggregate computation of (C1*Scale)*V + C2*Scale. - BaseOffs += IndexOffset.getSExtValue()*Scale; + BaseOffs += IndexOffset.getSExtValue() * Scale; Scale *= IndexScale.getSExtValue(); // If we already had an occurrence of this index variable, merge this @@ -388,23 +434,23 @@ DecomposeGEPExpression(const Value *V, int64_t &BaseOffs, // A[x][x] -> x*16 + x*4 -> x*20 // This also ensures that 'x' only appears in the index list once. for (unsigned i = 0, e = VarIndices.size(); i != e; ++i) { - if (VarIndices[i].V == Index && - VarIndices[i].Extension == Extension) { + if (VarIndices[i].V == Index && VarIndices[i].ZExtBits == ZExtBits && + VarIndices[i].SExtBits == SExtBits) { Scale += VarIndices[i].Scale; - VarIndices.erase(VarIndices.begin()+i); + VarIndices.erase(VarIndices.begin() + i); break; } } // Make sure that we have a scale that makes sense for this target's // pointer size. - if (unsigned ShiftBits = 64 - DL.getPointerSizeInBits(AS)) { + if (unsigned ShiftBits = 64 - PointerSize) { Scale <<= ShiftBits; Scale = (int64_t)Scale >> ShiftBits; } if (Scale) { - VariableGEPIndex Entry = {Index, Extension, + VariableGEPIndex Entry = {Index, ZExtBits, SExtBits, static_cast(Scale)}; VarIndices.push_back(Entry); } @@ -416,6 +462,7 @@ DecomposeGEPExpression(const Value *V, int64_t &BaseOffs, // If the chain of expressions is too deep, just return early. MaxLookupReached = true; + SearchLimitReached++; return V; } @@ -423,177 +470,24 @@ DecomposeGEPExpression(const Value *V, int64_t &BaseOffs, // BasicAliasAnalysis Pass //===----------------------------------------------------------------------===// -#ifndef NDEBUG -static const Function *getParent(const Value *V) { - if (const Instruction *inst = dyn_cast(V)) - return inst->getParent()->getParent(); - - if (const Argument *arg = dyn_cast(V)) - return arg->getParent(); - - return nullptr; -} - -static bool notDifferentParent(const Value *O1, const Value *O2) { - - const Function *F1 = getParent(O1); - const Function *F2 = getParent(O2); - - return !F1 || !F2 || F1 == F2; -} -#endif - -namespace { - /// BasicAliasAnalysis - This is the primary alias analysis implementation. - struct BasicAliasAnalysis : public ImmutablePass, public AliasAnalysis { - static char ID; // Class identification, replacement for typeinfo - BasicAliasAnalysis() : ImmutablePass(ID) { - initializeBasicAliasAnalysisPass(*PassRegistry::getPassRegistry()); - } - - bool doInitialization(Module &M) override; - - void getAnalysisUsage(AnalysisUsage &AU) const override { - AU.addRequired(); - AU.addRequired(); - AU.addRequired(); - } - - AliasResult alias(const MemoryLocation &LocA, - const MemoryLocation &LocB) override { - assert(AliasCache.empty() && "AliasCache must be cleared after use!"); - assert(notDifferentParent(LocA.Ptr, LocB.Ptr) && - "BasicAliasAnalysis doesn't support interprocedural queries."); - AliasResult Alias = aliasCheck(LocA.Ptr, LocA.Size, LocA.AATags, - LocB.Ptr, LocB.Size, LocB.AATags); - // AliasCache rarely has more than 1 or 2 elements, always use - // shrink_and_clear so it quickly returns to the inline capacity of the - // SmallDenseMap if it ever grows larger. - // FIXME: This should really be shrink_to_inline_capacity_and_clear(). - AliasCache.shrink_and_clear(); - VisitedPhiBBs.clear(); - return Alias; - } - - ModRefResult getModRefInfo(ImmutableCallSite CS, - const MemoryLocation &Loc) override; - - ModRefResult getModRefInfo(ImmutableCallSite CS1, - ImmutableCallSite CS2) override; - - /// pointsToConstantMemory - Chase pointers until we find a (constant - /// global) or not. - bool pointsToConstantMemory(const MemoryLocation &Loc, - bool OrLocal) override; - - /// Get the location associated with a pointer argument of a callsite. - ModRefResult getArgModRefInfo(ImmutableCallSite CS, - unsigned ArgIdx) override; - - /// getModRefBehavior - Return the behavior when calling the given - /// call site. - ModRefBehavior getModRefBehavior(ImmutableCallSite CS) override; - - /// getModRefBehavior - Return the behavior when calling the given function. - /// For use when the call site is not known. - ModRefBehavior getModRefBehavior(const Function *F) override; - - /// getAdjustedAnalysisPointer - This method is used when a pass implements - /// an analysis interface through multiple inheritance. If needed, it - /// should override this to adjust the this pointer as needed for the - /// specified pass info. - void *getAdjustedAnalysisPointer(const void *ID) override { - if (ID == &AliasAnalysis::ID) - return (AliasAnalysis*)this; - return this; - } - - private: - // AliasCache - Track alias queries to guard against recursion. - typedef std::pair LocPair; - typedef SmallDenseMap AliasCacheTy; - AliasCacheTy AliasCache; - - /// \brief Track phi nodes we have visited. When interpret "Value" pointer - /// equality as value equality we need to make sure that the "Value" is not - /// part of a cycle. Otherwise, two uses could come from different - /// "iterations" of a cycle and see different values for the same "Value" - /// pointer. - /// The following example shows the problem: - /// %p = phi(%alloca1, %addr2) - /// %l = load %ptr - /// %addr1 = gep, %alloca2, 0, %l - /// %addr2 = gep %alloca2, 0, (%l + 1) - /// alias(%p, %addr1) -> MayAlias ! - /// store %l, ... - SmallPtrSet VisitedPhiBBs; - - // Visited - Track instructions visited by pointsToConstantMemory. - SmallPtrSet Visited; - - /// \brief Check whether two Values can be considered equivalent. - /// - /// In addition to pointer equivalence of \p V1 and \p V2 this checks - /// whether they can not be part of a cycle in the value graph by looking at - /// all visited phi nodes an making sure that the phis cannot reach the - /// value. We have to do this because we are looking through phi nodes (That - /// is we say noalias(V, phi(VA, VB)) if noalias(V, VA) and noalias(V, VB). - bool isValueEqualInPotentialCycles(const Value *V1, const Value *V2); - - /// \brief Dest and Src are the variable indices from two decomposed - /// GetElementPtr instructions GEP1 and GEP2 which have common base - /// pointers. Subtract the GEP2 indices from GEP1 to find the symbolic - /// difference between the two pointers. - void GetIndexDifference(SmallVectorImpl &Dest, - const SmallVectorImpl &Src); - - // aliasGEP - Provide a bunch of ad-hoc rules to disambiguate a GEP - // instruction against another. - AliasResult aliasGEP(const GEPOperator *V1, uint64_t V1Size, - const AAMDNodes &V1AAInfo, - const Value *V2, uint64_t V2Size, - const AAMDNodes &V2AAInfo, - const Value *UnderlyingV1, const Value *UnderlyingV2); - - // aliasPHI - Provide a bunch of ad-hoc rules to disambiguate a PHI - // instruction against another. - AliasResult aliasPHI(const PHINode *PN, uint64_t PNSize, - const AAMDNodes &PNAAInfo, - const Value *V2, uint64_t V2Size, - const AAMDNodes &V2AAInfo); - - /// aliasSelect - Disambiguate a Select instruction against another value. - AliasResult aliasSelect(const SelectInst *SI, uint64_t SISize, - const AAMDNodes &SIAAInfo, - const Value *V2, uint64_t V2Size, - const AAMDNodes &V2AAInfo); - - AliasResult aliasCheck(const Value *V1, uint64_t V1Size, - AAMDNodes V1AATag, - const Value *V2, uint64_t V2Size, - AAMDNodes V2AATag); - }; -} // End of anonymous namespace - -// Register this pass... +// Register the pass... char BasicAliasAnalysis::ID = 0; INITIALIZE_AG_PASS_BEGIN(BasicAliasAnalysis, AliasAnalysis, "basicaa", - "Basic Alias Analysis (stateless AA impl)", - false, true, false) + "Basic Alias Analysis (stateless AA impl)", false, + true, false) INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker) INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass) INITIALIZE_AG_PASS_END(BasicAliasAnalysis, AliasAnalysis, "basicaa", - "Basic Alias Analysis (stateless AA impl)", - false, true, false) - + "Basic Alias Analysis (stateless AA impl)", false, true, + false) ImmutablePass *llvm::createBasicAliasAnalysisPass() { return new BasicAliasAnalysis(); } -/// pointsToConstantMemory - Returns whether the given pointer value -/// points to memory that is local to the function, with global constants being -/// considered local to all functions. +/// Returns whether the given pointer value points to memory that is local to +/// the function, with global constants being considered local to all +/// functions. bool BasicAliasAnalysis::pointsToConstantMemory(const MemoryLocation &Loc, bool OrLocal) { assert(Visited.empty() && "Visited must be cleared after use!"); @@ -671,56 +565,55 @@ static bool isMemsetPattern16(const Function *MS, return false; } -/// getModRefBehavior - Return the behavior when calling the given call site. -AliasAnalysis::ModRefBehavior +/// Returns the behavior when calling the given call site. +FunctionModRefBehavior BasicAliasAnalysis::getModRefBehavior(ImmutableCallSite CS) { if (CS.doesNotAccessMemory()) // Can't do better than this. - return DoesNotAccessMemory; + return FMRB_DoesNotAccessMemory; - ModRefBehavior Min = UnknownModRefBehavior; + FunctionModRefBehavior Min = FMRB_UnknownModRefBehavior; // If the callsite knows it only reads memory, don't return worse // than that. if (CS.onlyReadsMemory()) - Min = OnlyReadsMemory; + Min = FMRB_OnlyReadsMemory; + + if (CS.onlyAccessesArgMemory()) + Min = FunctionModRefBehavior(Min & FMRB_OnlyAccessesArgumentPointees); // The AliasAnalysis base class has some smarts, lets use them. - return ModRefBehavior(AliasAnalysis::getModRefBehavior(CS) & Min); + return FunctionModRefBehavior(AliasAnalysis::getModRefBehavior(CS) & Min); } -/// getModRefBehavior - Return the behavior when calling the given function. -/// For use when the call site is not known. -AliasAnalysis::ModRefBehavior +/// Returns the behavior when calling the given function. For