[X86][XOP] Add SimplifyDemandedVectorElts handling for xop shifts

Noticed while investigating how to improve funnel shift codegen

Author: RKSimon

llvm/lib/Target/X86/X86ISelLowering.cpp

@@ -40082,6 +40082,21 @@ bool X86TargetLowering::SimplifyDemandedVectorEltsForTargetNode(
             Op, TLO.DAG.getNode(Opc, SDLoc(Op), VT, NewSrc, Op.getOperand(1)));
     break;
   }
+  case X86ISD::VPSHA:
+  case X86ISD::VPSHL: {
+    APInt LHSUndef, LHSZero;
+    APInt RHSUndef, RHSZero;
+    SDValue LHS = Op.getOperand(0);
+    SDValue RHS = Op.getOperand(1);
+    if (SimplifyDemandedVectorElts(LHS, DemandedElts, LHSUndef, LHSZero, TLO,
+                                   Depth + 1))
+      return true;
+    if (SimplifyDemandedVectorElts(RHS, DemandedElts, RHSUndef, RHSZero, TLO,
+                                   Depth + 1))
+      return true;
+    KnownZero = LHSZero;
+    break;
+  }
   case X86ISD::KSHIFTL: {
     SDValue Src = Op.getOperand(0);
     auto *Amt = cast<ConstantSDNode>(Op.getOperand(1));

llvm/test/CodeGen/X86/combine-udiv.ll

@@ -675,10 +675,11 @@ define <16 x i8> @combine_vec_udiv_nonuniform4(<16 x i8> %x) {
 ;
 ; XOP-LABEL: combine_vec_udiv_nonuniform4:
 ; XOP:       # %bb.0:
-; XOP-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; XOP-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
-; XOP-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; XOP-NEXT:    vpperm {{.*#+}} xmm1 = xmm1[1,3,5,7,9,11,13,15],xmm2[1,3,5,7,9,11,13,15]
+; XOP-NEXT:    movl $171, %eax
+; XOP-NEXT:    vmovd %eax, %xmm1
+; XOP-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; XOP-NEXT:    vpmullw %xmm1, %xmm2, %xmm1
+; XOP-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; XOP-NEXT:    movl $249, %eax
 ; XOP-NEXT:    vmovd %eax, %xmm2
 ; XOP-NEXT:    vpshlb %xmm2, %xmm1, %xmm1

llvm/test/CodeGen/X86/xop-shifts.ll

@@ -4,7 +4,6 @@
 define <16 x i8> @demandedelts_vpshab(<16 x i8> %a0, <16 x i8> %a1) {
 ; CHECK-LABEL: demandedelts_vpshab:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
 ; CHECK-NEXT:    vpshab %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; CHECK-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
@@ -18,7 +17,6 @@ define <16 x i8> @demandedelts_vpshab(<16 x i8> %a0, <16 x i8> %a1) {
 define <4 x i32> @demandedelts_vpshld(<4 x i32> %a0, <4 x i32> %a1) {
 ; CHECK-LABEL: demandedelts_vpshld:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
 ; CHECK-NEXT:    vpshld %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
 ; CHECK-NEXT:    retq