ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
2 738 коммитов 3 Ветки 0 Теги
Граф коммитов

2738 Коммитов

This Branch
This Branch
Все ветки
Автор SHA1 Сообщение Дата
Tim Woodall
838066ab12 changed list iteration code to allow for an item to be removed 
from the list while iterating through it

This commit was SVN r2719.
 2004-09-16 14:34:03 +00:00
Tim Woodall
c32cb44a37 correct mca_oob_tcp_ping to cleanup correctly in the case of a timeout 
This commit was SVN r2718.
 2004-09-16 14:12:22 +00:00
Brian Barrett
46f755c016 * Fix missing comma in argument list to CHECK_FUNC 
This commit was SVN r2717.
 2004-09-16 13:46:24 +00:00
Brian Barrett
0716403174 * make sure to strip off quotes from the lines in post_configure.sh so that 
they don't cause escaping issues when executing the last eval.

This commit was SVN r2716.
 2004-09-16 13:45:40 +00:00
Tim Woodall
8d0851253a cleanup cleanup 
This commit was SVN r2715.
 2004-09-16 13:01:32 +00:00
Tim Woodall
cb48562fa5 temporary fix for mpirun2 - note this may break openmpi 
This commit was SVN r2714.
 2004-09-16 12:59:27 +00:00
Tim Woodall
7317caf2ca changed debug output - added sanity check 
This commit was SVN r2713.
 2004-09-16 12:58:50 +00:00
Tim Woodall
513129e816 fixed a recursive lock 
This commit was SVN r2712.
 2004-09-16 12:56:45 +00:00
Edgar Gabriel
5f29fe41fa fixing a stupid bug 
This commit was SVN r2711.
 2004-09-16 12:16:21 +00:00
Tim Woodall
30195b17cb - no need to register/unregister process info 
- cleanup event processing thread before anything else

This commit was SVN r2710.
 2004-09-16 11:30:44 +00:00
Edgar Gabriel
ed964097f5 a sever bug fix in comm_compare: we always comm1 with comm1, and ignored comm2. Strangely, we passed the Intel comm_compare test with this bug... 
This commit was SVN r2709.
 2004-09-16 10:08:40 +00:00
Edgar Gabriel
29110f6bc6 adapting some routines to the new interface of ompi_proc_get_namebuf etc... 
This commit was SVN r2708.
 2004-09-16 10:07:42 +00:00
Edgar Gabriel
3617f6d16e removing two unused functions. Furthermore, modifying the get_namebuf functions to use ompi_pack/unpack and thus work in heterogeneous environments. 
This commit was SVN r2707.
 2004-09-16 10:07:14 +00:00
Tim Woodall
3b3855fc23 move this out of rte_finalize - call explicitly from mpi_finalize 
This commit was SVN r2706.
 2004-09-16 09:25:27 +00:00
Tim Woodall
8f33ce8266 changed cleanup order slightly 
This commit was SVN r2705.
 2004-09-16 09:22:28 +00:00
George Bosilca
f5502e731d Work around differences between several versions of flex. Newer versions (3.5.21) seems to define YY_CURRENT_BUFFER_LVALUE as the correct value to be used on the left side assignments when older version can use the YY_CURRENT_BUFFER. 
This commit was SVN r2704.
 2004-09-16 09:14:01 +00:00
George Bosilca
c00407e5ff Add a temporary variable to keep the pml_wait happy as it does not support NULL. 
This commit was SVN r2703.
 2004-09-16 08:50:50 +00:00
George Bosilca
df80cc854e The correct name is DT_LONG_DOUBLE_INT, the other one is intended to disapear on the next version. 
This commit was SVN r2702.
 2004-09-16 08:49:34 +00:00
George Bosilca
4177ed8ab9 On left side of an assgnment YY_CURRENT_BUFFER should became YY_CURRENT_BUFFER_LVALUE. 
This commit was SVN r2701.
 2004-09-16 08:48:31 +00:00
George Bosilca
6a4df9ef5c Reading the documentation from the flex it look like if we use the YY_CURRENT on the left side 
of an assignment we should use the _LVALUE one.