use when the call +/// site is not known. +FunctionModRefBehavior BasicAliasAnalysis::getModRefBehavior(const Function *F) { // If the function declares it doesn't access memory, we can't do better. if (F->doesNotAccessMemory()) - return DoesNotAccessMemory; - - // For intrinsics, we can check the table. - if (Intrinsic::ID iid = F->getIntrinsicID()) { -#define GET_INTRINSIC_MODREF_BEHAVIOR -#include "llvm/IR/Intrinsics.gen" -#undef GET_INTRINSIC_MODREF_BEHAVIOR - } + return FMRB_DoesNotAccessMemory; - ModRefBehavior Min = UnknownModRefBehavior; + FunctionModRefBehavior Min = FMRB_UnknownModRefBehavior; // If the function declares it only reads memory, go with that. if (F->onlyReadsMemory()) - Min = OnlyReadsMemory; + Min = FMRB_OnlyReadsMemory; + + if (F->onlyAccessesArgMemory()) + Min = FunctionModRefBehavior(Min & FMRB_OnlyAccessesArgumentPointees); const TargetLibraryInfo &TLI = getAnalysis().getTLI(); if (isMemsetPattern16(F, TLI)) - Min = OnlyAccessesArgumentPointees; + Min = FMRB_OnlyAccessesArgumentPointees; // Otherwise be conservative. - return ModRefBehavior(AliasAnalysis::getModRefBehavior(F) & Min); + return FunctionModRefBehavior(AliasAnalysis::getModRefBehavior(F) & Min); } -AliasAnalysis::ModRefResult -BasicAliasAnalysis::getArgModRefInfo(ImmutableCallSite CS, unsigned ArgIdx) { +ModRefInfo BasicAliasAnalysis::getArgModRefInfo(ImmutableCallSite CS, + unsigned ArgIdx) { if (const IntrinsicInst *II = dyn_cast(CS.getInstruction())) switch (II->getIntrinsicID()) { default: @@ -730,7 +623,7 @@ BasicAliasAnalysis::getArgModRefInfo(ImmutableCallSite CS, unsigned ArgIdx) { case Intrinsic::memmove: assert((ArgIdx == 0 || ArgIdx == 1) && "Invalid argument index for memory intrinsic"); - return ArgIdx ? Ref : Mod; + return ArgIdx ? MRI_Ref : MRI_Mod; } // We can bound the aliasing properties of memset_pattern16 just as we can @@ -741,7 +634,7 @@ BasicAliasAnalysis::getArgModRefInfo(ImmutableCallSite CS, unsigned ArgIdx) { isMemsetPattern16(CS.getCalledFunction(), *TLI)) { assert((ArgIdx == 0 || ArgIdx == 1) && "Invalid argument index for memset_pattern16"); - return ArgIdx ? Ref : Mod; + return ArgIdx ? MRI_Ref : MRI_Mod; } // FIXME: Handle memset_pattern4 and memset_pattern8 also. @@ -761,13 +654,14 @@ bool BasicAliasAnalysis::doInitialization(Module &M) { return true; } -/// getModRefInfo - Check to see if the specified callsite can clobber the -/// specified memory object. Since we only look at local properties of this -/// function, we really can't say much about this query. We do, however, use -/// simple "address taken" analysis on local objects. -AliasAnalysis::ModRefResult -BasicAliasAnalysis::getModRefInfo(ImmutableCallSite CS, - const MemoryLocation &Loc) { +/// Checks to see if the specified callsite can clobber the specified memory +/// object. +/// +/// Since we only look at local properties of this function, we really can't +/// say much about this query. We do, however, use simple "address taken" +/// analysis on local objects. +ModRefInfo BasicAliasAnalysis::getModRefInfo(ImmutableCallSite CS, + const MemoryLocation &Loc) { assert(notDifferentParent(CS.getInstruction(), Loc.Ptr) && "AliasAnalysis query involving multiple functions!"); @@ -781,7 +675,7 @@ BasicAliasAnalysis::getModRefInfo(ImmutableCallSite CS, if (isa(Object)) if (const CallInst *CI = dyn_cast(CS.getInstruction())) if (CI->isTailCall()) - return NoModRef; + return MRI_NoModRef; // If the pointer is to a locally allocated object that does not escape, // then the call can not mod/ref the pointer unless the call takes the pointer @@ -810,38 +704,38 @@ BasicAliasAnalysis::getModRefInfo(ImmutableCallSite CS, } if (!PassedAsArg) - return NoModRef; + return MRI_NoModRef; } // While the assume intrinsic is marked as arbitrarily writing so that // proper control dependencies will be maintained, it never aliases any // particular memory location. if (isAssumeIntrinsic(CS)) - return NoModRef; + return MRI_NoModRef; // The AliasAnalysis base class has some smarts, lets use them. return AliasAnalysis::getModRefInfo(CS, Loc); } -AliasAnalysis::ModRefResult -BasicAliasAnalysis::getModRefInfo(ImmutableCallSite CS1, - ImmutableCallSite CS2) { +ModRefInfo BasicAliasAnalysis::getModRefInfo(ImmutableCallSite CS1, + ImmutableCallSite CS2) { // While the assume intrinsic is marked as arbitrarily writing so that // proper control dependencies will be maintained, it never aliases any // particular memory location. if (isAssumeIntrinsic(CS1) || isAssumeIntrinsic(CS2)) - return NoModRef; + return MRI_NoModRef; // The AliasAnalysis base class has some smarts, lets use them. return AliasAnalysis::getModRefInfo(CS1, CS2); } -/// \brief Provide ad-hoc rules to disambiguate accesses through two GEP -/// operators, both having the exact same pointer operand. -static AliasAnalysis::AliasResult -aliasSameBasePointerGEPs(const GEPOperator *GEP1, uint64_t V1Size, - const GEPOperator *GEP2, uint64_t V2Size, - const DataLayout &DL) { +/// Provide ad-hoc rules to disambiguate accesses through two GEP operators, +/// both having the exact same pointer operand. +static AliasResult aliasSameBasePointerGEPs(const GEPOperator *GEP1, + uint64_t V1Size, + const GEPOperator *GEP2, + uint64_t V2Size, + const DataLayout &DL) { assert(GEP1->getPointerOperand() == GEP2->getPointerOperand() && "Expected GEPs with the same pointer operand"); @@ -851,13 +745,13 @@ aliasSameBasePointerGEPs(const GEPOperator *GEP1, uint64_t V1Size, // We also need at least two indices (the pointer, and the struct field). if (GEP1->getNumIndices() != GEP2->getNumIndices() || GEP1->getNumIndices() < 2) - return AliasAnalysis::MayAlias; + return MayAlias; // If we don't know the size of the accesses through both GEPs, we can't // determine whether the struct fields accessed can't alias. - if (V1Size == AliasAnalysis::UnknownSize || - V2Size == AliasAnalysis::UnknownSize) - return AliasAnalysis::MayAlias; + if (V1Size == MemoryLocation::UnknownSize || + V2Size == MemoryLocation::UnknownSize) + return MayAlias; ConstantInt *C1 = dyn_cast(GEP1->getOperand(GEP1->getNumOperands() - 1)); @@ -868,7 +762,7 @@ aliasSameBasePointerGEPs(const GEPOperator *GEP1, uint64_t V1Size, // If they're identical, the other indices might be also be dynamically // equal, so the GEPs can alias. if (!C1 || !C2 || C1 == C2) - return AliasAnalysis::MayAlias; + return MayAlias; // Find the last-indexed type of the GEP, i.e., the type you'd get if // you stripped the last index. @@ -886,7 +780,7 @@ aliasSameBasePointerGEPs(const GEPOperator *GEP1, uint64_t V1Size, for (unsigned i = 1, e = GEP1->getNumIndices() - 1; i != e; ++i) { if (!isa(GetElementPtrInst::getIndexedType( GEP1->getSourceElementType(), IntermediateIndices))) - return AliasAnalysis::MayAlias; + return MayAlias; IntermediateIndices.push_back(GEP1->getOperand(i + 1)); } @@ -895,7 +789,7 @@ aliasSameBasePointerGEPs(const GEPOperator *GEP1, uint64_t V1Size, GEP1->getSourceElementType(), IntermediateIndices)); if (!LastIndexedStruct) - return AliasAnalysis::MayAlias; + return MayAlias; // We know that: // - both GEPs begin indexing from the exact same pointer; @@ -924,23 +818,21 @@ aliasSameBasePointerGEPs(const GEPOperator *GEP1, uint64_t V1Size, if (EltsDontOverlap(V1Off, V1Size, V2Off, V2Size) || EltsDontOverlap(V2Off, V2Size, V1Off, V1Size)) - return AliasAnalysis::NoAlias; + return NoAlias; - return AliasAnalysis::MayAlias; + return MayAlias; } -/// aliasGEP - Provide a bunch of ad-hoc rules to disambiguate a GEP instruction -/// against another pointer. We know that V1 is a GEP, but we don't know -/// anything about V2. UnderlyingV1 is GetUnderlyingObject(GEP1, DL), -/// UnderlyingV2 is the same for V2. +/// Provides a bunch of ad-hoc rules to disambiguate a GEP instruction against +/// another pointer. /// -AliasAnalysis::AliasResult -BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, - const AAMDNodes &V1AAInfo, - const Value *V2, uint64_t V2Size, - const AAMDNodes &V2AAInfo, - const Value *UnderlyingV1, - const Value *UnderlyingV2) { +/// We know that V1 is a GEP, but we don't know anything about V2. +/// UnderlyingV1 is GetUnderlyingObject(GEP1, DL), UnderlyingV2 is the same for +/// V2. +AliasResult BasicAliasAnalysis::aliasGEP( + const GEPOperator *GEP1, uint64_t V1Size, const AAMDNodes &V1AAInfo, + const Value *V2, uint64_t V2Size, const AAMDNodes &V2AAInfo, + const Value *UnderlyingV1, const Value *UnderlyingV2) { int64_t GEP1BaseOffset; bool GEP1MaxLookupReached; SmallVector GEP1VariableIndices; @@ -970,16 +862,16 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // derived pointer. if (const GEPOperator *GEP2 = dyn_cast(V2)) { // Do the base pointers alias? - AliasResult BaseAlias = aliasCheck(UnderlyingV1, UnknownSize, AAMDNodes(), - UnderlyingV2, UnknownSize, AAMDNodes()); + AliasResult BaseAlias = + aliasCheck(UnderlyingV1, MemoryLocation::UnknownSize, AAMDNodes(), + UnderlyingV2, MemoryLocation::UnknownSize, AAMDNodes()); // Check for geps of non-aliasing underlying pointers where the offsets are // identical. if ((BaseAlias == MayAlias) && V1Size == V2Size) { // Do the base pointers alias assuming type and size. - AliasResult PreciseBaseAlias = aliasCheck(UnderlyingV1, V1Size, - V1AAInfo, UnderlyingV2, - V2Size, V2AAInfo); + AliasResult PreciseBaseAlias = aliasCheck(UnderlyingV1, V1Size, V1AAInfo, + UnderlyingV2, V2Size, V2AAInfo); if (PreciseBaseAlias == NoAlias) { // See if the computed offset from the common pointer tells us about the // relation of the resulting pointer. @@ -1013,7 +905,8 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // If we get a No or May, then return it immediately, no amount of analysis // will improve this situation. - if (BaseAlias != MustAlias) return BaseAlias; + if (BaseAlias != MustAlias) + return BaseAlias; // Otherwise, we have a MustAlias. Since the base pointers alias each other // exactly, see if the computed offset from the common pointer tells us @@ -1032,8 +925,7 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // DecomposeGEPExpression and GetUnderlyingObject should return the // same result except when DecomposeGEPExpression has no DataLayout. if (GEP1BasePtr != UnderlyingV1 || GEP2BasePtr != UnderlyingV2) { - assert(!DL && - "DecomposeGEPExpression and GetUnderlyingObject disagree!"); + assert(!DL && "DecomposeGEPExpression and GetUnderlyingObject disagree!"); return MayAlias; } @@ -1062,11 +954,12 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // pointer, we know they cannot alias. // If both accesses are unknown size, we can't do anything useful here. - if (V1Size == UnknownSize && V2Size == UnknownSize) + if (V1Size == MemoryLocation::UnknownSize && + V2Size == MemoryLocation::UnknownSize) return MayAlias; - AliasResult R = aliasCheck(UnderlyingV1, UnknownSize, AAMDNodes(), - V2, V2Size, V2AAInfo); + AliasResult R = aliasCheck(UnderlyingV1, MemoryLocation::UnknownSize, + AAMDNodes(), V2, V2Size, V2AAInfo); if (R != MustAlias) // If V2 may alias GEP base pointer, conservatively returns MayAlias. // If V2 is known not to alias GEP base pointer, then the two values @@ -1082,8 +975,7 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // DecomposeGEPExpression and GetUnderlyingObject should return the // same result except when DecomposeGEPExpression has no DataLayout. if (GEP1BasePtr != UnderlyingV1) { - assert(!DL && - "DecomposeGEPExpression and GetUnderlyingObject disagree!"); + assert(!DL && "DecomposeGEPExpression and GetUnderlyingObject disagree!"); return MayAlias; } // If the max search depth is reached the result is undefined @@ -1106,7 +998,7 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // greater, we know they do not overlap. if (GEP1BaseOffset != 0 && GEP1VariableIndices.empty()) { if (GEP1BaseOffset >= 0) { - if (V2Size != UnknownSize) { + if (V2Size != MemoryLocation::UnknownSize) { if ((uint64_t)GEP1BaseOffset < V2Size) return PartialAlias; return NoAlias; @@ -1120,7 +1012,8 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // GEP1 V2 // We need to know that V2Size is not unknown, otherwise we might have // stripped a gep with negative index ('gep , -1, ...). - if (V1Size != UnknownSize && V2Size != UnknownSize) { + if (V1Size != MemoryLocation::UnknownSize && + V2Size != MemoryLocation::UnknownSize) { if (-(uint64_t)GEP1BaseOffset < V1Size) return PartialAlias; return NoAlias; @@ -1137,7 +1030,7 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // Grab the least significant bit set in any of the scales. We // don't need std::abs here (even if the scale's negative) as we'll // be ^'ing Modulo with itself later. - Modulo |= (uint64_t) GEP1VariableIndices[i].Scale; + Modulo |= (uint64_t)GEP1VariableIndices[i].Scale; if (AllPositive) { // If the Value could change between cycles, then any reasoning about @@ -1151,7 +1044,7 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // Zero-extension widens the variable, and so forces the sign // bit to zero. - bool IsZExt = GEP1VariableIndices[i].Extension == EK_ZeroExt; + bool IsZExt = GEP1VariableIndices[i].ZExtBits > 0 || isa(V); SignKnownZero |= IsZExt; SignKnownOne &= !IsZExt; @@ -1160,8 +1053,7 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // unsigned. int64_t Scale = GEP1VariableIndices[i].Scale; AllPositive = - (SignKnownZero && Scale >= 0) || - (SignKnownOne && Scale < 0); + (SignKnownZero && Scale >= 0) || (SignKnownOne && Scale < 0); } } @@ -1171,14 +1063,19 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, // mod Modulo. Check whether that difference guarantees that the // two locations do not alias. uint64_t ModOffset = (uint64_t)GEP1BaseOffset & (Modulo - 1); - if (V1Size != UnknownSize && V2Size != UnknownSize && - ModOffset >= V2Size && V1Size <= Modulo - ModOffset) + if (V1Size != MemoryLocation::UnknownSize && + V2Size != MemoryLocation::UnknownSize && ModOffset >= V2Size && + V1Size <= Modulo - ModOffset) return NoAlias; // If we know all the variables are positive, then GEP1 >= GEP1BasePtr. // If GEP1BasePtr > V2 (GEP1BaseOffset > 0) then we know the pointers // don't alias if V2Size can fit in the gap between V2 and GEP1BasePtr. - if (AllPositive && GEP1BaseOffset > 0 && V2Size <= (uint64_t) GEP1BaseOffset) + if (AllPositive && GEP1BaseOffset > 0 && V2Size <= (uint64_t)GEP1BaseOffset) + return NoAlias; + + if (constantOffsetHeuristic(GEP1VariableIndices, V1Size, V2Size, + GEP1BaseOffset, DL, AC1, DT)) return NoAlias; } @@ -1192,60 +1089,57 @@ BasicAliasAnalysis::aliasGEP(const GEPOperator *GEP1, uint64_t V1Size, return PartialAlias; } -static AliasAnalysis::AliasResult -MergeAliasResults(AliasAnalysis::AliasResult A, AliasAnalysis::AliasResult