1
1
Граф коммитов

4640 Коммитов

Автор SHA1 Сообщение Дата
Thara Angskun
e9cc404c33 - just checkpoint
This commit was SVN r5752.
2005-05-18 22:46:17 +00:00
George Bosilca
b41444ddcb Less talkative ....
This commit was SVN r5751.
2005-05-18 21:23:48 +00:00
Ralph Castain
d3029fbdb9 Fix a few bugs.
Remote launch of daemon now working. Bunch of forced diagnostic messages in it, though, which I'll leave until I release this for actual use.

This commit was SVN r5750.
2005-05-18 18:39:50 +00:00
Ralph Castain
bb9c90b6ae Fix a couple of minor bugs that prevented the clean handshake of contact info.
This commit was SVN r5749.
2005-05-18 18:24:14 +00:00
Ralph Castain
91c75cb7d8 Checkpoint the remote launch work.
Orted has been modified to take a new parameter - a file descriptor used as a pipe to pass the daemon's contact info back to the probe when the daemon is remotely launched.

This commit was SVN r5748.
2005-05-18 17:56:51 +00:00
Ralph Castain
d1fe8b6b90 Move an output message so it only comes out under specific conditions.
You will now receive a message indicating that an existing universe was detected, but connection to it was refused. The system will tell you the name it created for the new universe it will now be using.

This commit was SVN r5747.
2005-05-18 16:36:11 +00:00
Ralph Castain
1b42e973d5 Only orted was saving the universe contact info - this should actually be saved by whomever believes they are the "seed". Modified files to ensure this happens. Also includes a checkpoint of the probe and remote launch functions.
This commit was SVN r5746.
2005-05-18 16:31:03 +00:00
Thara Angskun
c71f3f7152 - just checkpoint
This commit was SVN r5745.
2005-05-18 15:55:34 +00:00
George Bosilca
81d1d0322a Remove all useless includes. They get included anyway from others include ... The way we
manage the dependencies still need some rework ...

This commit was SVN r5744.
2005-05-18 15:54:24 +00:00
Tim Woodall
ea8c91627d correc typo
This commit was SVN r5743.
2005-05-18 15:32:05 +00:00
Tim Woodall
3528471f22 correct debug output
addition error checking

This commit was SVN r5742.
2005-05-18 15:31:23 +00:00
Tim Woodall
75a43e84ba fixed ping
This commit was SVN r5741.
2005-05-18 15:12:04 +00:00
Ralph Castain
ef07d67946 Checkpoint
This commit was SVN r5740.
2005-05-17 20:21:59 +00:00
Ralph Castain
3232200e51 Probe now functional! At least, it now at least runs. Next need to complete its operations
This commit was SVN r5739.
2005-05-17 20:19:38 +00:00
Ralph Castain
56475f64a3 Checkpoint for transfer
This commit was SVN r5738.
2005-05-17 19:56:54 +00:00
Ralph Castain
20c325b6bd Checkpoint for transfer to remote machie
This commit was SVN r5737.
2005-05-17 19:35:49 +00:00
Ralph Castain
3e982534cd Make sure all string-to-name field conversions are done in decimal now.
This commit was SVN r5736.
2005-05-17 16:04:50 +00:00
Ralph Castain
3092e0f6cf Fix a minor compiler squawk....
This commit was SVN r5735.
2005-05-17 13:11:48 +00:00
George Bosilca
36fc07e93e A TAB manage to get inside ,,,
This commit was SVN r5734.
2005-05-17 07:19:20 +00:00
George Bosilca
8ea51d6491 Typo ...
This commit was SVN r5733.
2005-05-17 07:16:28 +00:00
George Bosilca
ff2c9d5f9e The stack now hold the type of the last converted element.
This commit was SVN r5732.
2005-05-17 07:15:44 +00:00
George Bosilca
b4314b0426 Improved case for basic datatype with gaps around.
This commit was SVN r5731.
2005-05-17 07:14:30 +00:00
George Bosilca
7381c1f9ee This function come from the DDT version 2. Remove it by now !!!
This commit was SVN r5730.
2005-05-17 07:10:31 +00:00
George Bosilca
f68e255933 The DDT engine reach a quite stable state. All of the IBM test pass, few of the Intel tests fails
(mostly UB/LB related and doing questionable things). Otherwise a lot of changes:
- cleaner code + more comments
- stronger bound checker (DDT error messages if we exceed the buffer capacity)
- differentiate the 3 internal types: loop, basic element and end_loop (using union)
- more MACROS to solve the repetitive problems
- more output in debug mode (and if requested by the user).
- correct some mismaching between usage of true_extent and extent
- improve the special cases (contiguous data, contiguous with gaps, no conversion ...)
- in order to allow gdb to break in the pack/unpack function they became real function (defined
  in dt_pack.c and dt_unpack.c) if OMPI_ENABLE_DEBUG is defined. Otherwise they are just macros.
- a new approach for the computation of the lower bound and upper bound. In same time the true_lb and
  true_ub has been modified to match the new algorithm.
- handle specific cases in the datatype optimization. In some cases the datatype definition increase,
  but the complexity decrease.

