ManuelJBrito added a comment. In D143287#4108165 <https://reviews.llvm.org/D143287#4108165>, @craig.topper wrote:
> In D143287#4107439 <https://reviews.llvm.org/D143287#4107439>, @ManuelJBrito > wrote: > >> In D143287#4103597 <https://reviews.llvm.org/D143287#4103597>, @craig.topper >> wrote: >> >>> We have a couple bugs that show (freeze (poison)) doesn't work past >>> SelectionDAG. Is that a concern here? The most recent >>> https://github.com/llvm/llvm-project/issues/60429 >> >> I don't think it's a concern . Here https://godbolt.org/z/1ecM8roYh the >> lowering seems ok. > > The bugs start occurring if you use the same nondeterministic value multiple > times and expect the value to be the same for all uses. I understand now ... so if the bug is present it works as if it were an undef. For these intrinsics we would just regress to the current behavior. So is it okay if we land this patch? Because it will still be an improvement. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D143287/new/ https://reviews.llvm.org/D143287 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