B) { +static AliasResult MergeAliasResults(AliasResult A, AliasResult B) { // If the results agree, take it. if (A == B) return A; // A mix of PartialAlias and MustAlias is PartialAlias. - if ((A == AliasAnalysis::PartialAlias && B == AliasAnalysis::MustAlias) || - (B == AliasAnalysis::PartialAlias && A == AliasAnalysis::MustAlias)) - return AliasAnalysis::PartialAlias; + if ((A == PartialAlias && B == MustAlias) || + (B == PartialAlias && A == MustAlias)) + return PartialAlias; // Otherwise, we don't know anything. - return AliasAnalysis::MayAlias; + return MayAlias; } -/// aliasSelect - Provide a bunch of ad-hoc rules to disambiguate a Select -/// instruction against another. -AliasAnalysis::AliasResult -BasicAliasAnalysis::aliasSelect(const SelectInst *SI, uint64_t SISize, - const AAMDNodes &SIAAInfo, - const Value *V2, uint64_t V2Size, - const AAMDNodes &V2AAInfo) { +/// Provides a bunch of ad-hoc rules to disambiguate a Select instruction +/// against another. +AliasResult BasicAliasAnalysis::aliasSelect(const SelectInst *SI, + uint64_t SISize, + const AAMDNodes &SIAAInfo, + const Value *V2, uint64_t V2Size, + const AAMDNodes &V2AAInfo) { // If the values are Selects with the same condition, we can do a more precise // check: just check for aliases between the values on corresponding arms. if (const SelectInst *SI2 = dyn_cast(V2)) if (SI->getCondition() == SI2->getCondition()) { - AliasResult Alias = - aliasCheck(SI->getTrueValue(), SISize, SIAAInfo, - SI2->getTrueValue(), V2Size, V2AAInfo); + AliasResult Alias = aliasCheck(SI->getTrueValue(), SISize, SIAAInfo, + SI2->getTrueValue(), V2Size, V2AAInfo); if (Alias == MayAlias) return MayAlias; AliasResult ThisAlias = - aliasCheck(SI->getFalseValue(), SISize, SIAAInfo, - SI2->getFalseValue(), V2Size, V2AAInfo); + aliasCheck(SI->getFalseValue(), SISize, SIAAInfo, + SI2->getFalseValue(), V2Size, V2AAInfo); return MergeAliasResults(ThisAlias, Alias); } // If both arms of the Select node NoAlias or MustAlias V2, then returns // NoAlias / MustAlias. Otherwise, returns MayAlias. AliasResult Alias = - aliasCheck(V2, V2Size, V2AAInfo, SI->getTrueValue(), SISize, SIAAInfo); + aliasCheck(V2, V2Size, V2AAInfo, SI->getTrueValue(), SISize, SIAAInfo); if (Alias == MayAlias) return MayAlias; AliasResult ThisAlias = - aliasCheck(V2, V2Size, V2AAInfo, SI->getFalseValue(), SISize, SIAAInfo); + aliasCheck(V2, V2Size, V2AAInfo, SI->getFalseValue(), SISize, SIAAInfo); return MergeAliasResults(ThisAlias, Alias); } -// aliasPHI - Provide a bunch of ad-hoc rules to disambiguate a PHI instruction -// against another. -AliasAnalysis::AliasResult -BasicAliasAnalysis::aliasPHI(const PHINode *PN, uint64_t PNSize, - const AAMDNodes &PNAAInfo, - const Value *V2, uint64_t V2Size, - const AAMDNodes &V2AAInfo) { +/// Provide a bunch of ad-hoc rules to disambiguate a PHI instruction against +/// another. +AliasResult BasicAliasAnalysis::aliasPHI(const PHINode *PN, uint64_t PNSize, + const AAMDNodes &PNAAInfo, + const Value *V2, uint64_t V2Size, + const AAMDNodes &V2AAInfo) { // Track phi nodes we have visited. We use this information when we determine // value equivalence. VisitedPhiBBs.insert(PN->getParent()); @@ -1274,9 +1168,9 @@ BasicAliasAnalysis::aliasPHI(const PHINode *PN, uint64_t PNSize, for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) { AliasResult ThisAlias = - aliasCheck(PN->getIncomingValue(i), PNSize, PNAAInfo, - PN2->getIncomingValueForBlock(PN->getIncomingBlock(i)), - V2Size, V2AAInfo); + aliasCheck(PN->getIncomingValue(i), PNSize, PNAAInfo, + PN2->getIncomingValueForBlock(PN->getIncomingBlock(i)), + V2Size, V2AAInfo); Alias = MergeAliasResults(ThisAlias, Alias); if (Alias == MayAlias) break; @@ -1289,8 +1183,9 @@ BasicAliasAnalysis::aliasPHI(const PHINode *PN, uint64_t PNSize, return Alias; } - SmallPtrSet UniqueSrc; - SmallVector V1Srcs; + SmallPtrSet UniqueSrc; + SmallVector V1Srcs; + bool isRecursive = false; for (Value *PV1 : PN->incoming_values()) { if (isa(PV1)) // If any of the source itself is a PHI, return MayAlias conservatively @@ -1298,12 +1193,33 @@ BasicAliasAnalysis::aliasPHI(const PHINode *PN, uint64_t PNSize, // sides are PHI nodes. In which case, this is O(m x n) time where 'm' // and 'n' are the number of PHI sources. return MayAlias; + + if (EnableRecPhiAnalysis) + if (GEPOperator *PV1GEP = dyn_cast(PV1)) { + // Check whether the incoming value is a GEP that advances the pointer + // result of this PHI node (e.g. in a loop). If this is the case, we + // would recurse and always get a MayAlias. Handle this case specially + // below. + if (PV1GEP->getPointerOperand() == PN && PV1GEP->getNumIndices() == 1 && + isa(PV1GEP->idx_begin())) { + isRecursive = true; + continue; + } + } + if (UniqueSrc.insert(PV1).second) V1Srcs.push_back(PV1); } - AliasResult Alias = aliasCheck(V2, V2Size, V2AAInfo, - V1Srcs[0], PNSize, PNAAInfo); + // If this