https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117642
--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> --- __sync_lock_test_and_set on __int128 (and similarly for both builtins on long long with -m32 -march=i486) results in a __sync_lock_test_and_set_16 (resp. _8) call which isn't implemented anywhere. I'm not aware of warnings being ever emitted.
