1
1
Граф коммитов

3238 Коммитов

Автор SHA1 Сообщение Дата
Jeff Squyres
8311428602 btl.h: whitespace cleanup
No code changes
2015-02-13 11:46:38 -07:00
Jeff Squyres
7971fd57f0 btl.h: add more description for reg/dereg functions 2015-02-13 11:46:38 -07:00
Nathan Hjelm
a3b739d117 btl/ugni: use pthread_join to wait on progress thread completion
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:38 -07:00
Nathan Hjelm
953efc3eb2 btl/openib: fix compilation issues with XRC
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:38 -07:00
Nathan Hjelm
a9763e123d add btl comment
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:38 -07:00
Nathan Hjelm
44fb8369ff opal/convertor: add a function to get the pointer for an offset (instead of the current offset)
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:37 -07:00
Nathan Hjelm
1e518504e4 btl/smcuda: update for BTL 3.0 interface 2015-02-13 11:46:37 -07:00
Nathan Hjelm
aba0675fe7 btl/vader: update for BTL 3.0 interface
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:37 -07:00
Nathan Hjelm
f8ac3fb1e8 btl/ugni: add support for atomic operations
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:37 -07:00
Nathan Hjelm
655604f509 btl/ugni: update for BTL 3.0 interface
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:37 -07:00
Nathan Hjelm
4972d97b8b btl/template: update for BTL 3.0 interface
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:37 -07:00
Nathan Hjelm
f241b6e0a7 btl/tcp: update for BTL 3.0 interface
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:36 -07:00
Nathan Hjelm
25176cad27 btl/sm: update for BTL 3.0 interface
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:36 -07:00
Nathan Hjelm
19abc19ad9 btl/self: update for BTL 3.0 interface
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:36 -07:00
Nathan Hjelm
f96d48a2e1 btl/scif: update for BTL 3.0 interface
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:36 -07:00
Nathan Hjelm
cf91156105 btl/openib: add atomic operation support
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:36 -07:00
Nathan Hjelm
74f1af4548 btl/openib: update for BTL 3.0 interface
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:36 -07:00
Nathan Hjelm
fc7397949c btl: require that btls handle descriptor = NULL in the btl_sendi function
The send inline optimization uses the btl_sendi function to achieve lower
latency and higher message rates. Before this commit BTLs were allowed to
assume the descriptor was non-NULL and were expected to return a valid
descriptor if the send could not be completed using btl_sendi. This
behavior was fine until the usage of btl_sendi was changed in ob1. This
commit allows the caller to specify NULL for the descriptor. The affected
btls have been updated to handle this case.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:36 -07:00
Nathan Hjelm
593f97ae92 btl: add support for 64-bit atomic operations
This commit adds an interface for btl's to export support for 64-bit atomic
operations on integers. BTL's that can support atomic operations should
implement these functions and set the appropriate btl_flags and btl_atomic_flags.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:36 -07:00
Nathan Hjelm
f8e15ca83d Update the interface to provide a cleaner interface for RDMA operations.
The old BTL interface provided support for RDMA through the use of
the btl_prepare_src and btl_prepare_dst functions. These functions were
expected to prepare as much of the user buffer as possible for the RDMA
operation and return a descriptor. The descriptor contained segment
information on the prepared region. The btl user could then pass the
RDMA segment information to a remote peer. Once the peer received that
information it then packed it into a similar descriptor on the other
side that could then be passed into a single btl_put or btl_get
operation.

Changes:

 - Added functions to register and deregister memory regions with the
   btl. If no registration is needed a btl should set these function
   pointers to NULL. These function take over for btl_prepare_src/dst
   and btl_free for RDMA operations. The caller should specify the
   maximum permissions needed on the memory.

 - Changed the function signatures for both btl_put and btl_get. In
   place of a prepared descriptor the caller should provide the source
   and destination addresses and registration handles as well as a
   new callback function. The callback will be provided with the local
   address and registration handle, callback context, callback data, and
   status. See mca_btl_base_rdma_completion_fn_t in btl.h.

 - Added a new btl constraint: MCA_BTL_REG_HANDLE_MAX_SIZE. This
   value specifies the maximum size of any btl's registration handle.

