navarro [Thu, 25 Oct 2012 11:53:07 +0000 (13:53 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
navarro [Thu, 25 Oct 2012 11:53:01 +0000 (13:53 +0200)]
Clean and add files for simgrid documentation.
Now the distrib can generate the documentation
Lucas Schnorr [Thu, 25 Oct 2012 10:18:02 +0000 (12:18 +0200)]
[trace] update ChangeLog with recent changes on the TRACE module
navarro [Thu, 25 Oct 2012 09:51:13 +0000 (11:51 +0200)]
Add header, footer and stylesheet to the distrib
navarro [Thu, 25 Oct 2012 09:31:19 +0000 (11:31 +0200)]
Add dev_guide files to the distrib.
paul bedaride [Thu, 25 Oct 2012 09:20:54 +0000 (11:20 +0200)]
add missing userguide files
navarro [Wed, 24 Oct 2012 13:52:08 +0000 (15:52 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
navarro [Wed, 24 Oct 2012 13:51:17 +0000 (15:51 +0200)]
Always checkout simgrid with git:// and not https://
Martin Quinson [Wed, 24 Oct 2012 13:44:54 +0000 (15:44 +0200)]
ops, forgot to document the changes of JB this summer
Martin Quinson [Wed, 24 Oct 2012 13:29:59 +0000 (15:29 +0200)]
some cleanups to the changelog. Still improvable
Martin Quinson [Wed, 24 Oct 2012 13:28:37 +0000 (15:28 +0200)]
proposal of NEWS file for the upcomming release
Martin Quinson [Wed, 24 Oct 2012 13:13:41 +0000 (15:13 +0200)]
Desactivate an historical MC_assert so that model-checking Chord get a chance to work
Leave a comment on this first historical battle for MC in SimGrid in
the source code :)
Martin Quinson [Wed, 24 Oct 2012 12:59:08 +0000 (14:59 +0200)]
Document some of Pierre's work
Arnaud Giersch [Wed, 24 Oct 2012 13:17:59 +0000 (15:17 +0200)]
Fix format string and args.
Maximiliano Geier [Wed, 24 Oct 2012 12:59:36 +0000 (14:59 +0200)]
Pull again, something was changed before I could push
Maximiliano Geier [Wed, 24 Oct 2012 12:57:28 +0000 (14:57 +0200)]
Pull changes to push BitTorrent fix
Maximiliano Geier [Wed, 24 Oct 2012 12:55:53 +0000 (14:55 +0200)]
BitTorrent changes:
Fix message sizes (used MB instead of bytes)
Fix tesh again to match this change
Define new message type MESSAGE_CANCEL (not implemented yet)
Arnaud Giersch [Wed, 24 Oct 2012 12:55:22 +0000 (14:55 +0200)]
Disable mmalloc_test when !HAVE_MMAP, since mmalloc is not compiled in.
navarro [Wed, 24 Oct 2012 11:45:18 +0000 (13:45 +0200)]
Update script for build default conf into build-def directory
navarro [Wed, 24 Oct 2012 11:12:36 +0000 (13:12 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
Augustin Degomme [Wed, 24 Oct 2012 09:36:24 +0000 (11:36 +0200)]
edit changelog with smpi changes
navarro [Wed, 24 Oct 2012 09:37:16 +0000 (11:37 +0200)]
Update pipol script
navarro [Wed, 24 Oct 2012 09:36:15 +0000 (11:36 +0200)]
Fix libunwind install on ubuntu
Martin Quinson [Wed, 24 Oct 2012 08:13:38 +0000 (10:13 +0200)]
make it work on x86 too. Other arches are still broken for MC and this is wrong
Arnaud Giersch [Tue, 23 Oct 2012 15:13:01 +0000 (17:13 +0200)]
Cosmetics: Having -> Have; write -> written.
Arnaud Giersch [Tue, 23 Oct 2012 14:30:58 +0000 (16:30 +0200)]
Regenerate with latest flexml.
navarro [Tue, 23 Oct 2012 19:16:04 +0000 (21:16 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
navarro [Tue, 23 Oct 2012 19:15:49 +0000 (21:15 +0200)]
Add dev_guide mecanism
Augustin Degomme [Tue, 23 Oct 2012 08:20:29 +0000 (10:20 +0200)]
number of elements can be 0 in type constructors
Augustin Degomme [Tue, 23 Oct 2012 08:13:47 +0000 (10:13 +0200)]
do not free the group of MPI_COMM_WORLD when asked, because it is a global in smpi (it is freed in the smpi_global_destroy function)
Augustin Degomme [Tue, 23 Oct 2012 08:11:47 +0000 (10:11 +0200)]
we should not check subtypes are commited before, as this is not required by the standard, only types directly used in MPI calls should be commited
Augustin Degomme [Tue, 23 Oct 2012 08:04:57 +0000 (10:04 +0200)]
add MPI_PACKED datatype (for compilation only, MPI_Pack and Unpack functions are still not implemented )
Augustin Degomme [Tue, 23 Oct 2012 08:01:39 +0000 (10:01 +0200)]
do not free the group when freeing the comm, there is a separate function for that, which may be called
navarro [Tue, 23 Oct 2012 15:45:18 +0000 (17:45 +0200)]
Fix compilation error since TAG tutorial
navarro [Tue, 23 Oct 2012 14:20:51 +0000 (16:20 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
navarro [Tue, 23 Oct 2012 14:19:55 +0000 (16:19 +0200)]
Add first steps for a TAG tutorial
Arnaud Giersch [Tue, 23 Oct 2012 11:28:17 +0000 (13:28 +0200)]
Set smpi/cpu_threshold to 1 for smpi-replay, and hope to have deterministic test results now.
