1
1
Граф коммитов

3776 Коммитов

Автор SHA1 Сообщение Дата
Gilles Gouaillardet
4d2c7f7de1 cuda: fix missing #include opal/util/argv.h 2015-12-07 14:10:32 +09:00
George Bosilca
4d00c59b2e Cleanup the memory handling for temporary buffers in
some of the collective modules. Added a new function
opan_datatype_span, to compute the memory span of
count number of datatype, excluding the gaps in the
beginning and at the end. If a memory allocation is
made using the returned value, the gap (also returned)
should be removed from the allocated pointer.
2015-12-02 20:42:18 -05:00
Ralph Castain
9803d69d02 Ensure the embedded PMIx respects an OMPI-level --disable-debug 2015-12-01 08:00:24 -08:00
rhc54
ac892a667a Merge pull request #1168 from rhc54/topic/libevent
Somehow, this got left out when merging PR #1109, so let's try it again.
2015-12-01 07:56:36 -08:00
Artem Polyakov
6a4ec3396a Merge pull request #1164 from hjelmn/mlx5_atomics_update
btl/openib: fix compile problems when using experimental verbs
2015-12-01 19:14:31 +05:00
Ralph Castain
7ac5c082ff Somehow, this got left out when merging PR #1109, so let's try it again. 2015-12-01 06:02:29 -08:00
igor-ivanov
d8c85738ab Merge pull request #1151 from igor-ivanov/pr/opal-abort-vars
Add new mca variables opal_abort_delay and opal_abort_print_stack
2015-12-01 16:27:11 +04:00
Nathan Hjelm
191aebb9c8 btl/openib: fix compile problems when using experimental verbs
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-11-30 22:21:26 -07:00
Nathan Hjelm
bb8e347371 btl/openib: update experimental verbs support
This update adds an additional check (if supported) to see if 8-byte
atomics are supported by the hardware. If 8-byte atomics are not
supported the atomics support is disabled.

This commit also includes some cleanup.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-11-30 12:32:04 -07:00
igor.ivanov@itseez.com
c15bf147bf opal: Add opal_abort_print_stack mca variable with aliases for ompi/oshmem
This commit allows to control output during abnormal oshmem/ompi application
termination.
Fixed issue in backtrace output. HAVE_BACKTRACE was never set so user was limited
in control of this variable.
Two related mca variables are moved to opal layer. Corresponding aliases are
added for ompi and oshmem.
2015-11-25 18:18:33 +02:00
Nathan Hjelm
02a6c6856d btl/openib: add support for mlx5 atomic operations
This commit adds support for fetch-and-add and compare-and-swap when
using the mlx5 driver. The support is only enabled if the expanded
verbs interface is detected. This is required because mlx5 HCAs return
the atomic result in network byte order. This support may need to be
tweaked if Mellanox commits their changes into upstream verbs.

Closes open-mpi/ompi#1077

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-11-23 16:07:12 -07:00
Francois WELLENREITER
251009e0aa BTL portals4: remove useless PtlMDBind PtlMDRelease calls for RMDA 2015-11-19 14:51:00 +01:00
Matias A Cabral
254a05dbbb Default values for Intel HFI1 (OmniPath gen1 device) in openib btl 2015-11-11 12:35:35 -08:00
Nathan Hjelm
2c02294389 opal_free_list: fix strange size check
OPAL free lists can be initialized with a fragment size that differs
from the size of objects from a class. This allows the free list code
to support OPAL objects that have flexible array members.

Unfortunately the free list code will throw out the desired length in
some cases. The code in question was committed in
open-mpi/ompi@90fb58de. The side effects of this are varied and can
cause segmentation faults, assert failures, hangs, etc. This commit
adds a check to ensure the requested size is at least as large as the
class size and makes opal_free_list allocations always honor the
requested fragment size (as long as it is larger than the class
size).

