https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78633
--- Comment #13 from Oleg Endo <olegendo at gcc dot gnu.org> --- (In reply to Kazumoto Kojima from comment #12) > Perhaps the splitter in problem > might have to take care of subreg case even when referencing > a reg rtx in the input operands. So it looks like a new rtx object should be created when referencing a subreg. That's ... a problem. I have been ignoring this for a while, so I guess there are several places which need patching. I can reproduce the error on sh-elf. I will look at it, but I'm very busy at the moment, so it will take some time.