1
1
Граф коммитов

572 Коммитов

Автор SHA1 Сообщение Дата
Brian Barrett
22cd78abb5 * add header required when debugging is not enabled
This commit was SVN r10155.
2006-06-01 01:26:52 +00:00
Josh Hursey
bb95df9bf2 Added some user friendly output to the hostfile RDS component.
This is more of a usability feature, but a very useful one. So I 
suggest that it go into the release branches.

This commit was SVN r10153.
2006-05-31 20:07:59 +00:00
Josh Hursey
2f20a38c98 This is a fix for bug Ticket #27
We were stuck in an infinite loop inside the rmaps round_robin
component when the user specified a host, then over subscribed it.
Instead of retuning an error, we looped forever.

For example:
 $ cat hostfile
  A slots=2 max-slots=2
  B slots=2 max-slots=2
 $ mpirun -np 3 --hostfile hostfile --host B
  <hang>

The loop would not terminate because both host A and B are in the 
'nodes' structure as they are both allocated to the job. However,
after allocating 2 slots to host B, we remove it from the node list
leaving us with a 'nodes' structure with just A in it. Since we can't
use host A, we keep looping here until we find a node that we can use.

This patch checks to make sure that if we get into this situation where
rmaps is looping over the list a second time without finding a node
during the first pass then we know that there are no nodes left to
use, so we have a resource allocation error, and should return to the user.

This patch should be moved to all of the release branches

This commit was SVN r10131.
2006-05-31 03:42:01 +00:00
Brian Barrett
7000cecf78 Fix for standard output / standard error truncation issue when in a shell
pipeline.  See lengthy comment in iof_base_endpoint.c for the details, but
the short version is that we shouldn't set O_NONBLOCK on standard I/O 
file descriptors, so we no longer do.

Closes ticket:9

This commit was SVN r9966.
2006-05-18 15:43:32 +00:00
Jeff Squyres
1d6902296c Additions to the tm, slurm, and rsh pls modules to handle the --prefix
option as discussed on the devel-core mailing list.  The Big
Difference is that instead of hard-coding the strings "/lib" and
"/bin" in to append to the prefix, we append the basename of the local
libdir and bindir.  Hence, if your libdir is $prefix/lib64, we'll
append /lib64 to construct the remote node's LD_LIBRARY_PATH (etc.).

Also appended the orterun.1 man page to include a description of
--prefix, how it is constructed, what it handles / what it does not,
etc.

This commit was SVN r9930.
2006-05-16 14:14:12 +00:00
Gleb Natapov
80dfe7e39b remove newline from environment
This commit was SVN r9892.
2006-05-11 13:15:48 +00:00
Brian Barrett
1c0c84cf67 If the urm gets a request to kill itself *and* it's a singleton, just
exit out, rather than trying to have the pls exit.  Since singletons
weren't started with a pls, there's no way the pls is going to be
able to kill the process.  So just exit and save the error message.

This commit was SVN r9859.
2006-05-09 13:40:41 +00:00
Brian Barrett
b76b46bcec * fix some compile issues on Red Storm
This commit was SVN r9812.
2006-05-04 14:08:36 +00:00
Brian Barrett
9276127c0d * add some extra sauce to make sure we close down our processes properly
This commit was SVN r9807.
2006-05-04 00:38:49 +00:00
Brian Barrett
52369307f8 Add a feature to the build system that Terry from Sun and I talked about
in San Jose.  Allow the configure option --disable-binaries to build OMPI,
but not build or install the support binaries (so basically, just build
the libraries).

This commit was SVN r9777.
2006-04-29 02:16:41 +00:00
Brian Barrett
5fed99c2c2 Sending SIZE_MAX from machines with different sizeof(size_t) causes big problems,
as the smaller machine's SIZE_MAX won't be SIZE_MAX on the bigger machine, which
can lead to failures along the way -- in this case, with GPR triggers being
improperly fired.

This commit was SVN r9776.
2006-04-28 21:09:42 +00:00
George Bosilca
1ea3a39372 The condition was wrong. The fact that it accept 0 length messages
is interpreted as a shutdown of the io channel on the next iteration.
Definitively not the good approach. The correct condition is
bigger than 0.

This commit was SVN r9770.
2006-04-28 04:57:07 +00:00
Jeff Squyres
bfcf3867fc Back out George's commit from earlier today; it seems to break stdout
forwarding. 

More detailed mail coming to devel-core shortly that explains.