This commit was SVN r5729.
2005-05-17 06:54:41 +00:00
George Bosilca
f712270f8b Do not correct the lb. The inner layer should compute the correct one.
This commit was SVN r5728.
2005-05-17 06:42:11 +00:00
George Bosilca
10a07e0eff More options allowing more specific datatype tests.
This commit was SVN r5727.
2005-05-17 06:40:45 +00:00
George Bosilca
b70ae8a2bb This one is pretty strange. The protoytpe is clear the 3th argument should be a bool. But it was a char*
and icc didn't complain ... only gcc 4.1 ...

This commit was SVN r5726.
2005-05-17 01:10:44 +00:00
George Bosilca
dde75fe032 One of the include was missing. That's a big problem on 64 bits architecture where the default return value
is int (mostly 4 bytes) and strdup normally return a char* (usually 8 bytes). The result: stack corrupted
and other weird things ...

This commit was SVN r5725.
2005-05-17 01:02:40 +00:00
Ralph Castain
54a481cc14 Fix an incorrect free...
This commit was SVN r5724.
2005-05-16 21:06:09 +00:00
Ralph Castain
89b6a97f0f Bring the resource discovery system's resource file component online so I can find the node I need to launch upon. I removed all reference to the xml library that was causing trouble, and wrote my own limited xml parser instead, so this will now compile just fine anywhere.
Need to do some refining of the component, but it meets basic requirements right now. Nobody else should notice any change - system basically ignores it unless you tell it to do something.

This commit was SVN r5723.
2005-05-16 21:01:09 +00:00
George Bosilca
2c4209f4cb Add the output utilities include.
This commit was SVN r5722.
2005-05-16 19:36:12 +00:00
Ralph Castain
a0393e9cb9 Fix a malloc/free problem when user doesn't specify a name for the universe. No impact on what George is seeing - still looking into that with Tim.
This commit was SVN r5721.
2005-05-16 18:58:22 +00:00
Brian Barrett
09057fe311 * massive cleanup of debugging output to make it much easier to match
messages
* use different event queues for send / recv, part of moving towards dealing
  with dropped fragments

This commit was SVN r5719.
2005-05-15 21:05:00 +00:00
Brian Barrett
ac7b97a0d9 * convert to an array of event handles - still of size 1 - as prep work for
adding event queues for dropped fragments and retransmit requests

This commit was SVN r5718.
2005-05-13 18:36:15 +00:00
Brian Barrett
5d57956a02 * rename event queue flag to prepare for more event queues
This commit was SVN r5717.
2005-05-13 18:27:27 +00:00
Brian Barrett
66a0c49e2b Run the event loop once after adding a new signal event to the event library. Events are only processed at the start of an event loop (not at event add), so there was a window of time between event_add() and event_loop() for the signal event in which the event existed, but was not active. During this window, signals that should have triggered a callback could be lost.
Reviewed by Jeff and Tim.

This commit was SVN r5715.
2005-05-13 17:59:36 +00:00
Jeff Squyres
72f86297c8 Somehow this functionality got lost over time: when a process aborts,
orterun should abort the rest.  Reviewed by Brian.

This commit was SVN r5713.
2005-05-13 17:52:50 +00:00
Josh Hursey
46810fd155 Some fixes to get the subset of mca directories compile under Windows.
Added a special case under the win_makefile for the gpr/replica directory
since it contains multiple dependant layers of directories.

Added a couple of OMPI_DECLSPECs. Change a conflicting variable name in
gpr_replica_dict_tl.c from 'new' to 'new_dict'.

