[X86] getScalarMaskingNode - if the mask is zero just return the blended passthrough and preserved source value (#153575)

We already handle the case if the mask is one, so I added the other case where the op is replaced with a MOVSH/S/D blend.

This assumes the scalar passthrough is op0.

I had to adjust the test case for #98306 as AFAICT it'd been over
reduced

Fixes #153570

1 files changed, 14 insertions, 4 deletions

diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index dd4c608..572cfda 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -26269,10 +26269,9 @@ static SDValue getScalarMaskingNode(SDValue Op, SDValue Mask,
                                     SDValue PreservedSrc,
                                     const X86Subtarget &Subtarget,
                                     SelectionDAG &DAG) {
-
-  if (auto *MaskConst = dyn_cast<ConstantSDNode>(Mask))
-    if (MaskConst->getZExtValue() & 0x1)
-      return Op;
+  auto *MaskConst = dyn_cast<ConstantSDNode>(Mask);
+  if (MaskConst && (MaskConst->getZExtValue() & 0x1))
+    return Op;
 
   MVT VT = Op.getSimpleValueType();
   SDLoc dl(Op);
@@ -26288,6 +26287,17 @@ static SDValue getScalarMaskingNode(SDValue Op, SDValue Mask,
 
   if (PreservedSrc.isUndef())
     PreservedSrc = getZeroVector(VT, Subtarget, DAG, dl);
+
+  if (MaskConst) {
+    assert((MaskConst->getZExtValue() & 0x1) == 0 && "Expected false mask");
+    // Discard op and blend passthrough with scalar op src/dst.
+    SmallVector<int, 16> ShuffleMask(VT.getVectorNumElements());
+    std::iota(ShuffleMask.begin(), ShuffleMask.end(), 0);
+    ShuffleMask[0] = VT.getVectorNumElements();
+    return DAG.getVectorShuffle(VT, dl, Op.getOperand(0), PreservedSrc,
+                                ShuffleMask);
+  }
+
   return DAG.getNode(X86ISD::SELECTS, dl, VT, IMask, Op, PreservedSrc);
 }