[clang] [WIP][Clang][bytecode] interp__builtin_elementwise_int_binop - allow RHS operand to be a scalar (PR #156017)

[Simon Pilgrim via cfe-commits]

================
@@ -2565,21 +2565,48 @@ static bool interp__builtin_elementwise_int_binop(
     return true;
   }
 
+  const auto *VT = Call->getArg(0)->getType()->castAs<VectorType>();
+  assert(VT->getElementType()->isIntegralOrEnumerationType());
+  PrimType ElemT = *S.getContext().classify(VT->getElementType());
+  unsigned NumElems = VT->getNumElements();
+
+  // Vector + Scalar case.
+  if (!Call->getArg(1)->getType()->isVectorType()) {
+    assert(Call->getArg(1)->getType()->isIntegralOrEnumerationType());
+
+    APSInt RHS = popToAPSInt(
+        S.Stk, *S.getContext().classify(Call->getArg(1)->getType()));
+    const Pointer &LHS = S.Stk.pop<Pointer>();
+    const Pointer &Dst = S.Stk.peek<Pointer>();
+
+  for (unsigned I = 0; I != NumElems; ++I) {
+      APSInt Elem1;
+      INT_TYPE_SWITCH_NO_BOOL(ElemT, {
+        Elem1 = LHS.elem<T>(I).toAPSInt();
+      });
+
+      APSInt Result =
+          APSInt(Fn(Elem1, RHS),
+                 Call->getType()->isUnsignedIntegerOrEnumerationType());
----------------
RKSimon wrote:

OK - @Arghnews is going to handle this part in #155542 - and then this patch will become a rotate constexpr patch afterward

https://github.com/llvm/llvm-project/pull/156017