https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119154
--- Comment #10 from Nathaniel Shead <nshead at gcc dot gnu.org> --- Created attachment 60689 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=60689&action=edit pr119154.patch A minimal fix would be to avoid doing 'note_vague_linkage_fn' for 'gnu_inline' functions in the postprocessing loop within module_state::read_cluster. But I'm not a fan of the way that DECL_INTERFACE_KNOWN is handled here; maybe something like the attached patch would clean up some other possible issues? We probably need to also copy across some duplicate_decls things to handle the gnu_inline case better as well.