 - Removed the btl_prepare_dst function. This reflects the fact that
   RDMA operations no longer depend on "prepared" descriptors.

 - Removed the btl_seg_size member. There is no need to btl's to
   subclass the mca_btl_base_segment_t class anymore.

 - Expose the btl's put/get limitations with new struct members:
   btl_put_limit, btl_put_alignment, btl_get_limit, btl_get_alignment.

 - Remove the mca_mpool_base_registration_t argument from the btl_prepare_src
   function. The argument was intended to support RDMA operations and is no
   longer necessary.

 - Remove des_remote/des_remote_count from the mca_btl_base_descriptor_t
   structure. This structure member was originally used to specify the remote
   segment for RDMA operations. Since the new btl interface no longer uses
   desriptors for RDMA this member no longer has a purpose. In addition
   to removing these members the local segment structure fields have been
   renamed to from des_local/des_local_count to des_segments/des_segment_count.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-13 11:46:36 -07:00
Howard Pritchard
6a275f4489 Merge pull request #395 from hppritcha/topic/pmix_cray_kvs
pmix/cray: remove workaround for OBJ_RELEASE
2015-02-13 11:25:50 -07:00
Howard Pritchard
bd9d185951 pmix/cray: remove workaround for OBJ_RELEASE
Per feedback from rhc, manually set the base_ptr member
of the opal_buffer_t variable to NULL prior to calling
OBJ_RELEASE.  A similar feature of opal_dss.load also
exists so likewise reset the base_ptr to NULL prior to
invoking it.

Hopefully the opal_buffer_t struct does not change
frequently.

Minor cleanups to reduce output when pmix_base_verbose
mca paramater is set.
2015-02-13 07:47:26 -08:00
Gilles Gouaillardet
ca3a275823 opal/util: fix misc memory leaks reported by Coverity
fixes CID 996174, 996920, 1196735, 1196769 and 1196770
2015-02-13 14:28:59 +09:00
Jeff Squyres
f7b4b23383 usnic: ensure to NULL-terminate the string/not overflow
This was CID 1269921.
2015-02-12 13:41:30 -08:00
Jeff Squyres
8febd41a39 usnic: fix minor memory leak
This was CID 1269859.
2015-02-12 13:41:30 -08:00
Jeff Squyres
4c074da1c2 usnic: fix minor memory leak
This was CID 1269853.
2015-02-12 13:41:30 -08:00
Jeff Squyres
a7ce2d406c usnic: don't bother comparing unsigned values for <0
This was CID 1269812.
2015-02-12 13:41:30 -08:00
Jeff Squyres
caacc6ad91 usnic: properly differentiate data pool vs. malloc
usnic_fls() can actually return 0, leading us to incorrectly free() a
buffer instead of OMPI_FREE_LIST_RETURN_MT'ing it.

So add an explicit bool in the struct that tracks whether the buffer
came from malloc or a freelist.

This was CID 1269660.
2015-02-12 13:41:30 -08:00
Jeff Squyres
3b39535ebb usnic: ensure that the string is NULL-terminated
This was CID 1269666.
2015-02-12 13:41:30 -08:00
Jeff Squyres
41c6e26a38 usnic: ensure the copied string is NULL-terminated
This was CID 1269667
2015-02-12 13:41:30 -08:00
Jeff Squyres
81585c0a7c usnic: strengthen the check-if-accept()-failed test
This was Coverity CID 1269801.
2015-02-12 13:41:30 -08:00
Jeff Squyres
117e6feaa1 shmem sysv: ensure we don't shmdt(NULL)
This was CID 71999.
2015-02-12 13:41:30 -08:00
Jeff Squyres
6d3a84514f mca_base_cmd_line.c: fix minor memory leak
This was CID 1269874.
2015-02-12 13:41:29 -08:00
Jeff Squyres
a1037cd70a if.c: fix minor memory leak
This was CID 1269846.
2015-02-12 13:41:29 -08:00
Jeff Squyres
4a85f759ec opal_info_support.c: prevent a NULL pointer
If NULL is passed in, then assume the caller meant "".

