navarrop [Wed, 2 Dec 2009 16:34:13 +0000 (16:34 +0000)]
Stop building a 3.3.3 version...
3.3.3 has been stable for a while. We're preparing the 3.3.4 here.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6863
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 2 Dec 2009 15:49:30 +0000 (15:49 +0000)]
Another try to get the make dist working with or without supernovae
This was not a bug of automake, but a feature: it seeks every files
which may be needed by a if/endif branch and makes sure they are in
the archive. Adding an extra variable didn't solve the issue since
automake follows variable definitions.
Declaring supernovae files as BUILT_SOURCE was uneffective (still
trying to add them to the archive) and counter productive (automake
tried to build them even in non-supernovae mode)
So the solution is to prevent automake from derivating the DIST files
from the lib_*_la_SOURCES. For that, I use nodist_lib_*_la_SOURCES
and take care of the EXTRA_DIST files manually.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6862
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 2 Dec 2009 11:01:31 +0000 (11:01 +0000)]
Circumvent a bug in automake (at least found in v1.11)
If lib*_la_SOURCES are defined in a if/endif alternative, it works for
compilation, but the dist target uses the first variante of the
variable, regardless of whether this is the used variante or not.
So, use yet another name in the alternative and define lib*_la_SOURCES
only once, out of the alternatives
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6861
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 2 Dec 2009 10:34:19 +0000 (10:34 +0000)]
include all the needed files in the archive so that the tar.gz compiles afterward
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6860
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 2 Dec 2009 10:34:17 +0000 (10:34 +0000)]
Replay is not mature enough for distribution, yet
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6859
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 2 Dec 2009 10:20:33 +0000 (10:20 +0000)]
Replay is not mature enough for distribution, yet
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6858
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 2 Dec 2009 10:18:30 +0000 (10:18 +0000)]
create the listener thread in joinable mode since we join it at the end
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6857
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Mon, 30 Nov 2009 15:39:35 +0000 (15:39 +0000)]
Modify the output of the action replay tool.
Each action now output a message wit the same structure as its input
PLUS an extra field giving the time spent to execute this action
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6856
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 24 Nov 2009 11:36:15 +0000 (11:36 +0000)]
default of supernovae should be the opposite of maintainer-mode
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6855
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 24 Nov 2009 11:28:18 +0000 (11:28 +0000)]
Force initialization of a variable to shut gcc 4.4 up
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6854
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 20:40:46 +0000 (20:40 +0000)]
more inlining
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6853
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 20:40:32 +0000 (20:40 +0000)]
Mark two more functions as inline in supernovae, and use selective update by default
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6852
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 20:40:17 +0000 (20:40 +0000)]
Force memory allocation functions to always inline
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6851
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 20:40:03 +0000 (20:40 +0000)]
kill useless code
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6850
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 11:50:32 +0000 (11:50 +0000)]
Mark a bunch of functions as candidates to inlining. Not quite sure that gcc does it, yet
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6849
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 11:50:08 +0000 (11:50 +0000)]
these are variables, not functions. Use the right macro
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6848
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 11:50:05 +0000 (11:50 +0000)]
In supernovae mode, declare every public functions as [potentially] inline
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6847
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
alegrand [Fri, 20 Nov 2009 09:59:00 +0000 (09:59 +0000)]
Take into account "recent" simix modifications
Namely, the use of semaphores for communications in SIMIX_process
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6846
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
alegrand [Fri, 20 Nov 2009 09:55:32 +0000 (09:55 +0000)]
remove wrong comment
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6845
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 09:47:53 +0000 (09:47 +0000)]
Change the code so that it compiles in supernovae mode
- _simgrid_log_channel__default is then a macro defined in
mk_supernovae.sh instead of a global static to the compilation unit
- lots and lots of static symbol renaming (prefixing) to avoid name
clashes. I love eclipse refactoring functions
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6844
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 09:47:29 +0000 (09:47 +0000)]
Try to compile in supernovae mode when requested to
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6843
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 20 Nov 2009 09:47:26 +0000 (09:47 +0000)]
A configure option to control the supernovae mode
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6842
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 18 Nov 2009 07:30:16 +0000 (07:30 +0000)]
Also include hostfile into the archive since makecheck needs that
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6841
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 18 Nov 2009 07:28:20 +0000 (07:28 +0000)]
Some files missing from the archive. This prevents makecheck from running well
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6840
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
donassbr [Mon, 16 Nov 2009 15:41:27 +0000 (15:41 +0000)]
Bugfix: Initialize variable and remove from heap when action is suspended.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6839
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 11 Nov 2009 16:09:33 +0000 (16:09 +0000)]
typo found by Medhi while doing the tutorial
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6838
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 10 Nov 2009 12:33:10 +0000 (12:33 +0000)]
don't trust gcc for cleaning your code after you. he's too dumb to be trusted
