This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] Fix pr23046
On Tue, 2005-08-09 at 10:36 -0600, Jeffrey A Law wrote:
> On Sat, 2005-08-06 at 01:51 -0400, James A. Morrison wrote:
> > Hi,
> >
> > This is the fold patch that was refered to recently on the GCC list for
> > this bug. It allows fold to fold X > TYPE_MAX_VALUE and X < TYPE_MIN_VALUE to
> > 0. This patch has been bootstrapped and regtested on ia64-linux with all
> > languages except Ada with no new regressions. Ok for mainline?
> OK. Though a part of me thinks that VRP should also be fixed to cope with
> less than optimal trees. ie, it seems wrong for VRP to ICE due to a
> failure to fold and expression.
>
This is not less than optimal trees so much as arguably illegal
expressions (depending on who you ask)
IE they are
X = something > TYPE_MAX_VALUE (TREE_TYPE (X)) or < TYPE_MIN_VALUE
(TREE_TYPE (X))