#include "llvm/IR/DataLayout.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PatternMatch.h"
+#include "llvm/IR/Statepoint.h"
#include "llvm/Transforms/Utils/BuildLibCalls.h"
#include "llvm/Transforms/Utils/Local.h"
using namespace llvm;
// If the memcpy has metadata describing the members, see if we can
// get the TBAA tag describing our copy.
if (MDNode *M = MI->getMetadata(LLVMContext::MD_tbaa_struct)) {
- if (M->getNumOperands() == 3 &&
- M->getOperand(0) &&
- isa<ConstantInt>(M->getOperand(0)) &&
- cast<ConstantInt>(M->getOperand(0))->isNullValue() &&
+ if (M->getNumOperands() == 3 && M->getOperand(0) &&
+ mdconst::hasa<ConstantInt>(M->getOperand(0)) &&
+ mdconst::extract<ConstantInt>(M->getOperand(0))->isNullValue() &&
M->getOperand(1) &&
- isa<ConstantInt>(M->getOperand(1)) &&
- cast<ConstantInt>(M->getOperand(1))->getValue() == Size &&
- M->getOperand(2) &&
- isa<MDNode>(M->getOperand(2)))
+ mdconst::hasa<ConstantInt>(M->getOperand(1)) &&
+ mdconst::extract<ConstantInt>(M->getOperand(1))->getValue() ==
+ Size &&
+ M->getOperand(2) && isa<MDNode>(M->getOperand(2)))
CopyMD = cast<MDNode>(M->getOperand(2));
}
}
if (LHSKnownNegative && RHSKnownNegative) {
// The sign bit is set in both cases: this MUST overflow.
// Create a simple add instruction, and insert it into the struct.
- Value *Add = Builder->CreateAdd(LHS, RHS);
- Add->takeName(&CI);
- Constant *V[] = {
- UndefValue::get(LHS->getType()),
- ConstantInt::getTrue(II->getContext())
- };
- StructType *ST = cast<StructType>(II->getType());
- Constant *Struct = ConstantStruct::get(ST, V);
- return InsertValueInst::Create(Struct, Add, 0);
+ return CreateOverflowTuple(II, Builder->CreateAdd(LHS, RHS), true,
+ /*ReUseName*/true);
}
if (LHSKnownPositive && RHSKnownPositive) {
// The sign bit is clear in both cases: this CANNOT overflow.
// Create a simple add instruction, and insert it into the struct.
- Value *Add = Builder->CreateNUWAdd(LHS, RHS);
- Add->takeName(&CI);
- Constant *V[] = {
- UndefValue::get(LHS->getType()),
- ConstantInt::getFalse(II->getContext())
- };
- StructType *ST = cast<StructType>(II->getType());
- Constant *Struct = ConstantStruct::get(ST, V);
- return InsertValueInst::Create(Struct, Add, 0);
+ return CreateOverflowTuple(II, Builder->CreateNUWAdd(LHS, RHS), false);
}
}
}
if (ConstantInt *RHS = dyn_cast<ConstantInt>(II->getArgOperand(1))) {
// X + 0 -> {X, false}
if (RHS->isZero()) {
- Constant *V[] = {
- UndefValue::get(II->getArgOperand(0)->getType()),
- ConstantInt::getFalse(II->getContext())
- };
- Constant *Struct =
- ConstantStruct::get(cast<StructType>(II->getType()), V);
- return InsertValueInst::Create(Struct, II->getArgOperand(0), 0);
+ return CreateOverflowTuple(II, II->getArgOperand(0), false,
+ /*ReUseName*/false);
}
}
if (II->getIntrinsicID() == Intrinsic::sadd_with_overflow) {
Value *LHS = II->getArgOperand(0), *RHS = II->getArgOperand(1);
if (WillNotOverflowSignedAdd(LHS, RHS, II)) {
- Value *Add = Builder->CreateNSWAdd(LHS, RHS);
- Add->takeName(&CI);
- Constant *V[] = {UndefValue::get(Add->getType()), Builder->getFalse()};
