ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
4 235 коммитов 3 Ветки 0 Теги 130 MiB
3c6fb03e9c
Граф коммитов

4235 Коммитов

This Branch
This Branch
Все ветки
Автор SHA1 Сообщение Дата
Tim Woodall
3c6fb03e9c - dont create shared memory pool if it won't be used 
This commit was SVN r5494.
 2005-04-22 16:18:41 +00:00
George Bosilca
88ca070e88 2 less compilation warnings. 
This commit was SVN r5492.
 2005-04-22 05:04:52 +00:00
George Bosilca
362b3ab3fa One less uninitialized variable. 
This commit was SVN r5491.
 2005-04-22 04:54:00 +00:00
George Bosilca
54d58df8d5 Adapt to the new prototypes in the datatype.h 
This commit was SVN r5490.
 2005-04-22 04:53:29 +00:00
George Bosilca
3f89dc79e2 Protect the mutex declarations. They should be undeclared except if the threads are enabled. 
Additionally it remove some compilation warnings.

This commit was SVN r5489.
 2005-04-22 04:52:40 +00:00
George Bosilca
e3c2da5db6 Cleanup the code and add some more output. 
Correct a bug where the displacement was incorrectly computed in some cases.

This commit was SVN r5488.
 2005-04-22 04:51:15 +00:00
George Bosilca
d6b1d9805f Correct a unpack bug related to the size of the unpacked object. 
This commit was SVN r5487.
 2005-04-22 04:49:57 +00:00
George Bosilca
9b090e7e71 One more test (struct containing a char followed by a double). 
Modification to match the new prototypes and remove all warnings.

This commit was SVN r5486.
 2005-04-22 04:48:53 +00:00
George Bosilca
72502a5d17 Match the new prototypes. 
Some cleanups.

This commit was SVN r5485.
 2005-04-22 04:47:55 +00:00
George Bosilca
6f92675c56 Get rid of dt_desc_t and use exclusively ompi_datatype_t. 
Correctly place the const on the functions prototypes and adapt all file to match the new prototypes.

This commit was SVN r5484.
 2005-04-22 04:46:01 +00:00
George Bosilca
67c905e76d Add a dump function for the convertor. 
This commit was SVN r5483.
 2005-04-22 04:43:28 +00:00
Brian Barrett
de128a69fb Skip test when on old LinuxThreads machines and using progress threads 
since you can't fork() in one thread and waitpid() on the child in another,
which is what this test expects you to do.  If Linux would just implement
the stupid POSIX standard already, this wouldn't be a problem.

This commit was SVN r5482.
 2005-04-21 19:33:18 +00:00
Prabhanjan Kambadur
02314bda9c Making some small changes 
This commit was SVN r5481.
 2005-04-21 17:17:40 +00:00
Brian Barrett
1ac858dcc9 * add information on whether progress or mpi threads are in the build for 
ompi_info

This commit was SVN r5480.
 2005-04-21 15:28:03 +00:00
Brian Barrett
591b8eebda * Default to only progressing the event library every 100 calls to 
ompi_progress, unless someone is actually using it (MPI-2 dynamic,
  TCP PTL).  This is only for end of MPI_Init to start of MPI_Finalize. 
  All other times, the event library will be progressed every call
  into ompi_progress().

This commit was SVN r5479.
 2005-04-21 14:58:25 +00:00
Prabhanjan Kambadur
ea0e1de399 Adding a dumb but essential check to see if any file was touched 
This commit was SVN r5478.
 2005-04-21 05:18:12 +00:00
Prabhanjan Kambadur
ed60597034 Making a minor change which goes a long way. This was a typo :) 
This commit was SVN r5477.
 2005-04-21 05:15:56 +00:00
Tim Woodall
ee52046876 make sure setup is complete before waiting on child 
This commit was SVN r5476.
 2005-04-20 15:59:22 +00:00
Tim Woodall
d885155bae comments 
This commit was SVN r5475.
 2005-04-20 15:57:06 +00:00
Tim Woodall
3a5b6b654e - dont register for sigchild callbacks until setup is complete 
- flush io output on sigchild

This commit was SVN r5474.
 2005-04-20 15:44:57 +00:00
Tim Woodall
99ca522d39 debug option to not execute ompid - just print out command line - to enable 
debug of the deamon

This commit was SVN r5473.
 2005-04-20 15:43:55 +00:00
Tim Woodall
cf7a1c7631 resolve a couple of locking/progress issues 
This commit was SVN r5472.
 2005-04-20 15:43:00 +00:00
Tim Woodall
7713012d92 change file name so that it's not removed on cleanup 
This commit was SVN r5471.
 2005-04-20 15:42:20 +00:00
Tim Woodall
cc6eae3eb9 resolve locking issue 
This commit was SVN r5470.
 2005-04-20 15:41:59 +00:00
Tim Woodall
445d20ccab remove unlock - this is done externally 
This commit was SVN r5469.
 2005-04-20 14:19:14 +00:00
Jeff Squyres
e541350b9a Oops -- meant to *delete* the CCASFLAGS, not comment them out... 
This commit was SVN r5468.
 2005-04-20 14:18:09 +00:00
Jeff Squyres
266344210f Rename src/util/libutil.la -> src/util/libompiutil.la because Libtool 
was getting confused with the recent addition of "-lutil" to get the
system-level libutil (i.e., Libtool was confusing the two and doing
Bad Things).

