This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [graphite] Fix PR38431: runtime test for CP2K
- From: Richard Guenther <richard dot guenther at gmail dot com>
- To: Sebastian Pop <sebpop at gmail dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>, Richard Guenther <rguenther at suse dot de>
- Date: Wed, 14 Jan 2009 21:39:43 +0100
- Subject: Re: [graphite] Fix PR38431: runtime test for CP2K
- References: <cb9d34b20901140633h73b377f5ode1504f9616e4dea@mail.gmail.com>
On Wed, Jan 14, 2009 at 3:33 PM, Sebastian Pop <sebpop@gmail.com> wrote:
> Hi,
>
> With this patch trunk compiles and passes all the run time tests of
> CP2K. The last error was that we were dropping the definition of a
> memory partition tag that has no vdefs in the scop, and thus not
> rewritten. After code generation of the scop, the following PHI node
> was generated:
>
> # MPT.1986_1463 = PHI <MPT.1986_914(D)(184), MPT.1986_477(185)>
>
> with the edge 185 now pointing from dead code, eliminated later by the
> CFG cleanup. With this patch, the default definition MPT.1986_914(D)
> is replaced by the definition before the scop.
>
> I committed the attached patch to graphite branch. The patch passed
> regstrap on trunk with BOOT_CFLAGS="-g -O2 -fgraphite-identity
> -floop-block" on amd64-linux. Okay for trunk?
Ok.
Thanks,
RIchard.
> Thanks,
> Sebastian Pop
> --
> AMD - GNU Tools
>