RPCSX/llvm
1
0
Fork 0
mirror of https://github.com/RPCSX/llvm.git synced 2024-11-28 14:10:41 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[InstCombine] avoid infinite loop from shuffle-extract-insert sequence (PR30923)

Browse Source 
Removing the limitation in visitInsertElementInst() causes several regressions
because we're not prepared to fold sequences of shuffles or inserts and extracts
separated by shuffles. Fixing that appears to be a difficult mission because we
are purposely trying to avoid creating shuffles with arbitrary shuffle masks
because some targets may choke on those.

https://llvm.org/bugs/show_bug.cgi?id=30923


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@286423 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Sanjay Patel 2016-11-10 00:15:14 +00:00
parent 027f4d03c6
commit 9c5e4bac4a
2 changed files with 35 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
lib/Transforms/InstCombine/InstCombineVectorOps.cpp
View File

@ -413,6 +413,14 @@ static void replaceExtractElements(InsertElementInst *InsElt,
if (InsertionBlock != InsElt->getParent()) if (InsertionBlock != InsElt->getParent())
return; return;
// TODO: This restriction matches the check in visitInsertElementInst() and
// prevents an infinite loop caused by not turning the extract/insert pair
// into a shuffle. We really should not need either check, but we're lacking
// folds for shufflevectors because we're afraid to generate shuffle masks
// that the backend can't handle.
if (InsElt->hasOneUse() && isa<InsertElementInst>(InsElt->user_back()))
return;
auto *WideVec = new ShuffleVectorInst(ExtVecOp, UndefValue::get(ExtVecType), auto *WideVec = new ShuffleVectorInst(ExtVecOp, UndefValue::get(ExtVecType),
ConstantVector::get(ExtendMask)); ConstantVector::get(ExtendMask));

27
test/Transforms/InstCombine/insert-extract-shuffle.ll
View File

@ -237,3 +237,30 @@ end:
ret double %mu ret double %mu
} }
; https://llvm.org/bugs/show_bug.cgi?id=30923
; Delete the widening shuffle if we're not going to reduce the extract/insert to a shuffle.
define <4 x float> @PR30923(<2 x float> %x) {
; CHECK-LABEL: @PR30923(
; CHECK-NEXT: bb1:
; CHECK-NEXT: [[EXT1:%.*]] = extractelement <2 x float> %x, i32 1
; CHECK-NEXT: store float [[EXT1]], float* undef, align 4
; CHECK-NEXT: br label %bb2
; CHECK: bb2:
; CHECK-NEXT: [[EXT2:%.*]] = extractelement <2 x float> %x, i32 0
; CHECK-NEXT: [[INS1:%.*]] = insertelement <4 x float> <float 0.000000e+00, float 0.000000e+00, float undef, float undef>, float [[EXT2]], i32 2
; CHECK-NEXT: [[INS2:%.*]] = insertelement <4 x float> [[INS1]], float [[EXT1]], i32 3
; CHECK-NEXT: ret <4 x float> [[INS2]]
;
bb1:
%ext1 = extractelement <2 x float> %x, i32 1
store float %ext1, float* undef, align 4
br label %bb2
bb2:
%widen = shufflevector <2 x float> %x, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
%ext2 = extractelement <4 x float> %widen, i32 0
%ins1 = insertelement <4 x float> <float 0.0, float 0.0, float undef, float undef>, float %ext2, i32 2
%ins2 = insertelement <4 x float> %ins1, float %ext1, i32 3
ret <4 x float> %ins2
}