#include <algorithm>
using namespace llvm;
+/// Enable analysis of recursive PHI nodes.
+static cl::opt<bool> EnableRecPhiAnalysis("basicaa-recphi",
+ cl::Hidden, cl::init(false));
+
/// 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
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
// reads a bit past the end given sufficient alignment.
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
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;
}
//===----------------------------------------------------------------------===//
//===----------------------------------------------------------------------===//
namespace {
+ enum ExtensionKind {
+ EK_NotExtended,
+ EK_SignExt,
+ EK_ZeroExt
+ };
-// A linear transformation of a Value; this class represents ZExt(SExt(V,
-// SExtBits), ZExtBits) * Scale + Offset.
struct VariableGEPIndex {
-
- // An opaque Value - we can't decompose this further.
const Value *V;
-
- // We need to track what extensions we've done as we consider the same Value
- // with different extensions as different variables in a GEP's linear
- // expression;
- // e.g.: if V == -1, then sext(x) != zext(x).
- unsigned ZExtBits;
- unsigned SExtBits;
-
+ ExtensionKind Extension;
int64_t Scale;
bool operator==(const VariableGEPIndex &Other) const {
- return V == Other.V && ZExtBits == Other.ZExtBits &&
- SExtBits == Other.SExtBits && Scale == Other.Scale;
+ return V == Other.V && Extension == Other.Extension &&
+ Scale == Other.Scale;
}
bool operator!=(const VariableGEPIndex &Other) const {
///
/// Note that this looks through extends, so the high bits may not be
/// represented in the result.
-static const Value *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) {
+static Value *GetLinearExpression(Value *V, APInt &Scale, APInt &Offset,
+ ExtensionKind &Extension,
+ const DataLayout &DL, unsigned Depth,
+ AssumptionCache *AC, DominatorTree *DT) {
assert(V->getType()->isIntegerTy() && "Not an integer value");
// Limit our recursion depth.
return V;
}
- if (const ConstantInt *Const = dyn_cast<ConstantInt>(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<SExtInst> &
- // isa<ZExtInst> cases below.
- Offset += Const->getValue().zextOrSelf(Offset.getBitWidth());
+ if (ConstantInt *Const = dyn_cast<ConstantInt>(V)) {
+ // if it's a constant, just convert it to an offset
+ // and remove the variable.
+ Offset += Const->getValue();
assert(Scale == 0 && "Constant values don't have a scale");
return V;
}
- if (const BinaryOperator *BOp = dyn_cast<BinaryOperator>(V)) {
+ if (BinaryOperator *BOp = dyn_cast<BinaryOperator>(V)) {
if (ConstantInt *RHSC = dyn_cast<ConstantInt>(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<SExtInst> / isa<ZExtInst> cases).
- APInt RHS = RHSC->getValue().zextOrSelf(Offset.getBitWidth());
-
switch (BOp->getOpcode()) {
- default:
- // We don't understand this instruction, so we can't decompose it any
- // further.
- Scale = 1;
- Offset = 0;
- return V;
+ default: break;
case Instruction::Or:
// X|C == X+C if all the bits in C are unset in X. Otherwise we can't
// analyze it.
break;
// FALL THROUGH.
case Instruction::Add:
- 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;
+ V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, Extension,
+ DL, Depth + 1, AC, DT);
+ Offset += RHSC->getValue();
+ return V;
case Instruction::Mul:
- V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, ZExtBits,
- SExtBits, DL, Depth + 1, AC, DT, NSW, NUW);
- Offset *= RHS;
- Scale *= RHS;
- break;
+ V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, Extension,
+ DL, Depth + 1, AC, DT);
+ Offset *= RHSC->getValue();
+ Scale *= RHSC->getValue();
+ return V;
case Instruction::Shl:
- 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;
+ V = GetLinearExpression(BOp->getOperand(0), Scale, Offset, Extension,
+ DL, Depth + 1, AC, DT);
+ Offset <<= RHSC->getValue().getLimitedValue();
+ Scale <<= RHSC->getValue().getLimitedValue();
return V;
}
-
- if (isa<OverflowingBinaryOperator>(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<SExtInst>(V) || isa<ZExtInst>(V)) {
+ if ((isa<SExtInst>(V) && Extension != EK_ZeroExt) ||
+ (isa<ZExtInst>(V) && Extension != EK_SignExt)) {
Value *CastOp = cast<CastInst>(V)->getOperand(0);
- unsigned NewWidth = V->getType()->getPrimitiveSizeInBits();
+ unsigned OldWidth = Scale.getBitWidth();
unsigned SmallWidth = CastOp->getType()->getPrimitiveSizeInBits();
- 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<SExtInst>(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;
- }
+ Scale = Scale.trunc(SmallWidth);
+ Offset = Offset.trunc(SmallWidth);
+ Extension = isa<SExtInst>(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);
return Result;
}
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) {
- const Value *Index = *I;
+ Value *Index = *I;
// Compute the (potentially symbolic) offset in bytes for this index.
if (StructType *STy = dyn_cast<StructType>(*GTI++)) {
// For a struct, add the member offset.
