[clang] [clang][X86] Support __attribute__((model("small"/"large"))) (PR #124834)

[Artem Belevich via cfe-commits]

================
@@ -5,7 +5,7 @@
 // RUN: %clang_cc1 -triple riscv64 -verify=expected,riscv64 -fsyntax-only %s
 // RUN: %clang_cc1 -triple x86_64 -verify=expected,x86_64 -fsyntax-only %s
----------------
Artem-B wrote:

You could do it here, too, as the test does not actually need extra CUDA SDK bits that we have in SemaCUDA/Inputs.

A RUN with  `-x cuda -triple nvptx64-unknown-cuda -fcuda-is-device` should be sufficient.

https://github.com/llvm/llvm-project/pull/124834