- StructType *ST = cast<StructType>(II->getType());
- Constant *Struct = ConstantStruct::get(ST, V);
- return InsertValueInst::Create(Struct, Add, 0);
+ return CreateOverflowTuple(II, Builder->CreateNSWAdd(LHS, RHS), false);
}
}
break;
case Intrinsic::usub_with_overflow:
- case Intrinsic::ssub_with_overflow:
+ case Intrinsic::ssub_with_overflow: {
+ Value *LHS = II->getArgOperand(0), *RHS = II->getArgOperand(1);
// undef - X -> undef
// X - undef -> undef
- if (isa<UndefValue>(II->getArgOperand(0)) ||
- isa<UndefValue>(II->getArgOperand(1)))
+ if (isa<UndefValue>(LHS) || isa<UndefValue>(RHS))
return ReplaceInstUsesWith(CI, UndefValue::get(II->getType()));
- if (ConstantInt *RHS = dyn_cast<ConstantInt>(II->getArgOperand(1))) {
+ if (ConstantInt *ConstRHS = dyn_cast<ConstantInt>(RHS)) {
// X - 0 -> {X, false}
- if (RHS->isZero()) {
- Constant *V[] = {
- UndefValue::get(II->getArgOperand(0)->getType()),
- ConstantInt::getFalse(II->getContext())
- };
- Constant *Struct =
- ConstantStruct::get(cast<StructType>(II->getType()), V);
- return InsertValueInst::Create(Struct, II->getArgOperand(0), 0);
+ if (ConstRHS->isZero()) {
+ return CreateOverflowTuple(II, LHS, false, /*ReUseName*/false);
+ }
+ }
+ if (II->getIntrinsicID() == Intrinsic::ssub_with_overflow) {
+ if (WillNotOverflowSignedSub(LHS, RHS, II)) {
+ return CreateOverflowTuple(II, Builder->CreateNSWSub(LHS, RHS), false);
+ }
+ } else {
+ if (WillNotOverflowUnsignedSub(LHS, RHS, II)) {
+ return CreateOverflowTuple(II, Builder->CreateNUWSub(LHS, RHS), false);
}
}
break;
+ }
case Intrinsic::umul_with_overflow: {
Value *LHS = II->getArgOperand(0), *RHS = II->getArgOperand(1);
unsigned BitWidth = cast<IntegerType>(LHS->getType())->getBitWidth();
bool Overflow;
LHSMax.umul_ov(RHSMax, Overflow);
if (!Overflow) {
- Value *Mul = Builder->CreateNUWMul(LHS, RHS, "umul_with_overflow");
- Constant *V[] = {
- UndefValue::get(LHS->getType()),
- Builder->getFalse()
- };
- Constant *Struct = ConstantStruct::get(cast<StructType>(II->getType()),V);
- return InsertValueInst::Create(Struct, Mul, 0);
+ return CreateOverflowTuple(II, Builder->CreateNUWMul(LHS, RHS), false);
}
} // FALL THROUGH
case Intrinsic::smul_with_overflow:
// X * 1 -> {X, false}
if (RHSI->equalsInt(1)) {
- Constant *V[] = {
- UndefValue::get(II->getArgOperand(0)->getType()),
- ConstantInt::getFalse(II->getContext())
- };
- Constant *Struct =
- ConstantStruct::get(cast<StructType>(II->getType()), V);
- return InsertValueInst::Create(Struct, II->getArgOperand(0), 0);
+ return CreateOverflowTuple(II, II->getArgOperand(0), false,
+ /*ReUseName*/false);
+ }
+ }
+ if (II->getIntrinsicID() == Intrinsic::smul_with_overflow) {
+ Value *LHS = II->getArgOperand(0), *RHS = II->getArgOperand(1);
+ if (WillNotOverflowSignedMul(LHS, RHS, II)) {
+ return CreateOverflowTuple(II, Builder->CreateNSWMul(LHS, RHS), false);
}
}
break;
return new LoadInst(Ptr);
}
break;
+ case Intrinsic::ppc_vsx_lxvw4x:
+ case Intrinsic::ppc_vsx_lxvd2x: {
+ // Turn PPC VSX loads into normal loads.
