https://gcc.gnu.org/bugzilla/show_bug.cgi?id=23666
--- Comment #12 from Andrew Pinski <pinskia at gcc dot gnu.org> --- (In reply to Jeffrey A. Law from comment #11) > The transformation asked for in this BZ is invalid except when we're > allowing signed overflows (aka -fwrapv). And with -fwrapv other code takes > care of this simplification. > > Closing as INVALID. Andrew/Richi, if y'all disagree, please chime in. I agree at this stage until move over to a plus expr which has overflow or wrapping behavior explicit rather than part of the type (i dont know when that will be implemented). So I am fine with closing this.
