goldsteinn wrote: > Okay, the bug is from the fact that we simplify the values in `VMap` which > can mean that even if `VMap.lookup(InnerCB)` exists, it doesn't imply that > `NewInnerCB` is actually the same function. This is really only an issue for > intrins which can simplify but is an active bug: > https://godbolt.org/z/r873jTdvM > > introduced in: #95888 :( > > I think this bug shows up in other places we propagate i.e: > https://godbolt.org/z/zzbMvW5cq > > Here we may create UB by propagating the `nonnull` back to `%p1()`. > > I think the necessary fix is something along the lines of checking that the > new callbase calls the same function. I will have a patch up shortly.
Fix at: https://github.com/llvm/llvm-project/pull/109347 https://github.com/llvm/llvm-project/pull/91101 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