This was CID 993714.
2015-02-12 13:41:29 -08:00
Jeff Squyres
29794af0e9 cmd_line.c: use strncat() instead of strcat()
Be safe about appending to the end of strings.

This was CID 71932 (and probably also others).
2015-02-12 13:41:29 -08:00
Jeff Squyres
f8e334357d mca_base_pvar.c: protect removal from list
Only remove it from the list if it is actually on the list.

This was CID 1269758.
2015-02-12 13:41:29 -08:00
Jeff Squyres
e188c75edc opal_environ.c: ensure "value" is a valid string for the setenv() case
This was CID 1269764.
2015-02-12 13:41:29 -08:00
Nathan Hjelm
f1dc29b145 btl/vader: fix modex size when xpmem is in use
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-02-12 14:06:24 -07:00
Nathan Hjelm
49ba150972 mca/base: fix path string parsing
CID 993709
2015-02-12 13:03:46 -07:00
Jeff Squyres
00c878957c mca_base_var.c: add debug check for another programming error
Coverity alerted us to the fact that there are places where
the synonym_for param is hard-coded to -1 when calling
register_variable().  It would be a coding error if synonym_for==-1
and (flags & MCA_BASE_VAR_FLAG_SYNONYM)>0, so let's add that to the
debug-only check at the top of the function.

This was CID 993717.
2015-02-12 10:24:02 -08:00
Jeff Squyres
167d72ec68 net.c: ensure to free the args in the error case
This was CID 710643.
2015-02-12 10:24:02 -08:00
Jeff Squyres
332943f1c3 pstat linux: ensure to close the file
This was CID 71983.
2015-02-12 10:24:02 -08:00
Jeff Squyres
6a64fe85a1 pstat linux: ensure read() returns >=0
This was CID 71182.
2015-02-12 10:24:02 -08:00
Jeff Squyres
8be0e0b0ca usnic: don't close fp upon error
Let the caller close fp.  Properly check for errors when calling
subroutines.

This was Coverity CID 1269995.
2015-02-12 10:24:01 -08:00
Nathan Hjelm
1c8f8c6694 opal_fifo: add a couple of memory barriers to the cswap2 implementation 2015-02-12 11:01:40 -07:00
Howard Pritchard
0cf2b478e0 Merge pull request #391 from hppritcha/topic/cray_pmi_kvs
pmix/cray: initial kvs removal work
2015-02-11 19:55:34 -07:00
Howard Pritchard
9955834ff1 pmix/cray: initial kvs removal work
Remove use of the Cray PMI KVS - which is designed for a lighweight
MPI that exchanges only a minimimal amount of connection info
(about 128 bytes per rank) - within cray/pmix.  Use Cray PMI
collective extensions instead.

