[clang] [llvm] [TargetVerifier][AMDGPU] Add TargetVerifier. (PR #123609)

[Eli Friedman via llvm-commits]

efriedma-quic wrote:

I don't think it makes sense to have a target-specific "verifier": verification sets the limits of what optimizations are allowed to do.  And that determination needs to be target-independent, or else target-independent optimizations can't exist.  Target-specific intrinsics or calling conventions can have rules in the target-independent verifier if you want... in fact, we already have some rules.

Running Lint between passes is an interesting idea, but not sure how useful it would be in practice; there would be a lot of false positives.

https://github.com/llvm/llvm-project/pull/123609