[dcmf] Re: BGP patch for MPICH2

Pavan Balaji balaji at mcs.anl.gov
Mon Feb 25 15:45:21 CST 2008


Thanks Joe. We'll go over these patches and get back to you. It might be 
useful to have a telecon at some point after we go over the patches.

  -- Pavan

On 02/25/2008 03:16 PM, Joe Ratterman wrote:
> I've broken the patch into 4 sections--and gzipped them to reduce the
> size for the list--which modify these parts only:
>  + device : mpich/src/mpid/
>  + romio : mpich/src/mpi/romio
>  + test-suite : mpich2/test/ (these are probably almost completely
> pointless for you)
>  + other : everything else
> 
> Git uses ignore files much like cvs did, but I've found that
> subversion does not.  Since there should be ignored files in the
> mpich/test/util/timer/ directory, it seems most simple to keep the
> .cvsignore file.  That file doesn't have to stay, but I don't know how
> to get that info into SVN.  Do you know how to handle that?
> 
> I hope this is helpful--please let us know if you need more help.  We
> are open to having a teleconference if you think that might be
> helpful.
> 
> Thanks,
> Joe Ratterman
> 
> 
> 
> On Mon, Feb 25, 2008 at 2:40 PM, Pavan Balaji <balaji at mcs.anl.gov> wrote:
>> Joe,
>>
>>  This is a huge patch. Will you be able to create separate patches for
>>  different components, so that we can integrate them independently? Here
>>  are some patches you can create:
>>
>>  1. ROMIO changes
>>
>>  2. MPICH2 test suite changes
>>
>>  3. Rest of the changes
>>
>>  FYI, the confdb/confdb directory is still present in your version and is
>>  showing up in your patch. This was a bug that has been corrected and can
>>  be removed. Also, mpich2-1.0.7rc1/test/util/timer/.cvsignore should not
>>  exist.
>>
>>  Thanks.
>>
>>   -- Pavan
>>
>>
>>
>>  On 02/25/2008 02:15 PM, Joseph Ratterman wrote:
>>  >
>>  > Hello everyone,
>>  >
>>  > These the differences between our version of MPICH2 in git (which will
>>  > be on the DCMF wiki soon), and the 1.0.7rc1 from subversion.  Both
>>  > source bases have undergone a fair bit of post-processing to ensure that
>>  > they are comparable.  Looking at the diff, I think that they have helped
>>  > eliminate incorrect changes (such as $Id$).
>>  >
>>  > This does change the existing files from ANL, so it will have to be
>>  > checked to ensure that you really want the changes.  The end of this
>>  > email lists all the changed files and added directories that are in the
>>  > patch.
>>  >
>>  >
>>  > SHA1 hash:
>>  > cbacbf97402932c7d8566c04985a0b73eed75292  bgp-mpich-1.0.7rc1.patch
>>  >
>>  >
>>  > Enjoy,
>>  > Joe Ratterman
>>  >
>>  >
>>  >
>>  > $ diff -rq mpich2-1.0.7rc1/ bgp-mpich2/  | sort
>>  > Files mpich2-1.0.7rc1/configure.in and bgp-mpich2/configure.in differ
>>  > Files mpich2-1.0.7rc1/maint/simplemake.in and
>>  > bgp-mpich2/maint/simplemake.in differ
>>  > Files mpich2-1.0.7rc1/src/binding/f77/buildiface and
>>  > bgp-mpich2/src/binding/f77/buildiface differ
>>  > Files mpich2-1.0.7rc1/src/include/mpiimpl.h and
>>  > bgp-mpich2/src/include/mpiimpl.h differ
>>  > Files mpich2-1.0.7rc1/src/include/mpiimplthread.h and
>>  > bgp-mpich2/src/include/mpiimplthread.h differ
>>  > Files mpich2-1.0.7rc1/src/mpi/comm/commutil.c and
>>  > bgp-mpich2/src/mpi/comm/commutil.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/datatype/get_elements.c and
