This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] New regressions as of 2003-11-04
- From: Andrew MacLeod <amacleod at redhat dot com>
- To: Michael Matz <matz at suse dot de>
- Cc: Jan Hubicka <jh at suse dot cz>, Jeff Law <law at redhat dot com>, Diego Novillo <dnovillo at redhat dot com>, Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>, Daniel Berlin <dberlin at dberlin dot org>, gcc mailing list <gcc at gcc dot gnu dot org>
- Date: 05 Nov 2003 10:38:12 -0500
- Subject: Re: [tree-ssa] New regressions as of 2003-11-04
- References: <Pine.LNX.4.44.0311051630190.16164-100000@wotan.suse.de>
On Wed, 2003-11-05 at 10:33, Michael Matz wrote:
> Hi,
>
> On 5 Nov 2003, Andrew MacLeod wrote:
>
> > > The point is that we *want* to create temporary registers so unless
> > > there is good reason to do so, we don't want single register to be set
> > > two times. That confuses RTL.
> > >
> >
> > Setting a single register more than once confuses RTL? Im not sure I
> > follow.
>
> It doesn't strictly confuse anything, but makes the work for RTL
> optimizers harder. So much that some of the optimizers give up one some
> possible improvements if disconnected live ranges are using the same REG.
>
Which ones I wonder... Too bad that confuses them... If its the passes
the tree-ssa is planning to make obsolete, then its not much of an
issue.
> > The register allocator ought to be smart enough to tell when two
> > disjoint live ranges use the same register, and rename one of the
> > registers to something else to allow them to prevent artifical
>
> The new one is, the old one not, for instance.
>
I expected as much :-)
Andrew