This commit was SVN r9769.
2006-04-28 03:32:27 +00:00
Sushant Sharma
7a6e0c9ebf Fixed remote environment setup. Submitted by: Tim Woodall
This commit was SVN r9759.
2006-04-27 20:07:56 +00:00
George Bosilca
bafc16f724 We don't need the len anymore as everything is not attached to the fragment.
This commit was SVN r9758.
2006-04-27 17:35:05 +00:00
George Bosilca
5df94f812e Aren't we supposed to release the value on all possible execution paths ?
This commit was SVN r9757.
2006-04-27 17:31:01 +00:00
Tim Woodall
0a56067509 Correction to resolve a problem related to partial reads. We were making a
copy of the receive buffer based on the iovec struct that may have been updated 
during partial reads to reflect the current offset. Need to make the copy using 
the base address of the buffer.

Thanks to Sven Stork for finding this.

This should be backported to 1.0.X and 1.1.X branches.

This commit was SVN r9749.
2006-04-27 14:27:02 +00:00
Tim Woodall
7a139d6cc8 - corrections to I/O forwarding - handling of incomplete writes
THESE CHANGES SHOULD BE PROPOGATED TO BOTH 1.0 and 1.1 BRANCHES

This commit was SVN r9734.
2006-04-26 15:36:06 +00:00
Tim Woodall
3e57a4ec48 remove debug code - not required
This commit was SVN r9715.
2006-04-25 19:05:57 +00:00
Brian Barrett
ce72140633 Remove dependency libraries from these Makefile.ams - the libraries will
automagically bring in the libraries through the top-level library (so
liborte automatically brings in libopal, etc.).  Otherwise, we get some
warnings on Solaris

This should go to the v1.1 branch

This commit was SVN r9666.
2006-04-20 17:53:43 +00:00
Brian Barrett
62afa63ded Initialize length to 0 instead of -1 (size_t might be unsigned and therefore
-1 is an issue).

This should go to the v1.1 branch...

This commit was SVN r9665.
2006-04-20 15:42:36 +00:00
Brian Barrett
e737b0a106 Fix a bunch of warnings the Sun compilers find:
- The constant 1 is a signed int by default.  Explicitly say that
    it is an unsigned value so we can't overflow
  - Fix unreachable statement warnings in dss_arith by breaking out
    of switch statements instead of returning - this should have
    no impact on performance, since it's a non-conditional jump
  - A couple of the GPR files had carriage returns and were in
    DOS mode - put them in unix mode...

These should all probably go to the v1.1 branch...

This commit was SVN r9664.
2006-04-20 15:35:58 +00:00
Ralph Castain
95c4795157 Try a different tack...
This commit was SVN r9658.
2006-04-19 15:33:34 +00:00
Ralph Castain
93115fdaea Try again with passing the right enviro variables.
This commit was SVN r9629.
2006-04-13 18:07:22 +00:00
Ralph Castain
480af1c150 Add the missing enviro variables
This commit was SVN r9627.
2006-04-13 16:41:47 +00:00
Ralph Castain
d7456c3d89 Fix the broken Doxyfile so people can generate what little code base documentation we have :-)
Copy the Doxyfile to orte so people can generate just that documentation. Adjust the properties on that directory so it ignores the resulting doxygen output tree.

This should go over to the release branch(es)

This commit was SVN r9625.
2006-04-13 12:52:17 +00:00
Sushant Sharma
642e33fb3e xcpu launcher updated to setup the environment on remote nodes before launching jobs.
This commit was SVN r9622.
2006-04-12 22:42:41 +00:00
Ralph Castain
424900068f Update the xcpu launcher to setup the environment
This commit was SVN r9620.
2006-04-12 15:41:54 +00:00
Sushant Sharma
9fe5870862 xcpu pls component fixed so that it will compile correctly.
This commit was SVN r9617.
2006-04-11 20:27:13 +00:00
Ralph Castain
9adc16130e Proposed revision of the xcpu launcher to correctly incorporate the OpenRTE and Open MPI environment
This commit was SVN r9612.
2006-04-11 14:33:17 +00:00
Brian Barrett
f37a77dd08 * Fix potential deadlock when mpi threads are enabled and progress threads are
not.  See lengthy comment in the body of commit.

This commit was SVN r9573.
2006-04-07 18:13:35 +00:00
Sushant Sharma
26d51d5041 Cleaned lots of dead code in xcpu soh component (soh_xcpu.c). Checked the fix submitted by Ralph Castain for completing processes in soh_xcpu. Its working fine now.
This commit was SVN r9554.
2006-04-06 16:26:25 +00:00
George Bosilca
ca75ff2569 In the case we have support for threads, then the opal library have it's own
thread, which will do progress independently of MPI. So in this case we 
have to call opal_event_loop instead of opal_progress.