}
// For an array/pointer, add the element offset, explicitly scaled.
- if (const ConstantInt *CIdx = dyn_cast<ConstantInt>(Index)) {
+ if (ConstantInt *CIdx = dyn_cast<ConstantInt>(Index)) {
if (CIdx->isZero()) continue;
BaseOffs += DL.getTypeAllocSize(*GTI) * CIdx->getSExtValue();
continue;
}
uint64_t Scale = DL.getTypeAllocSize(*GTI);
- unsigned ZExtBits = 0, SExtBits = 0;
+ ExtensionKind Extension = EK_NotExtended;
// If the integer type is smaller than the pointer size, it is implicitly
// sign extended to pointer size.
unsigned Width = Index->getType()->getIntegerBitWidth();
- unsigned PointerSize = DL.getPointerSizeInBits(AS);
- if (PointerSize > Width)
- SExtBits += PointerSize - Width;
+ if (DL.getPointerSizeInBits(AS) > Width)
+ Extension = EK_SignExt;
// Use GetLinearExpression to decompose the index into a C1*V+C2 form.
APInt IndexScale(Width, 0), IndexOffset(Width, 0);
- bool NSW = true, NUW = true;
- Index = GetLinearExpression(Index, IndexScale, IndexOffset, ZExtBits,
- SExtBits, DL, 0, AC, DT, NSW, NUW);
+ Index = GetLinearExpression(Index, IndexScale, IndexOffset, Extension, DL,
+ 0, AC, DT);
// 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.
// 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].ZExtBits == ZExtBits &&
- VarIndices[i].SExtBits == SExtBits) {
+ if (VarIndices[i].V == Index &&
+ VarIndices[i].Extension == Extension) {
Scale += VarIndices[i].Scale;
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 - PointerSize) {
+ if (unsigned ShiftBits = 64 - DL.getPointerSizeInBits(AS)) {
Scale <<= ShiftBits;
Scale = (int64_t)Scale >> ShiftBits;
}
if (Scale) {
- VariableGEPIndex Entry = {Index, ZExtBits, SExtBits,
+ VariableGEPIndex Entry = {Index, Extension,
static_cast<int64_t>(Scale)};
VarIndices.push_back(Entry);
}
AU.addRequired<TargetLibraryInfoWrapperPass>();
}
- AliasResult alias(const Location &LocA, const Location &LocB) override {
+ 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.");
return Alias;
}
- ModRefResult getModRefInfo(ImmutableCallSite CS,
- const Location &Loc) override;
+ ModRefInfo getModRefInfo(ImmutableCallSite CS,
+ const MemoryLocation &Loc) override;
- ModRefResult getModRefInfo(ImmutableCallSite CS1,
- ImmutableCallSite CS2) override;
+ ModRefInfo getModRefInfo(ImmutableCallSite CS1,
+ ImmutableCallSite CS2) override;
/// pointsToConstantMemory - Chase pointers until we find a (constant
/// global) or not.
- bool pointsToConstantMemory(const Location &Loc, bool OrLocal) override;
+ bool pointsToConstantMemory(const MemoryLocation &Loc,
+ bool OrLocal) override;
/// Get the location associated with a pointer argument of a callsite.
- Location getArgLocation(ImmutableCallSite CS, unsigned ArgIdx,
- ModRefResult &Mask) override;
+ ModRefInfo getArgModRefInfo(ImmutableCallSite CS, unsigned ArgIdx) override;
/// getModRefBehavior - Return the behavior when calling the given
/// call site.
- ModRefBehavior getModRefBehavior(ImmutableCallSite CS) override;
+ FunctionModRefBehavior 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;
+ FunctionModRefBehavior getModRefBehavior(const Function *F) override;
/// getAdjustedAnalysisPointer - This method is used when a pass implements
/// an analysis interface through multiple inheritance. If needed, it
private:
// AliasCache - Track alias queries to guard against recursion.
- typedef std::pair<Location, Location> LocPair;
+ typedef std::pair<MemoryLocation, MemoryLocation> LocPair;
typedef SmallDenseMap<LocPair, AliasResult, 8> AliasCacheTy;
AliasCacheTy AliasCache;
/// 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 A Heuristic for aliasGEP that searches for a constant offset
- /// between the variables.
