diff options
| author | Chandler Carruth <chandlerc@gmail.com> | 2014-06-28 05:46:28 +0000 |
|---|---|---|
| committer | Chandler Carruth <chandlerc@gmail.com> | 2014-06-28 05:46:28 +0000 |
| commit | bd0717d7cc9c3a6c028ea6f5e7e7fbabe74c59d2 (patch) | |
| tree | f021b1267cdbea8b31694e0f7e32a62cf77c8c1f | |
| parent | d5821f36d94523ac7711b983fc6b433e954497e0 (diff) | |
| download | bcm5719-llvm-bd0717d7cc9c3a6c028ea6f5e7e7fbabe74c59d2.tar.gz bcm5719-llvm-bd0717d7cc9c3a6c028ea6f5e7e7fbabe74c59d2.zip | |
[x86] Fix a bug in the v8i16 shuffling exposed by the new splat-like
lowering for v16i8.
ASan and some bots caught this bug with existing test cases. Fixing it
even fixed a miscompile with one of the test cases. I'm still a bit
suspicious of this test case as I've not taken a proper amount of time
to think about it, but the fix here is strict goodness.
llvm-svn: 211976
| -rw-r--r-- | llvm/lib/Target/X86/X86ISelLowering.cpp | 2 | ||||
| -rw-r--r-- | llvm/test/CodeGen/X86/vector-shuffle-128-v16.ll | 7 |
2 files changed, 5 insertions, 4 deletions
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp index 8a00d6e2bc8..8a12c03e072 100644 --- a/llvm/lib/Target/X86/X86ISelLowering.cpp +++ b/llvm/lib/Target/X86/X86ISelLowering.cpp @@ -7535,7 +7535,7 @@ static SDValue lowerV8I16BasicBlendVectorShuffle(SDLoc DL, SDValue V1, } else { // Otherwise pin the low inputs. for (int GoodInput : GoodInputs) - MoveMask[Mask[GoodInput]] = Mask[GoodInput] - MaskOffset; + MoveMask[Mask[GoodInput] - MaskOffset] = Mask[GoodInput] - MaskOffset; } int MoveMaskIdx = diff --git a/llvm/test/CodeGen/X86/vector-shuffle-128-v16.ll b/llvm/test/CodeGen/X86/vector-shuffle-128-v16.ll index f3a0931f731..f392a6221c5 100644 --- a/llvm/test/CodeGen/X86/vector-shuffle-128-v16.ll +++ b/llvm/test/CodeGen/X86/vector-shuffle-128-v16.ll @@ -63,14 +63,15 @@ define <16 x i8> @shuffle_v16i8_00_00_00_00_04_04_04_04_08_08_08_08_12_12_12_12( ; CHECK-SSE2-NEXT: pxor %xmm1, %xmm1 ; CHECK-SSE2-NEXT: movdqa %xmm0, %xmm2 ; CHECK-SSE2-NEXT: punpcklbw %xmm1, %xmm2 -; CHECK-SSE2-NEXT: pshufd {{.*}} # xmm2 = xmm2[2,1,2,3] +; CHECK-SSE2-NEXT: pshufd {{.*}} # xmm2 = xmm2[0,2,2,3] +; CHECK-SSE2-NEXT: pshuflw {{.*}} # xmm2 = xmm2[0,2,2,3,4,5,6,7] ; CHECK-SSE2-NEXT: punpckhbw %xmm1, %xmm0 ; CHECK-SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3] ; CHECK-SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,3,4,5,6,7] ; CHECK-SSE2-NEXT: punpcklwd %xmm2, %xmm0 -; CHECK-SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,3] +; CHECK-SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,2,1] ; CHECK-SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[1,0,2,3,4,5,6,7] -; CHECK-SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,5,5,6,7] +; CHECK-SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,5,6,7] ; CHECK-SSE2-NEXT: packuswb %xmm0, %xmm0 ; CHECK-SSE2-NEXT: punpcklbw %xmm0, %xmm0 ; CHECK-SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,0,1] |
