[llvm] [AMDGPU] Add AMDGPU support for llvm-objcopy (PR #92066)

[Pierre van Houtryve via llvm-commits]

Pierre-vh wrote:

> Hi, considering both of your feedback I'm trying to have amdgpu go into cross-arch-headers.test with a flag as suggested.
> 
> But, adding "Flags: [[FLAGS=<none>]]" to the FileHeader: gives- YAML:136:20: error: expected sequence of bit values Flags: <none>
> 
> Any suggestions on how I can go about incorporating the flag?

What happens if you use another default value for the flag, like 0?
I assume it's expecting some identifier or numbers, but it sees `<none>` which may not be valid YAML

Can you just update this patch with the cross-arch-headers.test instead, even if the test is broken? Just put a comment with the error. It'll be much easier to review this way

https://github.com/llvm/llvm-project/pull/92066