1
1
Граф коммитов

8894 Коммитов

Автор SHA1 Сообщение Дата
Nathan Hjelm
3245428e82 Merge pull request #1535 from kawashima-fj/pr/osc-pt2pt-header-fix
osc/pt2pt: Fix a struct name typo
2016-04-14 15:55:25 -06:00
Jeff Squyres
4566286b9a Merge pull request #1538 from kawashima-fj/pr/fortran-binding-fix
fortran: Fix many Fortran binding bugs
2016-04-14 17:18:59 -04:00
Nathan Hjelm
330302c4b4 Merge pull request #1534 from kawashima-fj/pr/parallel-rma-fix
osc/pt2pt: Fix tag conflicts on parallel RMA communications
2016-04-14 15:13:32 -06:00
Jeff Squyres
fdf33674b3 Merge pull request #1532 from kmroz/wip-hindexed-cleanup-1
romio,java: cleanup deprecated hindexed call
2016-04-14 17:07:31 -04:00
Jeff Squyres
2374d8fcf7 Merge pull request #1536 from kawashima-fj/pr/inplace-fix
mpi/c, mpi/fortran: Fix `MPI_IN_PLACE`-related bugs
2016-04-14 15:56:55 -04:00
Nathan Hjelm
b4e5b5c09e Merge pull request #1531 from hjelmn/bml
bml: always enable the bml
2016-04-14 10:22:33 -06:00
Nathan Hjelm
1e6b4f2f55 Merge pull request #1495 from hjelmn/new_hooks
Add new patcher memory hooks
2016-04-13 18:19:23 -06:00
Nathan Hjelm
11e2d7886e opal/memory: update component structure
This commit makes it possible to set relative priorities for
components. Before the addition of the patched component there was
only one component that would run on any system but that is no longer
the case. When determining which component to open each component's
query function is called and the one that returns the highest priority
is opened. The default priority of the patcher component is set
slightly higher than the old ptmalloc2/ummunotify component.

This commit fixes a long-standing break in the abstration of the
memory components. ompi_mpi_init.c was referencing the linux malloc
hook initilize function to ensure the hooks are initialized for
libmpi.so. The abstraction break has been fixed by adding a memory
base function that calls the open memory component's malloc hook init
function if it has one. The code is not yet complete but is intended
to support ptmalloc in 2.0.0. In that case the base function will
always call the ptmalloc hook init if exists.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2016-04-13 17:14:51 -06:00
KAWASHIMA Takahiro
4944ba7edc datatype: Fix incorrect predefined datatype names and other datatype bugs (#1537)
* datatype: Fix a incorrect datatype name of `MPI_UNSIGNED`

Name of predefined datatype for C `unsigned int` gotten by
`MPI_TYPE_GET_NAME` should be `MPI_UNSIGNED`, not `MPI_UNSIGNED_INT`.

* datatype: Fix incorrect datatype names of `MPI_C_BOOL` and `MPI_CXX_*`

Names of predefined datatypes gotten by `MPI_TYPE_GET_NAME` are:

after this commit (correct) | before this commit (incorrect)
-----------------------------------------------------------
MPI_C_BOOL                    MPI_BOOL
MPI_CXX_BOOL                  MPI_BOOL
MPI_CXX_FLOAT_COMPLEX         MPI_C_FLOAT_COMPLEX
MPI_CXX_DOUBLE_COMPLEX        MPI_C_DOUBLE_COMPLEX
MPI_CXX_LONG_DOUBLE_COMPLEX   MPI_C_LONG_DOUBLE_COMPLEX

* datatype: Fix a incorrect datatype name of `MPI_2DOUBLE_PRECISION`

Name of the predefined datatype for Fortran two `double precision`
gotten by `MPI_TYPE_GET_NAME` should be `MPI_2DOUBLE_PRECISION`,
not `MPI_2DBLPREC`.

This bug was caused by setting the name to `opal_datatype_t::name`
instead of `ompi_datatype_t::name`.

* datatype: Fix `MPI_UNSIGNED_CHAR` internal flag

`MPI_UNSIGNED_CHAR` is an integer type.

* ompi/cxx: Fix C++ `MPI::LONG_DOUBLE_INT` definition

Just a typo fix. Without this fix, `MPI::MAX_LOC` and `MPI::MIN_LOC`
cannot be used with `MPI::LONG_DOUBLE_INT` in C++ programs.

I know the C++ binding is obsolete, but fixing this is harmless.

