https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111310
--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> --- Note, there is no guarantee that _BitInt(32) or _BitInt(64) are passed/returned/laid out exactly like int32_t or int64_t (ditto for 8 and 16 bits), but on the current single target which supports them it is so far the case. _BitInt(128) has different alignment from __int128 though.