[llvm] [PowerPC] Fix vector_shuffle combines when inputs are scalar_to_vector of differing types. (PR #80784)

[Amy Kwan via llvm-commits]

================
@@ -15771,33 +15808,33 @@ SDValue PPCTargetLowering::combineVectorShuffle(ShuffleVectorSDNode *SVN,
     // the value into element zero. Since scalar size of LHS and RHS may differ
     // after isScalarToVec, this should be checked using their own sizes.
     if (SToVLHS) {
-      if (!IsLittleEndian && SToVLHS.getValueType().getScalarSizeInBits() >= 64)
+      int LHSScalarSize = SToVLHS.getValueType().getScalarSizeInBits();
+      if (!IsLittleEndian && LHSScalarSize >= 64)
         return Res;
-      // Set up the values for the shuffle vector fixup.
-      LHSMaxIdx = NumEltsOut / NumEltsIn;
-      SToVLHS = getSToVPermuted(SToVLHS, DAG, Subtarget);
-      if (SToVLHS.getValueType() != LHS.getValueType())
-        SToVLHS = DAG.getBitcast(LHS.getValueType(), SToVLHS);
-      LHS = SToVLHS;
+      LHS = generateSToVPermutedForVecShuffle(
+          LHSScalarSize, ShuffleEltWidth, LHSNumValidElts, LHSFirstElt,
+          LHSLastElt, LHS, SToVLHS, DAG, Subtarget);
     }
     if (SToVRHS) {
-      if (!IsLittleEndian && SToVRHS.getValueType().getScalarSizeInBits() >= 64)
+      int RHSScalarSize = SToVRHS.getValueType().getScalarSizeInBits();
+      if (!IsLittleEndian && RHSScalarSize >= 64)
----------------
amy-kwan wrote:

Do you mean, do we still need to run `isShuffleMaskInRange()` if we generate a `SCALAR_TO_VECTOR_PERMUTED` for the LHS, but not the RHS side?

I don't think so. I think this is due to the above comment:
```
    // On big endian systems, this only makes sense for element sizes smaller
    // than 64 bits since for 64-bit elements, all instructions already put
    // the value into element zero. Since scalar size of LHS and RHS may differ
    // after isScalarToVec, this should be checked using their own sizes.
```

https://github.com/llvm/llvm-project/pull/80784