[X86] canonicalizeShuffleWithBinOps - add X86ISD::PSHUFHW/PSHUFLW handling

Author: RKSimon

llvm/lib/Target/X86/X86ISelLowering.cpp

@@ -38776,6 +38776,8 @@ static SDValue canonicalizeShuffleWithBinOps(SDValue N, SelectionDAG &DAG,
   case X86ISD::VBROADCAST:
   case X86ISD::MOVDDUP:
   case X86ISD::PSHUFD:
+  case X86ISD::PSHUFHW:
+  case X86ISD::PSHUFLW:
   case X86ISD::VPERMI:
   case X86ISD::VPERMILPI: {
     if (N.getOperand(0).getValueType() == ShuffleVT &&

llvm/test/CodeGen/X86/vector-fshr-rot-128.ll

@@ -1066,9 +1066,9 @@ define <8 x i16> @splatvar_funnnel_v8i16(<8 x i16> %x, <8 x i16> %amt) nounwind
 ;
 ; XOPAVX1-LABEL: splatvar_funnnel_v8i16:
 ; XOPAVX1:       # %bb.0:
+; XOPAVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
 ; XOPAVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; XOPAVX1-NEXT:    vpsubw %xmm1, %xmm2, %xmm1
-; XOPAVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
 ; XOPAVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
 ; XOPAVX1-NEXT:    vprotw %xmm1, %xmm0, %xmm0
 ; XOPAVX1-NEXT:    retq

llvm/test/CodeGen/X86/vector-fshr-rot-256.ll

@@ -881,9 +881,9 @@ define <16 x i16> @splatvar_funnnel_v16i16(<16 x i16> %x, <16 x i16> %amt) nounw
 ;
 ; XOPAVX1-LABEL: splatvar_funnnel_v16i16:
 ; XOPAVX1:       # %bb.0:
+; XOPAVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
 ; XOPAVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; XOPAVX1-NEXT:    vpsubw %xmm1, %xmm2, %xmm1
-; XOPAVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
 ; XOPAVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
 ; XOPAVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; XOPAVX1-NEXT:    vprotw %xmm1, %xmm2, %xmm2

llvm/test/CodeGen/X86/vector-shuffle-sse4a.ll

@@ -364,8 +364,8 @@ define <16 x i8> @shuffle_8_18_uuuuuuuuuuuuuu(<16 x i8> %a, <16 x i8> %b) {
 ; AMD10H:       # %bb.0:
 ; AMD10H-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; AMD10H-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; AMD10H-NEXT:    andps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; AMD10H-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,1,2,3,4,5,6,7]
+; AMD10H-NEXT:    pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; AMD10H-NEXT:    packuswb %xmm0, %xmm0
 ; AMD10H-NEXT:    retq
 ;