[llvm-branch-commits] Thread Safety Analysis: Add guarded_by_any and pt_guarded_by_any (PR #185173)

[Aaron Puchert via llvm-branch-commits]

aaronpuchert wrote:

> As for the last point, that'd be a backwards-compatibility break - I don't think we should do so.

I don't think it would be that bad. We wouldn't emit any warnings that we didn't emit before, and generated code wouldn't change. It would only weaken the warning, but this happens all the time when trying to fix false positives. That said, I don't have a preference for changing behavior here. We can still do that if people end up being confused about it. However, much more likely this is pretty niche since not many code bases use multiple locks for the same data anyway.

https://github.com/llvm/llvm-project/pull/185173