https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119241
--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> --- Iain mentioned the COBOL FE using fmodf128, that can perhaps be done using fold_const_call (CFN_BUILT_IN_FMODF128, float128_type_node, real_cst1, real_cst2) Note, all of this is assuming you want to do constant evaluation in the _Float128 target type, if you want something else in some parts (e.g. using _Float64 or _Float32), then the type needs to be different.
