1
1

2991 Коммитов

Автор SHA1 Сообщение Дата
Eugene Loh
c4adfd1806 Increase mpool_sm_min_size default to 64M so osu_bw will run.
This commit was SVN r20944.
2009-04-06 18:37:00 +00:00
Nysal Jan
5032f59edf Fix checksum computation in the buffered send code
This commit was SVN r20935.
2009-04-03 07:09:24 +00:00
Ralph Castain
ba1a98c398 Fix a warning message by pointing to the correct header
This commit was SVN r20930.
2009-04-02 13:54:59 +00:00
Nysal Jan
e561a6c43a Add missing checksum calculation. This fixes a checksum mismatch failure while using TCP BTL
This commit was SVN r20927.
2009-04-01 20:48:35 +00:00
Jeff Squyres
fc8993ba87 mpool name is the first show_help param, not the last.
This commit was SVN r20925.
2009-04-01 18:42:01 +00:00
Jeff Squyres
0d52271cd6 Per http://www.open-mpi.org/community/lists/announce/2009/03/0029.php
and https://svn.open-mpi.org/trac/ompi/ticket/1853, mallopt() hints do
not always work -- it is possible for memory to be returned to the OS
and therefore OMPI's registration cache becomes invalid.

This commit removes all use of mallopt() and uses a different way to
integrate ptmalloc2 than we have done in the past.  In particular, we
use almost exactly the same technique as MX:

 * Remove all uses of mallopt, to include the opal/memory mallopt
   component.
 * Name-shift all of OMPI's internal ptmalloc2 public symbols (e.g.,
   malloc -> opal_memory_ptmalloc2_malloc).
 * At run-time, use the existing glibc allocator malloc hook function
   pointers to fully hijack the glibc allocator with our own
   name-shifted ptmalloc2.
 * Make the decision whether to hijack the glibc allocator ''at run
   time'' (vs. at link time, as previous ptmalloc2 integration
   attempts have done).  Look at the OMPI_MCA_mpi_leave_pinned
   and OMPI_MCA_mpi_leave_pinned_pipeline environment variables and
   the existence of /sys/class/infiniband to determine if we should
   install the hooks or not.
 * As an added bonus, we can now tell if libopen-pal is linked
   statically or dynamically, and if we're linked statically, we
   assume that munmap intercept support doesn't work.

See the opal/mca/memory/ptmalloc2/README-open-mpi.txt file for all the
gory details about the implementation.

Fixes trac:1853.

This commit was SVN r20921.

The following Trac tickets were found above:
  Ticket 1853 --> https://svn.open-mpi.org/trac/ompi/ticket/1853
2009-04-01 17:52:16 +00:00
George Bosilca
8221975490 We don't need the opal_bitmap definitions here.
This commit was SVN r20918.
2009-04-01 15:25:10 +00:00
Terry Dontje
4b43911c6a Remove superfluous spaces in manpages that were causing catman to
generate mangled windex files.  Made ompi-top.1 and ompi-iof.1 build
by default.  Also, added the orte-top synonym to the ompi-top manpage.

This commit was SVN r20915.
2009-04-01 14:40:27 +00:00
Nysal Jan
aff903f39c Don't print this message by default
This commit was SVN r20914.
2009-04-01 14:31:21 +00:00
George Bosilca
c5b1bdd57c Correctly deal with the error case. The problem is tricky: the MPI standard doesn't allow
MPI_ERR_IN_STATUS to be returned from any functions that return only one completed request
(few exception here: wait_some and wait_all and the test versions). As we use an wait_all
in these send_receive functions we should convert the MPI_ERR_IN_STATUS to the real
error, i.e. the one comming from the MPI_ERROR field in the status corresponding to the
failed request.

This commit was SVN r20907.
2009-03-31 23:44:59 +00:00
George Bosilca
12ce14ec8c A possible patch for the SM problems. I moved the synchronization
after each process create it's FIFOs but before they access the
peer's FIFOs. Second, replace a one way synchronization by a real
barrier, so we know that every process is really where we expect
them to be.

