[llvm] r267346 - [X86][XOP] Fixed VPPERM permute op decoding (PR27472).

[Simon Pilgrim via llvm-commits]

Author: rksimon
Date: Sun Apr 24 10:05:04 2016
New Revision: 267346

URL: http://llvm.org/viewvc/llvm-project?rev=267346&view=rev
Log:
[X86][XOP] Fixed VPPERM permute op decoding (PR27472).

Fixed issue with VPPERM target shuffle mask decoding that was incorrectly masking off the 3-bit permute op with a 2-bit mask.

Modified:
    llvm/trunk/lib/Target/X86/Utils/X86ShuffleDecode.cpp
    llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-xop.ll

Modified: llvm/trunk/lib/Target/X86/Utils/X86ShuffleDecode.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/Utils/X86ShuffleDecode.cpp?rev=267346&r1=267345&r2=267346&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/Utils/X86ShuffleDecode.cpp (original)
+++ llvm/trunk/lib/Target/X86/Utils/X86ShuffleDecode.cpp Sun Apr 24 10:05:04 2016
@@ -367,7 +367,7 @@ void DecodeVPPERMMask(ArrayRef<uint64_t>
       continue;
     }
 
-    uint64_t PermuteOp = (M >> 5) & 0x3;
+    uint64_t PermuteOp = (M >> 5) & 0x7;
     if (PermuteOp == 4) {
       ShuffleMask.push_back(SM_SentinelZero);
       continue;

Modified: llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-xop.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-xop.ll?rev=267346&r1=267345&r2=267346&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-xop.ll (original)
+++ llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-xop.ll Sun Apr 24 10:05:04 2016
@@ -23,7 +23,7 @@ define <16 x i8> @combine_vpperm_identit
 define <16 x i8> @combine_vpperm_zero(<16 x i8> %a0, <16 x i8> %a1) {
 ; CHECK-LABEL: combine_vpperm_zero:
 ; CHECK:       # BB#0:
-; CHECK-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vxorps %xmm0, %xmm0, %xmm0
 ; CHECK-NEXT:    retq
   %res0 = call <16 x i8> @llvm.x86.xop.vpperm(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> <i8 128, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>)
   %res1 = call <16 x i8> @llvm.x86.xop.vpperm(<16 x i8> %res0, <16 x i8> undef, <16 x i8> <i8 0, i8 128, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>)