* Add FUJITSU copyright
2016-04-12 20:17:46 +02:00
KAWASHIMA Takahiro
17d32acbb6 fortran: Add missing (P)MPI_Alloc_mem_cptr_{f,f08} symbols
This commit adds the following symbols

  MPI_Alloc_mem_cptr_f
  MPI_Alloc_mem_cptr_f08
  PMPI_Alloc_mem_cptr_f
  PMPI_Alloc_mem_cptr_f08

These are implemented in the same way as other `_cptr` routines.
2016-04-12 22:40:58 +09:00
KAWASHIMA Takahiro
d48c8525ed fortran: Fix incorrect weak symbol names 2016-04-12 22:16:32 +09:00
KAWASHIMA Takahiro
5d32a601ff fortran: Add missing interfaces (part 2) 2016-04-12 22:06:35 +09:00
KAWASHIMA Takahiro
6f09d53e34 fortran: Add missing interfaces 2016-04-12 21:44:33 +09:00
KAWASHIMA Takahiro
f3b9a49ad1 fortran: Add missing PMPI interfaces 2016-04-12 20:55:41 +09:00
KAWASHIMA Takahiro
b6cb0bc257 fortran: Fix an incorrect interface name 2016-04-12 20:48:08 +09:00
KAWASHIMA Takahiro
96e93a9c5f fortran: Sort declared subroutins in alphabetical order
And insert necessary empty lines and remove unnecessary empty lines.

No code change.
2016-04-12 20:36:46 +09:00
KAWASHIMA Takahiro
334c63cf0a fortran: Change subroutine declaration order
Same order for `comm`, `type`, and `win`.