This commit was SVN r5467.
 2005-04-20 14:14:07 +00:00
Jeff Squyres
6bbdf75df0 Plug a few memory leaks in order to test new libtool improvements from 
Ralf W (core libtool developer).  There are still a few more to plug
in ompi_info (mainly concerned with shutting down OMPI/ORTE
subsystems), but they can wait...

This commit was SVN r5466.
 2005-04-20 13:16:32 +00:00
Jeff Squyres
191f8b2781 Move the declaration up into the C_DECLS protected area (needed for 
ompi_info). 

This commit was SVN r5465.
 2005-04-20 13:15:41 +00:00
Brian Barrett
776d3c02c9 * mention to the friendly compiler that we are going to stomp on ebx and it 
should probably deal with that fact already.

This commit was SVN r5463.
 2005-04-20 03:30:12 +00:00
Brian Barrett
8b69a532b7 * the next (and hopefullly last!) ompi_config_bottom.h fixup for Windows. 
Last night's fix completely disregarded building the DSO components,
  which need to have the dllimport flag set for all the libmpi stuff, and
  need to export their component structures.  Unfortunately needed to
  add another #define to deal with the components for windows, but
  it required the least amount of work for all the non-Windows people.

This commit was SVN r5461.
 2005-04-20 01:30:06 +00:00
Tim Woodall
469672a361 build array of progress functions for only those ptls used 
This commit was SVN r5459.
 2005-04-19 22:46:11 +00:00
Tim Woodall
6a3099a1d0 correction 
This commit was SVN r5458.
 2005-04-19 22:01:18 +00:00
Tim Woodall
5d26f93149 redefine the macros to be empty if threading is not enabled 
This commit was SVN r5457.
 2005-04-19 20:50:44 +00:00
Josh Hursey
0c3bfad751 back out windows changes 
This commit was SVN r5451.
 2005-04-19 16:01:42 +00:00
George Bosilca
29ab011aba Check with the standard and correct the behaviour of this function. 
This commit was SVN r5450.
 2005-04-19 15:32:30 +00:00
Josh Hursey
24e3ae8b94 Remove extraneous DECLSPEC on enum. 
This commit was SVN r5442.
 2005-04-19 14:33:02 +00:00
Josh Hursey
c5338de1ac Add a couple of file for the Windows build. 
This commit was SVN r5441.
 2005-04-19 14:20:00 +00:00
Jeff Squyres
ab95c2836e Allow epoll*, not just epoll_* 
This commit was SVN r5439.
 2005-04-19 12:09:13 +00:00
Thara Angskun
6569d5d569 check for POE support 
This commit was SVN r5438.
 2005-04-19 06:39:27 +00:00
George Bosilca
ad861063c8 I was thinking that void* is the most generic type of object pointer. And I was wrong as the type void* is 
signed.

This commit was SVN r5437.
 2005-04-19 06:03:53 +00:00
George Bosilca
3e54fd5fc5 Explicit cast to avoid compiler warnings. 
This commit was SVN r5436.
 2005-04-19 05:54:49 +00:00
George Bosilca
fa24b8332f By default data_map should be set to false. 
This commit was SVN r5435.
 2005-04-19 05:45:25 +00:00
George Bosilca
6a5130e2c8 Give a value to the jobid variable. 
This commit was SVN r5434.
 2005-04-19 05:44:53 +00:00
Josh Hursey
d1b12501fb usleep is not defined in Windows land. 
sleep is the equlivant command where:
sleep(10) sleeps for 10 milliseconds.

This commit was SVN r5433.
 2005-04-19 04:38:48 +00:00
Josh Hursey
fbf1ec6f2e Inching towards a compiling Windows build. 
Some cleanup changes...

This commit was SVN r5432.
 2005-04-19 04:24:13 +00:00
Josh Hursey
14ada6ff91 Fixed up for Windows build... 
This commit was SVN r5431.
 2005-04-19 04:21:50 +00:00
Brian Barrett
0964152893 clean up the OMPI_BUILDING #define. Rather than being defined to 1 if 
we are part of the source tree and not defined otherwise, we are going
with an always defined if ompi_config.h is included policy.  If
ompi_config.h is included before mpi.h or before OMPI_BUILDING is set,
it will set OMPI_BUILDING to 1 and enable all the internal code that
is in ompi_config_bottom.h.  Otherwise, it will only include the
system configuration data (enough for defining the C and C++ interfaces
to MPI, but not perturbing the user environment).

This should fix the problems with bool and the like that the Eclipse
folks were seeing.  It also cleans up some build system hacks that
we had along the way.

Also, don't use int64_t as the default size of MPI_Offset, because it
requires us including stdint.h in mpi.h, which is something we really
shouldn't be doing.

And finally, fix a ROMIO Makefile that didn't set -DOMPI_BUILDING=1,
as ROMIO includes mpi.h, but not ompi_config.h

This commit was SVN r5430.
 2005-04-19 03:51:20 +00:00
Josh Hursey
80af0b41c7 Fix include... 
This commit was SVN r5429.
 2005-04-19 03:46:27 +00:00
Brian Barrett
5f4a433086 * remove unused file 
This commit was SVN r5428.
 2005-04-19 03:43:42 +00:00