PHI node is recursive, set the size of the accessed memory to + // unknown to represent all the possible values the GEP could advance the + // pointer to. + if (isRecursive) + PNSize = MemoryLocation::UnknownSize; + + AliasResult Alias = + aliasCheck(V2, V2Size, V2AAInfo, V1Srcs[0], PNSize, PNAAInfo); + // Early exit if the check of the first PHI source against V2 is MayAlias. // Other results are not possible. if (Alias == MayAlias) @@ -1314,8 +1230,8 @@ BasicAliasAnalysis::aliasPHI(const PHINode *PN, uint64_t PNSize, for (unsigned i = 1, e = V1Srcs.size(); i != e; ++i) { Value *V = V1Srcs[i]; - AliasResult ThisAlias = aliasCheck(V2, V2Size, V2AAInfo, - V, PNSize, PNAAInfo); + AliasResult ThisAlias = + aliasCheck(V2, V2Size, V2AAInfo, V, PNSize, PNAAInfo); Alias = MergeAliasResults(ThisAlias, Alias); if (Alias == MayAlias) break; @@ -1324,14 +1240,12 @@ BasicAliasAnalysis::aliasPHI(const PHINode *PN, uint64_t PNSize, return Alias; } -// aliasCheck - Provide a bunch of ad-hoc rules to disambiguate in common cases, -// such as array references. -// -AliasAnalysis::AliasResult -BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size, - AAMDNodes V1AAInfo, - const Value *V2, uint64_t V2Size, - AAMDNodes V2AAInfo) { +/// Provideis a bunch of ad-hoc rules to disambiguate in common cases, such as +/// array references. +AliasResult BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size, + AAMDNodes V1AAInfo, const Value *V2, + uint64_t V2Size, + AAMDNodes V2AAInfo) { // If either of the memory references is empty, it doesn't matter what the // pointer values are. if (V1Size == 0 || V2Size == 0) @@ -1356,7 +1270,7 @@ BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size, return MustAlias; if (!V1->getType()->isPointerTy() || !V2->getType()->isPointerTy()) - return NoAlias; // Scalars cannot alias each other + return NoAlias; // Scalars cannot alias each other // Figure out what objects these things are pointing to if we can. const Value *O1 = GetUnderlyingObject(V1, *DL, MaxLookupSearchDepth); @@ -1410,8 +1324,10 @@ BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size, // If the size of one access is larger than the entire object on the other // side, then we know such behavior is undefined and can assume no alias. if (DL) - if ((V1Size != UnknownSize && isObjectSmallerThan(O2, V1Size, *DL, *TLI)) || - (V2Size != UnknownSize && isObjectSmallerThan(O1, V2Size, *DL, *TLI))) + if ((V1Size != MemoryLocation::UnknownSize && + isObjectSmallerThan(O2, V1Size, *DL, *TLI)) || + (V2Size != MemoryLocation::UnknownSize && + isObjectSmallerThan(O1, V2Size, *DL, *TLI))) return NoAlias; // Check the cache before climbing up use-def chains. This also terminates @@ -1421,7 +1337,7 @@ BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size, if (V1 > V2) std::swap(Locs.first, Locs.second); std::pair Pair = - AliasCache.insert(std::make_pair(Locs, MayAlias)); + AliasCache.insert(std::make_pair(Locs, MayAlias)); if (!Pair.second) return Pair.first->second; @@ -1434,8 +1350,10 @@ BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size, std::swap(V1AAInfo, V2AAInfo); } if (const GEPOperator *GV1 = dyn_cast(V1)) { - AliasResult Result = aliasGEP(GV1, V1Size, V1AAInfo, V2, V2Size, V2AAInfo, O1, O2); - if (Result != MayAlias) return AliasCache[Locs] = Result; + AliasResult Result = + aliasGEP(GV1, V1Size, V1AAInfo, V2, V2Size, V2AAInfo, O1, O2); + if (Result != MayAlias) + return AliasCache[Locs] = Result; } if (isa(V2) && !isa(V1)) { @@ -1444,9 +1362,9 @@ BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size, std::swap(V1AAInfo, V2AAInfo); } if (const PHINode *PN = dyn_cast(V1)) { - AliasResult Result = aliasPHI(PN, V1Size, V1AAInfo, - V2, V2Size, V2AAInfo); - if (Result != MayAlias) return AliasCache[Locs] = Result; + AliasResult Result = aliasPHI(PN, V1Size, V1AAInfo, V2, V2Size, V2AAInfo); + if (Result != MayAlias) + return AliasCache[Locs] = Result; } if (isa(V2) && !isa(V1)) { @@ -1455,17 +1373,20 @@ BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size, std::swap(V1AAInfo, V2AAInfo); } if (const SelectInst *S1 = dyn_cast(V1)) { - AliasResult Result = aliasSelect(S1, V1Size, V1AAInfo, - V2, V2Size, V2AAInfo); - if (Result != MayAlias) return AliasCache[Locs] = Result; + AliasResult Result = + aliasSelect(S1, V1Size, V1AAInfo, V2, V2Size, V2AAInfo); + if (Result != MayAlias) + return AliasCache[Locs] = Result; } // If both pointers are pointing into the same object and one of them // accesses is accessing the entire object, then the accesses must // overlap in some way. if (DL && O1 == O2) - if ((V1Size != UnknownSize && isObjectSize(O1, V1Size, *DL, *TLI)) || - (V2Size != UnknownSize && isObjectSize(O2, V2Size, *DL, *TLI))) + if ((V1Size != MemoryLocation::UnknownSize && + isObjectSize(O1, V1Size, *DL, *TLI)) || + (V2Size != MemoryLocation::UnknownSize && + isObjectSize(O2, V2Size, *DL, *TLI))) return