Instruction *InstCombiner::visitAllocaInst(AllocaInst &AI) {
// Ensure that the alloca array size argument has type intptr_t, so that
// any casting is exposed early.
if (TD) {
const Type *IntPtrTy = TD->getIntPtrType(AI.getContext());
if (AI.getArraySize()->getType() != IntPtrTy) {
Value *V = Builder->CreateIntCast(AI.getArraySize(),
IntPtrTy, false);
AI.setOperand(0, V);
return &AI;
}
}
// Convert: alloca Ty, C - where C is a constant != 1 into: alloca [C x Ty], 1
if (AI.isArrayAllocation()) { // Check C != 1
if (const ConstantInt *C = dyn_cast<ConstantInt>(AI.getArraySize())) {
const Type *NewTy =
ArrayType::get(AI.getAllocatedType(), C->getZExtValue());
assert(isa<AllocaInst>(AI) && "Unknown type of allocation inst!");
AllocaInst *New = Builder->CreateAlloca(NewTy, 0, AI.getName());
New->setAlignment(AI.getAlignment());
// Scan to the end of the allocation instructions, to skip over a block of
// allocas if possible...also skip interleaved debug info
//
BasicBlock::iterator It = New;
while (isa<AllocaInst>(*It) || isa<DbgInfoIntrinsic>(*It)) ++It;
// Now that I is pointing to the first non-allocation-inst in the block,
// insert our getelementptr instruction...
//
Value *NullIdx =Constant::getNullValue(Type::getInt32Ty(AI.getContext()));
Value *Idx[2];
Idx[0] = NullIdx;
Idx[1] = NullIdx;
Value *V = GetElementPtrInst::CreateInBounds(New, Idx, Idx + 2,
New->getName()+".sub", It);
// Now make everything use the getelementptr instead of the original
// allocation.
return ReplaceInstUsesWith(AI, V);
} else if (isa<UndefValue>(AI.getArraySize())) {
return ReplaceInstUsesWith(AI, Constant::getNullValue(AI.getType()));
}
}
if (TD && isa<AllocaInst>(AI) && AI.getAllocatedType()->isSized()) {
// If alloca'ing a zero byte object, replace the alloca with a null pointer.
// Note that we only do this for alloca's, because malloc should allocate
// and return a unique pointer, even for a zero byte allocation.
if (TD->getTypeAllocSize(AI.getAllocatedType()) == 0)
return ReplaceInstUsesWith(AI, Constant::getNullValue(AI.getType()));
// If the alignment is 0 (unspecified), assign it the preferred alignment.
if (AI.getAlignment() == 0)
AI.setAlignment(TD->getPrefTypeAlignment(AI.getAllocatedType()));
}
return 0;
}
AllocaInst* Variables::changeLocal(Value* value, ArrayType* newType) {
AllocaInst* oldTarget = dyn_cast<AllocaInst>(value);
PointerType* oldPointerType = dyn_cast<PointerType>(oldTarget->getType());
ArrayType* oldType = dyn_cast<ArrayType>(oldPointerType->getElementType());
AllocaInst* newTarget = NULL;
errs() << "Changing the precision of variable \"" << oldTarget->getName() << "\" from " << *oldType
<< " to " << *newType << ".\n";
if (newType->getElementType()->getTypeID() != oldType->getElementType()->getTypeID()) {
newTarget = new AllocaInst(newType, getInt32(1), "", oldTarget);
// we are not calling getAlignment because in this case double requires 16. Investigate further.
unsigned alignment;
switch(newType->getElementType()->getTypeID()) {
case Type::FloatTyID:
alignment = 4;
break;
case Type::DoubleTyID:
alignment = 16;
break;
case Type::X86_FP80TyID:
alignment = 16;
break;
default:
alignment = 0;
}
newTarget->setAlignment(alignment); // depends on type? 8 for float? 16 for double?
newTarget->takeName(oldTarget);
// iterating through instructions using old AllocaInst
vector<Instruction*> erase;
Value::use_iterator it = oldTarget->use_begin();
for(; it != oldTarget->use_end(); it++) {
bool is_erased = Transformer::transform(it, newTarget, oldTarget, newType, oldType, alignment);
if (!is_erased)
erase.push_back(dyn_cast<Instruction>(*it));
}
// erasing uses of old instructions
for(unsigned int i = 0; i < erase.size(); i++) {
erase[i]->eraseFromParent();
}
// erase old instruction
//oldTarget->eraseFromParent();
}
else {
errs() << "\tNo changes required.\n";
}
return newTarget;
}
static Instruction *simplifyAllocaArraySize(InstCombiner &IC, AllocaInst &AI) {
// Check for array size of 1 (scalar allocation).
if (!AI.isArrayAllocation()) {
// i32 1 is the canonical array size for scalar allocations.
if (AI.getArraySize()->getType()->isIntegerTy(32))
return nullptr;
// Canonicalize it.
Value *V = IC.Builder->getInt32(1);
AI.setOperand(0, V);
return &AI;
}
// Convert: alloca Ty, C - where C is a constant != 1 into: alloca [C x Ty], 1
if (const ConstantInt *C = dyn_cast<ConstantInt>(AI.getArraySize())) {
Type *NewTy = ArrayType::get(AI.getAllocatedType(), C->getZExtValue());
AllocaInst *New = IC.Builder->CreateAlloca(NewTy, nullptr, AI.getName());
New->setAlignment(AI.getAlignment());
// Scan to the end of the allocation instructions, to skip over a block of
// allocas if possible...also skip interleaved debug info
//
BasicBlock::iterator It(New);
while (isa<AllocaInst>(*It) || isa<DbgInfoIntrinsic>(*It))
++It;
// Now that I is pointing to the first non-allocation-inst in the block,
// insert our getelementptr instruction...
