1
1

530 Коммитов

Автор SHA1 Сообщение Дата
Xin Zhao
c1ac0c00c5
Merge pull request #5185 from jjolly/fix-memcpy-size-mismatch
- Build warning: stringop-overflow in get_dynamic_win_info() at osc_ucx_comm.c
2018-06-29 19:37:53 -07:00
Yossi Itigin
aca61a6bfb
Merge pull request #5238 from hoopoepg/topic/fixed-coverity-issues-ucx-pml
UCX/PML: fixed few coverity issues
2018-06-27 11:14:06 +03:00
Nathan Hjelm
4c230683e7 osc/sm: fix a typo
This commit fixes a typo where a bcast is used instead of the intended
collective (barrier).

References #5262

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2018-06-26 12:53:12 -06:00
Sergey Oblomov
502d04bf12 UCX/PML/SPML: fixed few coverity issues
- fixed incorrect pointer manipulation/free
- cleaned dead code
- minor optimization on process delete routine
- fixed error handling - free pointers
- added debug output for woker flush failure

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-06-26 18:52:39 +03:00
Yossi Itigin
ee873f4f79
Merge pull request #5322 from hoopoepg/topic/mca-ucx-common
MCA/UCX: added common module
2018-06-26 13:54:12 +03:00
Joshua Ladd
256ad707f1
Merge pull request #5293 from yosefe/topic/osc-ucx-on-demand-progress
osc_ucx: register progress on-demand
2018-06-25 15:09:11 -04:00
Nathan Hjelm
e4989714c2 osc/rdma: fix data race on teardown
The osc/rdma module did not wait for all pending atomics to complete
before tearing down. This could lead to weird issues as the target
location may no longer be registered or allocated.

This commit also fixes an offset calculation issue in
ompi_osc_get_data_blocking ().

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2018-06-25 11:47:34 -06:00
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
Yossi Itigin
c2fbf3a3e8 osc_ucx: register progress on-demand
Signed-off-by: Yossi Itigin <yosefe@mellanox.com>
2018-06-19 12:47:08 +03:00
Howard Pritchard
64de269cc3 topo/treematch - quash compiler warning
quash a compiler warning showing up with gcc 7.3

Signed-off-by: Howard Pritchard <howardp@lanl.gov>
2018-06-13 16:34:17 -05:00
Nathan Hjelm
64a5baaa28
Merge pull request #5193 from hjelmn/osc_sm_location
Use /dev/shm for shared memory files in osc components
2018-06-05 09:42:14 -06:00
Nathan Hjelm
e9de42544e osc/sm: add support for controlling location of backing store
This commit adds a new MCA variable to set the location of the backing
store: osc_sm_backing_directory. The default on Linux has been
changed to use /dev/shm to improve performance in cases where /tmp is
not a tmpfs.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2018-05-29 21:44:01 -06:00
Nathan Hjelm
d0d59b1d7d osc/rdma: add support for controlling location of backing store
This commit adds a new MCA variable to set the location of the backing
store: osc_rdma_backing_directory. The default on Linux has been
changed to use /dev/shm to improve performance in cases where /tmp is
not a tmpfs.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2018-05-29 21:43:33 -06:00
Howard Pritchard
5b7c866f59 osc/pt2pt: disable when THREAD_MULTIPLE.
Per discussion at
https://github.com/open-mpi/ompi/issues/2614#issuecomment-392815654,
do not allow for selection of the OSC PT2PT when creating an MPI RMA
window when THREAD_MULTIPLE is active.  Print a helpful message and
return a not-supported error.

Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>

(cherry picked from commit d0ffd660841623c02d1dfa3151e7f7afd3327698)
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
2018-05-29 08:59:53 -07:00
John L. Jolly
36b9e15fb7 - Build warning: stringop-overflow in
get_dynamic_win_info() at osc_ucx_comm.c

In file included from /usr/include/string.h:494:0,
                 from ../../../../ompi/info/info.h:29,
                 from ../../../../ompi/mca/osc/base/base.h:24,
                 from osc_ucx_comm.c:13:
In function 'memcpy',
    inlined from 'get_dynamic_win_info' at osc_ucx_comm.c:359:5,
    inlined from 'ompi_osc_ucx_put' at osc_ucx_comm.c:401:18:
/usr/include/bits/string_fortified.h:34:10: warning: '__builtin___memcpy_chk' writing 8 bytes into a region of size 4 overflows the destination [-Wstringop-overflow=]
   return __builtin___memcpy_chk (__dest, __src, __len, __bos0 (__dest));
          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

This is caused by a type size mismatch in a call to memcpy

This fix corrects the type definition of the win_count variable.

Signed-off-by: John Jolly <jjolly@suse.com>
2018-05-22 10:11:57 -06:00
bosilca
2ab628b92e
Merge pull request #5074 from bosilca/topic/remove_warnings
Remove warnings identified by clang.
2018-05-15 11:15:23 -04:00
Nathan Hjelm
cf585d725c osc/rdma: fix SEGV will null origin in FOP in debug build
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2018-05-08 14:10:20 -06:00
Joshua Ladd
32ddc6af7e
Merge pull request #5094 from xinzhao3/topic/osc-win-fix-master
OMPI/OSC/UCX: fix issue in impl of MPI_Win_create_dynamic/MPI_Win_attach/MPI_Win_detach
2018-05-02 17:42:34 -04:00
Xin Zhao
3f5ac97649 OMPI/OSC/UCX: set priority to 0.
Signed-off-by: Xin Zhao <xinz@mellanox.com>
2018-05-02 21:40:06 +03:00
Xin Zhao
53bdfd1dcb OMPI/OSC/UCX: fix issue in impl of MPI_Win_create_dynamic/MPI_Win_attach/MPI_Win_detach
Signed-off-by: Xin Zhao <xinz@mellanox.com>
2018-04-24 23:09:52 +03:00
George Bosilca
6ff11267fb
Remove warnings identified by clang.
Plus minor spacing and indentation issues.

Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
2018-04-14 17:14:12 -04:00
Nathan Hjelm
e79debc320 osc/rdma: fix overflow in offset calculation
This commit fixes a bug is osc/rdma that can occur if the total size
of the shared memory segment gets larger than 4 GiB. The bug was
caused by a typo. The type of my_base_offset should have been size_t
not int.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2018-03-27 09:33:44 -06:00
Nathan Hjelm
f7faacca4e osc/rdma: fix 32-bit builds
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2018-03-27 09:16:04 -06:00
Jeff Squyres
06af6f1c4c
Merge pull request #4962 from jsquyres/pr/cid-fixes
A bunch of CID fixes
2018-03-26 22:30:31 -04:00
Jeff Squyres
08ceb66a19 osc/pt2pt: fix (effectively false positive) CID 1402113
This will almost certainly never happen, but be defensive and
guarantee that we never return an uninitialized variable.

Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
2018-03-26 14:26:17 -07:00
Jeff Squyres
124208198c osc/rdma: fix CID 1424327
Fix minor memory leak.

Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
2018-03-26 14:21:21 -07:00
Ralph Castain
3a93b535ec Silence the flood of OSC/RDMA warnings
Fixes #4950

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2018-03-25 16:12:41 -07:00
Nathan Hjelm
7f4872d483 osc/rdma: performance improvments and bug fixes
This commit is a large update to the osc/rdma component. Included in
this commit:

 - Add support for using hardware atomics for fetch-and-op and single
   count accumulate  when using the accumulate lock. This will improve
   the performance of these operations even when not setting the
   single intrinsic info key.

 - Rework how large accumulates are done. They now block on the get
   operation to fix some bugs discovered by an IBM one-sided test. I
   may roll back some of the changes if the underlying bug in the
   original design is discovered. There appear to be no real
   difference (on the hardware this was tested with) in performance so
   its probably a non-issue. References #2530.

