ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
3 985 коммитов 3 Ветки 0 Теги 130 MiB
c2898fd946
Граф коммитов

3985 Коммитов

This Branch
This Branch
Все ветки
Автор SHA1 Сообщение Дата
Josh Hursey
c2898fd946 Fixed MPI_Abort correctly by pulling through the orte_rmgr to use when terminating processes. 
This commit was SVN r5201.
 2005-04-07 13:05:02 +00:00
Jeff Squyres
398857aba0 Add missing file to make "make dist" work in tarballs 
This commit was SVN r5200.
 2005-04-07 03:28:46 +00:00
Jeff Squyres
ca9be51a2d ROMIO does not need to check for inline; we do it in the top-level 
configure already

This commit was SVN r5199.
 2005-04-07 02:50:41 +00:00
George Bosilca
aa421ffc45 Move the definition if INADDR_NONE in the unique file that use it. Anyway the 
ompi_config_bottom.h was a bad place to have it as normally this file is included
before everything else so we always define our own INADDR_NONE generating
a lot of warnings on some OSes.

This commit was SVN r5198.
 2005-04-07 02:18:41 +00:00
Jeff Squyres
e6ac18bd3b Don't "find ... rm ..."; that makes the version number have an "M" in 
it.  Instead, just lie about $USER so that autogen doesn't do any
.ompi_unignore stuff.

This commit was SVN r5197.
 2005-04-06 23:54:53 +00:00
Brian Barrett
e9bd716715 * prevent doing the array "condensing" when 1 or 0 things in the array, as 
size_t might be unsigned and 0 - 1 on an unsigned number is not <= 0.

This commit was SVN r5196.
 2005-04-06 20:17:45 +00:00
Josh Hursey
6046262c5b 10 second fix for MPI_Abort. 
This commit was SVN r5195.
 2005-04-06 18:44:35 +00:00
Josh Hursey
d495f5ee62 fixed a couple comments 
This commit was SVN r5194.
 2005-04-06 18:38:27 +00:00
George Bosilca
8e259ddea3 In mca_pml_base_bsend_init we dont need to create an allocator to detect the thread level 
as this level is now fixed by the MPI_Init. But later we should initialize the allocator
with the correct thread level.

This commit was SVN r5193.
 2005-04-06 17:32:18 +00:00
Thara Angskun
12b2a3f56d - include <fcntl.h>: Several variables used by fcntl define in this file. 
- check if we have member si_fd in siginfo_t before use it.

This commit was SVN r5192.
 2005-04-06 05:32:11 +00:00
Josh Hursey
ef54ce8846 Remove a useless comment. Sorry about that. 
This commit was SVN r5191.
 2005-04-06 02:00:47 +00:00
Josh Hursey
cedb1d46c4 Export any environment variable that starts with OMPI_* not just 
those that start with OMPI_MCA_*

This commit was SVN r5190.
 2005-04-06 01:58:30 +00:00
Jeff Squyres
2bd8347a7e Fix compiler warnings 
This commit was SVN r5189.
 2005-04-06 01:46:39 +00:00
Jeff Squyres
a387741277 Slight modification to Geroge's fix -- it seems that the right thing 
to do is really to pack in size_t number of bytes, since that's what
pack/unpack expect as arguments for the amount of data.

This commit was SVN r5188.
 2005-04-06 00:42:00 +00:00
George Bosilca
ce679e862e This bug make my life awfull for nearly 4 hours. 64 bits machines have a size_t equal to a 64 bits unsigned int. 
Not as an unsigned int as expected by the dps_unpack function...

This commit was SVN r5187.
 2005-04-05 23:00:16 +00:00
Jeff Squyres
5946d303d8 Make this test compile again 
This commit was SVN r5186.
 2005-04-05 20:40:22 +00:00
Jeff Squyres
c36eab4749 32/64 fixes (ensure that our classes really at 64 bit clean -- and 
they are!  Mostly the tests needed to be adjusted to run on 32 and 64
systems)

This commit was SVN r5185.
 2005-04-05 20:34:56 +00:00
Jeff Squyres
c19182a0d7 Remove compiler warning 
This commit was SVN r5184.
 2005-04-05 19:54:56 +00:00
Jeff Squyres
066dd5cbac Fix some minor issues: 
- (void*) <--> function pointer casting
- missing <string.h>

This commit was SVN r5183.
 2005-04-05 19:53:56 +00:00
Jeff Squyres
5dab8563ea Fix a 32/64 issue: get_proc()/set_proc() were dealing with (void*), 
and the value we were saving/retrieving was a unit32_t.

This commit was SVN r5182.
 2005-04-05 19:41:20 +00:00
Thara Angskun
eea0e4dbfb The src/threads/condition.h should also explicit include <time.h> because 
in some OS (e.g. AIX) the struct timespec is define in <time.h> and the <sys/time.h>
is not include <time.h>

This commit was SVN r5181.
 2005-04-05 18:56:32 +00:00
Jeff Squyres
b5459e9da6 Convert the rest of the GPR tests to use loading DSO's and function 
pointers instead of direct function invocations.

This commit was SVN r5180.
 2005-04-05 18:18:27 +00:00
Jeff Squyres
3380498e79 Properly initialize item. Thanks to George for finding this. 
This commit was SVN r5179.
 2005-04-05 17:59:49 +00:00
Brian Barrett
6ac3a9ffe2 * a few tweeks to the progress function to make it go just a tad bit faster. 
Not really noticable with NetPipe, unfortunately :/.