//
Type *IdxTy = IC.getDataLayout().getIntPtrType(AI.getType());
Value *NullIdx = Constant::getNullValue(IdxTy);
Value *Idx[2] = {NullIdx, NullIdx};
Instruction *GEP =
GetElementPtrInst::CreateInBounds(New, Idx, New->getName() + ".sub");
IC.InsertNewInstBefore(GEP, *It);
// Now make everything use the getelementptr instead of the original
// allocation.
return IC.ReplaceInstUsesWith(AI, GEP);
}
if (isa<UndefValue>(AI.getArraySize()))
return IC.ReplaceInstUsesWith(AI, Constant::getNullValue(AI.getType()));
// Ensure that the alloca array size argument has type intptr_t, so that
// any casting is exposed early.
Type *IntPtrTy = IC.getDataLayout().getIntPtrType(AI.getType());
if (AI.getArraySize()->getType() != IntPtrTy) {
Value *V = IC.Builder->CreateIntCast(AI.getArraySize(), IntPtrTy, false);
AI.setOperand(0, V);
return &AI;
}
return nullptr;
}
bool MisalignStackPass::runOnBasicBlock (BasicBlock &BB)
{
bool Changed = false;
const unsigned alignLimit = sizeof(uint32_t);
for (BasicBlock::iterator I = BB.begin(), E = BB.end(); I != E; ++I) {
AllocaInst *AI = dyn_cast<AllocaInst>(I);
if (AI && AI->getAlignment() > alignLimit) {
AI->setAlignment(alignLimit);
Changed = true;
}
}
return Changed;
}
// =============================================================================
// If the function had a byval struct ptr arg, say foo(%struct.x* byval %d),
// then add the following instructions to the first basic block:
//
// %temp = alloca %struct.x, align 8
// %tempd = addrspacecast %struct.x* %d to %struct.x addrspace(101)*
// %tv = load %struct.x addrspace(101)* %tempd
// store %struct.x %tv, %struct.x* %temp, align 8
//
// The above code allocates some space in the stack and copies the incoming
// struct from param space to local space.
// Then replace all occurrences of %d by %temp.
// =============================================================================
void NVPTXLowerArgs::handleByValParam(Argument *Arg) {
Function *Func = Arg->getParent();
Instruction *FirstInst = &(Func->getEntryBlock().front());
PointerType *PType = dyn_cast<PointerType>(Arg->getType());
assert(PType && "Expecting pointer type in handleByValParam");
Type *StructType = PType->getElementType();
AllocaInst *AllocA = new AllocaInst(StructType, Arg->getName(), FirstInst);
// Set the alignment to alignment of the byval parameter. This is because,
// later load/stores assume that alignment, and we are going to replace
// the use of the byval parameter with this alloca instruction.
AllocA->setAlignment(Func->getParamAlignment(Arg->getArgNo() + 1));
Arg->replaceAllUsesWith(AllocA);
Value *ArgInParam = new AddrSpaceCastInst(
Arg, PointerType::get(StructType, ADDRESS_SPACE_PARAM), Arg->getName(),
FirstInst);
LoadInst *LI = new LoadInst(ArgInParam, Arg->getName(), FirstInst);
new StoreInst(LI, AllocA, FirstInst);
}
void NVPTXLowerStructArgs::handleParam(Argument *Arg) {
Function *Func = Arg->getParent();
Instruction *FirstInst = &(Func->getEntryBlock().front());
PointerType *PType = dyn_cast<PointerType>(Arg->getType());
assert(PType && "Expecting pointer type in handleParam");
Type *StructType = PType->getElementType();
AllocaInst *AllocA = new AllocaInst(StructType, Arg->getName(), FirstInst);
/* Set the alignment to alignment of the byval parameter. This is because,
* later load/stores assume that alignment, and we are going to replace
* the use of the byval parameter with this alloca instruction.
*/
AllocA->setAlignment(Func->getParamAlignment(Arg->getArgNo() + 1));
Arg->replaceAllUsesWith(AllocA);
// Get the cvt.gen.to.param intrinsic
Type *CvtTypes[] = {
Type::getInt8PtrTy(Func->getParent()->getContext(), ADDRESS_SPACE_PARAM),
Type::getInt8PtrTy(Func->getParent()->getContext(),
ADDRESS_SPACE_GENERIC)};
Function *CvtFunc = Intrinsic::getDeclaration(
Func->getParent(), Intrinsic::nvvm_ptr_gen_to_param, CvtTypes);
Value *BitcastArgs[] = {
new BitCastInst(Arg, Type::getInt8PtrTy(Func->getParent()->getContext(),
ADDRESS_SPACE_GENERIC),
Arg->getName(), FirstInst)};
CallInst *CallCVT =
CallInst::Create(CvtFunc, BitcastArgs, "cvt_to_param", FirstInst);
BitCastInst *BitCast = new BitCastInst(
CallCVT, PointerType::get(StructType, ADDRESS_SPACE_PARAM),
Arg->getName(), FirstInst);
LoadInst *LI = new LoadInst(BitCast, Arg->getName(), FirstInst);
new StoreInst(LI, AllocA, FirstInst);
}
void ConstantInsertExtractElementIndex::fixNonConstantVectorIndices(
BasicBlock &BB, const Instructions &Instrs) const {
for (Instructions::const_iterator IB = Instrs.begin(), IE = Instrs.end();
IB != IE; ++IB) {
Instruction *I = *IB;
Value *Vec = I->getOperand(0);
Value *Idx = getInsertExtractElementIdx(I);
VectorType *VecTy = cast<VectorType>(Vec->getType());
Type *ElemTy = VecTy->getElementType();
unsigned ElemAlign = DL->getPrefTypeAlignment(ElemTy);
unsigned VecAlign = std::max(ElemAlign, DL->getPrefTypeAlignment(VecTy));
IRBuilder<> IRB(I);
AllocaInst *Alloca = IRB.CreateAlloca(
ElemTy, ConstantInt::get(Type::getInt32Ty(M->getContext()),
vectorNumElements(I)));
Alloca->setAlignment(VecAlign);
Value *AllocaAsVec = IRB.CreateBitCast(Alloca, VecTy->getPointerTo());
IRB.CreateAlignedStore(Vec, AllocaAsVec, Alloca->getAlignment());
Value *GEP = IRB.CreateGEP(Alloca, Idx);
Value *Res;
switch (I->getOpcode()) {
default:
llvm_unreachable("expected InsertElement or ExtractElement");
case Instruction::InsertElement:
IRB.CreateAlignedStore(I->getOperand(1), GEP, ElemAlign);
Res = IRB.CreateAlignedLoad(AllocaAsVec, Alloca->getAlignment());
break;
case Instruction::ExtractElement:
Res = IRB.CreateAlignedLoad(GEP, ElemAlign);
break;
}
I->replaceAllUsesWith(Res);
I->eraseFromParent();
}
}
void Preparer::expandAlloca(AllocaInst *AI) {
// Skip dynaa.slots which is added by AliasCheckerInstrumenter.