 - Add support for an additional lock-all algorithm: on-demand. The
   on-demand algorithm will attempt to acquire the peer lock when
   starting an RMA operation. The lock algorithm default has not
   changed. The algorithm can be selected by setting the
   osc_rdma_locking_mode MCA variable. The valid values are two_level
   and on_demand.

 - Make use of the btl_flush function if available. This can improve
   performance with some btls.

 - When using btl_flush do not keep track of the number of put
   operations. This reduces the number of atomic operations in the
   critical path.

 - Make the window buffers more friendly to multi-threaded
   applications. This was done by dropping support for multiple
   buffers per MPI window. I intend to re-add that support once the
   underlying performance bug under the old buffering scheme is
   fixed.

 - Fix a bug in request completion in the accumulate, get, and put
   paths. This also helps with #2530.

 - General code cleanup and fixes.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2018-03-15 14:53:53 -06:00
Jeff Squyres
e7f91f8068
Merge pull request #4527 from clementFoyer/osc-no-includes
Remove inter-dependencies between OSC modules.
2018-02-09 15:49:56 -05:00
Clement Foyer
f5b4fc05f8 Remove inter-dependencies between OSC modules.
The osc monitoring component needed to include other OSC components
header in order to be able tu access communicator through the
component specific ompi_osc_*_module_t structures. This commit remove
the dependency, and resolve the issue #4523.

Extend the common monitoring API.

  * Now it's possible to translate from local rank to world rank from
    both the communicator and the group.
  * Remove useless hashtable as we directly use the w_group contained
    in window structure.

Add automatic generation at config time.

The templates are expanded at configure time. It creates a new header
file that generates all the variables/functions needed. Adding this
during the autogen automagicaly generates for each of the available
modules the proper functions.

Only keep a generated argv-style array.

Following Jeff's advice, the configure.m4 file generate a simple array
of module variables to be iterated over to find the proper module.

Signed-off-by: Clement Foyer <clement.foyer@inria.fr>
2018-02-07 11:52:00 +00:00
Gilles Gouaillardet
34b45cc879 osc/sm: fix the osc_free callback
If component selection fails, then module->bases might be unallocated
when ompi_osc_sm_free() in invoked, so test it before trying to free()
module->bases[0].

Thanks Martin Binder for the report.

Refs open-mpi/ompi#4770

Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
2018-01-31 11:23:21 +09:00
Xin Zhao
72ff2b1135 OMPI/OSC/UCX: adding atomic lock for fetch_and_op and compare_and_swap
Signed-off-by: Xin Zhao <xinz@mellanox.com>
2018-01-18 00:36:22 +02:00
Matias A Cabral
009ba475e1 osc/rmda: fix missing opal_argv_free in mtls search.
Use asprintf in description message to avoid missing default values
Signed-off-by: Matias Cabral <matias.a.cabral@intel.com>
2018-01-11 14:29:16 -08:00
Nathan Hjelm
7893248c5a opal/asm: add fetch-and-op atomics
This commit adds support for fetch-and-op atomics. This is needed
because and and or are irreversible operations so there needs to be a
way to get the old value atomically. These are also the only semantics
supported by C11 (there is not atomic_op_fetch, just
atomic_fetch_op). The old op-and-fetch atomics have been defined in
terms of fetch-and-op.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2017-11-30 10:41:23 -07:00
Nathan Hjelm
1282e98a01 opal/asm: rename existing arithmetic atomic functions
This commit renames the arithmetic atomic operations in opal to
indicate that they return the new value not the old value. This naming
differentiates these routines from new functions that return the old
value.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2017-11-30 10:41:22 -07:00
Nathan Hjelm
9d0b3fe9f4 opal/asm: remove opal_atomic_bool_cmpset functions
This commit eliminates the old opal_atomic_bool_cmpset functions. They
have been replaced by the opal_atomic_compare_exchange_strong
functions.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2017-11-30 10:41:22 -07:00
Nathan Hjelm
45db3637af osc/rdma: bug fixes
This commit fixes the following bugs:

 - Allow a btl to be used for communication if it can communicate with
   all non-self peers and it supports global atomic visibility. In
   this case CPU atomics can be used for self and the btl for any
   other peer.

