1
1
Граф коммитов

6656 Коммитов

Автор SHA1 Сообщение Дата
Jeff Squyres
ed0fa9720d Incorporate fix suggested by Chris Gottbratch.
This commit was SVN r8750.
2006-01-19 15:21:53 +00:00
George Bosilca
317f8bc4d4 Indentation + print the shared memory file name if we fail to create it.
This commit was SVN r8749.
2006-01-19 07:09:44 +00:00
George Bosilca
1e48ab533a Remove some left-around from long time ago ...
This commit was SVN r8748.
2006-01-19 07:08:32 +00:00
George Bosilca
81de5e7a47 Get a meaningful error message on Windows when the shared memory file cannot
be created.

This commit was SVN r8747.
2006-01-19 07:07:47 +00:00
George Bosilca
8b50c3d8c3 Handle systems where the HOME directory is NULL (windows and friends).
A small and ugly workaround the path problem on windows (the absolute
path start with [a-z]: whn : is considered as separator for most of
the environment variables.

This commit was SVN r8746.
2006-01-19 07:05:34 +00:00
George Bosilca
e6e28460f1 Remove all windows code as fork is not available on windows. Instead a shinny new pls
will join the fun (handling process creation on windows).

This commit was SVN r8745.
2006-01-19 07:01:51 +00:00
George Bosilca
b58921c7ad Correctly compute the length of the path (including systems where the delimitator
is not just one char) and use the computed length. Handle Windows paths as they
don't have to start with /.

This commit was SVN r8744.
2006-01-19 06:58:49 +00:00
George Bosilca
e728fc0c15 I know windows is "different" ... but not at this point !!!
This commit was SVN r8743.
2006-01-19 06:54:23 +00:00
George Bosilca
770d530e83 Same bug as blm.h. I know that this macros are not used right now but I fix them for later ...
This commit was SVN r8742.
2006-01-19 06:48:45 +00:00
George Bosilca
be64d98687 We're thread safe again. The atomic compare-and-swap was not used in the correct way allowing
the descriptors to vanish. The PML was thinking that they are in the btl_cache when they
weren't ... It lead to memory consumption on most environments when compiled with
thread enabled. After modification the latency went down by nearly 0.5 microseconds.

Simple way to trigger the bug: limit the number of maximum items in the free list and run
any communication intensive application (like Netpipe).

This commit was SVN r8741.
2006-01-19 06:45:29 +00:00
Jeff Squyres
e6bd80b424 Per the commit message of r8514, change the search order to be "ssh :
rsh", *not* "rsh : ssh".

This commit was SVN r8736.

The following SVN revision numbers were found above:
  r8514 --> open-mpi/ompi@9c25bdc5ac
2006-01-18 22:00:34 +00:00
George Bosilca
9fe0969f0e Enable multi-threaded builds.
This commit was SVN r8735.
2006-01-18 21:48:40 +00:00
George Bosilca
992daf7522 Remove all unused defines from the Makefile.
This commit was SVN r8734.
2006-01-18 21:21:29 +00:00
Brian Barrett
17197666bf When heap is shrinking ptmalloc tries to be smart and uses mmap to change
page protection, which causes the pages to be droped, which causes problems
if we don't deregister the pages first.  Since memory is cheap in this case
(it is still usable, should ptmalloc2 want it back, and is limited in size),
we just mprotect the pages instead.  This solves the dropping pages problem,
and doesn't cause even more calls into the cache code.

Thanks to Gleb Natapov for both finding the problem and giving a fix.

This should go to the v1.0 branch

This commit was SVN r8732.
2006-01-18 16:23:52 +00:00
Galen Shipman
d657052510 misc cleanup..
This commit was SVN r8731.
2006-01-18 16:20:50 +00:00
Galen Shipman
84a09e4f4e use #if not #ifdef..
This commit was SVN r8720.
2006-01-17 21:07:34 +00:00
Brian Barrett
759bfc91a3 * George pointed out this should be OMPI_HAV_ESOLARIS_THREADS, not OPAL_HAVE_
SOLARIS_THREADS...

