https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50430
--- Comment #5 from Martin Jambor <jamborm at gcc dot gnu.org> --- Am I correct thinking that this has been addressed (long time ago)? The entire optimized dump of the testcase from comment #3 is now the following, so no missed devirtualization there: void _GLOBAL__sub_I_typelib_TypeClass () { <bb 2> [local count: 1073741824]: __ct_comp (&typelib_TypeClass, 0, 1); [tail call] return; }