[llvm] [Instrumentation] Add TSan instrumentation for AMDGPU targets (PR #207713)

[Marco Elver via llvm-commits]

melver wrote:

> > I optionally suggest extracting Parser changes into a separate PR.
> 
> So, one question I have is if we should just make the names different, since this does chance the proper 'ABI'. My concern as well is that we will need to add some more special handling for various AMDGPU intrinsics that touch things like LDS. Do you think it would be reasonable to put logic like that inside of the existing pass here? Should we change these to like `__csan_`?

There's an alternative to either extending the TSan pass or forking the TSan pass entirely: add a smaller instrumentation pass that is inserted either before or after TSan pass that translates the AMDGPU-specific intrinsics, and anything else AMDGPU-specific that doesn't belong into the TSan pass.

As for the prefix, you could add either a constructor param, or an cl::opt option to change the prefix. But that's cosmetic, and not a user-visible detail, so up to you if it's worth doing.

https://github.com/llvm/llvm-project/pull/207713