https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107346
--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> --- I think the issue was probably latent and the function assumes that bitfield accesses are rejected earlier but now are no longer(?). vect_analyze_data_refs is usually responsible for that, eventually we also rely on data-ref analysis itself to reject bitfield accesses.