if (AI->getName().startswith(DynAAUtils::SlotsName))
return;
if (AI->isArrayAllocation()) {
// e.g. %32 = alloca i8, i64 %conv164
Value *Size = AI->getArraySize();
Value *ExpandedSize = BinaryOperator::Create(
Instruction::Add,
Size,
ConstantInt::get(cast<IntegerType>(Size->getType()), 1),
"expanded.size",
AI);
AI->setOperand(0, ExpandedSize);
return;
}
Type *AllocatedType = AI->getAllocatedType();
if (ArrayType *ArrType = dyn_cast<ArrayType>(AllocatedType)) {
ArrayType *NewArrType = ArrayType::get(ArrType->getElementType(),
ArrType->getNumElements() + 1);
AllocaInst *NewAI = new AllocaInst(NewArrType, AI->getName(), AI);
// inherit the alignment as well
NewAI->setAlignment(AI->getAlignment());
BitCastInst *CastNewAI = new BitCastInst(NewAI,
AI->getType(),
AI->getName(),
AI);
AI->replaceAllUsesWith(CastNewAI);
AI->eraseFromParent();
return;
}
assert(AllocatedType->isSized());
IntegerType *PadType = IntegerType::get(AI->getContext(), 8);
new AllocaInst(PadType, "alloca_pad", AI);
}
Instruction *InstCombiner::visitAllocaInst(AllocaInst &AI) {
if (auto *I = simplifyAllocaArraySize(*this, AI))
return I;
if (AI.getAllocatedType()->isSized()) {
// If the alignment is 0 (unspecified), assign it the preferred alignment.
if (AI.getAlignment() == 0)
AI.setAlignment(DL.getPrefTypeAlignment(AI.getAllocatedType()));
// Move all alloca's of zero byte objects to the entry block and merge them
// together. Note that we only do this for alloca's, because malloc should
// allocate and return a unique pointer, even for a zero byte allocation.
if (DL.getTypeAllocSize(AI.getAllocatedType()) == 0) {
// For a zero sized alloca there is no point in doing an array allocation.
// This is helpful if the array size is a complicated expression not used
// elsewhere.
if (AI.isArrayAllocation()) {
AI.setOperand(0, ConstantInt::get(AI.getArraySize()->getType(), 1));
return &AI;
}
// Get the first instruction in the entry block.
BasicBlock &EntryBlock = AI.getParent()->getParent()->getEntryBlock();
Instruction *FirstInst = EntryBlock.getFirstNonPHIOrDbg();
if (FirstInst != &AI) {
// If the entry block doesn't start with a zero-size alloca then move
// this one to the start of the entry block. There is no problem with
// dominance as the array size was forced to a constant earlier already.
AllocaInst *EntryAI = dyn_cast<AllocaInst>(FirstInst);
if (!EntryAI || !EntryAI->getAllocatedType()->isSized() ||
DL.getTypeAllocSize(EntryAI->getAllocatedType()) != 0) {
AI.moveBefore(FirstInst);
return &AI;
}
// If the alignment of the entry block alloca is 0 (unspecified),
// assign it the preferred alignment.
if (EntryAI->getAlignment() == 0)
EntryAI->setAlignment(
DL.getPrefTypeAlignment(EntryAI->getAllocatedType()));
// Replace this zero-sized alloca with the one at the start of the entry
// block after ensuring that the address will be aligned enough for both
// types.
unsigned MaxAlign = std::max(EntryAI->getAlignment(),
AI.getAlignment());
EntryAI->setAlignment(MaxAlign);
if (AI.getType() != EntryAI->getType())
return new BitCastInst(EntryAI, AI.getType());
return ReplaceInstUsesWith(AI, EntryAI);
}
}
}
if (AI.getAlignment()) {
// Check to see if this allocation is only modified by a memcpy/memmove from
// a constant global whose alignment is equal to or exceeds that of the
// allocation. If this is the case, we can change all users to use
// the constant global instead. This is commonly produced by the CFE by
// constructs like "void foo() { int A[] = {1,2,3,4,5,6,7,8,9...}; }" if 'A'
// is only subsequently read.