AliasCache[Locs] = PartialAlias; AliasResult Result = @@ -1474,6 +1395,13 @@ BasicAliasAnalysis::aliasCheck(const Value *V1, uint64_t V1Size, return AliasCache[Locs] = Result; } +/// Check whether two Values can be considered equivalent. +/// +/// In addition to pointer equivalence of \p V1 and \p V2 this checks whether +/// they can not be part of a cycle in the value graph by looking at all +/// visited phi nodes an making sure that the phis cannot reach the value. We +/// have to do this because we are looking through phi nodes (That is we say +/// noalias(V, phi(VA, VB)) if noalias(V, VA) and noalias(V, VB). bool BasicAliasAnalysis::isValueEqualInPotentialCycles(const Value *V, const Value *V2) { if (V != V2) @@ -1506,10 +1434,10 @@ bool BasicAliasAnalysis::isValueEqualInPotentialCycles(const Value *V, return true; } -/// GetIndexDifference - Dest and Src are the variable indices from two -/// decomposed GetElementPtr instructions GEP1 and GEP2 which have common base -/// pointers. Subtract the GEP2 indices from GEP1 to find the symbolic -/// difference between the two pointers. +/// Computes the symbolic difference between two de-composed GEPs. +/// +/// Dest and Src are the variable indices from two decomposed GetElementPtr +/// instructions GEP1 and GEP2 which have common base pointers. void BasicAliasAnalysis::GetIndexDifference( SmallVectorImpl &Dest, const SmallVectorImpl &Src) { @@ -1518,14 +1446,14 @@ void BasicAliasAnalysis::GetIndexDifference( for (unsigned i = 0, e = Src.size(); i != e; ++i) { const Value *V = Src[i].V; - ExtensionKind Extension = Src[i].Extension; + unsigned ZExtBits = Src[i].ZExtBits, SExtBits = Src[i].SExtBits; int64_t Scale = Src[i].Scale; // Find V in Dest. This is N^2, but pointer indices almost never have more // than a few variable indexes. for (unsigned j = 0, e = Dest.size(); j != e; ++j) { if (!isValueEqualInPotentialCycles(Dest[j].V, V) || - Dest[j].Extension != Extension) + Dest[j].ZExtBits != ZExtBits || Dest[j].SExtBits != SExtBits) continue; // If we found it, subtract off Scale V's from the entry in Dest. If it @@ -1540,8 +1468,62 @@ void BasicAliasAnalysis::GetIndexDifference( // If we didn't consume this entry, add it to the end of the Dest list. if (Scale) { - VariableGEPIndex Entry = { V, Extension, -Scale }; + VariableGEPIndex Entry = {V, ZExtBits, SExtBits, -Scale}; Dest.push_back(Entry); } } } + +bool BasicAliasAnalysis::constantOffsetHeuristic( + const SmallVectorImpl &VarIndices, uint64_t V1Size, + uint64_t V2Size, int64_t BaseOffset, const DataLayout *DL, + AssumptionCache *AC, DominatorTree *DT) { + if (VarIndices.size() != 2 || V1Size == MemoryLocation::UnknownSize || + V2Size == MemoryLocation::UnknownSize || !DL) + return false; + + const VariableGEPIndex &Var0 = VarIndices[0], &Var1 = VarIndices[1]; + + if (Var0.ZExtBits != Var1.ZExtBits || Var0.SExtBits != Var1.SExtBits || + Var0.Scale != -Var1.Scale) + return false; + + unsigned Width = Var1.V->getType()->getIntegerBitWidth(); + + // We'll strip off the Extensions of Var0 and Var1 and do another round + // of GetLinearExpression decomposition. In the example above, if Var0 + // is zext(%x + 1) we should get V1 == %x and V1Offset == 1. + + APInt V0Scale(Width, 0), V0Offset(Width, 0), V1Scale(Width, 0), + V1Offset(Width, 0); + bool NSW = true, NUW = true; + unsigned V0ZExtBits = 0, V0SExtBits = 0, V1ZExtBits = 0, V1SExtBits = 0; + const Value *V0 = GetLinearExpression(Var0.V, V0Scale, V0Offset, V0ZExtBits, + V0SExtBits, *DL, 0, AC, DT, NSW, NUW); + NSW = true, NUW = true; + const Value *V1 = GetLinearExpression(Var1.V, V1Scale, V1Offset, V1ZExtBits, + V1SExtBits, *DL, 0, AC, DT, NSW, NUW); + + if (V0Scale != V1Scale || V0ZExtBits != V1ZExtBits || + V0SExtBits != V1SExtBits || !isValueEqualInPotentialCycles(V0, V1)) + return false; + + // We have a hit - Var0 and Var1 only differ by a constant offset! + + // If we've been sext'ed then zext'd the maximum difference between Var0 and + // Var1 is possible to calculate, but we're just interested in the absolute + // minumum difference between the two. The minimum distance may occur due to + // wrapping; consider "add i3 %i, 5": if %i == 7 then 7 + 5 mod 8 == 4, and so + // the minimum distance between %i and %i + 5 is 3. + APInt MinDiff = V0Offset - V1Offset, + Wrapped = APInt::getMaxValue(Width) - MinDiff + APInt(Width, 1); + MinDiff = APIntOps::umin(MinDiff, Wrapped); + uint64_t MinDiffBytes = MinDiff.getZExtValue() * std::abs(Var0.Scale); + + // We can't definitely say whether GEP1 is before or after V2 due to wrapping + // arithmetic (i.e. for some values of GEP1 and V2 GEP1 < V2, and for other + // values GEP1 > V2). We'll therefore only declare NoAlias if both V1Size and + // V2Size can fit in the MinDiffBytes gap. + return V1Size + std::abs(BaseOffset) <= MinDiffBytes && + V2Size + std::abs(BaseOffset) <= MinDiffBytes; +}