sdesmalen added a comment. Using metadata seems sensible, but did you also identify any downsides? I could imagine that we'd need to manually propagate metadata to any nodes after we do a combine (which can't be blindly copied?), e.g. add + mul -> mla, this new intrinsic would also need the metadata.
For intrinsics that don't have a directly corresponding (unpredicated) LLVM IR instruction, is there still a way to use this information in SelectionDAG? > the select instruction itself has strict rules relating to poison that > hampered the intent of this change For my understanding, can you elaborate what these strict rules regarding poison are that hamper such a change, and what it was that you tried? Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D141240/new/ https://reviews.llvm.org/D141240 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
