1
1
Граф коммитов

28672 Коммитов

Автор SHA1 Сообщение Дата
Sergey Oblomov
bf7fd480e9 MCA/COMMON/UCX: added non-blocking implementations of atomics
- added implementation of swap/cswap/fadd operations
- blocking add64 is replaced by non-blocking routine

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-06-25 12:25:31 +03:00
Sergey Oblomov
63e7ba6843 MCA/COMMON/UCX: added parameter for UCX/opal progress
- added parameter to set UCX/opal progresses
- minor refactoring of request wait routines

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-06-25 11:00:12 +03:00
Sergey Oblomov
d57ae62dee MCA/UCX: added common module
- implemented non-blocking routines for flush operations

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-06-22 16:41:09 +03:00
Ralph Castain
f875bfd082
Merge pull request #5311 from rhc54/topic/bind
Define a new binding method and qualifier
2018-06-21 05:40:06 -07:00
Yossi Itigin
db26c08336
Merge pull request #5307 from hoopoepg/topic/async-progress-on-mpi-fin
PML/UCX: fixed hang on MPI_Finalize
2018-06-21 13:44:14 +03:00
Sergey Oblomov
5f03628560 PML/UCX: removed uneeded flush
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-06-21 12:40:46 +03:00
Sergey Oblomov
2745da7dcc PML/UCX: use non-blocking fence instead of async progress
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-06-21 09:46:03 +03:00
Ralph Castain
f17d47087a Define a new binding method and qualifier
Allow users to request that procs be bound to a cpu in a given cpu-list based on their corresponding local rank

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-20 21:26:09 -07:00
Ralph Castain
151d13c248
Merge pull request #5310 from rhc54/topic/convert
Cover all the PMIx data types
2018-06-20 10:32:40 -07:00
Ralph Castain
5ac2ce6346 Cover all the PMIx data types
Cover all data types for OPAL-to-PMIx conversion, generating error logs when we hit something we don't support

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-20 09:06:19 -07:00
Edgar Gabriel
7bbeaf30ff
Merge pull request #5306 from edgargabriel/pr/minor-improvements
Pr/minor improvements
2018-06-20 08:43:41 -05:00
Sergey Oblomov
10f2d831ec PML/UCX: fixed hang on MPI_Finalize
- added async UCX progress thread to allow
  pending requests to complete

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-06-20 16:12:05 +03:00
Edgar Gabriel
0757cb11a8 fcoll/all components: minor updates
two minor updates:
 - in all components: use the fh->f_bytes_per_agg value
   (which might have been set by an info object) instead
   of re-reading the mca parameter
 - vulcan and dynamic_gen2: replace one allgather operation
   by an allreduce, since it is used to determine the sum
   of an array.

Signed-off-by: Edgar Gabriel <egabriel@central.uh.edu>
2018-06-20 07:47:29 -05:00
Ralph Castain
4bd745940e
Merge pull request #5305 from karasevb/fix_pmix_component
pmix/ext2x: fixed detection PMIx v2.0 by pmix component
2018-06-20 05:41:46 -07:00
Boris Karasev
39c9cb12bb pmix/ext2x: fixed detection PMIx v2.0 by pmix component
Signed-off-by: Boris Karasev <karasev.b@gmail.com>
2018-06-20 13:23:51 +03:00
Gilles Gouaillardet
6a504a1544
Merge pull request #5304 from rhc54/topic/resync
Sync to updated PMIx v3.0.0rc
2018-06-20 14:21:03 +09:00
Ralph Castain
97d4e2b578
Merge pull request #5303 from rhc54/topic/lock
Prevent thread lock when show_help msgs are emitted
2018-06-19 21:55:12 -07:00
Ralph Castain
08707c9762 Sync to updated PMIx v3.0.0rc
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-19 21:25:43 -07:00
Ralph Castain
97665d44cd Prevent thread log when show_help msgs are emitted
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-19 21:07:03 -07:00
Edgar Gabriel
df4431bd48 io/ompio: add support for some info objects
add support for the info objects cb_buffer_size and collective_buffering.
Also, introduce a new mca parameter that allows to give feedback
on whether an info object is recognized (and honored).

