https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79061
--- Comment #29 from Jakub Jelinek <jakub at gcc dot gnu.org> --- (In reply to Maxim Ostapenko from comment #28) > > In short, the easiest fix is just to disable the initialization order > > checking altogether for LTO (by forcing dynamically_initialized = 0 in LTO). > > Yeah, that's what I thought about. Should I revert the whole bunch of > related patches or just force vnode->dynamically_initialized = 0 for LTO > leaving a comment with link to this PR? I think it doesn't hurt to have reasonable DECL_NAME on TRANSLATION_UNIT_DECL. But the rest probably can be reverted, yes. It would be good to restore it for LTO at least for the cases that can be handled (where only one TU originally owns it), but that can wait for GCC 8.