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-11-09 19:47:55 -07:00
Ralph Castain
52ea538bc1 Per fix from Nysal: set the listener_active flag before starting the progress thread, and declare the flag to be volatile 2015-11-09 09:00:59 -08:00
Ralph Castain
ee9aa67483 Update the libevent renaming file to ensure that all public symbols are covered 2015-11-07 12:52:31 -08:00
rhc54
7a9b9325a8 Merge pull request #1107 from rhc54/topic/pmix
Work on cleaning up memory leaks that are causing orte-dvm to eventua…
2015-11-06 17:16:49 -07:00
Ralph Castain
fed28e4cfc Add missing file that was previously ignored 2015-11-06 14:37:09 -08:00
Ralph Castain
5f446570d8 Work on cleaning up memory leaks that are causing orte-dvm to eventually run out of memory. Still don't have everything plugged, but getting better. Sync to the PMIx master that includes removal of the pmix_common.h.in file that really didn't need to be generated, and update to the PMIx_server_init API. 2015-11-06 14:15:30 -08:00
Jeff Squyres
b35b708979 tcp BTL: fix inconsistent whitespace problems
No code/logic changes.
2015-11-06 12:41:13 -08:00
Jeff Squyres
300cff2b89 usnic: fix/update the usnic stats
1. Fix: old v1.6-era code reset the stats-emitting event to fire twice
   for each time period.
1. Add the usNIC device name to the output for differentiating the
   output in multi-rail scenarios.
2015-11-06 12:05:34 -08:00
Ralph Castain
bfdf08ae86 Fix intercomm_create by ensuring that both sides know how to translate jobid to/from nspace
Return something just to ensure that pack is happy
2015-11-06 02:19:45 -08:00
Nathan Hjelm
4ddbdad772 btl/openib: fix access flags
Per spec for ibv_reg_mr if remote write or remote atomic is requested also
need to specify local write.

Signed-off-by: Nathan Hjelm <hjelmn@me.com>
2015-11-04 15:23:11 -07:00
KAWASHIMA Takahiro
d4bdf405bd opal/threads: Correct nsec -> usec conversion. 2015-11-04 11:28:43 +09:00
KAWASHIMA Takahiro
60546c6418 opal/datatype: Fix a macro value typo for heterogeneous.
This affects behaiviors only on a heterogeneous environment.
2015-11-04 11:28:43 +09:00
KAWASHIMA Takahiro
2dcb2d711b Makefile: Move fd.c to SOURCES from headers.
And reorder fd.h and few.h in alphabetical order.
2015-11-04 11:28:43 +09:00
bosilca
f1a5362f94 Merge pull request #1072 from bosilca/topic/resized
Fix for the subarray and darray type creation issue.
2015-11-01 21:17:03 -05:00
George Bosilca
6c54e63ecb Allow the PVARs to be written as requested by the MPI standard. 2015-10-31 17:13:35 -04:00
George Bosilca
5224a7ce4d Allow the pvar to be written by invoking the associated callback.
Use a PVAR to generate the monitoring dump of the information into a
file.