This commit was SVN r2700.
 2004-09-16 08:47:14 +00:00
Tim Woodall
ad7db4e1cb restored call to ompi_rte_register 
This commit was SVN r2699.
 2004-09-16 08:38:24 +00:00
Tim Woodall
0ffa11b904 modified locking 
This commit was SVN r2698.
 2004-09-16 08:26:57 +00:00
Ralph Castain
d0e308fbc4 First attempt to thread safe the registry and name server subsystems. Comment out the duplicate calls to register processes in mpi_init and mpirun2. 
This commit was SVN r2697.
 2004-09-16 04:14:35 +00:00
Vishal Sahay
3aa0b648e2 More f77 file bindings 
This commit was SVN r2696.
 2004-09-16 01:20:19 +00:00
Jeff Squyres
eb4279559e Some fixes for the attribute code: 
- move the attribute init section in ompi_mpi_init() down below where
  communicators are setup (we need MPI_COMM_WORLD to be setup before
  attributes and keyvals are setup)
- removed confusing extra wrapper class around ompi_hash_table_t;
  looks like it was a victim of slow eroding of members so I put took
  it out back and put it out of its misery
- added preliminary definitions for all the pre-defined keyvals.
  Still need more work here to assign their initial values, but I
  think Edgar was running into an atrribute problem and it may have
  been that the pre-defined attrs didn't yet exist.
- removed some LAM-specific predefined keyvals from mpi.h

This commit was SVN r2695.
 2004-09-16 00:00:09 +00:00
Jeff Squyres
c89f61660a Fix minor warning -- no need to realloc here because we've never 
alloc'ed before.

This commit was SVN r2694.
 2004-09-15 23:56:40 +00:00
Ralph Castain
8f9b399b6d This is a checkpoint of some minor changes made in sequencing the startup - mainly to ensure that those helping me track a bug in mpirun2 are operating from an identical code base. 
For everyone else, this is transparent.

This commit was SVN r2693.
 2004-09-15 22:50:34 +00:00
Jeff Squyres
ae6ff056d1 Don't bother trying to use the fancy schmancy ERRHANDLER macros when 
invoking an errhandler in MPI_INIT when we don't yet have a
communicator to invoke them on.  Instead, remove some of the logic
from the ERRHANDLER macros and just invoke the back-end function
directly in MPI_INIT.

This commit was SVN r2692.
 2004-09-15 20:55:29 +00:00
Jeff Squyres
7a0c3fd490 Add some fflush()'s to help user output appear before the error output 
This commit was SVN r2691.
 2004-09-15 20:54:11 +00:00
Jeff Squyres
9bdf18443a First cut of MPI_ABORT. We might make it better someday (i.e., try to 
honor killing only the procs in the comm, do something with the
errorcode, etc.).

This commit was SVN r2690.
 2004-09-15 20:53:40 +00:00
Vishal Sahay
67bf0036b4 File bindings 
This commit was SVN r2689.
 2004-09-15 20:43:28 +00:00
Jeff Squyres
52a3a9b62b Slightly modify a few error messages (make them lower case, like the rest) 
This commit was SVN r2688.
 2004-09-15 20:00:21 +00:00
Jeff Squyres
1087508c85 Return MPI_ERR_COMM, not MPI_ERR_ARG if have invalid comm 
This commit was SVN r2687.
 2004-09-15 19:55:36 +00:00
Brian Barrett
c2747dc370 * forgot to include configure.stub 
This commit was SVN r2686.
 2004-09-15 19:49:45 +00:00
Brian Barrett
ab346859dc * dooh - add file accidently left out 
This commit was SVN r2685.
 2004-09-15 19:43:32 +00:00
Ralph Castain
c65619e294 Move the enviro parsing function a little bit. Doesn't seem to be causing a problem with a dynamic build under Mac, but Jeff noted problem with a static build under Linux. Weird.... 
This commit was SVN r2684.
 2004-09-15 19:43:32 +00:00
Brian Barrett
2d34826106 * add configure script for RMS to look for librms and librmsapi during 
configure so that we can be smarter about selection logic (note: we are
  not yet smarter about selection logic - coming soon).  To keep things
  sane, moved a bunch of files into rms/src/ from rms/.
* Since we were already forcing an autogen, move all the files from
  rsh/src/ to rsh/ for the rsh pcm component.  Makes things a bit quicker
  on the compile, with no real downside

NOTE: You will have to run autogen / configure after this update.

