ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
4 279 коммитов 3 Ветки 0 Теги
Граф коммитов

4279 Коммитов

This Branch
This Branch
Все ветки
Автор SHA1 Сообщение Дата
Jeff Squyres
cba83190af Remove extra environ declaration 
This commit was SVN r5365.
 2005-04-14 21:20:51 +00:00
Jeff Squyres
3f8e29136a Remove useless extern for environ 
This commit was SVN r5364.
 2005-04-14 21:19:15 +00:00
Tim Woodall
e157491589 back out this change from the am 
This commit was SVN r5363.
 2005-04-14 21:10:47 +00:00
Tim Woodall
ef20da3d12 cleanup environment 
This commit was SVN r5362.
 2005-04-14 21:08:36 +00:00
Tim Woodall
eb7bf1ca58 post non-blocking recv before processing callbacks - as callback 
could block - waiting for a gpr subscription

This commit was SVN r5361.
 2005-04-14 21:08:03 +00:00
Jeff Squyres
0ac08d5f49 A bunch of f90 build process fixes: 
- ensure to dist mpi_kinds.f90
- move a bunch of things inside the AM_CONDITIONAL for building f90
  stuff
  - make mpi-f90-interfaces.h depend on fortran_kinds.sh (the output
    from configure), so that if configure is run again with a different
    --with-90-max-array-dim, the file will be re-generated
  - only generate the .f90 files if we're building the f90 library
- remove some old kruft
- fully remove the use of BUILT_SOURCES

This commit was SVN r5353.
 2005-04-14 20:54:47 +00:00
Jeff Squyres
2a9c10260c Ensure that if we don't find what we're unregistering that we don't 
try to shrink the array.

This commit was SVN r5350.
 2005-04-14 20:10:27 +00:00
Jeff Squyres
7153cc6dbd Updates 
This commit was SVN r5346.
 2005-04-14 19:37:22 +00:00
Josh Hursey
78ae61c535 minor define for windows build 
This commit was SVN r5344.
 2005-04-14 19:04:16 +00:00
Brian Barrett
815964f491 * start of code to dynamically decide whether to have the sched_yield or not 
in the ompi_progress() function.  Not yet fully operational, as there 
  is something going on with the subscription not getting called.  E-mail
  coming to those interested.

This commit was SVN r5343.
 2005-04-14 18:55:53 +00:00
Tim Woodall
9b57669593 not an error if the endpoint already exists 
This commit was SVN r5342.
 2005-04-14 18:54:55 +00:00
Brian Barrett
d22594ec48 * set default value for the want MPI param checking test to be the value of 
the configure-time param checking variable, rather than "true".  Removes
  the warning about not being able to enable MPI param checking during
  MPI_INIT.

This commit was SVN r5341.
 2005-04-14 16:35:56 +00:00
Edgar Gabriel
6b6aa6431a remove the warnings which George pointed out 
This commit was SVN r5340.
 2005-04-14 16:28:15 +00:00
Edgar Gabriel
9b620f6a36 first cut on the spawn stuff - based on Ralphs commit earlier this week. 
This commit was SVN r5339.
 2005-04-14 16:05:17 +00:00
Tim Woodall
e6ea2c757b rollback earlier change 
This commit was SVN r5337.
 2005-04-14 15:32:21 +00:00
George Bosilca
425f184ce6 Send request should return OMPI_SUCCESS even if it does nothing to the request !!! 
This commit was SVN r5336.
 2005-04-14 14:35:47 +00:00
George Bosilca
f60bb44a6e Seems like nobody want to add the cast ... well now it's done. 
This commit was SVN r5335.
 2005-04-14 14:33:57 +00:00
George Bosilca
a2de24e8ce All buffered requests should be detached as soon as they are completed. 
This commit was SVN r5334.
 2005-04-14 14:30:22 +00:00
George Bosilca
2186b865c2 Construct the request before use them otherwise ... strange things will happend. 
This commit was SVN r5333.
 2005-04-14 14:29:40 +00:00
Tim Woodall
7209b35b32 fix test 
This commit was SVN r5332.
 2005-04-14 14:27:28 +00:00
Jeff Squyres
b1e7abf408 Fix bug 1255: if the basename of the agent is ssh, and we don't have 
"-x" or "-X" already in the agent executable command line, and
debugging is not enabled, then automatically add "-x" to the agent
command line to disable X forward (and avoid locking issues with
$HOME/.Xauthority, etc.).

This commit was SVN r5331.
 2005-04-14 14:08:21 +00:00
Jeff Squyres
2a26f402e1 - Use the new ompi_basename() utility function 
- Fix bug 1270 (which incorrectly identified the rsh pls as the
  problem): if we "mpirun -np 2 ./test", we would find /bin/test if it
  appeared in the PATH earlier than ".".  Now, if there is a "/" in
  the executable name anywhere, don't do a PATH search to find the
  executable. 

This commit was SVN r5330.
 2005-04-14 14:06:58 +00:00
Jeff Squyres
0cdcba3403 Add new utility function ompi_basename() for a portable version of 
basename with stronger guarantees on its semantics.  See the doxygen
comments for how to call and use it.

This commit was SVN r5329.
 2005-04-14 14:04:41 +00:00
Tim Woodall
2c0207f409 allow jobid to be allocated and passed into spawn - note that this requires 
the jobid to be initialized to ORTE_JOBID_MAX if it is to be allocated by spawn

This commit was SVN r5328.
 2005-04-14 13:12:17 +00:00
George Bosilca
551b5b6a05 We have to free the pointer not the content of the pointer (confirmed 
with the standard).