This is the first of several steps to accelerate launch of
Open MPI on Cray systems using either native aprun or nativized
slurm.
2015-02-11 15:14:55 -08:00
Rolf vandeVaart
08dceda2c0 Fix logic for handling priority and eager RDMA. There was some refactoring that was done
in this code and it ended up changing the logic that is used to set up eager RDMA.
Rather than setting up eager RDMA with a high priority message, it did it the other
way around.  For some reason, CUDA-aware support did not like this.  So, basically,
restore the logic to the way it was prior to the refactoring.  The refactoring did not
intend to change this.  Lightly reviewed by hjelmn.
2015-02-11 16:38:36 -05:00
Jeff Squyres
08285c6361 lt_interface: properly check OPAL_HAVE_LTDL_ADVISE 2015-02-11 12:25:20 -08:00
Jeff Squyres
4f1996df5d various: remove $(LTDLINCL) from Makefile.am's that didn't need it 2015-02-11 12:25:20 -08:00
Ralph Castain
3de8c5c7c6 Cleanup the munge support - the credential cannot be reused for multiple connections 2015-02-10 20:34:35 -08:00
George Bosilca
e173f9b0c0 Somehow we lost one of the most critical parameter
allowing the PML to decide how to order the different
interconnects. Bring it back !
2015-02-10 20:32:05 -05:00
George Bosilca
7f4c5fa96f Add the displacement of the element to the
safeguard check.
2015-02-10 20:13:36 -05:00
Ralph Castain
3ae3b96c17 Fix master compilation - a buried header dependency must have been removed. 2015-02-10 07:22:10 -08:00
Mike Dubman
6816e3421f Merge pull request #377 from regrant/ib_wr_fix
fix problem with get_pathrecord posting too many recv requests
2015-02-10 08:47:23 +02:00
Ralph Castain
bef830efef Fix debug output 2015-02-09 20:49:04 -08:00
Ralph Castain
07134f5b17 Add munge security 2015-02-09 20:49:03 -08:00
Ralph Castain
a3275aa867 Once again, fix the blasted singleton comm_spawn 2015-02-05 17:34:25 -08:00
Jeff Squyres
0dbbffb753 pmix_base_frame: use the "= { 0 }" initializer
Per open-mpi/ompi#381, convert the specific intialization of opal_pmix
to use the generic "= { 0 }" initializer.  This form can be used to
initialize any type when the intent is just to zero out / assign
*some* value.
2015-02-05 17:51:06 -05:00
Ralph Castain
f28238af59 Fix a race condition seen by Absoft during finalize. Stop the orte progress thread without cleaning it up, thus allowing the frameworks to still cancel their posted recv's. Then cleanup the memory footprint afterwards. 2015-02-05 11:41:37 -08:00
Ralph Castain
4d882796b6 Silence warnings 2015-02-05 11:41:00 -08:00
Howard Pritchard
e508a4078e Merge pull request #376 from regrant/ib_error_fix
fixes OpenIB connect error reporting for ibv_* calls that return an errn...
2015-02-04 10:22:03 -07:00
Jeff Squyres
621af3aa07 pmix_base: fix global opal_pmix symbol for static linking on OS X
OS X has weirdness when static linking.  If a symbol is not
initialized, it is put into the common block section, and Weird Things
happen (linking when trying to using that global symbol will fail).
If you initialize the variable, it goes into a different section (and
linking to it will work).

This link (that might go stale someday) has some information about OS
X linker scope and treatment of symbol definitions:
https://developer.apple.com/library/mac/documentation/DeveloperTools/Conceptual/MachOTopics/1-Articles/executing_files.html#//apple_ref/doc/uid/TP40001829-98432-TPXREF120