This commit was SVN r20906.
2009-03-31 21:46:27 +00:00
Patrick Geoffray
278d508fa4 Missing send immediate function (NULL) in btl module definition.
Fixes trac:1859.

This commit was SVN r20905.

The following Trac tickets were found above:
  Ticket 1859 --> https://svn.open-mpi.org/trac/ompi/ticket/1859
2009-03-31 18:44:10 +00:00
George Bosilca
2b1804b0f4 Remove useless header files.
This commit was SVN r20897.
2009-03-30 23:09:18 +00:00
Jeff Squyres
b95a3d0eb9 * Remove an extraneous OBJ_CONSTRUCT
* Ensure we don't try to do opal_list_get_next() on an item we just
   deleted
 * set myaddrs = NULL when we're done with it, just for good measure

Once this is ported to OMPI v1.3 branch, it fixes
https://bugs.openfabrics.org/show_bug.cgi?id=1579.

This commit was SVN r20896.
2009-03-30 20:24:31 +00:00
Ralph Castain
cba3708893 Cleanup debugging output, remove an unnecessary re-compute of the checksum
This commit was SVN r20895.
2009-03-30 17:09:32 +00:00
Ralph Castain
d5e6104035 Continue to cleanup the csum pml module. Some minor corrections and debug output added.
This commit was SVN r20894.
2009-03-29 23:27:06 +00:00
Donald Kerr
47dc1bd493 fix #1828; rework the private data connection establishment process; reviewed by terry d.
This commit was SVN r20889.
2009-03-26 17:54:44 +00:00
Donald Kerr
27ed29a0a1 wrap linux specific steps in __linux__ define
This commit was SVN r20888.
2009-03-26 15:09:01 +00:00
Rainer Keller
6ad07dbffa - First have the _config.h header,
then include system headers based on what's defined.

This commit was SVN r20886.
2009-03-26 12:56:22 +00:00
Donald Kerr
a1ba2e2164 add Sun vendor_id to Tavor Infinihost settings
This commit was SVN r20883.
2009-03-25 19:13:55 +00:00
Donald Kerr
cd7940e208 add vendor_id
This commit was SVN r20882.
2009-03-25 17:57:24 +00:00
Pavel Shamis
d25b7203a2 Adding send_immediate (sendi) implementation to openib btl.
This commit was SVN r20881.
2009-03-25 16:53:26 +00:00
Pavel Shamis
8888c9831c Prevent segfault for case when we release SRQ before srq_create.
This commit was SVN r20872.
2009-03-25 14:18:05 +00:00
Ralph Castain
f72e3ba9f9 Update the PML base send init macro to take a converter_flag field (discussed with George).
Update the csum pml module - still not quite right, but closer.

Modify the LANL platform files to keep pace.

This commit was SVN r20859.
2009-03-24 19:12:53 +00:00
Ralph Castain
d88df53a86 A touch more cleanup. Also, bring over the peruse cleanups from r20844
This commit was SVN r20849.

The following SVN revision numbers were found above:
  r20844 --> open-mpi/ompi@daba352af4
2009-03-24 01:36:31 +00:00
Ralph Castain
78323fd6b2 Minor cleanups to compile without warnings
This commit was SVN r20848.
2009-03-24 00:54:16 +00:00
Ralph Castain
75ca19d1d1 Turn off a function that hasn't been added to the code base yet...
This commit was SVN r20847.
2009-03-23 23:56:11 +00:00
Ralph Castain
17f51a0389 Add a new PML module that acts as a "mini-dr" - when requested, it performs a dr-like checksum on messages for BTL's that require it, as specified by MCA params.
Add two new configure options that specify:

1. when to add padding to the openib control header - this *only* happens when the configure option is specified

2. when to use the dr-like checksum as opposed to the memcpy checksum. Not selectable at runtime - to eliminate performance impacts, this is a configure-only option

Also removed an unused checksum version from opal/util/crc.h.

The new component still needs a little cleanup and some sync with recent ob1 bug fixes. It was created as a separate module to avoid performance hits in ob1 itself, though most of the code is duplicative. The component is only selectable by either specifying it directly, or configuring with the dr-like checksum -and- setting -mca pml_csum_enable_checksum 1.

