[X86][SSE] Fixed PALIGNR target shuffle decode

The PALIGNR target shuffle decode was not taking into account that DecodePALIGNRMask (rather oddly) expects the operands to be in reverse order, nor was it detecting unary patterns, causing combines to combine with the incorrect input.

The cgbuiltin, auto upgrade and instruction comments code correctly swap the operands so are not affected.

llvm-svn: 278494

diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 4c913bf..694af1d 100644 (file)
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -4914,6 +4914,9 @@ static bool getTargetShuffleMask(SDNode *N, MVT VT, bool AllowSentinelZero,
     assert(VT.getScalarType() == MVT::i8 && "Byte vector expected");
     ImmN = N->getOperand(N->getNumOperands()-1);
     DecodePALIGNRMask(VT, cast<ConstantSDNode>(ImmN)->getZExtValue(), Mask);
+    IsUnary = IsFakeUnary = N->getOperand(0) == N->getOperand(1);
+    Ops.push_back(N->getOperand(1));
+    Ops.push_back(N->getOperand(0));
     break;
   case X86ISD::VSHLDQ:
     assert(VT.getScalarType() == MVT::i8 && "Byte vector expected");

diff --git a/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll b/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
index 7335ddd..2a27483 100644 (file)
--- a/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
@@ -142,12 +142,12 @@ define <4 x float> @combine_pshufb_movsldup(<4 x float> %a0) {
 define <16 x i8> @combine_pshufb_palignr(<16 x i8> %a0, <16 x i8> %a1) {
 ; SSE-LABEL: combine_pshufb_palignr:
 ; SSE:       # BB#0:
-; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,3,2,3]
+; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_pshufb_palignr:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm1[2,3,2,3]
+; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
 ; AVX-NEXT:    retq
   %1 = shufflevector <16 x i8> %a0, <16 x i8> %a1, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
   %2 = tail call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %1, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7>)