https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121873
--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> --- (In reply to Jakub Jelinek from comment #2) > It fails with -m32 -std=c23 without -mfpmath=sse -msse2, but it should fail > in that case, for -fexcess-precision=standard (implied by -std=c??) on > targets with FLT_EVAL_METHOD that is the right behavior, constants are to be > evaluated in that case with precision of long double (i.e. 64 bits). I mean with FLT_EVAL_METHOD 2.