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6837
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 10 Nov 2009 12:32:54 +0000 (12:32 +0000)]
Revert "try to port the gras simulation side to the new smx_network infrastructure (not yet working)"
This git branch is not ready for public consumption yet (sorry)
This reverts commit
063c63642a29000a011c0d6176d30eb62a4e0dca.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6836
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 10 Nov 2009 12:28:14 +0000 (12:28 +0000)]
try to port the gras simulation side to the new smx_network infrastructure (not yet working)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6835
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 10 Nov 2009 11:52:05 +0000 (11:52 +0000)]
Kill portability to MAC for sake of compiling on linux. See #8852 for a proper solution (which will want for the IJD to arrive)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6834
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Thu, 5 Nov 2009 22:48:55 +0000 (22:48 +0000)]
First implementation of allReduce action (Refactoring will be needed
at some point)
All involved processes exit at the same time
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6833
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
donassbr [Wed, 4 Nov 2009 12:14:20 +0000 (12:14 +0000)]
Fix test. Remove timestep parameter.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6832
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
donassbr [Wed, 4 Nov 2009 12:14:17 +0000 (12:14 +0000)]
Remove parameter timestep.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6831
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
donassbr [Wed, 4 Nov 2009 12:13:53 +0000 (12:13 +0000)]
Changes implementation of CPU TI. Now, it uses the simple integration of trace and does a binary search to find the points in the array.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6830
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
donassbr [Wed, 4 Nov 2009 12:13:50 +0000 (12:13 +0000)]
Add 2 tests to handle problems when using timesteps too big.
This part of code will disappear soon, but this commit can be used to retrieve this implementation of trace integration feature (with multi levels).
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6829
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
donassbr [Wed, 4 Nov 2009 12:13:26 +0000 (12:13 +0000)]
Add tests to skip actions that aren't running or have priority = 0.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6828
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
donassbr [Wed, 4 Nov 2009 12:13:23 +0000 (12:13 +0000)]
Remove old code and useless code.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6827
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Fri, 30 Oct 2009 17:27:34 +0000 (17:27 +0000)]
Not using set/get functions is bad, but raising an errer when a set
modifies an existing data is worse.
Remove the paranoid assert
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6826
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Thu, 29 Oct 2009 22:24:49 +0000 (22:24 +0000)]
add a reduce X bytes Y flops action
once the root has received all the contributions of X bytes from the
other processes, it computes Y flops. The other processes exit the
function as soon as they their send is completed.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6825
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Thu, 29 Oct 2009 21:15:57 +0000 (21:15 +0000)]
Refactor spawned_send and bcast_spawned_send
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6824
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Thu, 29 Oct 2009 20:59:50 +0000 (20:59 +0000)]
add a flat broadcast action
requires a comm_size action to set the size of the MPI_COMM_WORLD
communicator
update the actions_with_isend.txt to include the Isend action
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6823
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 29 Oct 2009 16:10:40 +0000 (16:10 +0000)]
plug a memleak
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6822
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Thu, 29 Oct 2009 10:52:55 +0000 (10:52 +0000)]
forgot to remove the cleaning of a remove data
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6821
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Thu, 29 Oct 2009 10:51:46 +0000 (10:51 +0000)]
add an Isend, just the prototype for the barrier action
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6820
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 29 Oct 2009 07:00:59 +0000 (07:00 +0000)]
try harder not to overflow that buffer
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6819
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 28 Oct 2009 22:36:02 +0000 (22:36 +0000)]
try to ensure that stdio is loaded when str.h is loaded, so that __USE_GNU is defined on glibc
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6818
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 28 Oct 2009 22:17:55 +0000 (22:17 +0000)]
Try to fix #8852 (SimGrid won't compile on Ubuntu 9.10) by betterly checking my arguments to standard functions
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6817
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Fri, 23 Oct 2009 05:59:09 +0000 (05:59 +0000)]
small improvements suggested by Arnaud
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6813
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Thu, 22 Oct 2009 18:12:25 +0000 (18:12 +0000)]
Add Irecv and wait actions and corresponding example
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6812
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:10:35 +0000 (13:10 +0000)]
Change the waitany functions on semaphore and communications to return the rank of the object in the dynar instead of the object itself
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6811
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:10:31 +0000 (13:10 +0000)]
Implement smx_comm_t SIMIX_network_waitany(xbt_dynar_t comms)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6810
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:10:07 +0000 (13:10 +0000)]
Implement smx_sem_t SIMIX_sem_acquire_any(xbt_dynar_t sems)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6809
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:10:04 +0000 (13:10 +0000)]
when requested to acquire a cond or semaphore with max_duration=0, call the regular blocking function instead of the internal code so that the dummy action in charge of detecting host failures get created
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6808
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:09:40 +0000 (13:09 +0000)]
Implement Asynchronous send/recv in SIMIX (you dreamt of it? It's here now)
Too bad that nobody cared enough to propagate the change to MSG (yet).
