[clang] [CIR][AArch64] Upstream addition-across-vector (incl. widening) NEON builtins (PR #193396)

[Andrzej Warzyński via cfe-commits]

================
@@ -2701,14 +2722,6 @@ CIRGenFunction::emitAArch64BuiltinExpr(unsigned builtinID, const CallExpr *expr,
   case NEON::BI__builtin_neon_vminnmv_f16:
   case NEON::BI__builtin_neon_vminnmvq_f16:
   case NEON::BI__builtin_neon_vmul_n_f64:
-  case NEON::BI__builtin_neon_vaddlv_u8:
-  case NEON::BI__builtin_neon_vaddlv_u16:
-  case NEON::BI__builtin_neon_vaddlvq_u8:
-  case NEON::BI__builtin_neon_vaddlvq_u16:
-  case NEON::BI__builtin_neon_vaddlv_s8:
-  case NEON::BI__builtin_neon_vaddlv_s16:
-  case NEON::BI__builtin_neon_vaddlvq_s8:
-  case NEON::BI__builtin_neon_vaddlvq_s16:
----------------
banach-space wrote:

We avoid moving cases - that's to keep things in-sync with https://github.com/llvm/llvm-project/blob/main/clang/lib/CodeGen/TargetBuiltins/ARM.cpp. Also, by adding new calls to `cgm.errorNYI` it is easy to see what is missing.

https://github.com/llvm/llvm-project/pull/193396