https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127358
--- Comment #3 from Drea Pinski <pinskia at gcc dot gnu.org> --- I see the issue with r14-2299, I removed the "prefer unsigned types from the two with the same precision." part as I thought the overflow code would have handled that case but it seems like the overflow check falls down when the precision are the same.