This commit was SVN r8713.
2006-01-17 17:17:25 +00:00
Galen Shipman
0c81c0a6ce use ibv_get_device_list if present
(submitted from roland)

This commit was SVN r8712.
2006-01-17 16:23:35 +00:00
Brian Barrett
e830339098 * turns out nodist_ and dist_ mean different things... We actually want to
include help-opal-wrapper.txt in the dist tarball directly.  We don't want
  to include the wrapper-data.txt files, because we actually want to ship the
  .in versions of those files, and AM automatically ships aything needed for
  configure's output

This commit was SVN r8711.
2006-01-17 15:23:46 +00:00
George Bosilca
bd5bcdc8a3 Add missing include files for x86 builds.
This commit was SVN r8710.
2006-01-17 13:22:08 +00:00
Jeff Squyres
61daf52344 Remove defunct parameter
This commit was SVN r8709.
2006-01-17 11:38:07 +00:00
Brian Barrett
22c50c2cca * allow loading of configure options from an external file, for use in
cross-compile environment

This commit was SVN r8706.
2006-01-16 23:38:42 +00:00
Andrew Friedley
5ccab7bcda Checkpoint:
- Move mca_btl_udapl_error/mca_btl_module_init to mca_btl_udapl.c and rename it
 - White space cleanups
 - Free the uDAPL evd and ia handles in mca_btl_udapl_finalize

This commit was SVN r8705.
2006-01-16 21:54:50 +00:00
Brian Barrett
be13362899 * add missing headers for the WIF macros
This commit was SVN r8704.
2006-01-16 15:46:17 +00:00
Brian Barrett
187f1b21bb remove reference to file that doesn't exist
This commit was SVN r8703.
2006-01-16 14:00:52 +00:00
Brian Barrett
17d2c907fc Some changes to better support cross-compiling OMPI:
- fall back to compile test for windows paffinity component
    when cross compiling
  - fall back to platform guess when checking for threads having
    different pids with pthreads (yes on linux, no elsewhere)
  - pass the proper host, target, and build flags to the
    ROMIO configure script

With these changes, cross-compiling should be possible with the exception
of the Fortran 77 and Fortran 90 bindings.  Fortran 77 can be cross-
compiled if cache values are provided for type sizes and alignment.

This commit was SVN r8702.
2006-01-16 04:00:44 +00:00
Andrew Friedley
a4abe3bdbe Checkpoint:
- Borrow configure.m4 from the mvapi btl.  One of the uDAPL headers emits a
   warning when -pedantic is enabled, so strip it out.
 - Change function check in ompi_check_dapl.m4 from dat_ia_open to
   dat_registry_list_providers.. dat_ia_open wasn't working right
 - Make the references to prepare_dst, put, and get NULL for now
 - Add opal_output() calls in all the udapl interface functions for debugging
 - Add evd_qlen component parameter to control event dispatcher queue length
 - First stab at component_init and module_init
 - Misc cleanups - whitespace, dead code removal
 - Update copyrights to 2006

This commit was SVN r8701.
2006-01-16 03:01:12 +00:00
Brian Barrett
edfdf13096 fix missing SUBDIRS
This commit was SVN r8700.
2006-01-16 02:00:10 +00:00
Brian Barrett
c96f870674 * Merge of wrapper compiler updates from the bwb-wrapper-fix branch (r8690 -
r8698), with changes below:

  - Split wrapper flags into those required for each of the three projects,
    and cleaned up some cruft (including the LIBMPI_EXTRA_*FLAGS) through-
    out the build system
  - Added opal_init_util and opal_finalize_util to allow init / cleanup
    of all the opal code that doesn't require the MCA system
  - Create standalone key=value file parser, based on the one that used
    to be in the mca param parser, so that it can be shared in multiple
    places
  - Add wrapper datafiles for opal, orte, and ompi wrappers, and add
    wrapper compiler with support for all the old features

This commit was SVN r8699.

The following SVN revisions from the original message are invalid or
inconsistent and therefore were not cross-referenced:
  r8690
  r8698
2006-01-16 01:48:03 +00:00
George Bosilca
d4699037f7 Protect an assert if the endpoint cache is not activated.
This commit was SVN r8695.
2006-01-14 21:10:09 +00:00
George Bosilca
3317bf81ad A better implementation for the TCP endpoint cache + few comments.
This commit was SVN r8692.
2006-01-14 20:21:44 +00:00
George Bosilca
5f37949b10 Correct few typos and make the "no fortran" a valid choice.
This commit was SVN r8689.
2006-01-14 00:20:13 +00:00
Jeff Squyres
74d759dc47 Patch submitted by Roland Dreier to look for a change in the OpenIB
API.

This commit was SVN r8687.
2006-01-13 23:20:49 +00:00
George Bosilca
bf266c6109 Rollback the 8682 commit until we figure out the correct way to do it. It break several things
inside (like MPI_Wait* functions).

