RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 20:23:11 +01:00
Code Issues Projects Releases Wiki Activity

[Hexagon] Reconsider getMask fix, return original mask, convert later

Browse Source 
The getPayload/getMask/getPassThrough functions should return values
that could be composed into a masked load/store without any additional
type casts. The previous fix violated that.
Instead, convert scalar mask to a vector right before rescaling.
This commit is contained in:
Krzysztof Parzyszek 2020-12-12 13:19:36 -06:00
parent 313d9ab376
commit 577e1b0232
1 changed files with 13 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
lib/Target/Hexagon/HexagonVectorCombine.cpp
View File

@ -416,8 +416,7 @@ auto AlignVectors::getMask(Value *Val) const -> Value * {
int ElemCount = VecTy->getElementCount().getFixedValue();
return HVC.getFullValue(HVC.getBoolTy(ElemCount));
}
// For scalars, return a vector <1 x i1>.
return HVC.getFullValue(HVC.getBoolTy(1));
return HVC.getFullValue(HVC.getBoolTy());
}
auto AlignVectors::getPassThrough(Value *Val) const -> Value * {
@ -811,14 +810,24 @@ auto AlignVectors::realignGroup(const MoveGroup &Move) const -> bool {
// Stores.
ByteSpan ASpanV, ASpanM;
// Return a vector value corresponding to the input value Val:
// either <1 x Val> for scalar Val, or Val itself for vector Val.
auto MakeVec = [](IRBuilder<> &Builder, Value *Val) -> Value * {
Type *Ty = Val->getType();
if (Ty->isVectorTy())
return Val;
auto *VecTy = VectorType::get(Ty, 1, /*Scalable*/ false);
return Builder.CreateBitCast(Val, VecTy);
};
for (int i = -1; i != NumSectors; ++i) {
ByteSpan Section = VSpan.section(i * ScLen, ScLen).normalize();
Value *AccumV = UndefValue::get(SecTy);
Value *AccumM = HVC.getNullValue(SecTy);
for (ByteSpan::Block &S : Section) {
Value *Pay = getPayload(S.Seg.Val);
Value *Mask = HVC.rescale(Builder, getMask(S.Seg.Val), Pay->getType(),
HVC.getByteTy());
Value *Mask = HVC.rescale(Builder, MakeVec(Builder, getMask(S.Seg.Val)),
Pay->getType(), HVC.getByteTy());
AccumM = HVC.insertb(Builder, AccumM, HVC.vbytes(Builder, Mask),
S.Seg.Start, S.Seg.Size, S.Pos);
AccumV = HVC.insertb(Builder, AccumV, HVC.vbytes(Builder, Pay),