- ///
- /// GetLinearExpression has some limitations, as generally zext(%x + 1)
- /// != zext(%x) + zext(1) if the arithmetic overflows. GetLinearExpression
- /// will therefore conservatively refuse to decompose these expressions.
- /// However, we know that, for all %x, zext(%x) != zext(%x + 1), even if
- /// the addition overflows.
- bool
- constantOffsetHeuristic(const SmallVectorImpl<VariableGEPIndex> &VarIndices,
- uint64_t V1Size, uint64_t V2Size,
- int64_t BaseOffset, const DataLayout *DL,
- AssumptionCache *AC, DominatorTree *DT);
-
/// \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
/// 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.
-bool
-BasicAliasAnalysis::pointsToConstantMemory(const Location &Loc, bool OrLocal) {
+bool BasicAliasAnalysis::pointsToConstantMemory(const MemoryLocation &Loc,
+ bool OrLocal) {
assert(Visited.empty() && "Visited must be cleared after use!");
unsigned MaxLookup = 8;
Visited.clear();
return AliasAnalysis::pointsToConstantMemory(Loc, OrLocal);
}
- for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
- Worklist.push_back(PN->getIncomingValue(i));
+ for (Value *IncValue : PN->incoming_values())
+ Worklist.push_back(IncValue);
continue;
}
return Worklist.empty();
}
+// FIXME: This code is duplicated with MemoryLocation and should be hoisted to
+// some common utility location.
static bool isMemsetPattern16(const Function *MS,
const TargetLibraryInfo &TLI) {
if (TLI.has(LibFunc::memset_pattern16) &&
}
/// getModRefBehavior - Return the behavior when calling the given call site.
-AliasAnalysis::ModRefBehavior
+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
+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;
+ return FMRB_DoesNotAccessMemory;
// For intrinsics, we can check the table.
- if (unsigned iid = F->getIntrinsicID()) {
+ if (Intrinsic::ID iid = F->getIntrinsicID()) {
#define GET_INTRINSIC_MODREF_BEHAVIOR
#include "llvm/IR/Intrinsics.gen"
#undef GET_INTRINSIC_MODREF_BEHAVIOR
}
- 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<TargetLibraryInfoWrapperPass>().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::Location
-BasicAliasAnalysis::getArgLocation(ImmutableCallSite CS, unsigned ArgIdx,
- ModRefResult &Mask) {
- Location Loc = AliasAnalysis::getArgLocation(CS, ArgIdx, Mask);
- const TargetLibraryInfo &TLI =
- getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
- const IntrinsicInst *II = dyn_cast<IntrinsicInst>(CS.getInstruction());
- if (II != nullptr)
+ModRefInfo BasicAliasAnalysis::getArgModRefInfo(ImmutableCallSite CS,
+ unsigned ArgIdx) {
+ if (const IntrinsicInst *II = dyn_cast<IntrinsicInst>(CS.getInstruction()))
switch (II->getIntrinsicID()) {
- default: break;
+ default:
+ break;
case Intrinsic::memset:
case Intrinsic::memcpy:
- case Intrinsic::memmove: {
+ case Intrinsic::memmove:
assert((ArgIdx == 0 || ArgIdx == 1) &&
"Invalid argument index for memory intrinsic");
- if (ConstantInt *LenCI = dyn_cast<ConstantInt>(II->getArgOperand(2)))
- Loc.Size = LenCI->getZExtValue();
- assert(Loc.Ptr == II->getArgOperand(ArgIdx) &&
- "Memory intrinsic location pointer not argument?");
- Mask = ArgIdx ? Ref : Mod;
- break;
- }
- case Intrinsic::lifetime_start:
- case Intrinsic::lifetime_end:
- case Intrinsic::invariant_start: {
- assert(ArgIdx == 1 && "Invalid argument index");
- assert(Loc.Ptr == II->getArgOperand(ArgIdx) &&
- "Intrinsic location pointer not argument?");
- Loc.Size = cast<ConstantInt>(II->getArgOperand(0))->getZExtValue();
- break;
- }
- case Intrinsic::invariant_end: {
- assert(ArgIdx == 2 && "Invalid argument index");
- assert(Loc.Ptr == II->getArgOperand(ArgIdx) &&
- "Intrinsic location pointer not argument?");
- Loc.Size = cast<ConstantInt>(II->getArgOperand(1))->getZExtValue();
- break;
- }
- case Intrinsic::arm_neon_vld1: {
- assert(ArgIdx == 0 && "Invalid argument index");
- assert(Loc.Ptr == II->getArgOperand(ArgIdx) &&
- "Intrinsic location pointer not argument?");
- // LLVM's vld1 and vst1 intrinsics currently only support a single
- // vector register.
