[clang] [Headers][X86] VectorExprEvaluator::VisitCallExpr / InterpretBuiltin - allow PALIGNR byte shift intrinsics to be used in constexpr (PR #162005)

[Simon Pilgrim via cfe-commits]

================
@@ -4718,6 +4718,30 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
           return APInt(8, 0);
         });
 
+  case X86::BI__builtin_ia32_palignr128:
+  case X86::BI__builtin_ia32_palignr256:
+  case X86::BI__builtin_ia32_palignr512:
+    return interp__builtin_ia32_shuffle_generic(
+        S, OpPC, Call, [](unsigned DstIdx, unsigned Shift) {
+          // Default to -1 → zero-fill this destination element
+          unsigned VecIdx = 1;
+          int ElemIdx = -1;
+
+          int Lane = DstIdx / 16;
+          int Offset = DstIdx % 16;
+
+          // Elements come from VecB first, then VecA after the shift boundary
+          unsigned ShiftedIdx = Offset + Shift;
+          if (ShiftedIdx < 16) { // from VecB
+            ElemIdx = ShiftedIdx + Lane * 16;
----------------
RKSimon wrote:

```suggestion
            ElemIdx = ShiftedIdx + (Lane * 16);
```

https://github.com/llvm/llvm-project/pull/162005