>>  > bgp-mpich2/src/mpi/datatype/get_elements.c differ
>>  > Files
>>  > mpich2-1.0.7rc1/src/mpid/common/datatype/mpid_type_create_pairtype.c and
>>  > bgp-mpich2/src/mpid/common/datatype/mpid_type_create_pairtype.c differ
>>  > Files mpich2-1.0.7rc1/src/mpid/Makefile.sm and
>>  > bgp-mpich2/src/mpid/Makefile.sm differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/ad_testfs/ad_testfs_hints.c and
>>  > bgp-mpich2/src/mpi/romio/adio/ad_testfs/ad_testfs_hints.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/ad_ufs/ad_ufs.c and
>>  > bgp-mpich2/src/mpi/romio/adio/ad_ufs/ad_ufs.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/ad_close.c and
>>  > bgp-mpich2/src/mpi/romio/adio/common/ad_close.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/ad_fstype.c and
>>  > bgp-mpich2/src/mpi/romio/adio/common/ad_fstype.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/ad_get_sh_fp.c and
>>  > bgp-mpich2/src/mpi/romio/adio/common/ad_get_sh_fp.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/ad_open.c and
>>  > bgp-mpich2/src/mpi/romio/adio/common/ad_open.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/ad_read_coll.c and
>>  > bgp-mpich2/src/mpi/romio/adio/common/ad_read_coll.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/ad_set_sh_fp.c and
>>  > bgp-mpich2/src/mpi/romio/adio/common/ad_set_sh_fp.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/cb_config_list.c and
>>  > bgp-mpich2/src/mpi/romio/adio/common/cb_config_list.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/flatten.c and
>>  > bgp-mpich2/src/mpi/romio/adio/common/flatten.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/lock.c and
>>  > bgp-mpich2/src/mpi/romio/adio/common/lock.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/common/Makefile.in and
>>  > bgp-mpich2/src/mpi/romio/adio/common/Makefile.in differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/include/adio.h and
>>  > bgp-mpich2/src/mpi/romio/adio/include/adio.h differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/include/adioi_errmsg.h and
>>  > bgp-mpich2/src/mpi/romio/adio/include/adioi_errmsg.h differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/include/adioi_fs_proto.h and
>>  > bgp-mpich2/src/mpi/romio/adio/include/adioi_fs_proto.h differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/include/adioi.h and
>>  > bgp-mpich2/src/mpi/romio/adio/include/adioi.h differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/adio/include/mpio_error.h and
>>  > bgp-mpich2/src/mpi/romio/adio/include/mpio_error.h differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/configure.in and
>>  > bgp-mpich2/src/mpi/romio/configure.in differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/Makefile.in and
>>  > bgp-mpich2/src/mpi/romio/Makefile.in differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/mpi2-other/info/fortran/Makefile.in
>>  > and bgp-mpich2/src/mpi/romio/mpi2-other/info/fortran/Makefile.in differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/mpi-io/get_amode.c and
>>  > bgp-mpich2/src/mpi/romio/mpi-io/get_amode.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/romio/test/file_info.c and
>>  > bgp-mpich2/src/mpi/romio/test/file_info.c differ
>>  > Files mpich2-1.0.7rc1/src/mpi/timer/mpidtime.c and
>>  > bgp-mpich2/src/mpi/timer/mpidtime.c differ
>>  > Files mpich2-1.0.7rc1/test/Makefile.sm and bgp-mpich2/test/Makefile.sm
>>  > differ