- if (DL)
- Loc.Size = DL->getTypeStoreSize(II->getType());
- break;
- }
- case Intrinsic::arm_neon_vst1: {
- assert(ArgIdx == 0 && "Invalid argument index");
- assert(Loc.Ptr == II->getArgOperand(ArgIdx) &&
- "Intrinsic location pointer not argument?");
- if (DL)
- Loc.Size = DL->getTypeStoreSize(II->getArgOperand(1)->getType());
- break;
- }
+ return ArgIdx ? MRI_Ref : MRI_Mod;
}
// We can bound the aliasing properties of memset_pattern16 just as we can
// for memcpy/memset. This is particularly important because the
// LoopIdiomRecognizer likes to turn loops into calls to memset_pattern16
// whenever possible.
- else if (CS.getCalledFunction() &&
- isMemsetPattern16(CS.getCalledFunction(), TLI)) {
+ if (CS.getCalledFunction() &&
+ isMemsetPattern16(CS.getCalledFunction(), *TLI)) {
assert((ArgIdx == 0 || ArgIdx == 1) &&
"Invalid argument index for memset_pattern16");
- if (ArgIdx == 1)
- Loc.Size = 16;
- else if (const ConstantInt *LenCI =
- dyn_cast<ConstantInt>(CS.getArgument(2)))
- Loc.Size = LenCI->getZExtValue();
- assert(Loc.Ptr == CS.getArgument(ArgIdx) &&
- "memset_pattern16 location pointer not argument?");
- Mask = ArgIdx ? Ref : Mod;
+ return ArgIdx ? MRI_Ref : MRI_Mod;
}
// FIXME: Handle memset_pattern4 and memset_pattern8 also.
- return Loc;
+ return AliasAnalysis::getArgModRefInfo(CS, ArgIdx);
}
static bool isAssumeIntrinsic(ImmutableCallSite CS) {
/// 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 Location &Loc) {
+ModRefInfo BasicAliasAnalysis::getModRefInfo(ImmutableCallSite CS,
+ const MemoryLocation &Loc) {
assert(notDifferentParent(CS.getInstruction(), Loc.Ptr) &&
"AliasAnalysis query involving multiple functions!");
if (isa<AllocaInst>(Object))
if (const CallInst *CI = dyn_cast<CallInst>(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
// is impossible to alias the pointer we're checking. If not, we have to
// assume that the call could touch the pointer, even though it doesn't
// escape.
- if (!isNoAlias(Location(*CI), Location(Object))) {
+ if (!isNoAlias(MemoryLocation(*CI), MemoryLocation(Object))) {
PassedAsArg = true;
break;
}
}
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) {
+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");
// 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<ConstantInt>(GEP1->getOperand(GEP1->getNumOperands() - 1));
// 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.
for (unsigned i = 1, e = GEP1->getNumIndices() - 1; i != e; ++i) {
if (!isa<ArrayType>(GetElementPtrInst::getIndexedType(
GEP1->getSourceElementType(), IntermediateIndices)))
- return AliasAnalysis::MayAlias;
+ return MayAlias;
IntermediateIndices.push_back(GEP1->getOperand(i + 1));
}
GEP1->getSourceElementType(), IntermediateIndices));
if (!LastIndexedStruct)
- return AliasAnalysis::MayAlias;
+ return MayAlias;
// We know that:
// - both GEPs begin indexing from the exact same pointer;
if (EltsDontOverlap(V1Off, V1Size, V2Off, V2Size) ||
EltsDontOverlap(V2Off, V2Size, V1Off, V1Size))
- return AliasAnalysis::NoAlias;
-
- return AliasAnalysis::MayAlias;
-}
-
-bool BasicAliasAnalysis::constantOffsetHeuristic(
- const SmallVectorImpl<VariableGEPIndex> &VarIndices, uint64_t V1Size,
- uint64_t V2Size, int64_t BaseOffset, const DataLayout *DL,
- AssumptionCache *AC, DominatorTree *DT) {
- if (VarIndices.size() != 2 || V1Size == UnknownSize ||
- V2Size == 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, 1),
- V1Offset(Width, 1);
- 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;
+ return NoAlias;
- // 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;
+ return MayAlias;
}
/// aliasGEP - Provide a bunch of ad-hoc rules to disambiguate a GEP instruction
/// anything about V2. UnderlyingV1 is GetUnderlyingObject(GEP1, DL),
/// UnderlyingV2 is the same for V2.