Signed-off-by: Edgar Gabriel <egabriel@central.uh.edu>
2018-06-19 19:34:36 -05:00
Ralph Castain
30ffdc9efc
Merge pull request #5301 from rhc54/topic/nodis
Fix the no-disconnect test
2018-06-19 14:56:12 -07:00
Ralph Castain
98b4ed9a3a Fix the no-disconnect test
A race condition exists based on whether or not the userdata object attached to a hwloc_obj_t has been initialized. These objects are setup whenever we scan for resources under that location. You therefore must not set a variable to the pointer to the userdata object and then call a function that will initialize the data in it - you need to set the variable after the function call, and protect against a NULL pointer

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-19 13:52:34 -07:00
Mikhail Kurnosov
66bc86a25b Change the tree_next to a flexible array member
Signed-off-by: Mikhail Kurnosov <mkurnosov@gmail.com>
2018-06-19 13:01:26 -06:00
Mikhail Kurnosov
6547b58316 coll/base: add knomial tree algorithm for MPI_Bcast
Signed-off-by: Mikhail Kurnosov <mkurnosov@gmail.com>
2018-06-19 13:01:26 -06:00
Noah Evans
a64abadf97 Fix mca_base_var_files separator
In the opal list parsing behavior paths should be separated by ':' while files are separated by ','. In the opal and pmix code (the pmix fix is in a separate commit) there was a mistake in the parsing such that files were being separated by ':' when they should be separated by ','s. This commit attempts to address this mismatch.

Signed-off-by: Noah Evans <noah.evans@gmail.com>
2018-06-19 12:59:07 -06:00
Edgar Gabriel
e59f58a57d
Merge pull request #5297 from edgargabriel/topic/sharedfp-revamp
sharedfp/sm and lockedfile: fix coverty warnings
2018-06-19 11:27:20 -05:00
Ralph Castain
cf48baf606
Merge pull request #5298 from rhc54/topic/cov
Silence Coverity warnings
2018-06-19 09:26:33 -07:00
Edgar Gabriel
c3ac06dc1b sharedfp/sm and lockedfile: fix coverty warnings
this commit fixes the coverty warnings CID 1437402 and
CID 1437401

Signed-off-by: Edgar Gabriel <egabriel@central.uh.edu>
2018-06-19 10:04:51 -05:00
Ralph Castain
cdb3d798f0 Silence Coverity warnings
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-19 07:47:50 -07:00
Gilles Gouaillardet
d44e0409c7
Merge pull request #4822 from ggouaillardet/topic/opal_setup_cc
configury: define C11 macros once
2018-06-19 13:42:28 +09:00
Gilles Gouaillardet
c9d3c81cbf configury: define C11 macros once
Revamp OPAL_PROG_CC_C11 macro in order to define macros only once.
Otherwise, macros get redefined during the configure process and
issue a bunch of warning in config.log. That would also cause
Open MPI fail to build if compiled with "-Werror"

Refs. open-mpi/ompi#5190

Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
2018-06-19 11:18:41 +09:00
Edgar Gabriel
f301cabff3
Merge pull request #5291 from edgargabriel/topic/sharedfp-revamp
sharedfp/individual: only complain about fseek if sharedfp operations…
2018-06-18 19:16:47 -05:00
Edgar Gabriel
9986a15b57 sharedfp/individual: only complain about fseek if sharedfp operations are really in use
this component can only be used in very specific scenarios. However, since some file systems do not support file locking and processes might be distributed over multiple nodes (hence the sm sharedfp component is also inelligible), the component might be selected in some scenarios, even if an application does not intend to use shared file pointers.