SmallVector<Instruction *, 4> ToDelete;
if (MemTransferInst *Copy = isOnlyCopiedFromConstantGlobal(&AI, ToDelete)) {
unsigned SourceAlign = getOrEnforceKnownAlignment(
Copy->getSource(), AI.getAlignment(), DL, &AI, AC, DT);
if (AI.getAlignment() <= SourceAlign) {
DEBUG(dbgs() << "Found alloca equal to global: " << AI << '\n');
DEBUG(dbgs() << " memcpy = " << *Copy << '\n');
for (unsigned i = 0, e = ToDelete.size(); i != e; ++i)
EraseInstFromFunction(*ToDelete[i]);
Constant *TheSrc = cast<Constant>(Copy->getSource());
Constant *Cast
= ConstantExpr::getPointerBitCastOrAddrSpaceCast(TheSrc, AI.getType());
Instruction *NewI = ReplaceInstUsesWith(AI, Cast);
EraseInstFromFunction(*Copy);
++NumGlobalCopies;
return NewI;
}
}
}
// At last, use the generic allocation site handler to aggressively remove
// unused allocas.
return visitAllocSite(AI);
}
/// If it is possible to inline the specified call site,
/// do so and update the CallGraph for this operation.
///
/// This function also does some basic book-keeping to update the IR. The
/// InlinedArrayAllocas map keeps track of any allocas that are already
/// available from other functions inlined into the caller. If we are able to
/// inline this call site we attempt to reuse already available allocas or add
/// any new allocas to the set if not possible.
static bool InlineCallIfPossible(CallSite CS, InlineFunctionInfo &IFI,
InlinedArrayAllocasTy &InlinedArrayAllocas,
int InlineHistory, bool InsertLifetime) {
Function *Callee = CS.getCalledFunction();
Function *Caller = CS.getCaller();
// Try to inline the function. Get the list of static allocas that were
// inlined.
if (!InlineFunction(CS, IFI, InsertLifetime))
return false;
AdjustCallerSSPLevel(Caller, Callee);
// Look at all of the allocas that we inlined through this call site. If we
// have already inlined other allocas through other calls into this function,
// then we know that they have disjoint lifetimes and that we can merge them.
//
// There are many heuristics possible for merging these allocas, and the
// different options have different tradeoffs. One thing that we *really*
// don't want to hurt is SRoA: once inlining happens, often allocas are no
// longer address taken and so they can be promoted.
//
// Our "solution" for that is to only merge allocas whose outermost type is an
// array type. These are usually not promoted because someone is using a
// variable index into them. These are also often the most important ones to
// merge.
//
// A better solution would be to have real memory lifetime markers in the IR
// and not have the inliner do any merging of allocas at all. This would
// allow the backend to do proper stack slot coloring of all allocas that
// *actually make it to the backend*, which is really what we want.
//
// Because we don't have this information, we do this simple and useful hack.
//
SmallPtrSet<AllocaInst*, 16> UsedAllocas;
// When processing our SCC, check to see if CS was inlined from some other
// call site. For example, if we're processing "A" in this code:
// A() { B() }
// B() { x = alloca ... C() }
// C() { y = alloca ... }
// Assume that C was not inlined into B initially, and so we're processing A
// and decide to inline B into A. Doing this makes an alloca available for
// reuse and makes a callsite (C) available for inlining. When we process
// the C call site we don't want to do any alloca merging between X and Y
// because their scopes are not disjoint. We could make this smarter by
// keeping track of the inline history for each alloca in the
// InlinedArrayAllocas but this isn't likely to be a significant win.
if (InlineHistory != -1) // Only do merging for top-level call sites in SCC.
return true;
// Loop over all the allocas we have so far and see if they can be merged with
// a previously inlined alloca. If not, remember that we had it.
for (unsigned AllocaNo = 0, e = IFI.StaticAllocas.size();
AllocaNo != e; ++AllocaNo) {
AllocaInst *AI = IFI.StaticAllocas[AllocaNo];
// Don't bother trying to merge array allocations (they will usually be
// canonicalized to be an allocation *of* an array), or allocations whose
// type is not itself an array (because we're afraid of pessimizing SRoA).
ArrayType *ATy = dyn_cast<ArrayType>(AI->getAllocatedType());
if (!ATy || AI->isArrayAllocation())
continue;
// Get the list of all available allocas for this array type.
std::vector<AllocaInst*> &AllocasForType = InlinedArrayAllocas[ATy];
// Loop over the allocas in AllocasForType to see if we can reuse one. Note
// that we have to be careful not to reuse the same "available" alloca for
// multiple different allocas that we just inlined, we use the 'UsedAllocas'
// set to keep track of which "available" allocas are being used by this
// function. Also, AllocasForType can be empty of course!
bool MergedAwayAlloca = false;
for (unsigned i = 0, e = AllocasForType.size(); i != e; ++i) {
AllocaInst *AvailableAlloca = AllocasForType[i];
unsigned Align1 = AI->getAlignment(),
Align2 = AvailableAlloca->getAlignment();
// The available alloca has to be in the right function, not in some other
// function in this SCC.
if (AvailableAlloca->getParent() != AI->getParent())
continue;
// If the inlined function already uses this alloca then we can't reuse
// it.
if (!UsedAllocas.insert(AvailableAlloca).second)
continue;
// Otherwise, we *can* reuse it, RAUW AI into AvailableAlloca and declare
// success!
DEBUG(dbgs() << " ***MERGED ALLOCA: " << *AI << "\n\t\tINTO: "
<< *AvailableAlloca << '\n');
AI->replaceAllUsesWith(AvailableAlloca);
if (Align1 != Align2) {
if (!Align1 || !Align2) {
const DataLayout &DL = Caller->getParent()->getDataLayout();
unsigned TypeAlign = DL.getABITypeAlignment(AI->getAllocatedType());
Align1 = Align1 ? Align1 : TypeAlign;
Align2 = Align2 ? Align2 : TypeAlign;
}
if (Align1 > Align2)
AvailableAlloca->setAlignment(AI->getAlignment());
}
AI->eraseFromParent();
MergedAwayAlloca = true;
++NumMergedAllocas;
IFI.StaticAllocas[AllocaNo] = nullptr;
break;
}
// If we already nuked the alloca, we're done with it.
if (MergedAwayAlloca)
continue;
// If we were unable to merge away the alloca either because there are no
// allocas of the right type available or because we reused them all
// already, remember that this alloca came from an inlined function and mark
// it used so we don't reuse it for other allocas from this inline
// operation.
