1
1

55 Коммитов

Автор SHA1 Сообщение Дата
Joseph Schuchart
557ae80858 UCX osc: allow for overlap with (some) request-based atomic operations
Signed-off-by: Joseph Schuchart <schuchart@hlrs.de>
2020-06-23 12:41:52 +02:00
Joseph Schuchart
1a3c6bbf35 UCX osc: re-use value returned by cswap to save additional get
Signed-off-by: Joseph Schuchart <schuchart@hlrs.de>
2020-06-23 12:41:52 +02:00
Joseph Schuchart
8606a02b87 UCX osc: fix macro parameter name usage in OMPI_OSC_UCX_REQUEST_RETURN
Signed-off-by: Joseph Schuchart <schuchart@hlrs.de>
2020-06-23 12:41:52 +02:00
Joseph Schuchart
d448efd49c UCX osc: properly clean up requests in case of errors
Signed-off-by: Joseph Schuchart <schuchart@hlrs.de>
2020-06-23 12:41:52 +02:00
Joseph Schuchart
73a183408f UCX osc: add support for acc_single_intrinsic info key / mca param
Signed-off-by: Joseph Schuchart <schuchart@hlrs.de>
2020-06-23 12:41:52 +02:00
Joseph Schuchart
a5cc380416 UCX osc: properly release exclusive lock to avoid lockup
Signed-off-by: Joseph Schuchart <schuchart@hlrs.de>
2019-08-27 23:01:35 +02:00
Tomislav Janjusic
d5f6b088ae osc/ucx: Fix error path
Signed-off-by: Tomislav Janjusic <tomislavj@mellanox.com>
2019-08-12 21:54:01 +03:00
Nysal Jan K.A
3529d44702 osc/ucx: Fix data corruption with non-contiguous accumulates
Signed-off-by: Nysal Jan K.A <jnysal@in.ibm.com>
2019-07-24 13:07:59 +05:30
Nysal Jan K.A
14808922cf osc/ucx: Add support for the no_locks info key
Signed-off-by: Nysal Jan K.A <jnysal@in.ibm.com>
2019-07-18 17:29:01 +05:30
Artem Polyakov
6678ac0f55 osc/ucx: Fix possible win creation/destruction race condition
To avoid fully initializing the osc/ucx component for MPI application
that are not using One-Sided functionality, the initialization happens
at the first MPI window creation.

This commit ensures atomicity of global state modifications.

Signed-off-by: Artem Polyakov <artpol84@gmail.com>
2019-06-20 09:05:03 -07:00
Artem Polyakov
0857742624 osc/ucx: Fix worker pool finalization
Signed-off-by: Artem Polyakov <artpol84@gmail.com>
2019-06-20 09:05:03 -07:00
Yossi Itigin
9d1994b906 OSC/UCX: Fix deadlock with atomic lock
Atomic lock must progress local worker while obtaining the remote lock,
otherwise an active message which actually releases the lock might not
be processed while polling on local memory location.

Signed-off-by: Yossi Itigin <yosefe@mellanox.com>
2019-05-19 20:10:09 +03:00
Mikhail Brinskii
2ef5bd8b36 SPML/UCX: Add shmemx_alltoall_global_nb routine to shmemx.h
The new routine transfers the data asynchronously from the source PE to all
PEs in the OpenSHMEM job. The routine returns immediately. The source and
target buffers are reusable only after the completion of the routine.
After the data is transferred to the target buffers, the counter object
is updated atomically. The counter object can be read either using atomic
operations such as shmem_atomic_fetch or can use point-to-point synchronization
routines such as shmem_wait_until and shmem_test.

Signed-off-by: Mikhail Brinskii <mikhailb@mellanox.com>
2019-04-26 14:47:58 +03:00
Valentin Petrov
30970bdfdf OSC/UCX: correctly handle NULL origin addr and MPI_NO_OP
Addtional bugfix: origin_addr -> result_addr for no_op, replace_op
    and sum_op fetch destination.

Signed-off-by: Valentin Petrov <valentinp@mellanox.com>
2019-04-17 10:30:21 +03:00
Artem Polyakov
19e2ae2efb opal/common/ucx: Switch to opal/tsd
Signed-off-by: Artem Polyakov <artpol84@gmail.com>
2019-02-19 14:22:07 -08:00
Artem Polyakov
7984d7d997 opal/common/ucx: Remove unused debugging macro
Will be reintroduced later if needed and after adaptation to the OMPI
infrastructure.

Signed-off-by: Artem Polyakov <artpol84@gmail.com>
2019-02-19 14:22:07 -08:00
Artem Polyakov
43f16d8796 opal/common/ucx: Remove common_ucx_int.h
Place the content of common_ucx_int.h back to the common_ucx.h and
include common_ucx_wpool.h explicitly.

