[X86] Add AddToWorklist(N) after calls to SimplifyDemandedBits/Simpli…

…fyDemandedVectorElts that are called on an operand of N.

If a simplication occurs the operand will be added to the worklist.
But since the demanded mask was based on N, we need to make sure
we revisit N in case there are more simplifications to be done.
Returning SDValue(N, 0) as we do, only tells DAG combine that
something changed, but that won't make it add anything to the
worklist.

Found while playing around with using VEXTRACT_STORE in more cases.
But I guess this doesn't affect any of our existing tests.

llvm/lib/Target/X86/X86ISelLowering.cpp

@@ -42126,8 +42126,10 @@ static SDValue combineMaskedStore(SDNode *N, SelectionDAG &DAG,
   SDValue Mask = Mst->getMask();
   if (Mask.getScalarValueSizeInBits() != 1) {
     APInt DemandedBits(APInt::getSignMask(VT.getScalarSizeInBits()));
-    if (TLI.SimplifyDemandedBits(Mask, DemandedBits, DCI))
+    if (TLI.SimplifyDemandedBits(Mask, DemandedBits, DCI)) {
+      DCI.AddToWorklist(N);
       return SDValue(N, 0);
+    }
     if (SDValue NewMask =
             TLI.SimplifyMultipleUseDemandedBits(Mask, DemandedBits, DAG))
       return DAG.getMaskedStore(Mst->getChain(), SDLoc(N), Mst->getValue(),
@@ -42391,8 +42393,10 @@ static SDValue combineVEXTRACT_STORE(SDNode *N, SelectionDAG &DAG,
   APInt KnownUndef, KnownZero;
   const TargetLowering &TLI = DAG.getTargetLoweringInfo();
   if (TLI.SimplifyDemandedVectorElts(StoredVal, DemandedElts, KnownUndef,
-                                     KnownZero, DCI))
+                                     KnownZero, DCI)) {
+    DCI.AddToWorklist(N);
     return SDValue(N, 0);
+  }
 
   return SDValue();
 }
@@ -43738,8 +43742,10 @@ static SDValue combineBT(SDNode *N, SelectionDAG &DAG,
   // BT ignores high bits in the bit index operand.
   unsigned BitWidth = N1.getValueSizeInBits();
   APInt DemandedMask = APInt::getLowBitsSet(BitWidth, Log2_32(BitWidth));
-  if (DAG.getTargetLoweringInfo().SimplifyDemandedBits(N1, DemandedMask, DCI))
+  if (DAG.getTargetLoweringInfo().SimplifyDemandedBits(N1, DemandedMask, DCI)) {
+    DCI.AddToWorklist(N);
     return SDValue(N, 0);
+  }
 
   return SDValue();
 }
@@ -43753,8 +43759,10 @@ static SDValue combineCVTPH2PS(SDNode *N, SelectionDAG &DAG,
     const TargetLowering &TLI = DAG.getTargetLoweringInfo();
     APInt DemandedElts = APInt::getLowBitsSet(8, 4);
     if (TLI.SimplifyDemandedVectorElts(Src, DemandedElts, KnownUndef, KnownZero,
-                                       DCI))
+                                       DCI)) {
+      DCI.AddToWorklist(N);
       return SDValue(N, 0);
+    }
 
     if (ISD::isNormalLoad(Src.getNode()) && Src.hasOneUse()) {
       LoadSDNode *LN = cast<LoadSDNode>(N->getOperand(0));
@@ -44655,8 +44663,10 @@ static SDValue combineX86GatherScatter(SDNode *N, SelectionDAG &DAG,
   if (Mask.getScalarValueSizeInBits() != 1) {
     const TargetLowering &TLI = DAG.getTargetLoweringInfo();
     APInt DemandedMask(APInt::getSignMask(Mask.getScalarValueSizeInBits()));
-    if (TLI.SimplifyDemandedBits(Mask, DemandedMask, DCI))
+    if (TLI.SimplifyDemandedBits(Mask, DemandedMask, DCI)) {
+      DCI.AddToWorklist(N);
       return SDValue(N, 0);
+    }
   }
 
   return SDValue();
@@ -44745,8 +44755,10 @@ static SDValue combineGatherScatter(SDNode *N, SelectionDAG &DAG,
   if (Mask.getScalarValueSizeInBits() != 1) {
     const TargetLowering &TLI = DAG.getTargetLoweringInfo();
     APInt DemandedMask(APInt::getSignMask(Mask.getScalarValueSizeInBits()));
-    if (TLI.SimplifyDemandedBits(Mask, DemandedMask, DCI))
+    if (TLI.SimplifyDemandedBits(Mask, DemandedMask, DCI)) {
+      DCI.AddToWorklist(N);
       return SDValue(N, 0);
+    }
   }
 
   return SDValue();