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6807
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:09:36 +0000 (13:09 +0000)]
Enough is enough. Add semaphores as SIMIX synchronization mechanism.
Check the SIMIX_sem_release_forever() function, which allows something
between conditions and pure semaphores.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6806
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:09:13 +0000 (13:09 +0000)]
Adapt tests to the fact that we can now detect that there was a timeout during communication
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6805
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:09:10 +0000 (13:09 +0000)]
use a bit more informative actions'names. Helpful while debuging
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6804
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:08:47 +0000 (13:08 +0000)]
Add SIMIX_rdv_{set,get}_data; tiny fixups
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6803
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:08:44 +0000 (13:08 +0000)]
New function: SIMIX_action_set_name()
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6802
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:08:20 +0000 (13:08 +0000)]
useless cosmetics
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6801
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Tue, 20 Oct 2009 13:08:16 +0000 (13:08 +0000)]
a script to generate a strassen workflow (+cosmetics on other script)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6800
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 16 Oct 2009 09:06:55 +0000 (09:06 +0000)]
Fix a vicious bug in dictionaries inducing that some elements were not freed on xbt_dict_free()
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6799
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 16 Oct 2009 07:54:18 +0000 (07:54 +0000)]
Integrate patch #8636: Obey DESTDIR when installing documentation (thanks to Robson Peixoto)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6798
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 16 Oct 2009 07:37:44 +0000 (07:37 +0000)]
Introduce a MSG_TIMEOUT_FAILURE return code and use it for communication timeouts (closes #8563)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6797
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Fri, 16 Oct 2009 07:37:42 +0000 (07:37 +0000)]
Fix #8569: XBT/synchro.h has redundant declarations
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6796
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:26:55 +0000 (16:26 +0000)]
A script to recompile and rerun the whole chain from platfile+daxfile to actual replay on simulator
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6795
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:26:32 +0000 (16:26 +0000)]
A script to generate neat DAX files of fork/joins
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6794
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:26:30 +0000 (16:26 +0000)]
Use private headers to help debugging
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6793
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:26:07 +0000 (16:26 +0000)]
Various fixes. Working (in SG) for ForkJoins, but not for Montage.
* Change synchronization scheme for ending: workers send go to master
instead of having the master calling amok_pm_group_end() right after
the setup. This killed the workers ways too early.