Modify the LANL platform files to take advantage of the new module.

This commit was SVN r20846.
2009-03-23 23:52:05 +00:00
Ralph Castain
fb2b41d40a Give up on the pcie BTL and blow it away. The drivers for this initial implementation have been too customized by IBM - too hard to re-integrate the code.
Maybe someday, someone with enough interest/time can start over...

This commit was SVN r20845.
2009-03-23 23:27:57 +00:00
George Bosilca
daba352af4 As the request is not yet updated (i.e. _MATCHED cannot be called as we don't yet know the
expected length of the message) we should use the source and tag from the message header
instead of the value from the status structure attached to the request.
-This line, and those below, will be ignored--

M    pml_ob1_recvreq.c

This commit was SVN r20844.
2009-03-23 20:25:53 +00:00
Jeff Squyres
804eb94f5f Fix the MCA param name to use the non-deprecated name.
This commit was SVN r20832.
2009-03-20 01:44:40 +00:00
Aurelien Bouteiller
fa9b6e729b Fix missing file in Makefile.am and the "CREATE FAILURE".
This commit was SVN r20821.
2009-03-18 13:42:48 +00:00
Rainer Keller
6f808d9b05 Preparation work for another commit (after RFC):
- This patch solely _adds_ required headers and is rather localized
   The next patch (after RFC) heavily removes headers (based on script)
 - ompi/communicator/communicator.h: For sources that use
   ompi_mpi_comm_world, don't require them to include "mpi.h"
 - ompi/debuggers/ompi_common_dll.c: mca_topo_base_comm_1_0_0_t needs
   #include "ompi/mca/topo/topo.h"
 - ompi/errhandler/errhandler_predefined.h:
   ompi/communicator/communicator.h depends on this header file!
   To prevent recursion just have fwd declarations.
   #include "ompi/types.h" for fwd declarations of the main structs.
 - ompi/mca/btl/btl.h: #include "opal/types.h" for ompi_ptr_t 
 - ompi/mca/mpool/base/mpool_base_tree.c: We use ompi_free_list_t and
   ompi_rb_tree_t, so have the proper classes
 - ompi/mca/op/op.h:
   Op is pretty self-contained: Nobody up to now has done
   #include "opal/class/opal_object.h"
 - ompi/mca/osc/pt2pt/osc_pt2pt_replyreq.h:
   #include "opal/types.h" for ompi_ptr_t 
 - ompi/mca/pml/base/base.h:
   We use opal_lists  
 - ompi/mca/pml/dr/pml_dr_vfrag.h:
   #include "opal/types.h" for ompi_ptr_t
 - ompi/mca/pml/ob1/pml_ob1_hdr.h:
   #include "ompi/mca/btl/btl.h" for mca_btl_base_segment_t
 - opal/dss/dss_unpack.c:
   #include "opal/types.h"
 - opal/mca/base/base.h:
   #include "opal/util/cmd_line.h" for opal_cmd_line_t
 - orte/mca/oob/tcp/oob_tcp.c:
   #include "opal/types.h" for opal_socklen_t
 - orte/mca/oob/tcp/oob_tcp.h:
   #include "opal/threads/threads.h" for opal_thread_t
 - orte/mca/oob/tcp/oob_tcp_msg.c:
   #include "opal/types.h" 
 - orte/mca/oob/tcp/oob_tcp_peer.c:
   #include "opal/types.h"  for opal_socklen_t
 - orte/mca/oob/tcp/oob_tcp_send.c:
   #include "opal/types.h" 
 - orte/mca/plm/base/plm_base_proxy.c:
   #include "orte/util/name_fns.h" for ORTE_NAME_PRINT
 - orte/mca/rml/base/rml_base_receive.c:
   #include "opal/util/output.h" for OPAL_OUTPUT_VERBOSE
 - orte/mca/rml/oob/rml_oob_recv.c:
   #include "opal/types.h" for ompi_iov_base_ptr_t
 - orte/mca/rml/oob/rml_oob_send.c:
   #include "opal/types.h" for ompi_iov_base_ptr_t
 - orte/runtime/orte_data_server.c
   #include "opal/util/output.h" for OPAL_OUTPUT_VERBOSE
 - orte/runtime/orte_globals.h:
   #include "orte/util/name_fns.h" for ORTE_NAME_PRINT

