https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119966
--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> --- > That paradoxical hardware register is not valid. It was not validated > with pru_hard_regno_mode_ok because the latter would have rejected it. The patch uses rtl_hooks.gen_lowpart_no_emit which is used all over simplify-rtl.cc so that is not the issue directly. Do you have a preprocessed source, I can take a look. Because I don't think we should have gotten a subreg of a hard register anyways.