Fixes #375.
2015-02-04 12:12:31 -05:00
Ryan Grant
de93497789 fix problem with get_pathrecord posting too many recv requests 2015-02-04 09:53:58 -07:00
Ryan Grant
5d5e9bc1f8 fixes OpenIB connect error reporting for ibv_* calls that return an errno 2015-02-04 09:09:14 -07:00
Jeff Squyres
a3728f09af libfabric: add another missing file to the Makefile.am 2015-02-04 04:02:27 -08:00
Jeff Squyres
66a680879e libfabric: fix header file name in Makefile.am 2015-02-03 19:41:25 -08:00
Jeff Squyres
cb7cc171f9 usnic: update README.txt notes
Update notes about copying the usnic BTL between master and the v1.8
branch.
2015-02-03 15:54:36 -08:00
Jeff Squyres
edf7232e00 usnic: enable building with an external libfabric 2015-02-03 13:46:06 -08:00
Jeff Squyres
bfa54d5d7b usnic: update to match new libfabric 2015-02-03 13:46:06 -08:00
Jeff Squyres
d2490d2fd8 libfabric: update Makefile.am to match new libfabric drop 2015-02-03 13:46:05 -08:00
Jeff Squyres
3dc0abfbc4 libfabric: update to (just past) 1.0rc1
Updated to Github ofiwg/libfabric@6b005d0d19.
2015-02-03 13:46:05 -08:00
Ralph Castain
d3267c200f Add missing OMPI-changes to libevent 2.0.22 2015-02-02 20:57:40 -08:00
Jeff Squyres
965ccab6cc libfabric: remove a few warnings
Embedding libfabric is a temporary measure; I'm removing some warning
notifications so that the output isn't so cluttered (we're getting
the real warnings fixed upstream, but the OMPI community doesn't
really care/need to see the warnings in the meantime).
2015-01-29 17:38:02 -08:00
Todd Kordenbrock
37e6096fe7 Copyright update. 2015-01-29 11:08:13 -06:00
Todd Kordenbrock
ca30e129e8 Add the option to use the Portals4 logical to physical table.
This commit adds an MCA variable to select Portals4 logical
addressing, populates the logical-to-physical mapping table and
initializes the NI in this mode.
2015-01-29 11:08:13 -06:00
George Bosilca
b9a63cbe7a One less warning. 2015-01-27 13:25:55 -05:00
Ralph Castain
294ebc907a Fix singleton operations so they can work inside a slurm environment 2015-01-27 09:29:42 -06:00
Ralph Castain
ba25e8a0ce Fix singletons 2015-01-27 09:29:42 -06:00
Ralph Castain
028b00154d Complete implementation of the schizo framework to support OMPI component 2015-01-27 09:29:42 -06:00
Jeff Squyres
436223959d usnic: update to match new libfabric APIs 2015-01-24 05:49:36 -08:00
Jeff Squyres
7d5755f62b libfabric: update to ofiwg/libfabric@b3f7af4c67
Pull down a new embedded copy of libfabric from
https://github.com/ofiwg/libfabric.
2015-01-24 05:48:48 -08:00
Howard Pritchard
4de512af66 Merge pull request #358 from hppritcha/topic/ugni_spawn_issue
btl/ugni: use PMIX_GLOBAL for modex_send in ugni
2015-01-22 12:55:46 -06:00
Howard Pritchard
056daa05bf btl/ugni: use PMIX_GLOBAL for modex_send in ugni
Using PMIX_REMOTE is not the right thing for ugni
BTL when its possible that spawned ranks end up
on the same node as some of the spawnee ranks.
2015-01-22 06:53:45 -08:00
Bert Wesarg
0d0a754c42 Remove VampirTrace. 2015-01-22 08:08:07 +01:00
Gilles Gouaillardet
9f80aa2d28 btl/openib: regression fix when rdmacm or udcm are disabled
This fixes a regression introduced in open-mpi/ompi@661c35ca67

Thanks to Mark Santcroos for reporting this issue
2015-01-20 11:31:50 +09:00
George Bosilca
da83b084f5 Shifting the datatype around should alter it's true LB and UB. 2015-01-19 02:28:17 -05:00
George Bosilca
3ae89dc686 Clarify some of the comments. 2015-01-19 02:26:59 -05:00
Rolf vandeVaart
66f6026214 Improve error message to help user figure out what to do 2015-01-16 13:55:27 -05:00
Jeff Squyres
65a279019e usnic: fix typo in memchecker usage 2015-01-16 09:42:19 -08:00
Jeff Squyres
3969fe3a94 libfabric: ensure wrapper libs are loaded for static builds
For static builds, we need to also set
<framework>_<component>_WRAPPER_EXTRA_LIBS so that the wrappers know
what other libraries to add to link executables.
2015-01-16 09:29:52 -08:00
Gilles Gouaillardet
661c35ca67 cleanup dead code caused by the removal of the --with-threads configure option 2015-01-16 19:13:59 +09:00
Gilles Gouaillardet
ac16970d21 opal_tree: use a safer syntax
intel compiler incorrectly inline this function,
so use a safer syntax to get correct generated code.
2015-01-16 18:45:55 +09:00
Gilles Gouaillardet
5687ce8a07 Revert "opal/lifo: fix type declaration when cmpset_128 is available"
This reverts commit 1ba36175be.
2015-01-16 15:18:07 +09:00
Gilles Gouaillardet
1ba36175be opal/lifo: fix type declaration when cmpset_128 is available 2015-01-16 15:12:29 +09:00
Gilles Gouaillardet
b23126497c Merge branch 'master' of https://github.com/open-mpi/ompi 2015-01-16 10:55:35 +09:00
Nathan Hjelm
006074c48d Merge pull request #332 from hjelmn/openib_updates
Openib updates
2015-01-15 15:05:18 -06:00
Jeff Squyres
d13c14ec82 CSCus22527: fix off-by-one error in checking the number of VFs
Ensure to count *this* process when checking for how many VFs we need
on the local server.

