https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71436
--- Comment #9 from ktkachov at gcc dot gnu.org --- (In reply to Jakub Jelinek from comment #8) > Or reload_completed || lra_in_progress, or punt on pseudos in the predicate. I tried punting on pseudos in the predicate (in ldm_stm_operation_p) and it fixed the testcase. I'm tempted by either of these solutions but I'm a bit cautious about the load_multiple expander that may try to generate a load_multiple early on during expand and expect a matching pattern. It only generates between 2 and 4 loads though, so those should be caught by the patterns in ldmstm.md so I think it will work out okay.
