ZequanWu wrote: > Richard had a comment on the original which I think it still relevant: > > > There are 496 calls to Builder.CreateCall in clang's CodeGen. Do they all > > need this change? If not, how can we be confident we've found all the ones > > that do? (From a quick check, at least MSVC's __fastfail builtin seems like > > it would also benefit from this handling.) > > Would it be reasonable to make clang's CGBuilder do this for every call > > instruction we create? > > I think we could still do this, but we should fix this long outstanding bug > first.
Yeah, it would be easier if there's one place to attach statement attributes on all expressions inside that statement. https://github.com/llvm/llvm-project/pull/101549 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits