https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119244
--- Comment #18 from Jakub Jelinek <jakub at gcc dot gnu.org> --- I think s390x-linux is similar to aarch64-linux here, neither has libquadmath support. Just long double is always IEEE quad on aarch64-linux and on s390x-linux it is configurable, one can choose between IEEE double and IEEE quad and for the latter needs some glibc support (though years older than the *f128 APIs).