No code change.
2016-04-12 20:10:15 +09:00
KAWASHIMA Takahiro
10c11ff5b5 fortran: Add missing MPI_DUP_FN subroutine
Though the `MPI_DUP_FN` subroutine is depricated, it is not yet removed
as of MPI-3.1.
2016-04-12 20:06:50 +09:00
KAWASHIMA Takahiro
35ea9e5c3c Add FUJITSU copyright 2016-04-12 13:47:53 +09:00
KAWASHIMA Takahiro
39bcbe439a osc/pt2pt: Fix a struct name typo
Fortunately the sizes of `ompi_osc_pt2pt_header_put_t` and
`ompi_osc_pt2pt_header_get_t` are same. So this doesn't affect
the behavior.
2016-04-11 20:55:22 +09:00
KAWASHIMA Takahiro
d3d6386578 mpi/forran: Support MPI_IN_PLACE on (I)ALLTOALLW and (I)EXSCAN
`MPI_IN_PLACE` support for `MPI_ALLTOALLW` and `MPI_EXSCAN` was
added in MPI-2.2 but it was missed in OMPI Fortran binding code.
2016-04-11 20:38:28 +09:00
KAWASHIMA Takahiro
28a0577364 osc/pt2pt: Insert breaks in long lines 2016-04-11 19:06:01 +09:00
KAWASHIMA Takahiro
5ac95df9dc osc/pt2pt: use two distinct "namespaces" for tags - revised
Before this commit, a same PML tag may be used for distinct
communications for long messages. For example, consider a condition
where rank A calls ```MPI_PUT``` targeting rank B and rank B calls
```MPI_GET``` targeting rank A simultaneously.
A PML tag for the ```MPI_PUT``` is acquired on rank A and is used
for the long-message communication from rank A to rank B.
A PML tag for the ```MPI_GET``` is acquired on rank B and is used
for the long-message communication from rank A to rank B.
These two tags may become a same value because they are managed
independently on each rank. This will cause a data corruption.

This commit separates the tag used in a single RMA communication
call, one for communication from an origin to a target, and one
for communication from a target to an origin. A "base" tag
is acquired using ```get_tag``` function and PML tag is caluculated
from the base tag by ```tag_to_target``` and ```tag_to_origin```
function.
2016-04-11 19:05:20 +09:00
KAWASHIMA Takahiro
3576ecafa7 Revert "osc/pt2pt: use two distinct "namespaces" for tags"
This reverts commit 06ecdb6aa7
to reimplement the fix completely.
2016-04-11 19:04:11 +09:00
KAWASHIMA Takahiro
eb5c31521b mpi/c: Fix MPI_IALLTOALLW memchecker 2016-04-11 18:47:30 +09:00
KAWASHIMA Takahiro
1ced7f213c mpi/c: Fix IALLTOALL{V|W} + MPI_IN_PLACE param check
`sendcounts`, `sdispls`, and `sendtype(s)` must be ignored
if `MPI_IN_PLACE` is specified for `sendbuf`.
This commit makes the param check code same as the blocking
`ALLTOALL{V|W}` function.
2016-04-11 18:34:11 +09:00
Karol Mroz
f8ecdbd623 java: replace deprecated hindexed call
Signed-off-by: Karol Mroz <mroz.karol@gmail.com>
2016-04-10 19:56:22 +02:00
Karol Mroz
5c54184986 romio: replace deprecated hindexed call
Signed-off-by: Karol Mroz <mroz.karol@gmail.com>
2016-04-10 19:56:22 +02:00
Nathan Hjelm
c6b19818be bml: always enable the bml
This commit ensures the bml is always enabled whether or not it will
be used. This ensures that any available btls communicate their modex
so that they can be used for one-sided communication.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2016-04-08 21:14:17 -06:00
George Bosilca
896f857fc4 Thanks @hjelmn for catching up the typo. 2016-04-07 13:56:26 -04:00
Thananon Patinyasakdikul
92290b94e0 Fixed Coverity reports 1358014-1358018 (DEADCODE and CHECK_RETURN) 2016-04-07 12:52:17 -04:00
Ryan Grant
7cdf50533c Merge pull request #1314 from francois-wellenreiter/osc_disable_portals4_evt_send
OSC portals4 : do not generate an EVENT_SEND to avoid to filter it
2016-04-07 10:04:27 -06:00
Gilles Gouaillardet
7b803ac557 MPI_Unpack: fix error code when insize <= 0
this fixes a regression from open-mpi/ompi@f2e33c725f
2016-04-06 09:47:21 +09:00
Karol Mroz
a468c3ba1a opal_info_support: pass component map when handling params
Pass component_map to opal_info_do_params(). It will be needed to output
component versions.

Signed-off-by: Karol Mroz <mroz.karol@gmail.com>
2016-04-02 21:17:44 +02:00
Gilles Gouaillardet
f2e33c725f MPI_Unpack: fix return status
this regression was previously introduced in open-mpi/ompi@221e6e2eab
2016-03-31 09:56:54 +09:00
Gilles Gouaillardet
5932287cef datatype/[un]pack_external[_size]: move subroutines down to ompi/datatype
so it can be directly used by test/datatype/external32
2016-03-30 13:01:33 +09:00
Gilles Gouaillardet
221e6e2eab Add the datatype checks to the pack/unpack functions.
The datatype must satisfy the same constraints as for the
corresponding communication function (send for pack and
recv for unpack).
2016-03-30 11:40:08 +09:00
Gilles Gouaillardet
a89f113507 mpi/c: add missing OPAL_CR_EXIT_LIBRARY() in [un]pack[_external] 2016-03-30 11:25:21 +09:00
George Bosilca
004c0cc05b Fix issues identified by @derbeyn. 2016-03-29 15:50:32 -04:00
Jeff Squyres
91c54d7a07 Merge pull request #1491 from ICLDisco/progress_thread
BTL TCP async progress
2016-03-29 06:26:10 -04:00
George Bosilca
f69eba1bc4 Update the copyright and cleanup the code.
Per @jsquyres suggestion remove all trailing spaces.
Credit to `sed -i.bak 's/ *$//' */[ch]`.
2016-03-28 14:41:01 -04:00
Thananon Patinyasakdikul
92062492b9 Enable Threading in the BTL TCP
Added mca parameter to turn progress thread on/off
Add a flag to check if we have btl progress thread.
Added macro for ob1 matching lock.
Update the AUTHORS file.
2016-03-28 14:41:01 -04:00
Nathan Hjelm
9d5eeecb8a pml/ob1: detect unreachable errors
This commit adds code to detect when procs are unreachable when using
the dynamic add_procs functionality.

Fixes #1501

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2016-03-28 10:52:40 -06:00
Gilles Gouaillardet
1baed498b6 win: silence a warning in alloc_window(...) 2016-03-28 14:57:31 +09:00
Nathan Hjelm
d6e90f24b1 Merge pull request #1483 from hjelmn/flag_enum_2
RFC: Add support for flag enumerators for MCA variables
2016-03-26 11:43:33 -06:00
George Bosilca
57eadb0dd6 Fix for Coverity CID 1357152.
Or at least that was the origin of the issue. It turns out
we were freeing the wrong buffer (but as it only happen in the
case of an error we never noticed).
2016-03-24 00:53:30 -04:00
George Bosilca
4b38b6bd0c Fix multiple issues with the collective requests.
This patch addresses most (if not all) @derbeyn concerns
expressed on #1015. I added checks for the requests allocation
in all functions, ompi_coll_base_free_reqs is called with the
right number of requests, I removed the unnecessary basic_module_comm_t
and use the base_module_comm_t instead, I remove all uses of the
COLL_BASE_BCAST_USE_BLOCKING define, and other minor fixes.
2016-03-23 18:35:41 -04:00
Nathan Hjelm
a1420003b6 ompi/comm: clean up includes in comm_request.h
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2016-03-22 09:17:38 -06:00
Nathan Hjelm
b15a45088c mca: add support for flag enumerators
This commit adds a new type of enumerator meant to support flag
values. The enumerator parses comma-delimited strings and matches
each string or value to a list of valid flags. Additionally, the
enumerator does some basic checks to see if 1) a flag is valid in the
enumerator, and 2) if any conflicting flags are specified.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2016-03-21 15:20:56 -06:00
Todd Kordenbrock
2122a15217 Merge pull request #1443 from francois-wellenreiter/fix_trig_rndv
MTL portals4 : fix around triggered rndv operations
2016-03-21 08:16:33 -05:00