AllocasForType.push_back(AI);
UsedAllocas.insert(AI);
}
return true;
}
Instruction *InstCombiner::visitAllocaInst(AllocaInst &AI) {
// Ensure that the alloca array size argument has type intptr_t, so that
// any casting is exposed early.
if (DL) {
Type *IntPtrTy = DL->getIntPtrType(AI.getType());
if (AI.getArraySize()->getType() != IntPtrTy) {
Value *V = Builder->CreateIntCast(AI.getArraySize(),
IntPtrTy, false);
AI.setOperand(0, V);
return &AI;
}
}
// Convert: alloca Ty, C - where C is a constant != 1 into: alloca [C x Ty], 1
if (AI.isArrayAllocation()) { // Check C != 1
if (const ConstantInt *C = dyn_cast<ConstantInt>(AI.getArraySize())) {
Type *NewTy =
ArrayType::get(AI.getAllocatedType(), C->getZExtValue());
AllocaInst *New = Builder->CreateAlloca(NewTy, nullptr, AI.getName());
New->setAlignment(AI.getAlignment());
// Scan to the end of the allocation instructions, to skip over a block of
// allocas if possible...also skip interleaved debug info
//
BasicBlock::iterator It = New;
while (isa<AllocaInst>(*It) || isa<DbgInfoIntrinsic>(*It)) ++It;
// Now that I is pointing to the first non-allocation-inst in the block,
// insert our getelementptr instruction...
//
Type *IdxTy = DL
? DL->getIntPtrType(AI.getType())
: Type::getInt64Ty(AI.getContext());
Value *NullIdx = Constant::getNullValue(IdxTy);
Value *Idx[2] = { NullIdx, NullIdx };
Instruction *GEP =
GetElementPtrInst::CreateInBounds(New, Idx, New->getName() + ".sub");
InsertNewInstBefore(GEP, *It);
// Now make everything use the getelementptr instead of the original
// allocation.
return ReplaceInstUsesWith(AI, GEP);
} else if (isa<UndefValue>(AI.getArraySize())) {
return ReplaceInstUsesWith(AI, Constant::getNullValue(AI.getType()));
}
}
if (DL && AI.getAllocatedType()->isSized()) {
// If the alignment is 0 (unspecified), assign it the preferred alignment.
if (AI.getAlignment() == 0)
AI.setAlignment(DL->getPrefTypeAlignment(AI.getAllocatedType()));
// Move all alloca's of zero byte objects to the entry block and merge them
// together. Note that we only do this for alloca's, because malloc should
// allocate and return a unique pointer, even for a zero byte allocation.
if (DL->getTypeAllocSize(AI.getAllocatedType()) == 0) {
// For a zero sized alloca there is no point in doing an array allocation.
// This is helpful if the array size is a complicated expression not used
// elsewhere.
if (AI.isArrayAllocation()) {
AI.setOperand(0, ConstantInt::get(AI.getArraySize()->getType(), 1));
return &AI;
}
// Get the first instruction in the entry block.
BasicBlock &EntryBlock = AI.getParent()->getParent()->getEntryBlock();
Instruction *FirstInst = EntryBlock.getFirstNonPHIOrDbg();
if (FirstInst != &AI) {
// If the entry block doesn't start with a zero-size alloca then move
// this one to the start of the entry block. There is no problem with
// dominance as the array size was forced to a constant earlier already.
AllocaInst *EntryAI = dyn_cast<AllocaInst>(FirstInst);
if (!EntryAI || !EntryAI->getAllocatedType()->isSized() ||
DL->getTypeAllocSize(EntryAI->getAllocatedType()) != 0) {
AI.moveBefore(FirstInst);
return &AI;
}
// If the alignment of the entry block alloca is 0 (unspecified),
// assign it the preferred alignment.
if (EntryAI->getAlignment() == 0)
EntryAI->setAlignment(
DL->getPrefTypeAlignment(EntryAI->getAllocatedType()));
// Replace this zero-sized alloca with the one at the start of the entry
// block after ensuring that the address will be aligned enough for both
// types.
unsigned MaxAlign = std::max(EntryAI->getAlignment(),
AI.getAlignment());
EntryAI->setAlignment(MaxAlign);
if (AI.getType() != EntryAI->getType())
return new BitCastInst(EntryAI, AI.getType());
return ReplaceInstUsesWith(AI, EntryAI);
}
}
}
if (AI.getAlignment()) {
// Check to see if this allocation is only modified by a memcpy/memmove from
// a constant global whose alignment is equal to or exceeds that of the
// allocation. If this is the case, we can change all users to use
// the constant global instead. This is commonly produced by the CFE by
// constructs like "void foo() { int A[] = {1,2,3,4,5,6,7,8,9...}; }" if 'A'
// is only subsequently read.