* make sure that workers do create all mandatory sockets to others
since dopar threads have real difficulties speaking to the listener
thread to add new sockets
* instead of having a callback to the go message in the worker which
does all the work, do explicitely wait for that message and move the
code in worker's main()
Like I said, there is still a deadlock when trying to replay
Montage_25, but this may related to the fact that this workflow seems
somehow broken. the "cimages.tbl" file, generated by root and consumed
by ID00021@mImgTbl is mangled in the dot output, and thus presumably
also in memory representation. More to come tomorrow.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6792
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:26:05 +0000 (16:26 +0000)]
conditions are not semaphores. signals can get lost if delivered before you are waiting on the cond. So only wait on the condition if the message were not delevered already, and ignore all this cruft in other cases
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6791
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:25:42 +0000 (16:25 +0000)]
start threads only if I have more than 1 element to handle
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6790
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:25:40 +0000 (16:25 +0000)]
Revert the change inducing that GRAS processes keep quiet when leaving on simulator, since it's helpful sometimes, and since it breaks several tesh files
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6789
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:25:17 +0000 (16:25 +0000)]
plug various memleaks
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6788
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:25:15 +0000 (16:25 +0000)]
More informative error message
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6787
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:24:51 +0000 (16:24 +0000)]
Do not call xbt_thread_exit at the end of the dopar worker since this call forcefully kills the process in SG, preventing the wrapper to cleanup around before leaving
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6786
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 16:24:49 +0000 (16:24 +0000)]
unify the log channels for SG and RL implementations of xbt_synchro
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6785
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:54:49 +0000 (11:54 +0000)]
Try to do all communications in parallel (not quite working)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6784
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:54:46 +0000 (11:54 +0000)]
Implement xbt_thread_join and xbt_thread_yield in SG also
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6783
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:54:23 +0000 (11:54 +0000)]
Change the prototype of xbt_thread_create(), sorry.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6782
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:54:18 +0000 (11:54 +0000)]
Ignore (in git) more example binaries
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6781
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:53:55 +0000 (11:53 +0000)]
malloc the expected size (and announce the malloced size)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6780
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:53:53 +0000 (11:53 +0000)]
Give a name to the dopar worker threads since logging from them segfault otherwise
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6779
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:53:30 +0000 (11:53 +0000)]
do not create execution action of size 0 since it seems to lock surf down
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6778
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:53:28 +0000 (11:53 +0000)]
Make SIMIX_process_yield() public
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6777
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:53:05 +0000 (11:53 +0000)]
allow to specify tracefilename on cmdline; use workstations in lexical order; return 0 on success
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6776
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Thu, 15 Oct 2009 11:53:02 +0000 (11:53 +0000)]
Add a note to flatifier to the ChangeLog. That tool is really neat
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6775
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 19:14:00 +0000 (19:14 +0000)]
Move my favorite test first
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6774
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 19:13:53 +0000 (19:13 +0000)]
Start to fix the <cluster> tag (see ChangeLog)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6773
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 19:13:46 +0000 (19:13 +0000)]
This file was written when <route:multi> were not symmetric by default. Specify it now
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6772
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 19:13:38 +0000 (19:13 +0000)]
Sanitize static functions' names; kill unused ones
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6771
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
velho [Wed, 14 Oct 2009 18:34:16 +0000 (18:34 +0000)]
Bug fix, corrected jitter works for each linkk separately.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6770
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 10:09:31 +0000 (10:09 +0000)]
Sanitize a bit the <foreach> handling (cosmetics, no bug fixed so far)
Do all the work in the starting tag of <host> and <link> instead of
saving values in starting tag and using them in closing tag.
This allows to:
- have only one function in the interception mecanism and no global
value
- not have to intercept the closing tags and run directly the
original functions
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6769
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 10:09:28 +0000 (10:09 +0000)]
More debug
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6768
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
suter [Wed, 14 Oct 2009 09:14:51 +0000 (09:14 +0000)]
correction/verfication of the tesh up to one_link_state_file
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6767
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 09:11:46 +0000 (09:11 +0000)]
Kill a 'unused variable' warning when compiling without GTNets
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6766
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 09:11:44 +0000 (09:11 +0000)]
I really need this to be printed as an int to fit DOT taste. Let's assume that longs are as long as pointer on any arch
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6765
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 09:07:09 +0000 (09:07 +0000)]
Make sure it works to compile only one file of the java example directories by adding . to the classpath
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6764
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 09:06:45 +0000 (09:06 +0000)]
Kill a warning in java example: unused variable
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6763
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
mquinson [Wed, 14 Oct 2009 09:06:42 +0000 (09:06 +0000)]
Make sure this variable is always initialized
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6762
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
cristianrosa [Wed, 14 Oct 2009 08:50:58 +0000 (08:50 +0000)]
Fix invalid memory references in SIMIX network.
Fix compilation issues in SimDag on 64bit platforms
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk@6761
48e7efb5-ca39-0410-a469-
dd3cf9ba447f