Since the fseek_shared function is involved as part of the File_set_view operation, only complain about the inability to perform the seek_shared operation if actual shared file pointer operations are being used. This avoid spurious error values being returned.

Signed-off-by: Edgar Gabriel <egabriel@central.uh.edu>
2018-06-18 18:25:29 -05:00
Ralph Castain
d6a81ecf41
Merge pull request #5287 from rhc54/topic/tools
Correct accounting for tools
2018-06-18 14:57:17 -07:00
Edgar Gabriel
bb1522472f
Merge pull request #5286 from edgargabriel/topic/sharedfp-revamp
sharedfp/all components: revamp internal operations
2018-06-18 16:09:54 -05:00
Ralph Castain
081a0d98eb Ignore the ud/oob component
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-18 13:25:16 -07:00
Ralph Castain
f0a0d606a0 Correct accounting for tools
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
(cherry picked from commit 1be080f7b92bad39745f42628a8cb6afefad2d2a)
2018-06-18 13:24:25 -07:00
Edgar Gabriel
bc0f60dfd9 sharedfp/all components: revamp internal operations
this commit revamps the internal operations of the sharedfp components.
Specifically, it is focused around removing the second file_open
operation for shared file pointers. This makes the code more efficient.
Because of that, there is no necessity anymore for the sharedfp_lazy_open
mca parameter.

Signed-off-by: Edgar Gabriel <egabriel@central.uh.edu>
2018-06-18 14:34:05 -05:00
Yossi Itigin
733cac864a
Merge pull request #5282 from yosefe/topic/pml-ucx-opal-mem-hooks
pml_ucx: add option to use opal memhooks instead of ucx internal hooks
2018-06-18 19:07:01 +03:00
Jeff Squyres
cd8d169599
Merge pull request #5276 from jsquyres/pr/info-key-len-fix
util/info: tighten up error detection on key length
2018-06-18 12:03:21 -04:00
Thananon Patinyasakdikul
13f58f3191
Merge pull request #5274 from thananon/ofi_sep
btl/ofi: add scalable endpoint support.
2018-06-18 08:41:06 -07:00
Jeff Squyres
266d5b2110
Merge pull request #5277 from jsquyres/pr/cygwin-patch
external libevent: fix for Cygwin
2018-06-18 11:08:30 -04:00
Gilles Gouaillardet
a4bfa11425
Merge pull request #5283 from ggouaillardet/topic/spc
spc: remove ompi_spc_get_count() prototype from ompi_spc.h
2018-06-18 16:56:37 +09:00
Gilles Gouaillardet
3f874c9857 spc: remove ompi_spc_get_count() prototype from ompi_spc.h
This function is only used in ompi_spc.c and is hence declared as static.
Remove its prototype from the header file in order to silence compiler warnings who will typically consider ompi_spc_get_count() as a declared but not defined function.

Fixes open-mpi/ompi#5279
Fixes open-mpi/ompi#5273

Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
2018-06-18 16:07:11 +09:00
Ralph Castain
1ec813e712
Merge pull request #5281 from rhc54/topic/ext4
Correct typo to properly check for PMIx 4
2018-06-17 11:45:44 -07:00
Ralph Castain
7981818b84 Update PMIx atomics
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-17 10:03:49 -07:00
Yossi Itigin
564f80d362 pml_ucx: add option to use opal memhooks instead of ucx internal hooks
Signed-off-by: Yossi Itigin <yosefe@mellanox.com>
2018-06-17 15:30:44 +03:00
Ralph Castain
795140e590 Make use of "instant-on" feature optional
The PMIx support for "instant on" remains experimental, so disable it by default. Provide an MCA param and corresponding command line option to enable it at runtime.

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-17 02:42:00 -07:00
Ralph Castain
fa18ba395d Sync to latest PMIx v3.0rc
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-17 02:41:46 -07:00
Ralph Castain
ea21f7175a Silence warnings and remove unused code
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-06-16 17:42:48 -07:00