[llvm] [clang] [compiler-rt] [X86] Support "f16c" and "avx512fp16" for __builtin_cpu_supports (PR #78384)

[Fangrui Song via cfe-commits]

================
@@ -80,17 +80,17 @@ OutOfLineDefs::foo(int, int, int) {
 // LINUX: define dso_local noundef i32 @_ZN13OutOfLineDefs3fooEiii.S
 // LINUX: define dso_local noundef i32 @_ZN13OutOfLineDefs3fooEiii.R
 // LINUX: define weak_odr ptr @_ZN13OutOfLineDefs3fooEiii.resolver()
-// LINUX: ret ptr @_ZN13OutOfLineDefs3fooEiii.R
 // LINUX: ret ptr @_ZN13OutOfLineDefs3fooEiii.S
+// LINUX: ret ptr @_ZN13OutOfLineDefs3fooEiii.R
----------------
MaskRay wrote:

Reordering `X86_FEATURE` causes the change here. This should not matter. Ideally, the test can be improved to be check the relevant `__cpu_model` members.

https://github.com/llvm/llvm-project/pull/78384