 Tested on Linux/x86-64

This commit was SVN r20817.
2009-03-17 21:34:30 +00:00
Rainer Keller
b9b84a9c29 - ompi/mca/mpool/rdma/mpool_rdma_module.c: At this level
without mpi.h we have no notion of MPI_SUCCESS...
 - ompi/mca/btl/sm/btl_sm.h: ptrdiff_t needs stddef.h 
 - ompi/mca/mpool/base/: If we use opal_pointer_array_t,
   better include the class header.

This commit was SVN r20816.
2009-03-17 20:21:36 +00:00
Aurelien Bouteiller
3cd5a0d833 Support for the MPI event logger improving event logging perfs.
This commit was SVN r20804.
2009-03-17 17:35:28 +00:00
Pavel Shamis
c6d038a8e8 Adding vendor_error code to error report.
This commit was SVN r20803.
2009-03-17 15:47:34 +00:00
Pavel Shamis
5afa2988f1 Updating RNR/IB timeout for openib btl
This commit was SVN r20801.
2009-03-17 15:03:06 +00:00
Donald Kerr
d29a5e57c1 remove superfluous define
This commit was SVN r20785.
2009-03-16 02:24:01 +00:00
George Bosilca
a9be1b1dde Set the mem_node to a more meaningful value, as suggested by Ake Sandgren.
This commit was SVN r20780.
2009-03-14 22:08:26 +00:00
Eugene Loh
64f52b0168 Clean up in response to code review on CMR 1825:
minor changes in comments and edge-case handling.

This commit was SVN r20774.
2009-03-13 18:11:41 +00:00
Rainer Keller
d8cf4c0fec - Get pgcc on XT to complain less:
In case we use memcmp, strlen, strup and friends include <string.h>
   Also several constants.h are not included directly
 - Let's have mca_topo_base_cart_create  return ompi-errors in
   ompi/mca/topo/base/topo_base_cart_create.c

This commit was SVN r20773.
2009-03-13 02:10:32 +00:00
Donald Kerr
ef55aae401 fix #1829 : udapl btl support for relaxed ordering
This commit was SVN r20772.
2009-03-13 01:01:00 +00:00
Rainer Keller
6fca443a71 - No, we don't want to have a notion of an MPI_Comm in this layer
We want ompi_communicator_t instead, rrrrrr.

This commit was SVN r20770.
2009-03-12 22:38:14 +00:00
George Bosilca
b29da4744f Amazing that there is only one compiler complaining about this ...
This commit was SVN r20768.
2009-03-12 21:30:08 +00:00
Rainer Keller
74b3acd4bd - No need to declare struct mca_mpool_base_resources_t;
Already in
   #include "ompi/mca/mpool/mpool.h"

This commit was SVN r20767.
2009-03-12 20:27:16 +00:00
Rainer Keller
296a6fb275 - So much fun along the way:
we normally don't do opal/include/opal/...
   Just use the std. opal/...

This commit was SVN r20766.
2009-03-12 19:21:11 +00:00
Rainer Keller
29b1b205fd - Remove two headers (and actually include rml.h) prior to test of
removal script...

This commit was SVN r20765.
2009-03-12 17:58:39 +00:00
Jeff Squyres
14ee1b7ba2 Refs trac:1826: remove barriers before all non-rooted collective ops.
This commit was SVN r20763.

The following Trac tickets were found above:
  Ticket 1826 --> https://svn.open-mpi.org/trac/ompi/ticket/1826
2009-03-12 02:23:08 +00:00
Ralph Castain
a8002c0f04 Remove missing files from Makefile.am so make dist will succeed
This commit was SVN r20743.
2009-03-06 02:57:51 +00:00
Rainer Keller
ec0ed48718 - Revert r20739
This commit was SVN r20742.

The following SVN revision numbers were found above:
  r20739 --> open-mpi/ompi@781caee0b6
2009-03-05 21:56:03 +00:00