This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug optimization/13951] [tree-ssa] cselib ICE during x86-64 kernel compile with treessa
- From: "jh at suse dot cz" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 5 Feb 2004 08:36:33 -0000
- Subject: [Bug optimization/13951] [tree-ssa] cselib ICE during x86-64 kernel compile with treessa
- References: <20040131205930.13951.ak@muc.de>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Additional Comments From jh at suse dot cz 2004-02-05 08:35 -------
Subject: Re: [tree-ssa] cselib ICE during x86-64 kernel compile with treessa
> On Thu, Feb 05, 2004 at 12:56:58AM +0100, Jan Hubicka wrote:
> > > I think CSELIB ICE because dummy is set twice. I think it makes sense
> > > to do so...
>
> Why? I think an ICE because of such isn't a good idea.
We care to not have undefined insns of this kind.
One never knows what of the sets will be winner and thus never knows the
effect of instructions. Various optimizers may miss the detail and
choose one of the sets to be "the right one"
To deal with this correctly, we would have to teach cselib to track down
such a multi sets and invalidate result of the register immediately
after processing.
Not terribly dificult, but anoying.
Honza
>
>
> r~
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13951