https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106137
--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> --- Created attachment 53224 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=53224&action=edit gcc13-pr106137.patch Perhaps this patch could fix this? We no longer need __float128 type in libgfortran (except I think for the powerpc64le __*ieee128 prototypes), but want to keep HAVE_FLOAT128 being defined where it was defined before, which is targets where _Float128 is supported and long double isn't IEEE quad (which is equivalent to targets where __float128 is supported, the latter is only supported if long double isn't IEEE quad and _Float128 is supported).
