zygoloid wrote: > So an optimization pass that replaces `%pred` with `true` is sound under my > proposed rule. Am I missing anything?
It's currently correct to replace ```llvm load %p ``` with ```llvm icmp eq %p, <something> ... load %p ``` and profile-driven optimizations may wish to do exactly that. Your rule makes that transformation incorrect. https://github.com/llvm/llvm-project/pull/197855 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