Signed-off-by: Artem Polyakov <artpol84@gmail.com>
2019-02-19 14:22:07 -08:00
Xin Zhao
c6de09940f ompi/osc/ucx: Switch osc/ucx code to use Worker Pool.
Signed-off-by: Artem Polyakov <artpol84@gmail.com>
2019-02-19 14:22:07 -08:00
Yossi Itigin
83cca9d52a ucx: add owner.txt for components
Signed-off-by: Yossi Itigin <yosefe@mellanox.com>
2018-12-01 17:14:03 +02:00
Sergey Oblomov
2d230b3aac OSC/UCX: set max level value to 60
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-11-27 14:20:28 +02:00
Sergey Oblomov
e91f214982 OSC/UCX: added UCX version evaluation
- added UCX version evaluation to set OSC UCX priority

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-11-14 10:03:13 +02:00
Sergey Oblomov
36934a8bb2 OSC: set UCX module used by default
- OSC/UCX module set priority to 200 to be used by default

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-11-12 15:08:22 +02:00
Sergey Oblomov
1099d5f023 COMMON/UCX: added error code to log output
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-10-21 11:37:25 +03:00
Sergey Oblomov
df765595e3 COMMON/UCX: suppressed coverity warnings
- suppressed coverity warnings - added log messages on failed calls

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-10-17 16:11:03 +03:00
Yossi Itigin
b8e1af6fcb osc_ucx: add worker flush before osc module free
Make sure all pending communications are done on all ranks before
closing the window. This way it will be safe to close the endpoints when
closing the component.

Signed-off-by: Yossi Itigin <yosefe@mellanox.com>
2018-10-10 20:47:16 +03:00
Yossi Itigin
bcc48515e4 Revert "osc_ucx: fix hang/timeout in component finalize"
This reverts commit 438d13b4ca1e7333b789ca3fb536fda17b0feb38.

Signed-off-by: Yossi Itigin <yosefe@mellanox.com>
2018-10-10 20:47:13 +03:00
Yossi Itigin
a012ee91d8
Merge pull request #5886 from yosefe/topic/osc-ucx-fix-finalize-hang
osc_ucx: fix hang/timeout in component finalize
2018-10-10 16:29:29 +03:00
Yossi Itigin
dc6809495d osc_ucx: fix hang/timeout in component finalize
Add barrier to make sure all endpoints are destroyed before destroying
the worker.

Signed-off-by: Yossi Itigin <yosefe@mellanox.com>
2018-10-10 14:38:06 +03:00
Sergey Oblomov
ae6f81983f OSC/UCX: fixed zero-size window processing
- added processing of zero-size MPI window

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-10-10 13:08:01 +03:00
Brian Barrett
e9e4d2a4bc Handle asprintf errors with opal_asprintf wrapper
The Open MPI code base assumed that asprintf always behaved like
the FreeBSD variant, where ptr is set to NULL on error.  However,
the C standard (and Linux) only guarantee that the return code will
be -1 on error and leave ptr undefined.  Rather than fix all the
usage in the code, we use opal_asprintf() wrapper instead, which
guarantees the BSD-like behavior of ptr always being set to NULL.
In addition to being correct, this will fix many, many warnings
in the Open MPI code base.

Signed-off-by: Brian Barrett <bbarrett@amazon.com>
2018-10-08 16:43:53 -07:00
Sergey Oblomov
b72dd83f05 MCA/COMMON/UCX: added synonims for common ucx variables
- added synonims for atomic/osc modules

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-08-26 18:25:21 +03:00
Sergey Oblomov
fa33e322e7 OSC/UCX: code deduplication
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-07-19 12:39:15 +03:00
Sergey Oblomov
6f0a7a2005 OSC/UCX: opal progress register/unregister optimization
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-07-19 12:07:26 +03:00
Sergey Oblomov
55b934bacf OSC/UCX: enable progress when at least one window is allocated
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-07-18 17:52:30 +03:00
Sergey Oblomov
a081fba046 OSC/UCX: fixed hang on OSC init
- there worked progress was missed on startup which caused hang
  on one of ranks

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-07-18 17:01:53 +03:00
Xin Zhao
74ef51af1b OMPI/OSC/UCX: move memory hooks init in osc to win creation.
Move memory hooks init (for request based operation) in osc ucx to window
creation time, to avoid performance issue in MPI initialization.

Signed-off-by: Xin Zhao <xinz@mellanox.com>
2018-07-12 15:03:02 -07:00
Yossi Itigin
e77e31b50b
Merge pull request #5378 from hoopoepg/topic/unify-ucx-logging
MCA/COMMON/UCX: unified logging across all UCX modules
2018-07-08 12:45:26 +03:00
Sergey Oblomov
240670152e MCA/COMMON/UCX: code beautify - alignment
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-07-06 19:40:58 +03:00
Sergey Oblomov
eb7010933d OSC/UCX: suppressed compilation warnings
- suppressed sing/unsign-compare warnings

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-07-06 10:58:09 +03:00
Sergey Oblomov
bef47b792c MCA/COMMON/UCX: unified logging across all UCX modules
- added common logging infrastructure for all
  UCX modules
- all UCX modules are switched to new infra

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-07-05 16:25:39 +03:00
Sergey Oblomov
8080283b3d MCA/COMMON/UCX: changed return type for wait_request
- for now wait_request returns OMPI status
- updated callers

Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
2018-07-04 23:29:38 +03:00
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
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
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
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
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