This commit was SVN r5178.
 2005-04-05 17:58:43 +00:00
George Bosilca
f60f23898d Remove some unused variables. 
This commit was SVN r5177.
 2005-04-05 04:11:08 +00:00
George Bosilca
0cc19e2d75 Reorder the functions prototypes to match the comments. 
This commit was SVN r5176.
 2005-04-05 04:08:08 +00:00
Brian Barrett
6446127f47 * remove directives that annoy ICC 
This commit was SVN r5175.
 2005-04-05 04:08:05 +00:00
George Bosilca
81b548272f The correct error code is ORTE_ERR_OUT_OF_RESOURCE not ORTE_SUCCESS (the value inherited from the previous 
initialization).

This commit was SVN r5174.
 2005-04-05 04:07:37 +00:00
Jeff Squyres
394be4e6cb Ensure that the output is printed 
This commit was SVN r5173.
 2005-04-05 02:32:57 +00:00
Jeff Squyres
447201de15 Bunches of cleanup in dps: 
- remove all tabs from source code (replace with spaces)
- use size of size_t and int to determine some equivalent ORTE types
  in a header file so that it's always done consistently (at compile
  time)
- be consistent about using size_t for number of elements that are
  packed
- don't assume that size_t is uint32_t (it is uint64_t on some
  platforms)
- simplify some logic
- implement packing/unpacking for INT64 and UINT64 (so that we can do
  size_t on platforms where it's 64 bits)

This commit was SVN r5172.
 2005-04-05 02:32:36 +00:00
Jeff Squyres
205ed1d9d9 @#@#$ Accidentally reverse-backed out the patch in this file, causing 
replicated code.  Doh!

This commit was SVN r5171.
 2005-04-05 02:12:44 +00:00
Jeff Squyres
acb3f3a5cc Update svn:ignore 
This commit was SVN r5170.
 2005-04-05 01:58:55 +00:00
Jeff Squyres
631436517a Zero out this memory when debugging so that memory-checking debuggers 
don't give read-from-unit warnings (they may not have visibility into
the ioctl writing into the memory).

This commit was SVN r5169.
 2005-04-05 01:57:56 +00:00
Jeff Squyres
a8f370f140 Remove duplicate "gpr" -- thanks Paul Hargrove. 
This commit was SVN r5168.
 2005-04-05 01:49:06 +00:00
Jeff Squyres
26164bcf03 Remove romioconf.h. Thanks to Paul Hargrove for finding this. 
This commit was SVN r5167.
 2005-04-05 01:42:41 +00:00
Jeff Squyres
1701010301 Back out this patch; it appears to break in at least 64 bit 
environments.  Working on the fix, but don't break everyone's unit
tests while I'm working on it -- will re-commit once ompi_setenv() and
ompi_unsetenv() are fixed.

This commit was SVN r5166.
 2005-04-04 22:55:26 +00:00
Jeff Squyres
4e23a21803 Erroneously listed errmgr twice in SUBDIRS. Thanks to Paul Hargrove 
for finding this.

This commit was SVN r5165.
 2005-04-04 22:35:50 +00:00
Jeff Squyres
38b814b0cc Convert to use portable ompi_setenv() and ompi_unsetenv() 
This commit was SVN r5164.
 2005-04-04 22:17:48 +00:00
Jeff Squyres
11ea2b7e4e Check for the bozo case 
This commit was SVN r5163.
 2005-04-04 22:03:37 +00:00
Jeff Squyres
69f5aa1117 Add missing <netdb.h> 
This commit was SVN r5162.
 2005-04-04 22:03:18 +00:00
Jeff Squyres
ebb5ad37ef So that others don't have to declare it. 
This commit was SVN r5161.
 2005-04-04 21:54:38 +00:00
Jeff Squyres
bb36ebc6bc More function signature changes; thanks to Paul Hargrove for finding 
these.

This commit was SVN r5160.
 2005-04-04 19:34:15 +00:00
Brian Barrett
92e0de2a89 * make sure to ship the IA64 ASM base 
This commit was SVN r5159.
 2005-04-04 19:08:13 +00:00
Jeff Squyres
2753522f70 Change the test to properly check uint64_t in ompi_value_array -- it's 
a better test to check that instead of int's

This commit was SVN r5157.
 2005-04-04 17:53:58 +00:00
George Bosilca
dd50d36423 Improuve the perfs a little bit by changing the rendez-vous protocol. Now we start sending burst 
of data while waiting for the rendez-vous to establish.

This commit was SVN r5154.
 2005-04-04 16:02:08 +00:00
Brian Barrett
453b449596 * move sparc64 directory to sparcv9, which is slightly more descriptive 
of what is in there.  Closes bug 1167 

This commit was SVN r5153.
 2005-04-04 14:23:35 +00:00
Brian Barrett
4dc864c2a4 Some fixes to help Solaris (and others, I'm sure) compile: 
- the dirent.dt_type field is non-portable.  Use stat() if there isn't
    a dt_type field in the dirent struct
  - Make sure -laio is added to LIBS in the romio component configure
    script so that static builds don't result in missing symbols
  - Add missing header file for the signal constants in pls_fork_module

This commit was SVN r5152.
 2005-04-04 14:19:34 +00:00
Jeff Squyres
0c61e93108 Use ompi_setenv() instead of the non-portable setenv(). Thanks to 
Brian for catching this.

This commit was SVN r5151.
 2005-04-04 13:42:31 +00:00
Brian Barrett
a00db1a1e4 * return the number of tests failed so that make check aborts properly 
This commit was SVN r5150.
 2005-04-04 13:29:03 +00:00
George Bosilca
7907ac3de1 Remove useless fields on the convertor struct. 
All initializations should be done ONCE (and only once).
Remove some double checks.

This commit was SVN r5148.
 2005-04-03 02:15:36 +00:00