SmallVector<Instruction *, 4> ToDelete;
if (MemTransferInst *Copy = isOnlyCopiedFromConstantGlobal(&AI, ToDelete)) {
unsigned SourceAlign = getOrEnforceKnownAlignment(Copy->getSource(),
AI.getAlignment(), DL);
if (AI.getAlignment() <= SourceAlign) {
DEBUG(dbgs() << "Found alloca equal to global: " << AI << '\n');
DEBUG(dbgs() << " memcpy = " << *Copy << '\n');
for (unsigned i = 0, e = ToDelete.size(); i != e; ++i)
EraseInstFromFunction(*ToDelete[i]);
Constant *TheSrc = cast<Constant>(Copy->getSource());
Constant *Cast
= ConstantExpr::getPointerBitCastOrAddrSpaceCast(TheSrc, AI.getType());
Instruction *NewI = ReplaceInstUsesWith(AI, Cast);
EraseInstFromFunction(*Copy);
++NumGlobalCopies;
return NewI;
}
}
}
// At last, use the generic allocation site handler to aggressively remove
// unused allocas.
return visitAllocSite(AI);
}
static Value *julia_to_native(Type *ty, jl_value_t *jt, Value *jv,
jl_value_t *aty, bool addressOf,
bool byRef, bool inReg,
bool needCopy,
int argn, jl_codectx_t *ctx,
bool *needStackRestore)
{
Type *vt = jv->getType();
// We're passing any
if (ty == jl_pvalue_llvmt) {
return boxed(jv,ctx);
}
if (ty == vt && !addressOf && !byRef) {
return jv;
}
if (vt != jl_pvalue_llvmt) {
// argument value is unboxed
if (addressOf || (byRef && inReg)) {
if (ty->isPointerTy() && ty->getContainedType(0)==vt) {
// pass the address of an alloca'd thing, not a box
// since those are immutable.
*needStackRestore = true;
Value *slot = builder.CreateAlloca(vt);
builder.CreateStore(jv, slot);
return builder.CreateBitCast(slot, ty);
}
}
else if ((vt->isIntegerTy() && ty->isIntegerTy()) ||
(vt->isFloatingPointTy() && ty->isFloatingPointTy()) ||
(vt->isPointerTy() && ty->isPointerTy())) {
if (vt->getPrimitiveSizeInBits() ==
ty->getPrimitiveSizeInBits()) {
if (!byRef) {
return builder.CreateBitCast(jv, ty);
}
else {
*needStackRestore = true;
Value *mem = builder.CreateAlloca(ty);
builder.CreateStore(jv,builder.CreateBitCast(mem,vt->getPointerTo()));
return mem;
}
}
}
else if (vt->isStructTy()) {
if (!byRef) {
return jv;
}
else {
*needStackRestore = true;
Value *mem = builder.CreateAlloca(vt);
builder.CreateStore(jv,mem);
return mem;
}
}
emit_error("ccall: argument type did not match declaration", ctx);
}
if (jl_is_tuple(jt)) {
return emit_unbox(ty,jv,jt);
}
if (jl_is_cpointer_type(jt) && addressOf) {
assert(ty->isPointerTy());
jl_value_t *ety = jl_tparam0(jt);
if (aty != ety && ety != (jl_value_t*)jl_any_type && jt != (jl_value_t*)jl_voidpointer_type) {
std::stringstream msg;
msg << "ccall argument ";
msg << argn;
emit_typecheck(jv, ety, msg.str(), ctx);
}
if (jl_is_mutable_datatype(ety)) {
// no copy, just reference the data field
return builder.CreateBitCast(jv, ty);
}
else if (jl_is_immutable_datatype(ety) && jt != (jl_value_t*)jl_voidpointer_type) {
// yes copy
Value *nbytes;
if (jl_is_leaf_type(ety))
nbytes = ConstantInt::get(T_int32, jl_datatype_size(ety));
else
nbytes = tbaa_decorate(tbaa_datatype, builder.CreateLoad(
builder.CreateGEP(builder.CreatePointerCast(emit_typeof(jv), T_pint32),
ConstantInt::get(T_size, offsetof(jl_datatype_t,size)/sizeof(int32_t))),
false));
*needStackRestore = true;
AllocaInst *ai = builder.CreateAlloca(T_int8, nbytes);
ai->setAlignment(16);
builder.CreateMemCpy(ai, builder.CreateBitCast(jv, T_pint8), nbytes, 1);
return builder.CreateBitCast(ai, ty);
}
// emit maybe copy
*needStackRestore = true;
Value *jvt = emit_typeof(jv);
BasicBlock *mutableBB = BasicBlock::Create(getGlobalContext(),"is-mutable",ctx->f);
BasicBlock *immutableBB = BasicBlock::Create(getGlobalContext(),"is-immutable",ctx->f);
BasicBlock *afterBB = BasicBlock::Create(getGlobalContext(),"after",ctx->f);
Value *ismutable = builder.CreateTrunc(
tbaa_decorate(tbaa_datatype, builder.CreateLoad(
builder.CreateGEP(builder.CreatePointerCast(jvt, T_pint8),
ConstantInt::get(T_size, offsetof(jl_datatype_t,mutabl))),
false)),
T_int1);
builder.CreateCondBr(ismutable, mutableBB, immutableBB);
builder.SetInsertPoint(mutableBB);
Value *p1 = builder.CreatePointerCast(jv, ty);
builder.CreateBr(afterBB);
builder.SetInsertPoint(immutableBB);
Value *nbytes = tbaa_decorate(tbaa_datatype, builder.CreateLoad(
builder.CreateGEP(builder.CreatePointerCast(jvt, T_pint32),
ConstantInt::get(T_size, offsetof(jl_datatype_t,size)/sizeof(int32_t))),
false));
AllocaInst *ai = builder.CreateAlloca(T_int8, nbytes);
ai->setAlignment(16);
builder.CreateMemCpy(ai, builder.CreatePointerCast(jv, T_pint8), nbytes, 1);
Value *p2 = builder.CreatePointerCast(ai, ty);
builder.CreateBr(afterBB);
builder.SetInsertPoint(afterBB);
PHINode *p = builder.CreatePHI(ty, 2);
p->addIncoming(p1, mutableBB);
p->addIncoming(p2, immutableBB);
return p;
}
if (addressOf)
jl_error("ccall: unexpected & on argument"); // the only "safe" thing to emit here is the expected struct
assert(jl_is_datatype(jt));
if (aty != jt) {
std::stringstream msg;
msg << "ccall argument ";
msg << argn;
emit_typecheck(jv, jt, msg.str(), ctx);
}
Value *p = data_pointer(jv);
Value *pjv = builder.CreatePointerCast(p, PointerType::get(ty,0));
if (byRef) {
if (!needCopy) {
return pjv;
}
else {
*needStackRestore = true;
Value *mem = builder.CreateAlloca(ty);
builder.CreateMemCpy(mem,pjv,(uint64_t)jl_datatype_size(jt),(uint64_t)((jl_datatype_t*)jt)->alignment);
return mem;
}
}
else {
return builder.CreateLoad(pjv,false);
}
}
static void convertInstruction(Instruction *Inst, ConversionState &State) {
if (SExtInst *Sext = dyn_cast<SExtInst>(Inst)) {
Value *Op = Sext->getOperand(0);
Value *NewInst = NULL;
// If the operand to be extended is illegal, we first need to fill its
// upper bits (which are zero) with its sign bit.