This commit was SVN r2683.
 2004-09-15 19:23:59 +00:00
Brian Barrett
07b3693ff0 * woops - forgot to sed one file 
This commit was SVN r2682.
 2004-09-15 19:10:00 +00:00
Brian Barrett
15d7c000c4 * add shell of a wmi (Windows Management Interface?) pcm for Anju to hack 
on.  Does not work at all, so .ompi_ignored

This commit was SVN r2681.
 2004-09-15 19:00:14 +00:00
Brian Barrett
b3aca01d1a * commit .ompi_ignored shell of a bproc component for the pcm 
This commit was SVN r2680.
 2004-09-15 18:54:57 +00:00
Sayantan Sur
0d85c2b5b9 Dynamic connection management is fully setup now. MPI_Init works with 
dynamic connection management.

This commit was SVN r2679.
 2004-09-15 16:45:31 +00:00
Ralph Castain
70dae461e4 MPI_Init will now detect and join a persistent universe - hooray! Fixed the session_dir cleanup process so it is kinder to the universe-setup file (i.e., leaves it alone), thus allowing persistent universes to retain their contact info on the session_dir tree. Adjusted mpirun2, ompid, and ompiconsole accordingly. 
Put some error protection in ompi_rte_monitor.

This commit was SVN r2678.
 2004-09-15 16:33:36 +00:00
Jeff Squyres
5de6a8c211 Fix minor logic error -- need to group the "or" together so that it 
doesn't get grouped with the "and".  Precedence, schmecedence...

This commit was SVN r2677.
 2004-09-15 09:15:50 +00:00
Vishal Sahay
b5e26196fa More bindings with macros 
This commit was SVN r2676.
 2004-09-15 07:40:58 +00:00
Tim Woodall
6221cac318 corrected documentation for send 
This commit was SVN r2675.
 2004-09-15 07:26:03 +00:00
Sayantan Sur
dcd69d5c6d Changed ptl_first_frag_size. For totally supporting dynamic 
connection management, the first *REAL* infiniband send has
to be triggered from the progress engine. Taking care of this
case with this code.

This commit was SVN r2674.
 2004-09-15 00:13:50 +00:00
Sayantan Sur
c46f5556c1 More send frag management code. 
This commit was SVN r2673.
 2004-09-14 19:49:06 +00:00
Vishal Sahay
148580acad More bindings 
This commit was SVN r2672.
 2004-09-14 19:17:52 +00:00
Jeff Squyres
a2667dae11 Remove this outdated file -- automated report pretty much covers it now 
This commit was SVN r2671.
 2004-09-14 17:08:29 +00:00
Jeff Squyres
f6a06fcbb5 There is a very confusing situation dealing with MPI-defined functions 
such as MPI_DUP_FN: MPI says that they have to be available via that
name in both C and Fortran.  However, for implementation reasons, we
have to have them as separate functions.  But if a Fortran compiler's
naming convention is ALL_CAPS, this is not possible.  Hence, we need
to put in #define's in mpi.h to change the C names something like
this: MPI_DUP_FN -> OMPI_C_MPI_DUP_FN.  This was actually done a long
time ago.

However, the source code where those C functions actually live
(src/mpi/c/attr_fn.c) still reflected the old names (e.g.,
MPI_DUP_FN).  This is fine, actually -- mpi.h would come in and
#define them to their real names (OMPI_C_MPI_DUP_FN).  So it was
functionally correct, but confusing at hell (and it just bit Edgar and
me today).  So I'm changing src/mpi/c/attr_fn.c to use the real names
(OMPI_C_MPI_DUP_FN) so that tools like ctags and grep can find them in
the source code when you go looking.  Plus, the code is just more
clear that way.  I also put in massive comments about this in
src/mpi/c/attr_fn.c and src/mpi/f77/attr_fn_f.c so that we remember
why the heck we did this.  :-)

This commit was SVN r2670.
 2004-09-14 17:03:41 +00:00