https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126072
--- Comment #4 from Hans-Peter Nilsson <hp at gcc dot gnu.org> --- (In reply to Nathan Myers from comment #2) > What "other bad assumptions" do you refer to? I probably inferred the plural a bit prematurely while looking at the next test, but now while calculating for the case I had in mind; a theoretical target with structs required to be aligned to 4 or 8, it seems I was wrong about that one, so I'll be quiet.
