[clang] [llvm] [AArch64][NEON] NEON intrinsic compilation error with -fno-lax-vector-conversion flag fix (PR #149329)

[via llvm-commits]

================
@@ -1401,14 +1401,18 @@ void Intrinsic::emitBodyAsBuiltinCall() {
       if (LocalCK == ClassB || (T.isHalf() && !T.isScalarForMangling())) {
         CastToType.makeInteger(8, true);
         Arg = "__builtin_bit_cast(" + CastToType.str() + ", " + Arg + ")";
-      } else if (LocalCK == ClassI) {
-        if (CastToType.isInteger()) {
----------------
Lukacma wrote:

Is it not enough to amend the check here to this ?

`` if (CastToType.isInteger() && CastToType.isPoly())``

https://github.com/llvm/llvm-project/pull/149329