https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128292
--- Comment #1 from Drea Pinski <pinskia at gcc dot gnu.org> --- Note that actually might be the fix. I will look later today. The backprop pass might be messing the global range by a removal of a neg or fabs and since before that patch it was reusing a ssa name, the range on it is broken. Using a new ssa name instead of reusing it fixes the range and it works correctly on the trunk and didnt becomes a latent bug.