+ Value *Ptr = Builder->CreateBitCast(II->getArgOperand(0),
+ PointerType::getUnqual(II->getType()));
+ return new LoadInst(Ptr, Twine(""), false, 1);
+ }
case Intrinsic::ppc_altivec_stvx:
case Intrinsic::ppc_altivec_stvxl:
// Turn stvx -> store if the pointer is known aligned.
return new StoreInst(II->getArgOperand(0), Ptr);
}
break;
+ case Intrinsic::ppc_vsx_stxvw4x:
+ case Intrinsic::ppc_vsx_stxvd2x: {
+ // Turn PPC VSX stores into normal stores.
+ Type *OpPtrTy = PointerType::getUnqual(II->getArgOperand(0)->getType());
+ Value *Ptr = Builder->CreateBitCast(II->getArgOperand(1), OpPtrTy);
+ return new StoreInst(II->getArgOperand(0), Ptr, false, 1);
+ }
case Intrinsic::x86_sse_storeu_ps:
case Intrinsic::x86_sse2_storeu_pd:
case Intrinsic::x86_sse2_storeu_dq:
// TODO: eventually we should lower this intrinsic to IR
if (auto CIWidth = dyn_cast<ConstantInt>(II->getArgOperand(2))) {
if (auto CIStart = dyn_cast<ConstantInt>(II->getArgOperand(3))) {
- if (CIWidth->equalsInt(64) && CIStart->isZero()) {
+ unsigned Index = CIStart->getZExtValue();
+ // From AMD documentation: "a value of zero in the field length is
+ // defined as length of 64".
+ unsigned Length = CIWidth->equalsInt(0) ? 64 : CIWidth->getZExtValue();
+
+ // From AMD documentation: "If the sum of the bit index + length field
+ // is greater than 64, the results are undefined".
+
+ // Note that both field index and field length are 8-bit quantities.
+ // Since variables 'Index' and 'Length' are unsigned values
+ // obtained from zero-extending field index and field length
+ // respectively, their sum should never wrap around.
+ if ((Index + Length) > 64)
+ return ReplaceInstUsesWith(CI, UndefValue::get(II->getType()));
+
+ if (Length == 64 && Index == 0) {
Value *Vec = II->getArgOperand(1);
Value *Undef = UndefValue::get(Vec->getType());
const uint32_t Mask[] = { 0, 2 };
return EraseInstFromFunction(*II);
}
+ // assume( (load addr) != null ) -> add 'nonnull' metadata to load
+ // (if assume is valid at the load)
+ if (ICmpInst* ICmp = dyn_cast<ICmpInst>(IIOperand)) {
+ Value *LHS = ICmp->getOperand(0);
+ Value *RHS = ICmp->getOperand(1);
+ if (ICmpInst::ICMP_NE == ICmp->getPredicate() &&
+ isa<LoadInst>(LHS) &&
+ isa<Constant>(RHS) &&
+ RHS->getType()->isPointerTy() &&
+ cast<Constant>(RHS)->isNullValue()) {
+ LoadInst* LI = cast<LoadInst>(LHS);
+ if (isValidAssumeForContext(II, LI, DL, DT)) {
+ MDNode *MD = MDNode::get(II->getContext(), None);
+ LI->setMetadata(LLVMContext::MD_nonnull, MD);
+ return EraseInstFromFunction(*II);
+ }
+ }
+ // TODO: apply nonnull return attributes to calls and invokes
+ // TODO: apply range metadata for range check patterns?
+ }
// If there is a dominating assume with the same condition as this one,
// then this one is redundant, and should be removed.
APInt KnownZero(1, 0), KnownOne(1, 0);
if (!CI->isLosslessCast())
return false;
+ // If this is a GC intrinsic, avoid munging types. We need types for
+ // statepoint reconstruction in SelectionDAG.
+ // TODO: This is probably something which should be expanded to all
+ // intrinsics since the entire point of intrinsics is that
+ // they are understandable by the optimizer.
+ if (isStatepoint(CS) || isGCRelocate(CS) || isGCResult(CS))
+ return false;
+
// The size of ByVal or InAlloca arguments is derived from the type, so we
// can't change to a type with a different size. If the size were
// passed explicitly we could avoid this check.
projects / oota-llvm.git / blobdiff