This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug middle-end/26835] ICE in f951 on valid code at -O2 -ftree-loop-linear
- From: "dberlin at dberlin dot org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 24 Mar 2006 12:27:37 -0000
- Subject: [Bug middle-end/26835] ICE in f951 on valid code at -O2 -ftree-loop-linear
- References: <bug-26835-8649@http.gcc.gnu.org/bugzilla/>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Comment #2 from dberlin at gcc dot gnu dot org 2006-03-24 12:27 -------
Subject: Re: ICE in f951 on valid code at -O2
-ftree-loop-linear
On Fri, 2006-03-24 at 10:30 +0000, rguenth at gcc dot gnu dot org wrote:
>
> ------- Comment #1 from rguenth at gcc dot gnu dot org 2006-03-24 10:30 -------
> Confirmed. We ICE after IVOPTs during computation of dominance frontiers.
> Other than that, tree-loop-linear seems to be somewhat buggy in general.
>
The code generation portion of tree-loop-linear is pretty solid (there
may be a few off by one errors lurking, however).
The perfect nest conversion is a rats nest in a lot of ways, because
it's a really really bad hack to avoid real loop distribution.
I've seen cases where it breaks the cfg and dominance info.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26835