1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-22 18:54:02 +01:00

[NFC] Bail out for scalable vectors before calling getNumElements

Summary:
Move the bail out logic to before constructing the Result and Lane
vectors. This is both potentially faster, and avoids calling
getNumElements on a potentially scalable vector

Reviewers: efriedma, sunfish, chandlerc, c-rhodes, fpetrogalli

Reviewed By: fpetrogalli

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D81619
This commit is contained in:
Christopher Tetreault 2020-06-16 13:19:15 -07:00
parent 650da30b37
commit f0aae37e1d

View File

@ -2689,24 +2689,26 @@ static Constant *ConstantFoldVectorCall(StringRef Name,
const DataLayout &DL, const DataLayout &DL,
const TargetLibraryInfo *TLI, const TargetLibraryInfo *TLI,
const CallBase *Call) { const CallBase *Call) {
SmallVector<Constant *, 4> Result(VTy->getNumElements());
SmallVector<Constant *, 4> Lane(Operands.size());
Type *Ty = VTy->getElementType();
// Do not iterate on scalable vector. The number of elements is unknown at // Do not iterate on scalable vector. The number of elements is unknown at
// compile-time. // compile-time.
if (isa<ScalableVectorType>(VTy)) if (isa<ScalableVectorType>(VTy))
return nullptr; return nullptr;
auto *FVTy = cast<FixedVectorType>(VTy);
SmallVector<Constant *, 4> Result(FVTy->getNumElements());
SmallVector<Constant *, 4> Lane(Operands.size());
Type *Ty = FVTy->getElementType();
if (IntrinsicID == Intrinsic::masked_load) { if (IntrinsicID == Intrinsic::masked_load) {
auto *SrcPtr = Operands[0]; auto *SrcPtr = Operands[0];
auto *Mask = Operands[2]; auto *Mask = Operands[2];
auto *Passthru = Operands[3]; auto *Passthru = Operands[3];
Constant *VecData = ConstantFoldLoadFromConstPtr(SrcPtr, VTy, DL); Constant *VecData = ConstantFoldLoadFromConstPtr(SrcPtr, FVTy, DL);
SmallVector<Constant *, 32> NewElements; SmallVector<Constant *, 32> NewElements;
for (unsigned I = 0, E = VTy->getNumElements(); I != E; ++I) { for (unsigned I = 0, E = FVTy->getNumElements(); I != E; ++I) {
auto *MaskElt = Mask->getAggregateElement(I); auto *MaskElt = Mask->getAggregateElement(I);
if (!MaskElt) if (!MaskElt)
break; break;
@ -2732,12 +2734,12 @@ static Constant *ConstantFoldVectorCall(StringRef Name,
return nullptr; return nullptr;
} }
} }
if (NewElements.size() != VTy->getNumElements()) if (NewElements.size() != FVTy->getNumElements())
return nullptr; return nullptr;
return ConstantVector::get(NewElements); return ConstantVector::get(NewElements);
} }
for (unsigned I = 0, E = VTy->getNumElements(); I != E; ++I) { for (unsigned I = 0, E = FVTy->getNumElements(); I != E; ++I) {
// Gather a column of constants. // Gather a column of constants.
for (unsigned J = 0, JE = Operands.size(); J != JE; ++J) { for (unsigned J = 0, JE = Operands.size(); J != JE; ++J) {
// Some intrinsics use a scalar type for certain arguments. // Some intrinsics use a scalar type for certain arguments.