(cherry picked from commit 386c01934e98cb8dcb48ff648ecdfb0c8677baa9)
2015-01-15 11:44:29 -08:00
Jeff Squyres
4685767b2d libfabric: update usnic configury
Use new common m4 macro for choosing between libnl3 and libnl.
2015-01-15 07:12:39 -08:00
Jeff Squyres
400b02e566 libfabric: update to github:ofiwg/libfabric HEAD
Specifically: bbf0f3ea8e92c92a7cee56473ecdbbbb34cceb7d (15 Jan 2015)
2015-01-15 07:11:54 -08:00
Gilles Gouaillardet
bf6adedd70 atomic/ia32: silence warnings 2015-01-15 18:53:58 +09:00
Aurélien Bouteiller
f49981bb2a Disable coalescing until pull request #332 gets in. 2015-01-14 14:12:47 -05:00
Nathan Hjelm
cf4975501d rcache/vma: fix parent class of mca_rcache_vma_t
There was a mismatch between the structure for mca_rcache_vma_t and
the OBJ_CLASS_INSTANCE. One was opal_list_item_t and the other was
ompi_free_list_item_t. The super class in the structure looks like it
is the correct one. Changed the superclass in OBJ_CLASS_INSTANCE to
match.
2015-01-14 10:21:24 -07:00
Jeff Squyres
e4e5e7dbc0 usnic: ensure to clean up nicely in case of low resources
If there are not enough resources (e.g., low VFs), we can end up
calling finalize_one_channel() on the same channel multiple times.  So
ensure to NULL out fields that we have freed already so that we do not
try to free them a second time.

Fixes CSCus26648.
2015-01-13 14:37:31 -08:00
Jeff Squyres
8807ae2497 usnic libfabric: also set the us_netmask_be field.
From libfabric upstream commit ofiwg/libfabric@3976745.

Part of the fix for CSCus22495.
2015-01-13 12:04:57 -08:00
Jeff Squyres
d00cede718 usnic: fix if_include/exclude of CIDR-specified networks
Fix the ordering so that we obtain the usnic netmask information
*before* we do the filtering based on CIDR-specified networks.

Also requires upstream Github libfabric commit 3976745.

Fixes CSCus22495.
2015-01-13 12:04:51 -08:00
Jeff Squyres
a220b92cf8 usnic: fix function name in opal_output 2015-01-13 12:04:07 -08:00
Gilles Gouaillardet
955f3c2730 configury: check existence of the atomic_init function in libfabric
intel compilers implements atomic_init in c++ only,
so disable c11 atomic in libfabric for now
2015-01-13 16:39:41 +09:00
Gilles Gouaillardet
cbe0d26b2d configury: do test the __STDC_NO_ATOMICS__ macro for libfabric 2015-01-13 16:06:37 +09:00
Jeff Squyres
5ed688a074 usnic: enusre that we only get "usnic"-named providers
Also, a minor update to a verbose message.
2015-01-12 13:21:22 -08:00
Jeff Squyres
881b1dcf19 usnic: document libfabric abstractions
Handy tips to remember the libfabric abstractions and what they
correspond to in usnic/VIC terms.
2015-01-09 15:21:51 -08:00
Gilles Gouaillardet
194d9f84d3 btl/usnic: move call to check_reg_mem_basics()
avoid annoying memlock related messages when there is no usnic device.
2015-01-09 11:37:45 +09:00
George Bosilca
1344097d35 Turn OFF the TCP dump mechanism. 2015-01-08 18:50:49 -05:00
George Bosilca
8ddd3b3b09 Cleanup the TCP dump mechanism. 2015-01-08 18:50:05 -05:00
Nathan Hjelm
c65f026fee btl/vader: fix typo in xpmem setup 2015-01-08 12:52:38 -07:00
Nathan Hjelm
9f6faadd91 opal_fifo: add missing memory barrier in pop
Thanks to Adrian Reber for reporting this.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-01-08 09:14:56 -07:00
Gilles Gouaillardet
4c29d8e247 btl/openib: silence warning (unused code) 2015-01-08 17:18:07 +09:00
Gilles Gouaillardet
8ab605d9c5 btl/tcp: fix overflow in mca_btl_tcp_endpoint_dump() 2015-01-08 15:40:16 +09:00
Nathan Hjelm
7d206ae769 btl/ugni: fix a couple of bugs
Two fixes:

 - Do not try to return a mailbox to the free list if one wasn't
   allocated.