Use the PVAR to instruct the PML monitoring when to do the dump.
2015-10-31 17:13:35 -04:00
Nathan Hjelm
70186f9145 mca/base: revive mca_base_component_repository_retain_component
This commit revives the component retention functionality that was
removed as part of the component repository rewrite. The new
mca_base_component_repository_retain_component function works by
preventing the dlclosing of a dynamic component until a matching call
to mca_base_component_repository_release is made.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-10-31 17:13:35 -04:00
Rolf vandeVaart
7a25214911 Merge pull request #1082 from rolfv/pr/fix-cuda-require-41
Fix a few more places that utilized CUDA 4.1 checks
2015-10-30 12:40:34 -04:00
Rolf vandeVaart
2e2e175f13 Fix a few more places that utilized CUDA 4.1 checks 2015-10-30 09:43:24 -04:00
Ralph Castain
206e9a011e Add a couple of missing translations to/from PMIx internal and OPAL error constants 2015-10-29 12:33:02 -07:00
Rolf vandeVaart
f2ff6e03ab Make CUDA 4.1 a requirement for CUDA-aware support.
Remove all related preprocessor conditionals.
2015-10-29 11:24:02 -04:00
Ralph Castain
8ad9b450c4 Silence Coverity warning 2015-10-28 20:10:28 -07:00
George Bosilca
c2b7c2d904 Revert d667e53b. 2015-10-28 09:09:19 -04:00
George Bosilca
d667e53b30 Add a missing header (for sscanf). 2015-10-28 00:55:15 -04:00
George Bosilca
c9d0fffab3 Add a missing include. 2015-10-28 00:50:58 -04:00
Ralph Castain
267ca8fcd3 Cleanup the PMIx direct modex support. Add an MCA parameter pmix_base_async_modex that will cause the async modex to be used when set to 1. Default it to 0 for now
to continue current default behavior.

Also add an MCA param pmix_base_collect_data to direct that the blocking fence shall return all data to each process. Obviously, this param has no effect if async_
modex is used.
2015-10-27 17:31:56 -07:00
Ralph Castain
a7045352e2 Modify the OPAL_LIST_RELEASE and OPAL_LIST_DESTRUCT macros to release the objects only when the list object is refcounted down to 1, which will then reach zero when destructed/released at the end of the macro
`
2015-10-27 16:42:46 -07:00
Ralph Castain
02cdd046bd Revert " Releasing the list items when list destructor is called"
This reverts commit 7579ae3086.
2015-10-27 15:24:48 -07:00
Jeff Squyres
e85406721a verbs_usnic: add missing <stdio.h> 2015-10-27 05:22:38 -07:00
George Bosilca
679dc9b437 Fix the subarray and darray type creation. Include a
small patch provided by Gilles.
2015-10-26 23:44:26 -04:00
Jeff Squyres
c28324c5c8 common_usnic: move fake IBV provider to libopen-pal
Move the fake usnic IBV provider out of common/verbs and into a new
common/verbs_usnic component that is always statically linked into
libopen-pal.  The fake provider is registered with libibverbs at run
time, but there is no *un*register IBV API.  Hence, we can't let the
code containing this provider be dlclosed -- which means it needs to
be statically linked into libopen-pal.

Fixes open-mpi/ompi#1060.
2015-10-26 14:29:54 -07:00
George Bosilca
6c28f114f1 Silence a warning regarding the format str for snprintf. 2015-10-24 15:24:40 -04:00
Jeff Squyres
b43fcb7695 Merge pull request #1028 from ggouaillardet/poc/pmix1xx_configury
pmix1xx configury: invoke sub-configure with CFLAGS and CPPFLAGS on t…
2015-10-24 13:19:33 -04:00
Ralph Castain
43e0c5bfef Release the allocator list items from the list before destructing 2015-10-23 12:26:49 -07:00
Ralph Castain
4c12022a50 Silence a couple of warnings from valgrind and compilers. Since some pmix components may return success with a NULL value from a "get", check for that situation before attempting to unload the data. Preset the hostname before calling modex_recv to get it so unload properly checks for NULL. Cast a returned value to the correct ompi_proc_t pointer 2015-10-22 20:56:02 -07:00
Nathan Hjelm
0b9a0c2063 Merge pull request #1053 from hjelmn/openib_fixes
udcm: fix bugs
2015-10-22 15:39:11 -06:00
Gilles Gouaillardet
0221f59197 pmix1xx configury: invoke sub-configure with CFLAGS and CPPFLAGS on the command line
if CFLAGS and/or CPPFLAGS are passed to the ompi configure command line, pmix1xx
configure will not use the correct ones previously passed in the environment
see discussion started at http://www.open-mpi.org/community/lists/devel/2015/10/18159.php
Thanks Siegmar Gross for bringing this to our attention
2015-10-22 10:13:52 +09:00