///
-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) {
+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<VariableGEPIndex, 4> GEP1VariableIndices;
// derived pointer.
if (const GEPOperator *GEP2 = dyn_cast<GEPOperator>(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.
// 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
// 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;
// GEP1 V2
// We need to know that V2Size is not unknown, otherwise we might have
// stripped a gep with negative index ('gep <ptr>, -1, ...).
- if (V1Size != UnknownSize && V2Size != UnknownSize) {
+ if (V1Size != MemoryLocation::UnknownSize &&
+ V2Size != MemoryLocation::UnknownSize) {
if (-(uint64_t)GEP1BaseOffset < V1Size)
return PartialAlias;
return NoAlias;
// Zero-extension widens the variable, and so forces the sign
// bit to zero.
- bool IsZExt = GEP1VariableIndices[i].ZExtBits > 0 || isa<ZExtInst>(V);
+ bool IsZExt = GEP1VariableIndices[i].Extension == EK_ZeroExt;
SignKnownZero |= IsZExt;
SignKnownOne &= !IsZExt;
// 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.
// don't alias if V2Size can fit in the gap between V2 and GEP1BasePtr.
if (AllPositive && GEP1BaseOffset > 0 && V2Size <= (uint64_t) GEP1BaseOffset)
return NoAlias;
-
- if (constantOffsetHeuristic(GEP1VariableIndices, V1Size, V2Size,
- GEP1BaseOffset, DL, AC1, DT))
- return NoAlias;
}
// Statically, we can see that the base objects are the same, but the
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) {
+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<SelectInst>(V2))
// 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) {
+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());
// on corresponding edges.
if (const PHINode *PN2 = dyn_cast<PHINode>(V2))
if (PN2->getParent() == PN->getParent()) {
- LocPair Locs(Location(PN, PNSize, PNAAInfo),
- Location(V2, V2Size, V2AAInfo));
+ LocPair Locs(MemoryLocation(PN, PNSize, PNAAInfo),
+ MemoryLocation(V2, V2Size, V2AAInfo));
if (PN > V2)
std::swap(Locs.first, Locs.second);
// Analyse the PHIs' inputs under the assumption that the PHIs are
SmallPtrSet<Value*, 4> UniqueSrc;
SmallVector<Value*, 4> V1Srcs;
- for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
- Value *PV1 = PN->getIncomingValue(i);
+ bool isRecursive = false;
+ for (Value *PV1 : PN->incoming_values()) {
if (isa<PHINode>(PV1))
// If any of the source itself is a PHI, return MayAlias conservatively
// to avoid compile time explosion. The worst possible case is if both
// 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<GEPOperator>(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<ConstantInt>(PV1GEP->idx_begin())) {
+ isRecursive = true;
+ continue;
+ }
+ }
+
if (UniqueSrc.insert(PV1).second)
V1Srcs.push_back(PV1);
}
+ // 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)
// 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) {
+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)
// 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
// otherwise infinitely recursive queries.
- LocPair Locs(Location(V1, V1Size, V1AAInfo),
- Location(V2, V2Size, V2AAInfo));
+ LocPair Locs(MemoryLocation(V1, V1Size, V1AAInfo),
+ MemoryLocation(V2, V2Size, V2AAInfo));
if (V1 > V2)
std::swap(Locs.first, Locs.second);
std::pair<AliasCacheTy::iterator, bool> Pair =
// 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 =
- AliasAnalysis::alias(Location(V1, V1Size, V1AAInfo),
- Location(V2, V2Size, V2AAInfo));
+ AliasAnalysis::alias(MemoryLocation(V1, V1Size, V1AAInfo),
+ MemoryLocation(V2, V2Size, V2AAInfo));
return AliasCache[Locs] = Result;
}
for (unsigned i = 0, e = Src.size(); i != e; ++i) {
const Value *V = Src[i].V;
- unsigned ZExtBits = Src[i].ZExtBits, SExtBits = Src[i].SExtBits;
+ ExtensionKind Extension = Src[i].Extension;
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].ZExtBits != ZExtBits || Dest[j].SExtBits != SExtBits)
+ Dest[j].Extension != Extension)
continue;
// If we found it, subtract off Scale V's from the entry in Dest. If it
// If we didn't consume this entry, add it to the end of the Dest list.
if (Scale) {
- VariableGEPIndex Entry = {V, ZExtBits, SExtBits, -Scale};
+ VariableGEPIndex Entry = { V, Extension, -Scale };
Dest.push_back(Entry);
}
}