https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68404
--- Comment #10 from Bill Schmidt <wschmidt at gcc dot gnu.org> --- Looking at scan_rtx_address () in regrename.c, it indeed looks like a PLUS nested inside a PLUS isn't handled. I'm not sure if this is a deficiency in regrename.c, or whether the definition of fusion_gpr_load_di is just playing with fire here. Richard S., do you have an opinion? fusion_gpr_load_di will not be present through most of the back end passes, because it is introduced during peephole2. So it seems like we'd probably be ok just adding some more handling to regrename, but I'd like to hear other opinions.