RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-21 18:22:53 +01:00
Code Issues Projects Releases Wiki Activity

[InstCombine] Fixed select + masked load fold failure

Browse Source 
Fixed type assertion failure caused by trying to fold a masked load with a
select where the select condition is a scalar value

Reviewed By: sdesmalen, lebedev.ri

Differential Revision: https://reviews.llvm.org/D107372

(cherry picked from commit 3943a74666cbe718b74e06092ce3b4c20e85fde1)
This commit is contained in:
Dylan Fleming 2021-08-03 18:40:23 +01:00 committed by Tom Stellard
parent 7563d8dfe5
commit a558de374f
2 changed files with 15 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
lib/Transforms/InstCombine/InstCombineSelect.cpp
View File

@ -3230,7 +3230,8 @@ Instruction *InstCombinerImpl::visitSelectInst(SelectInst &SI) {
Value *Mask;
if (match(TrueVal, m_Zero()) &&
match(FalseVal, m_MaskedLoad(m_Value(), m_Value(), m_Value(Mask),
m_CombineOr(m_Undef(), m_Zero())))) {
m_CombineOr(m_Undef(), m_Zero()))) &&
(CondVal->getType() == Mask->getType())) {
// We can remove the select by ensuring the load zeros all lanes the
// select would have. We determine this by proving there is no overlap
// between the load and select masks.

13
test/Transforms/InstCombine/select-masked_load.ll
View File

@ -94,5 +94,18 @@ define <4 x float> @masked_load_and_zero_inactive_8(<4 x float>* %ptr, <4 x i1>
ret <4 x float> %masked
}
define <8 x float> @masked_load_and_scalar_select_cond(<8 x float>* %ptr, <8 x i1> %mask, i1 %cond) {
; CHECK-LABEL: @masked_load_and_scalar_select_cond(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* [[PTR:%.*]], i32 32, <8 x i1> [[MASK:%.*]], <8 x float> undef)
; CHECK-NEXT: [[TMP1:%.*]] = select i1 [[COND:%.*]], <8 x float> zeroinitializer, <8 x float> [[TMP0]]
; CHECK-NEXT: ret <8 x float> [[TMP1]]
entry:
%0 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* %ptr, i32 32, <8 x i1> %mask, <8 x float> undef)
%1 = select i1 %cond, <8 x float> zeroinitializer, <8 x float> %0
ret <8 x float> %1
}
declare <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>*, i32 immarg, <8 x i1>, <8 x float>)
declare <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>*, i32 immarg, <4 x i1>, <4 x i32>)
declare <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>*, i32 immarg, <4 x i1>, <4 x float>)