shay-kl wrote: @matthias-springer I gave assume alignment a result and changed all its uses accordingly and it does seem cleaner this way. Thanks for the suggestion. https://github.com/llvm/llvm-project/pull/139521