if (shouldConvert(Op)) {
NewInst = getSignExtend(State.getConverted(Op), Op, Sext);
}
// If the converted type of the operand is the same as the converted
// type of the result, we won't actually be changing the type of the
// variable, just its value.
if (getPromotedType(Op->getType()) !=
getPromotedType(Sext->getType())) {
NewInst = new SExtInst(
NewInst ? NewInst : State.getConverted(Op),
getPromotedType(cast<IntegerType>(Sext->getType())),
Sext->getName() + ".sext", Sext);
}
// Now all the bits of the result are correct, but we need to restore
// the bits above its type to zero.
if (shouldConvert(Sext)) {
NewInst = getClearUpper(NewInst, Sext->getType(), Sext);
}
assert(NewInst && "Failed to convert sign extension");
State.recordConverted(Sext, NewInst);
} else if (ZExtInst *Zext = dyn_cast<ZExtInst>(Inst)) {
Value *Op = Zext->getOperand(0);
Value *NewInst = NULL;
// TODO(dschuff): Some of these zexts could be no-ops.
if (shouldConvert(Op)) {
NewInst = getClearUpper(State.getConverted(Op),
Op->getType(),
Zext);
}
// If the converted type of the operand is the same as the converted
// type of the result, we won't actually be changing the type of the
// variable, just its value.
if (getPromotedType(Op->getType()) !=
getPromotedType(Zext->getType())) {
NewInst = CastInst::CreateZExtOrBitCast(
NewInst ? NewInst : State.getConverted(Op),
getPromotedType(cast<IntegerType>(Zext->getType())),
"", Zext);
}
assert(NewInst);
State.recordConverted(Zext, NewInst);
} else if (TruncInst *Trunc = dyn_cast<TruncInst>(Inst)) {
Value *Op = Trunc->getOperand(0);
Value *NewInst = NULL;
// If the converted type of the operand is the same as the converted
// type of the result, we won't actually be changing the type of the
// variable, just its value.
if (getPromotedType(Op->getType()) !=
getPromotedType(Trunc->getType())) {
NewInst = new TruncInst(
State.getConverted(Op),
getPromotedType(cast<IntegerType>(Trunc->getType())),
State.getConverted(Op)->getName() + ".trunc",
Trunc);
}
// Restoring the upper-bits-are-zero invariant effectively truncates the
// value.
if (shouldConvert(Trunc)) {
NewInst = getClearUpper(NewInst ? NewInst : Op,
Trunc->getType(),
Trunc);
}
assert(NewInst);
State.recordConverted(Trunc, NewInst);
} else if (AllocaInst *Alloc = dyn_cast<AllocaInst>(Inst)) {
// Don't handle arrays of illegal types, but we could handle an array
// with size specified as an illegal type, as unlikely as that seems.
if (shouldConvert(Alloc) && Alloc->isArrayAllocation())
report_fatal_error("Can't convert arrays of illegal type");
AllocaInst *NewInst = new AllocaInst(
getPromotedType(Alloc->getAllocatedType()),
State.getConverted(Alloc->getArraySize()),
"", Alloc);
NewInst->setAlignment(Alloc->getAlignment());
State.recordConverted(Alloc, NewInst);
} else if (BitCastInst *BCInst = dyn_cast<BitCastInst>(Inst)) {
// Only handle pointers. Ints can't be casted to/from other ints
Type *DestType = shouldConvert(BCInst) ?
getPromotedType(BCInst->getDestTy()) : BCInst->getDestTy();
BitCastInst *NewInst = new BitCastInst(
State.getConverted(BCInst->getOperand(0)),
DestType,
"", BCInst);
State.recordConverted(BCInst, NewInst);
} else if (LoadInst *Load = dyn_cast<LoadInst>(Inst)) {
if (shouldConvert(Load)) {
splitLoad(Load, State);
}
} else if (StoreInst *Store = dyn_cast<StoreInst>(Inst)) {
if (shouldConvert(Store->getValueOperand())) {
splitStore(Store, State);
}
} else if (isa<CallInst>(Inst)) {
report_fatal_error("can't convert calls with illegal types");
} else if (BinaryOperator *Binop = dyn_cast<BinaryOperator>(Inst)) {
Value *NewInst = NULL;
if (Binop->getOpcode() == Instruction::AShr) {
// The AShr operand needs to be sign-extended to the promoted size
// before shifting. Because the sign-extension is implemented with
// with AShr, it can be combined with the original operation.