This commit was SVN r8686.
2006-01-13 22:02:40 +00:00
Brian Barrett
92aaaad611 * Fix issue when a type existed but we didn't have a corresponding C type -
the failure wasn't being properly propogated up from ompi_find_type and
  resulting in bad #define values in ompi_config.h
* Fix issue where we could emit illegal sh code if we were checking for a
  type with no corresponding C types listed.  Thanks to Ralf for tracking
  this one down.
* Fix a couple more messages to match all the others.

This commit was SVN r8685.
2006-01-13 14:51:20 +00:00
Brian Barrett
48f82db838 Convert all Fortran 77 tests to use config cache so that we have a way to
determine values like Fortran alignment (which can only be determined by
running a program) when cross-compiling.  By providing cache values, the
programs will not be run at all, and life will be good.  Also clean up
some macro interfaces so that they are a bit easier to use, at the cost
of horrid internals ;).

This commit was SVN r8684.
2006-01-13 04:08:40 +00:00
Tim Woodall
a584c60dbe re-worked flow control logic to take into account the return
of credits from the peer prior to local completion, so that
we don't overrun the number of send wqes available.

This commit was SVN r8683.
2006-01-12 23:42:44 +00:00
Rainer Keller
95f886b6ab - Protect callers of opal/ompi_condition_wait from spurious wakeups,
possible when with building with pthreads.
   Compiled on Linux ia32 with and without
   --enable-progress-threads

This commit was SVN r8682.
2006-01-12 17:13:08 +00:00
Andrew Friedley
c0bad339af - Use the GM BTL as a template instead, per Tim's suggestion
- Begin adding uDAPL-specific stuff
- Added config/ompi_check_udapl.m4 - hopefully I did this right

This commit was SVN r8681.
2006-01-12 04:05:02 +00:00
Jeff Squyres
f5efb70252 Sync with v1.0 NEWS.
This commit was SVN r8679.
2006-01-12 03:39:51 +00:00
Brian Barrett
6afdc3a65d * per the Autoconf manual, work around sys/socket.h requirement on Darwin
for the net/if.h header file. 

  This should probably go to the trunk...

This commit was SVN r8676.
2006-01-12 03:19:37 +00:00
George Bosilca
698b9b52fe There is no need for the ret variable when DEBUG is not enabled.
This commit was SVN r8675.
2006-01-11 20:56:22 +00:00
Brian Barrett
234c6b96c3 commit missing file from r8672
This commit was SVN r8674.

The following SVN revision numbers were found above:
  r8672 --> open-mpi/ompi@60ac1cb5f4
2006-01-11 05:02:15 +00:00
Brian Barrett
80c5e5c8ba * fix assembly issues for linux, which wants a notation to make the stack
non-executable.

  This should go to the v1.0 branch

This commit was SVN r8673.
2006-01-11 04:49:40 +00:00
Brian Barrett
60ac1cb5f4 print stack traces (when available) for opal and orte processes, as well as
ompi processes.  Also add SIGABRT to the list of signals that are intercepted
to print out pretty messages.

This commit was SVN r8672.
2006-01-11 04:36:39 +00:00
Brian Barrett
9310fd6f85 When debugging code is turned on with --enable-debug, try to use error checking
mutexes instead of fast mutexes.  If an error occurs, a message will be
printed, and abort() will be called.

This commit was SVN r8671.
2006-01-11 04:34:29 +00:00
David Daniel
c2ee847184 Missing header file.
This commit was SVN r8670.
2006-01-10 21:58:21 +00:00
Brian Barrett
5fdf753e2d remove unneeded second copy of the PML_CALL macros
This commit was SVN r8668.
2006-01-10 18:44:21 +00:00
Jeff Squyres
dd142673f4 Add missing MPI::Request::Get_status methods.
This commit was SVN r8663.
2006-01-09 18:10:12 +00:00
Graham Fagg
25375759c3 arrgh. reduce could for very small message sizes and proc counts call a linear function
this was implemented using a chain (tree followed with pipeline) by setting the chain fanout to a factor of size etc but the chain datastructure was fixed in length and if exceeded the topo create returned a null which isn't helpfull in cid next function of comdup...
Anyway two fixes, first we do have a real linear function so changed the decision function and second altered the
topo chain create to force chain fanouts of less than 1 to 1 and fanouts bigger than max to max.
next check in will change chain to dynamically allocd array (reallocable) but we shouldn't ever use a chain fanout for a linear tree anyway. 
(lession must rerun all tests for all data sizes when changing decision functions)

This commit was SVN r8662.
2006-01-08 02:41:09 +00:00