 - It was possible to get into a state where different threads of an
   MPI process could issue conflicting accumulate operations to a
   remote peer. To eliminate this race we now update the peer flags
   atomically.

 - Queue up and re-issue put operations that failed during a BTL
   callback. This can occur during an accumulate operation. This was
   an unhandled error case.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2017-11-29 12:43:58 -07:00
Nathan Hjelm
67e26b6e5a
Merge pull request #4482 from matcabral/osc_rdma_skip_mtls
osc/rdma: mca parameter to list MTLs that lower osc rdma priority
2017-11-29 09:34:33 -07:00
Ralph Castain
3906aaf41a Silence warnings
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2017-11-25 11:50:18 -08:00
Matias A Cabral
80c8858c5a osc/rdma: add an mca parameter to list MTLs for which osc pt2pt should have
higher priority than rdma and default to psm2.

Context: the Intel Omni-path driver (hfi1) has verbs support, so the openib
btl is available to use. However, at a bad performance. Without this
change osc rdma using btl openib is the default choice when running on Intel
Omni-path, with a lower performance than osc pt2pt over mtl psm2.

Signed-off-by: Matias A Cabral <matias.a.cabral@intel.com>
2017-11-09 11:54:11 -08:00
Thomas Naughton
86bb6f8bac
Merge pull request #4444 from naughtont3/tjn-fix-plm-monitoring-configury
configury: single quote to avoid trouble with BSD
2017-11-08 10:56:37 -05:00
Thomas Naughton
c5dc41ee1a configury: single quote to avoid trouble with BSD
Signed-off-by: Thomas Naughton <naughtont@ornl.gov>
2017-11-03 11:34:28 -04:00
bosilca
3e1f771045
Merge pull request #4403 from naughtont3/tjn-fix-plm-monitoring-configury
fix PML monitoring configury to compile DSOs
2017-11-02 19:08:20 -04:00
Nathan Hjelm
3ff34af355 opal: rename opal_atomic_cmpset* to opal_atomic_bool_cmpset*
This commit renames the atomic compare-and-swap functions to indicate
the return value. This is in preperation for adding support for a
compare-and-swap that returns the old value. At the same time the
return type has been changed to bool.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2017-10-31 12:47:23 -06:00
Thomas Naughton
86d282d6dd fix PML monitoring configury to compile DSOs
Signed-off-by: Thomas Naughton <naughtont@ornl.gov>
2017-10-26 15:53:11 -04:00
Geoff Paulsen
be7b0af5d9 Merge pull request #3609 from markalle/pr/single_type_with_LB_UB
single_predefined_type with MPI_LB/UB
2017-10-04 15:13:09 -05:00
Mark Allen
e24d5ccb7e single_predefined_type with MPI_LB/UB
The ompi_datatype_get_single_predefined_type_from_args() recurses down
into a constructed type to identify what base datatype it's built from
if it's built from a single type.  But if the type has MPI_LB/MPI_UB,
for example
    lens[0] = 1;
    lens[1] = 1;
    disps[0] = 0;
    disps[1] = 0;
    types[0] = MPI_LB;
    types[1] = MPI_INT;
    MPI_Type_create_struct(2, lens, disps, types, &mydt);
then this function will see the base type MPI_LB as differing from MPI_INT
and will identify mydt as not being constructed from a single base type, so
the type will be rejected for calls like MPI_Accumulate.

I think those "meta data" types shouldn't result in rejection like that, and
the above mydt should still be identified as having a single base type
of MPI_INT.

Addition: boslica wanted another change discussed here
    https://github.com/open-mpi/ompi/pull/3609
relating to the calculation for "count" after identifying the
predefined_type that was being used.

Signed-off-by: Mark Allen <markalle@us.ibm.com>
2017-10-03 19:08:18 -04:00
George Bosilca
64bff0e326
Disable monitoring if we compile statically.
Protect all components against compilation on static builds.

Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
2017-09-25 12:18:23 -04:00