Value *Op = Binop->getOperand(0);
Value *ShiftAmount = NULL;
APInt SignShiftAmt = APInt(
getPromotedType(Op->getType())->getIntegerBitWidth(),
getPromotedType(Op->getType())->getIntegerBitWidth() -
Op->getType()->getIntegerBitWidth());
NewInst = BinaryOperator::Create(
Instruction::Shl,
State.getConverted(Op),
ConstantInt::get(getPromotedType(Op->getType()), SignShiftAmt),
State.getConverted(Op)->getName() + ".getsign",
Binop);
if (ConstantInt *C = dyn_cast<ConstantInt>(
State.getConverted(Binop->getOperand(1)))) {
ShiftAmount = ConstantInt::get(getPromotedType(Op->getType()),
SignShiftAmt + C->getValue());
} else {
ShiftAmount = BinaryOperator::Create(
Instruction::Add,
State.getConverted(Binop->getOperand(1)),
ConstantInt::get(
getPromotedType(Binop->getOperand(1)->getType()),
SignShiftAmt),
State.getConverted(Op)->getName() + ".shamt", Binop);
}
NewInst = BinaryOperator::Create(
Instruction::AShr,
NewInst,
ShiftAmount,
Binop->getName() + ".result", Binop);
} else {
// If the original operation is not AShr, just recreate it as usual.
NewInst = BinaryOperator::Create(
Binop->getOpcode(),
State.getConverted(Binop->getOperand(0)),
State.getConverted(Binop->getOperand(1)),
Binop->getName() + ".result", Binop);
if (isa<OverflowingBinaryOperator>(NewInst)) {
cast<BinaryOperator>(NewInst)->setHasNoUnsignedWrap
(Binop->hasNoUnsignedWrap());
cast<BinaryOperator>(NewInst)->setHasNoSignedWrap(
Binop->hasNoSignedWrap());
}
}
// Now restore the invariant if necessary.
// This switch also sanity-checks the operation.
switch (Binop->getOpcode()) {
case Instruction::And:
case Instruction::Or:
case Instruction::Xor:
case Instruction::LShr:
// These won't change the upper bits.
break;
// These can change the upper bits, unless we are sure they never
// overflow. So clear them now.
case Instruction::Add:
case Instruction::Sub:
if (!(Binop->hasNoUnsignedWrap() && Binop->hasNoSignedWrap()))
NewInst = getClearUpper(NewInst, Binop->getType(), Binop);
break;
case Instruction::Shl:
if (!Binop->hasNoUnsignedWrap())
NewInst = getClearUpper(NewInst, Binop->getType(), Binop);
break;
// We modified the upper bits ourselves when implementing AShr
case Instruction::AShr:
NewInst = getClearUpper(NewInst, Binop->getType(), Binop);
break;
// We should not see FP operators here.
// We don't handle mul/div.
case Instruction::FAdd:
case Instruction::FSub:
case Instruction::Mul:
case Instruction::FMul:
case Instruction::UDiv:
case Instruction::SDiv:
case Instruction::FDiv:
case Instruction::URem:
case Instruction::SRem:
case Instruction::FRem:
case Instruction::BinaryOpsEnd:
errs() << *Inst << "\n";
llvm_unreachable("Cannot handle binary operator");
break;
}
State.recordConverted(Binop, NewInst);
} else if (ICmpInst *Cmp = dyn_cast<ICmpInst>(Inst)) {
Value *Op0, *Op1;
// For signed compares, operands are sign-extended to their
// promoted type. For unsigned or equality compares, the comparison
// is equivalent with the larger type because they are already
// zero-extended.
if (Cmp->isSigned()) {
Op0 = getSignExtend(State.getConverted(Cmp->getOperand(0)),
Cmp->getOperand(0),
Cmp);
Op1 = getSignExtend(State.getConverted(Cmp->getOperand(1)),
Cmp->getOperand(1),
Cmp);
} else {
Op0 = State.getConverted(Cmp->getOperand(0));
Op1 = State.getConverted(Cmp->getOperand(1));
}
ICmpInst *NewInst = new ICmpInst(
Cmp, Cmp->getPredicate(), Op0, Op1, "");
State.recordConverted(Cmp, NewInst);
} else if (SelectInst *Select = dyn_cast<SelectInst>(Inst)) {
SelectInst *NewInst = SelectInst::Create(
Select->getCondition(),
State.getConverted(Select->getTrueValue()),
State.getConverted(Select->getFalseValue()),
"", Select);
State.recordConverted(Select, NewInst);
} else if (PHINode *Phi = dyn_cast<PHINode>(Inst)) {
PHINode *NewPhi = PHINode::Create(
getPromotedType(Phi->getType()),
Phi->getNumIncomingValues(),
"", Phi);
for (unsigned I = 0, E = Phi->getNumIncomingValues(); I < E; ++I) {
NewPhi->addIncoming(State.getConverted(Phi->getIncomingValue(I)),
Phi->getIncomingBlock(I));
}
State.recordConverted(Phi, NewPhi);
} else if (SwitchInst *Switch = dyn_cast<SwitchInst>(Inst)) {
SwitchInst *NewInst = SwitchInst::Create(
State.getConverted(Switch->getCondition()),
Switch->getDefaultDest(),
Switch->getNumCases(),
Switch);
for (SwitchInst::CaseIt I = Switch->case_begin(),
E = Switch->case_end();
I != E; ++I) {
// Build a new case from the ranges that map to the successor BB. Each
// range consists of a high and low value which are typed, so the ranges
// must be rebuilt and a new case constructed from them.
IntegersSubset CaseRanges = I.getCaseValueEx();
IntegersSubsetToBB CaseBuilder;
for (unsigned RI = 0, RE = CaseRanges.getNumItems(); RI < RE; ++RI) {
CaseBuilder.add(
IntItem::fromConstantInt(cast<ConstantInt>(convertConstant(
CaseRanges.getItem(RI).getLow().toConstantInt()))),
IntItem::fromConstantInt(cast<ConstantInt>(convertConstant(
CaseRanges.getItem(RI).getHigh().toConstantInt()))));
}
IntegersSubset Case = CaseBuilder.getCase();
NewInst->addCase(Case, I.getCaseSuccessor());
}
Switch->eraseFromParent();
} else {
errs() << *Inst<<"\n";
llvm_unreachable("unhandled instruction");
}
}