This commit was SVN r9551.
2006-04-06 14:31:38 +00:00
Brian Barrett
7408de0bfb When progress threads are enabled, opal_progress() doesn't call the
event library (since the event library has its own thread).  So when
we are using progress threads, we really want to call opal_event_loop()
and not opal_progres(). 

This commit was SVN r9549.
2006-04-06 12:58:09 +00:00
Ralph Castain
895c2ade8b Proposed fix for completing processes
This commit was SVN r9543.
2006-04-06 08:18:42 +00:00
Ralph Castain
c79c1714de Okaaayyy....let's see if this restores the "prefix" command line option. No idea what the problem was with the other option, but it isn't critical right now, so I'll figure it out later.
This commit was SVN r9542.
2006-04-06 07:53:38 +00:00
Ralph Castain
0ba8851a47 Fix the univ_exist option
This commit was SVN r9535.
2006-04-05 17:18:06 +00:00
Ralph Castain
b9bdb2125e Fix and upgrade the console to support better debugging. Activate "dump" commands to display registry content. Remove the blasted opal_output default prefix that made the dump output illegible. Properly connect to existing daemons and/or start new ones.
This commit was SVN r9528.
2006-04-04 11:05:52 +00:00
Sushant Sharma
8d5289b2b8 Corrected Makefile.am files for pls and soh xcpu-components as per Brian's suggestion.
This commit was SVN r9519.
2006-04-03 17:14:47 +00:00
Brian Barrett
4ea8790342 * Don't try to call tcgetprgp on platforms that don't have that function
* Some more stuff to ignore / do in Red Storm build

This commit was SVN r9511.
2006-04-01 05:46:15 +00:00
Brian Barrett
2c64ab562e More fixes to try to get Red Storm port going again....
* Add a platform spec for using the portals reference implementation's
  RTE instead of our own to make local testing easier.
* Add a cnos rmgr component so that 1) we don't have to build nearly
  as many components (no need for ras,rds,pls,etc.) and 2) calls
  to MPI_ABORT() won't print error messages about not being able to
  contact the daemon.  Still need to fill in some of the terminate
  stuff with calls from cnos, but will come in time.
* Make gpr_null use the base code for creating value and keyval
  structures so that we don't segfault in ompi_mpi_init().

This commit was SVN r9510.
2006-04-01 04:54:46 +00:00
Jeff Squyres
858612fd06 Face the possibilty that the child may have already died.
This commit was SVN r9508.
2006-04-01 02:23:10 +00:00
Sushant Sharma
46f84b1e8e Added xcpu component in pls and soh.
This commit was SVN r9491.
2006-03-31 02:19:52 +00:00
Ralph Castain
8ba453b866 Modify the rmgr_proxy component so it includes the automatic wire-up of stdio.
This commit was SVN r9483.
2006-03-30 19:44:28 +00:00
George Bosilca
2b3779cd6e Correct some of the casting issues. By default the compilers attach an signed type
to the defines. As our internal types (job_id and co.) are unsigned that generate
several errors (integer overflow in expression and comparison between signed and
unsigned). Casting the defines to the correct type solve these problems.

This commit was SVN r9481.
2006-03-30 19:28:17 +00:00
Tim Woodall
637511e759 correct cleanup of callbacks
This commit was SVN r9479.
2006-03-30 16:55:02 +00:00
Brian Barrett
e0eb9a19e7 * make orte_process_name_t be of fixed size (rather than depending on the size of
size_t).  This should be the last piece of the puzzle required to get 32/64
  interoperability working for ORTE.

This commit was SVN r9476.
2006-03-30 14:59:41 +00:00
Brian Barrett
6be35fb604 * Use the ORTE_<type> constants instead of internal DSS_TYPE_<type>_T constants
for the type to be packed / unpacked when dealing with sized types (like
  size_t) so that the dss_unpack code to deal with types of different sizes is
  activated.  Necessary for proper 32/64 interoperability.

This commit was SVN r9475.
2006-03-30 14:33:25 +00:00
Brian Barrett
02c8a51b76 * fix endian encoding for 64 bit numbers to use hton64
* cleanup the unpack_size_mismatch macros a little bit
* ad comment about endianness of size_mismatch cleanup code so that I don't
  think I've found a bug that really isn't and lose an hour tracking it down
  again...

This commit was SVN r9458.
2006-03-29 18:58:02 +00:00
Brian Barrett
cf425f6289 * trick the stupid compiler (GCC in this case) into shutting up about not
being able to convert from object pointer to function pointer. 

This commit was SVN r9445.
2006-03-29 01:26:16 +00:00