This commit was SVN r5327.
 2005-04-14 04:01:23 +00:00
Jeff Squyres
2be1a1a2f3 Fixes provided by Ralph to help cleanup of the session directory, 
especially upon abnormal termination of a process.  Not yet integrated
into the fork pls; pending more discussion with other developers.

This commit was SVN r5326.
 2005-04-14 01:04:26 +00:00
George Bosilca
1fac78adb4 Temporary commit to fix the ibm tests. I still have to take a deeper look in the datatype black magic ... *just kidding* 
This commit was SVN r5325.
 2005-04-13 21:58:56 +00:00
George Bosilca
44373667ce Use the correct fields to initialize the convertor. Not the one specified by the user but the one 
initialized by the PML. We will now correctly act for the buffered sends.

This commit was SVN r5324.
 2005-04-13 21:55:02 +00:00
Jeff Squyres
e15d0778ae Don't test for a bad condition 
This commit was SVN r5323.
 2005-04-13 21:25:25 +00:00
Tim Woodall
986da5f0b9 correct size based on alignment 
This commit was SVN r5322.
 2005-04-13 21:02:19 +00:00
Jeff Squyres
59798aad75 Fix overwriting problem -- when writing to a bool, write a bool, not 
an int (oops!)

This commit was SVN r5321.
 2005-04-13 20:42:55 +00:00
George Bosilca
86ac9eab99 If the gm_get is not supported do not even compile the long rendez-vous protocol. In same time if gm_get 
is not supported remove some of the MCA parameters.

This commit was SVN r5320.
 2005-04-13 20:15:17 +00:00
George Bosilca
dfabcbde43 Even for contiguous data we should take in account the initial displacement of the user datatype. 
This commit was SVN r5319.
 2005-04-13 19:24:18 +00:00
Jeff Squyres
c8c0ad8666 Clarification of the DEBUG stuff 
This commit was SVN r5318.
 2005-04-13 19:09:49 +00:00
Tim Woodall
3ba515dafb fix this correctly 
This commit was SVN r5317.
 2005-04-13 19:08:25 +00:00
Tim Woodall
777a680e3b cleanup 
This commit was SVN r5316.
 2005-04-13 18:50:56 +00:00
Edgar Gabriel
b591791d30 saving the latest version, which should not compile or work yet. It's just for being able to delete my last copy of the tree using the before Tim-merge infrastructure... 
This commit was SVN r5315.
 2005-04-13 18:35:07 +00:00
George Bosilca
685ab2a54e Do not return errors in add_proc. If something bad happend or if one of the node is unreacheable 
just dont create the ptl_peer struct for this node. Later the PML have to take in account the
management of the reacheable nodes.

This commit was SVN r5314.
 2005-04-13 18:25:07 +00:00
Jeff Squyres
befac00371 Fix for bug Josh found -- if we kill as a result of the 
OMPI_ERR_INIT_FINALIZE macro (meaning that MPI_INIT has not yet been
invoked or MPI_FINALIZE has already been invoked), then the
communicator argument that is passed to ompi_mpi_abort() is
meaningless and should not be consulted.

This commit was SVN r5313.
 2005-04-13 18:07:55 +00:00
Tim Woodall
3b5a9e7377 removed duplicate free 
This commit was SVN r5312.
 2005-04-13 17:51:43 +00:00
Tim Woodall
55f9f800b1 attach buffer in start and detach when the request completes - this adds some 
overhead to the critical path - but ensures buffer reference counts are released
when requests are not released by the application

This commit was SVN r5311.
 2005-04-13 17:28:06 +00:00
George Bosilca
682af1ee0f Make sure that the match is called just once after completly reading the header. This bug was raised if we didn't 
manage to read any data from the message body, just the header. Then the match was done twice ...

This commit was SVN r5310.
 2005-04-13 16:11:20 +00:00
Jeff Squyres
9e0b6baea0 Correct fix for basename() -- rather than play portability games, 
let's just have a short for loop that does effectively the same thing,
and gives us better guarantees about [not] modifying the source
buffer, and that one definitely has to free() the output when done
(the implementations of basename() on different platforms vary on
these guarantees). :-) 

This commit was SVN r5307.
 2005-04-13 15:26:33 +00:00
George Bosilca
6c79cc3e04 Now we are able to cancel some requests. Not the sends only the receives if they are not matched. 
This commit was SVN r5306.
 2005-04-13 15:25:28 +00:00
Tim Woodall
d1327e616f fix proxy 
This commit was SVN r5305.
 2005-04-13 15:24:14 +00:00
Jeff Squyres
4c67d136d8 quick workaround to fix f90 building issue -- will work properly for 
--disable-f90, but am working on a proper fix for --enable-f90

This commit was SVN r5304.
 2005-04-13 14:45:12 +00:00
Brian Barrett
6609c2e1bb * quick fix to make tree compile on OS X. See bug #1295 for the right 
solution

This commit was SVN r5303.
 2005-04-13 14:44:46 +00:00
Edgar Gabriel
4ef292717e minor fix in scatterv_inter. We should now pass all ibm inter-collective tests. 
This commit was SVN r5302.
 2005-04-13 14:34:57 +00:00
Tim Woodall
a2b1f98444 init tag 
This commit was SVN r5301.
 2005-04-13 14:09:40 +00:00
Jeff Squyres
e761e49f14 A few minor touchups (reviewed by George and Brian). 
This commit was SVN r5300.
 2005-04-13 14:04:15 +00:00