This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] Constant PHI nodes and out-of-ssa patch.
- From: law at redhat dot com
- To: Andrew MacLeod <amacleod at redhat dot com>
- Cc: gcc-patches <gcc-patches at gcc dot gnu dot org>
- Date: Mon, 28 Jul 2003 17:20:11 -0600
- Subject: Re: [tree-ssa] Constant PHI nodes and out-of-ssa patch.
- Reply-to: law at redhat dot com
In message <1059428857.25555.26.camel@p4>, Andrew MacLeod writes:
>
>
>The initial cut at handling constant in the out of ssa pass ended up
>emitting all the constant *before* any of the other copies required.
>Since all the PHI's are suppose to be processed simultaneously, they
>should be emitted last. This patch queues up the copies in the
>elimination graph structure, and emits all the constant copies after any
>other ones have been emitted.
>
>I've checked in the following patch. It bootstraps on x86 and causes no
>new regressions.
>
>Andrew
>
>
> * tree-ssa.c (struct _elim_graph): Add varray for constant copies.
> (new_elim_graph): Initialize constant copy array..
> (eliminate_build): Push constant copies onta stack instead of emitting.
> (eliminate_phi): Emit any pending constant copies.
Cool. I've submitted the changes to enable these propagations as well as
the test which initially got me looking at the issue.
Jeff