>>  > Files mpich2-1.0.7rc1/test/mpi/cxx/io/ioharness.tlt and
>>  > bgp-mpich2/test/mpi/cxx/io/ioharness.tlt differ
>>  > Files mpich2-1.0.7rc1/test/mpi/errors/coll/rerr.c and
>>  > bgp-mpich2/test/mpi/errors/coll/rerr.c differ
>>  > Files mpich2-1.0.7rc1/test/mpi/f77/io/ioharness.tlt and
>>  > bgp-mpich2/test/mpi/f77/io/ioharness.tlt differ
>>  > Files mpich2-1.0.7rc1/test/mpi/f77/io/shpositionf.f and
>>  > bgp-mpich2/test/mpi/f77/io/shpositionf.f differ
>>  > Files mpich2-1.0.7rc1/test/mpi/init/testlist and
>>  > bgp-mpich2/test/mpi/init/testlist differ
>>  > Files mpich2-1.0.7rc1/test/mpi/Makefile.sm and
>>  > bgp-mpich2/test/mpi/Makefile.sm differ
>>  > Files mpich2-1.0.7rc1/test/mpi/pt2pt/testlist and
>>  > bgp-mpich2/test/mpi/pt2pt/testlist differ
>>  > Files mpich2-1.0.7rc1/test/mpi/runtests.in and
>>  > bgp-mpich2/test/mpi/runtests.in differ
>>  > Files mpich2-1.0.7rc1/test/TestResults.xsl and
>>  > bgp-mpich2/test/TestResults.xsl differ
>>  > Files mpich2-1.0.7rc1/test/util/Makefile.sm and
>>  > bgp-mpich2/test/util/Makefile.sm differ
>>  > Files mpich2-1.0.7rc1/test/util/timer/.cvsignore and
>>  > bgp-mpich2/test/util/timer/.cvsignore differ
>>  > Only in bgp-mpich2/confdb: confdb
>>  > Only in bgp-mpich2/src/env: genmpxl.pl
>>  > Only in bgp-mpich2/src/env: mpixl.conf.in
>>  > Only in bgp-mpich2/src: mpe
>>  > Only in bgp-mpich2/src/mpid/common/datatype: gen_type_blockindexed.c
>>  > Only in bgp-mpich2/src/mpid: dcmf
>>  > Only in bgp-mpich2/src/mpi/romio/adio: ad_bgl
>>  > Only in bgp-mpich2/src/mpi/romio/adio/common: ad_write_nolock.c
>>  > Only in bgp-mpich2/src/pmi: uni
>>  > Only in bgp-mpich2/test: iotestlist
>>  > Only in bgp-mpich2/test: long.xsl
>>  > Only in bgp-mpich2/test/mpi/cxx/io: iotestlist
>>  > Only in bgp-mpich2/test/mpi/cxx: iotestlist
>>  > Only in bgp-mpich2/test/mpi/f77/io: iotestlist
>>  > Only in bgp-mpich2/test/mpi/f77: iotestlist
>>  > Only in bgp-mpich2/test/mpi/io: iotestlist
>>  > Only in bgp-mpich2/test/mpi: iotestlist
>>  > Only in bgp-mpich2/test: short.xsl
>>  >
>>  >
>>  >
>>  >
>>  >
>>  >
>>  >
>>  >
>>  >
>>  > *Pavan Balaji <balaji at mcs.anl.gov>*
>>  >
>>  > 02/25/08 01:39 PM
>>  >
>>  >
>>  > To
>>  >       Joseph Ratterman/Rochester/IBM at IBMUS
>>  > cc
>>  >       mpich2-core at mcs.anl.gov
>>  > Subject
>>  >       BGP patch for MPICH2
>>  >
>>  >
>>  >
>>  >
>>  >
>>  >
>>  >
>>  >
>>  > Hi Joe,
>>  >
>>  > Just pinging to check the status of the BG/P patch for MPICH2. We are
>>  > working on a few bugs and will start our final testing as soon as they
>>  > are fixed. If you think the patch will take much longer, we can add it
>>  > to the trunk after the 1.0.7 release and work on it for 1.0.8.
>>  > Alternatively, if you'd like to contribute some generic parts now and
>>  > more later, that can be worked out as well. Please let us know how you'd
>>  > like to go about with this integration.
>>  >
>>  > Thanks.
>>  >
>>  >  -- Pavan
>>  >
>>  > --
>>  > Pavan Balaji
>>  > http://www.mcs.anl.gov/~balaji
>>  >
>>
>>  --
>>  Pavan Balaji
>>  http://www.mcs.anl.gov/~balaji
>>  _______________________________________________
>>  dcmf mailing list
>>  dcmf at lists.anl-external.org
>>  http://lists.anl-external.org/cgi-bin/mailman/listinfo/dcmf
>>  http://dcmf.anl-external.org/wiki
>>

-- 
Pavan Balaji
http://www.mcs.anl.gov/~balaji



More information about the dcmf mailing list