Hi Jiawei, this as well as 4/4 was still outstanding. I had a cursory look and the changes look mostly sensible to me. Rather a comment for the whole series and not on these particular files but I'd like us to have altfmt handling to be as "static" as possible. This means we shouldn't make it dependent on a specific operand but rather on an insn type or so. To distinguish between signed and unsigned we should have two insn (like mul and mulu).
RE overlap: The RA bits are in now and Pan Li is working on initial patches. Even though we're still seeing some ICEs I hope this approach is generally the right way. You can have a look at his patches for inspiration. -- Regards Robin
