[SelectionDAG] Allow vselect in foldBinOpIntoSelect

It seems vselect was also meant to be an option given the comment and the fact vectors are allowed and the kind is checked too.

Author: AZero13

llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

@@ -2490,7 +2490,8 @@ SDValue DAGCombiner::foldBinOpIntoSelect(SDNode *BO) {
   unsigned SelOpNo = 0;
   SDValue Sel = BO->getOperand(0);
   auto BinOpcode = BO->getOpcode();
-  if (Sel.getOpcode() != ISD::SELECT || !Sel.hasOneUse()) {
+  if ((Sel.getOpcode() != ISD::SELECT && Sel.getOpcode() != ISD::VSELECT) ||
+      !Sel.hasOneUse()) {
     SelOpNo = 1;
     Sel = BO->getOperand(1);
 
@@ -2506,7 +2507,8 @@ SDValue DAGCombiner::foldBinOpIntoSelect(SDNode *BO) {
     }
   }
 
-  if (Sel.getOpcode() != ISD::SELECT || !Sel.hasOneUse())
+  if ((Sel.getOpcode() != ISD::SELECT && Sel.getOpcode() != ISD::VSELECT) ||
+      !Sel.hasOneUse())
     return SDValue();
 
   SDValue CT = Sel.getOperand(1);

llvm/test/CodeGen/X86/extract-vselect-setcc.ll

@@ -5,11 +5,11 @@ define void @PR117684(i1 %cond, <8 x float> %vec, ptr %ptr1, ptr %ptr2) #0 {
 ; CHECK-LABEL: PR117684:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; CHECK-NEXT:    vcmpnltss %xmm1, %xmm0, %k1
-; CHECK-NEXT:    vbroadcastss {{.*#+}} xmm0 = [NaN,NaN,NaN,NaN]
-; CHECK-NEXT:    vinsertf32x4 $0, %xmm0, %ymm0, %ymm0 {%k1} {z}
-; CHECK-NEXT:    vmulss %xmm1, %xmm0, %xmm0
-; CHECK-NEXT:    vmulss {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm2
+; CHECK-NEXT:    vmovss {{.*#+}} xmm2 = [NaN,0.0E+0,0.0E+0,0.0E+0]
+; CHECK-NEXT:    vcmpltss %xmm1, %xmm0, %k1
+; CHECK-NEXT:    vmovaps %xmm2, %xmm0
+; CHECK-NEXT:    vmovss %xmm1, %xmm0, %xmm0 {%k1}
+; CHECK-NEXT:    vmulss %xmm2, %xmm0, %xmm2
 ; CHECK-NEXT:    vbroadcastss %xmm2, %ymm2
 ; CHECK-NEXT:    testb $1, %dil
 ; CHECK-NEXT:    cmoveq %rdx, %rsi