[clang] [llvm] [mlir] [AMDGPU] Fix missing cache bypass scope on seq_cst buffer fat pointer atomics (PR #213585)

[Arseniy Obolenskiy via cfe-commits]

aobolensk wrote:

> I think we just land through the aarch64 issue.
> 
> I'm inclined to land this ... let's say Thursday, if we don't hear too much more?
> 
> (But yeah, we should also have a followup about those fences we're adding - if we're encoding atomicity on the operation, we shouldn't need the fences)

@krzysz00 shall we proceed? 

https://github.com/llvm/llvm-project/pull/213585