 - Do not try to tear down IRQ CQs if they were not created.
2015-01-07 13:48:17 -07:00
Dave Goodell
49069bc661 usnic: fix fi_av_insert (ARP resolution) bugs
We had several problems in the old code:

1. We were specifying an arbitrary timeout (100 ms) and then abandoning
   all remaining pending AV insert operations.  We would then free the
   endpoint buffer that we gave to fi_av_insert(), usually causing
   libfabric's progress thread to write to a freed buffer.

2. We were claiming in a show_help message that the timeout was
   controllable via an MCA parameter.  This commit removes that
   parameter, since there's no good method for us to specify a timeout
   like this to libfabric right now.

3. We also weren't waiting for the correct number of fi_av_insert()
   operations to complete.  We were waiting for nprocs, which is
   accidentally fine for 2 procs on separate hosts, but not for most
   other proc counts.

Reviewed-by: Jeff Squyres <jsquyres@cisco.com>
2015-01-07 08:25:17 -08:00
Gilles Gouaillardet
06e071454e btl/openib: cleanup duplicate code 2015-01-07 14:07:30 +09:00
Gilles Gouaillardet
135ecce0eb btl/openib: rename OPAL_HAVE_XRCD macro into OPAL_HAVE_CONNECTX_XRC_DOMAINS 2015-01-07 13:27:25 +09:00
George Bosilca
bf62bed65f Typo in the poll/epoll ops declaration. 2015-01-06 21:21:25 -05:00
Ralph Castain
a7c5ff2ace Update to libevent 2.0.22-stable 2015-01-06 16:37:25 -08:00
Nathan Hjelm
6733d89cf9 btl/vader: fix return code check when opening ptrace_scope file 2015-01-06 15:17:56 -07:00
Nathan Hjelm
cde79bfa60 btl/openib: misc cleanup (tabs, etc) and put credit code into a common place (was duplicated in the send and sendi paths) 2015-01-06 11:39:23 -07:00
Nathan Hjelm
9bae131589 btl/openib: fix message coalescing
There was a bug in the openib btl handling this valid sequence of
calls:

desc = btl_alloc ();
btl_free (desc);

When triggered the bug would cause either fragment loss or undefined
behavior (SEGV, etc). The problem occured because btl_alloc contained
the logic to modify the pending fragment (length, etc) and these
changes were not corrected if the fragment was freed instead of sent.

To fix this issue I 1) moved some of the coalescing logic to the
btl_send function, and 2) retry the coalesced fragment on btl_free
if it was never sent. This appears to completely address the issue.
2015-01-06 11:39:16 -07:00
Nathan Hjelm
9aaac11648 btl/openib: fix recieve queue source detection 2015-01-06 11:39:11 -07:00
Howard Pritchard
7df648f1cf btl/openib: fix problems from commit b3617e73
For systems with OFED's lacking XRC support, commit b3617e73
broke the build of the openib btl.  This commit addresses
the issues introduced by this commit.
2015-01-06 11:31:12 -07:00
Ralph Castain
4c38c31ccf Actually copy buffer contents when dss.copy of a buffer is requested 2015-01-06 09:09:06 -08:00
Gilles Gouaillardet
b3617e736e btl/openib: add XRC support with OFED 3.12+
based on an original patch contributed by Bull.
2015-01-06 15:30:52 +09:00
Howard Pritchard
c857cc926c Merge pull request #327 from hppritcha/topic/async_progress
Topic/async progress
2015-01-05 16:20:44 -07:00
Dave Goodell
8afd8487f8 opal_stdint.h: fix "#pragma GCC" warnings
This was more complicated than I would like, but it's just an
unfortunate GCC/clang difference.  I don't have access to all the C
compilers out there, so this may still have problems with other
compilers that implement some form of `#pragma GCC diagnostic` support
but don't actually behave the same as some versions of GCC.

fixes #323
2015-01-05 14:44:46 -08:00
Gilles Gouaillardet
9e9261e90a pmix: correctly set locality flags in proc_flags
do not use opal_process_info.cpuset which is not
set at that time.
2014-12-26 15:37:08 +09:00
Howard Pritchard
0a6f841d5f xpmem/config: simple xpmem search on Cray's
Use the pkg-config related m4 functions to find out where
Cray's xpmem.h and libxpmem are located on a system.

