RPCS3/llvm
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm.git synced 2025-04-02 21:31:41 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[x86] Fix wrong shuffle mask in test 'combine-vec-shuffle-3.ll'. No functional change.

Browse Source 
Function @test3c should check that the DAGCombiner is able to fold a pair of
shuffles into a new shuffle with a permute mask of <6,7,2,3>. However, one of
the shuffles in @test3c had a wrong permute mask; this prevented the DAGCombiner
from folding the shuffles into the expected result.
Now that the shuffle mask is fixed, the backend correctly folds the two shuffles
in function @test3c into a single movhlps instruction.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213451 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Andrea Di Biagio 2014-07-19 07:52:58 +00:00
parent 07adbf3a87
commit 5bc21c3b57
1 changed files with 3 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
test/CodeGen/X86/combine-vec-shuffle-3.ll
View File

@ -273,19 +273,14 @@ define <4 x i8> @test2c(<4 x i8>* %a, <4 x i8>* %b) {
define <4 x i8> @test3c(<4 x i8>* %a, <4 x i8>* %b) {
%A = load <4 x i8>* %a
%B = load <4 x i8>* %b
%1 = shufflevector <4 x i8> %A, <4 x i8> %B, <4 x i32> <i32 0, i32 1, i32 5, i32 5>
%1 = shufflevector <4 x i8> %A, <4 x i8> %B, <4 x i32> <i32 2, i32 3, i32 5, i32 5>
%2 = shufflevector <4 x i8> %1, <4 x i8> %B, <4 x i32> <i32 6, i32 7, i32 0, i32 1>
ret <4 x i8> %2
}
; FIXME: this should be lowered as a single movhlps. However, the backend
; wrongly thinks that shuffle mask [6,7,2,3] is not legal. Therefore, we end up
; with a sub-optimal sequence of 'shufps+palignr'.
; CHECK-LABEL: test3c
; Mask: [6,7,2,3]
; CHECK: shufps $84
; CHECK: palignr $8
; CHECK: ret
; CHECK: movhlps
; CHECK-NEXT: ret
define <4 x i8> @test4c(<4 x i8>* %a, <4 x i8>* %b) {
%A = load <4 x i8>* %a