This commit was SVN r5712.
2005-05-13 16:55:14 +00:00
Josh Hursey
f176f85e55 Fix a couple of library checks, and some Windows related code
This commit was SVN r5711.
2005-05-13 15:05:07 +00:00
Jeff Squyres
c12d5c8c88 While waiting for fortran compiles...
Fixes for orterun in handling different MCA params for different
processes (reviewed by Brian):
- By design, if you run the following:
    mpirun --mca foo aaa --mca foo bbb a.out
  a.out will get a single MCA param for foo with value "aaa,bbb".
- However, if you specify multiple apps with different values for the
  same MCA param, you should expect to get the different values for
  each app.  For example:
    mpirun --mca foo aaa a.out : --mca foo bbb b.out
  Should yield a.out with a "foo" param with value "aaa" and b.out
  with a "foo" param with a value "bbb".  
- This did not work -- both a.out and b.out would get a "foo" with
  "aaa,bbb".
- This commit fixes this behavior -- now a.out will get aaa and b.out
  will get bbb.
- Additionally, if you mix --mca and and app file, you can have
  "global" params and per-line-in-the-appfile params.  For example:
    mpirun --mca foo zzzz --app appfile
  where "appfile" contains:
    -np 1 --mca bar aaa a.out
    -np 1 --mca bar bbb b.out
  In this case, a.out will get foo=zzzz and bar=aaa, and b.out will
  get foo=zzzz and bar=bbb.
Spiffy.

Ok, fortran build is done... back to Fortran... sigh...

This commit was SVN r5710.
2005-05-13 14:36:36 +00:00
Jeff Squyres
0fe7168823 Add missing header file.
This commit was SVN r5709.
2005-05-13 12:36:05 +00:00
Brian Barrett
f64e52a28d * more refactoring to reduce duplicate code
This commit was SVN r5708.
2005-05-13 04:04:08 +00:00
Brian Barrett
a242d5ad4f * Add OMPI_OUTPUT_VERBOSE macro that (like OMPI_OUTPUT) is a no-op when
debugging is disabled.
* convert one more output to only happen when debugging is enabled

This commit was SVN r5707.
2005-05-13 03:01:02 +00:00
Brian Barrett
a7fd494448 * start cleaning up output statements
* start refactoring duplicate code into inline functions (probably will
  have to become macros, but not until debugging is done)
* general code cleanup

This commit was SVN r5706.
2005-05-13 02:54:06 +00:00
Ralph Castain
fdfe457578 Bring in the remote launch changes. This still isn't fully functional, but impacted a few other places that were worth fixing.
1. Added a new function to launch head node processes on remote nodes.

2. Added new tool "orteprobe" that checks to see if a daemon is running on a node. If so, it reports the contact info back to the requestor. If not, it will (eventually - but not now) fork/exec a daemon on the node, report the contact info back to requestor, and then die.

3. Modified orted to handle universe name parameters, and added separate command line flags for debugging the daemon and saving daemon debugging output in a file. The "debug" flag now turns on the runtime debug info instead of the daemon debug - thus, you can now just get daemon debug info if you like.

4. Fix the dps to handle zero length strings correctly.

5. Modify the fork and rsh launchers to pass required environmental variables to the daemons and processes

6. Pulled the redirection of stdin/stdout/stderr for the daemon out of orted and put it into the daemon_init function to simplify orted logic.

7. Modified sys_info to correctly deal with passed mca param

8. Modified univ_info to parse incoming universe location information.

This commit was SVN r5705.
2005-05-12 21:44:23 +00:00
Brian Barrett
0c6eaaebe3 * start cleaning up debugging output (still much to do)
* make buffers really big so that we pass allocmem until we figure out
  why we're not flow controlling as I expected
* set event queue to invalid intially and use that as the enabled test
  rather than a seperate bool - shrinks the module a bit
* add dropped count checks, with a panic if one occurs.  Still need to
  implement some type of retransmit logic.

This commit was SVN r5704.
2005-05-12 21:28:48 +00:00
Brian Barrett
e2c2c72b84 Changes to pass allocmem IBM test
- don't free the send buffer unless the converter tells us we need to
  - properly do the math to determine when the receive buffer has been
    fully used and unlinked itself

This commit was SVN r5703.
2005-05-12 19:52:51 +00:00
Jeff Squyres
f5657fb8ee For the rsh pls, if the launch is on the local node, just exec it --
don't bother using the launching agent (typically rsh or ssh).

This commit was SVN r5702.
2005-05-12 19:12:53 +00:00
Jeff Squyres
544f9dd780 Fix silly string error (missing +2 in the len calculation, so just
replace it with asprintf).  Reviewed by Brian.

This commit was SVN r5700.
2005-05-12 18:56:05 +00:00
Jeff Squyres
f96d763aa7 /trunk is working towards 1.0
/branches/v0.9 is working towards 0.9

This commit was SVN r5699.
2005-05-12 17:56:42 +00:00