With this commit, there is no longer any need to have to
explicitly indicate an xpmem install location on the configure
line, at least for Cray systems running CLE 4.X and 5.X.
2014-12-24 14:40:06 -07:00
Howard Pritchard
065c756860 btl/ugni: improve error handling
Improve error handling when pthread functions return errors.
Remove stale debug code.
2014-12-24 11:50:24 -07:00
Howard Pritchard
f8e354ce00 btl/ugni: add a request_progress_thread mca param
Replace temporary environment variables with a MCA
parameter for the ugni btl.  A user wishing to
use the ugni btl async. progress thread needs to
set the request_progress_thread param to true.
For example, using env. variable format:

export OMPI_MCA_btl_ugni_request_progress_thread=1
2014-12-24 11:50:24 -07:00
Howard Pritchard
8b250cc15b btl/ugni: more debug cleanup 2014-12-24 11:50:24 -07:00
Howard Pritchard
f0c519517b btl/ugni: switch to using opal_progress
Switch to invoking opal_progress from the async progress
thread, rather than calling ugni btl specific progress.
2014-12-24 11:50:24 -07:00
Howard Pritchard
47747c1b27 btl/ugni: remove some debug output 2014-12-24 11:50:24 -07:00
Howard Pritchard
2d14c2a204 btl/ugni: switch to using tx cq irqs for rdma
Verified via testing with unit tests, etc. that
in fact BTE TX descriptors using CQs configured to
generate IRQs were in fact working correctly on Cray XC.  Disable
send message back to self and just use IRQs generated
by completion of TX descriptors posted to BTE.
2014-12-24 11:50:24 -07:00
Howard Pritchard
acd07d98da btl/ugni: turn off chatty debug in irq cq setup 2014-12-24 11:50:24 -07:00
Howard Pritchard
0dec2f4af7 btl/ugni: mark btl frags for irqs as btl owned
Make sure frags allocated to generate irqs to wake
the progress thread, etc. set the MCA_BTL_DES_FLAGS_BTL_OWNERSHIP
flag.
2014-12-24 11:50:23 -07:00
Howard Pritchard
d188f0bc6f btl/ugni: honor enable_mpi_threads
Honor enable_mpi_threads setting to enable the ugni btl
async progress thread.  If the app doesn't request thread-multiple
the thread will not be created.
2014-12-24 11:50:23 -07:00
Howard Pritchard
43cdcb745f btl/ugni: add missing mutex lock 2014-12-24 11:50:23 -07:00
Howard Pritchard
83bcbd1cf9 btl/ugni: compilation fixes
Fix compilation problems in ugni btl associated with
async progress additions.
2014-12-24 11:50:23 -07:00
Howard Pritchard
13ab8a9e5a btl/ugni: use MCA_BTL_DES_FLAGS_SIGNAL
Use MCA_BTL_DES_FLAGS_SIGNAL frag flag to indicate
whether or not an interrupt needs to be delivered
along with a control message going through smsg.
2014-12-24 11:50:23 -07:00
Howard Pritchard
3fc7b389ff initial async progress changes for gni 2014-12-24 11:50:23 -07:00
Devendar Bureddy
ccafc62c07 OMPI: btl openib: fix max registarable memory caluclation
- by default allow to register maximum possible (i.e 2 * total_memory)
      memory. This beheviour can be turned off using mca parameter
      "btl_openib_allow_max_memory_registration"

    - In fallback case, use device specific parameters to calulate
      memory limit.
2014-12-23 23:35:54 +02:00