paperchalice wrote: > is this purely a compile time optimization? or does this actually have functional change (in which case should have a test) It is a pass to lower GC intrinsics, GC related passes are lack of testing. https://github.com/llvm/llvm-project/pull/84421