This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [Patch, fortran] PR20938 and where_19 - equivalence dependencies
- From: Steve Kargl <sgk at troutmask dot apl dot washington dot edu>
- To: Paul Thomas <paulthomas2 at wanadoo dot fr>
- Cc: Tobias Schl?ter <tobias dot schlueter at physik dot uni-muenchen dot de>, "'fortran at gcc dot gnu dot org'" <fortran at gcc dot gnu dot org>, patch <gcc-patches at gcc dot gnu dot org>
- Date: Tue, 28 Feb 2006 12:15:42 -0800
- Subject: Re: [Patch, fortran] PR20938 and where_19 - equivalence dependencies
- References: <440011AD.2070305@wanadoo.fr>
On Sat, Feb 25, 2006 at 09:13:33AM +0100, Paul Thomas wrote:
>
> It stores some of the data in the segment_info lists, produced in
> trans-common.c(add_equivalences), in new structures and uses the
> information in dependency.c to check for dependencies through direct and
> indirect equivalences. The segment offsets are used in this checking.
> The check is conservative, in that a small group of cases, where the
> offset and the array references are such that, in reality, there is no
> dependency, will still produce temporaries; eg. an offset of -1 and an
> index difference of +1. The clean up occurs in gfc_free_namespace,
> partly for consistency and partly because analysis showed that there was
> no great advantage in doing it earlier.
>
> The testcases are all as originally submitted.
>
> Regtested on FC3/Athlon. OK for trunk and, eventually, 4.1?
(snip)
> 2005-02-25 Paul Thomas <pault@gcc.gnu.org>
>
> PR fortran/20938
> * gfortran.dg/dependency_2.f90: New test.
> * gfortran.fortran-torture/execute/where17.f90: New test.
> * gfortran.fortran-torture/execute/where18.f90: New test.
> * gfortran.fortran-torture/execute/where19.f90: New test.
> * gfortran.fortran-torture/execute/where_20.f90: New test.
These tests are missing.
I've read the patch and I've done a regression test on
the current testsuite. I believe the patch is OK.
--
Steve