Arnaud Giersch [Tue, 23 Oct 2012 11:24:03 +0000 (13:24 +0200)]
Call gap_remove in generic_update_actions_state_lazy too.
*** Please double-check this commit ! ***
Try to mimic what's done in generic_update_actions_state_full.
Fixes an "use after free" error with smpi_replay.
Arnaud Giersch [Tue, 23 Oct 2012 08:54:14 +0000 (10:54 +0200)]
Define an additional parameter for the hack around smpi_replay in smpirun.
Relying on the command name is too fragile.
Arnaud Giersch [Tue, 23 Oct 2012 08:46:10 +0000 (10:46 +0200)]
Cosmetics: rewrite usage() in smpirun.in.
Remove duplicates, use a "here document" instead of a series of echo's.
Arnaud Giersch [Tue, 23 Oct 2012 13:44:01 +0000 (15:44 +0200)]
Cannot use xbt_die when log layout format contains errors, since it uses log facilities.
Use fprintf(stderr) instead, then abort.
Arnaud Giersch [Tue, 23 Oct 2012 11:39:56 +0000 (13:39 +0200)]
PLATFGEN_SRC is useless in DefinePackage.cmake.
Fixes a cmake warning about source file "src/surf/platf_generator.c" listed multiple times for target "simgrid".
Arnaud Giersch [Mon, 22 Oct 2012 11:05:24 +0000 (13:05 +0200)]
Add missing calls to XBT_LOG_CONNECT.
suter [Mon, 22 Oct 2012 13:24:06 +0000 (15:24 +0200)]
move the xbt_mallocator_release after the free of the tracing stuff.
Arnaud Giersch [Fri, 19 Oct 2012 21:20:25 +0000 (23:20 +0200)]
Exclude smpi_process_argc and smpi_glogal_size when !SMPI_F2C.
This completes commit
4fb1f678c8da43cdf1bfcc6108883ef2a4cea952,
and fixes compilation errors in paranoid mode.
navarro [Fri, 19 Oct 2012 15:53:38 +0000 (17:53 +0200)]
Fix compilation error for smpi.
Can enable smpi whitout f2c
navarro [Fri, 19 Oct 2012 14:31:56 +0000 (07:31 -0700)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr/gitroot/simgrid/simgrid
navarro [Fri, 19 Oct 2012 14:31:42 +0000 (07:31 -0700)]
Compil smpi on win32.
Need to see why when compiling examples WinMain is not found
paul bedaride [Fri, 19 Oct 2012 12:34:48 +0000 (14:34 +0200)]
disable index for user guide
paul bedaride [Fri, 19 Oct 2012 12:12:56 +0000 (14:12 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
paul bedaride [Fri, 19 Oct 2012 12:12:45 +0000 (14:12 +0200)]
clean user guide
navarro [Fri, 19 Oct 2012 10:20:11 +0000 (03:20 -0700)]
Clean for win32
navarro [Fri, 19 Oct 2012 10:15:07 +0000 (12:15 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
navarro [Fri, 19 Oct 2012 10:14:25 +0000 (12:14 +0200)]
Need this file for windows. But still need to clean (not delete)
This reverts commit
2b86f38f9d331d921062c054ed5db7fba93c7583.
Maximiliano Geier [Fri, 19 Oct 2012 08:45:34 +0000 (10:45 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
navarro [Thu, 18 Oct 2012 18:26:47 +0000 (20:26 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
navarro [Thu, 18 Oct 2012 18:26:27 +0000 (20:26 +0200)]
Add missing files for distrib
Martin Quinson [Wed, 17 Oct 2012 21:36:34 +0000 (23:36 +0200)]
plug some memleaks in that example
This induce using static memory for the processes that are forcefully
stopped, preventing the end of their main to get executed.
Arnaud Giersch [Thu, 18 Oct 2012 14:18:33 +0000 (16:18 +0200)]
Copy hostfile for smpi examples moved to teshsuite.
Arnaud Giersch [Thu, 18 Oct 2012 14:12:17 +0000 (16:12 +0200)]
Don't ignore return value of 'system'.
It is sometimes declared with attribute warn_unused_result.
Arnaud Giersch [Thu, 18 Oct 2012 13:49:51 +0000 (15:49 +0200)]
Remove memcheck_tests.cmake from repository.
It's a pain to keep it up-to-date, and it already gets regenerated
automatically with enable_memcheck=ON.
Augustin Degomme [Thu, 18 Oct 2012 11:00:37 +0000 (13:00 +0200)]
move smpi examples to teshsuite
Maximiliano Geier [Thu, 18 Oct 2012 13:27:07 +0000 (15:27 +0200)]
BitTorrent: fixed message sizes, based on the actual protocol sizes
Fixed tesh to match this change
navarro [Thu, 18 Oct 2012 13:10:12 +0000 (15:10 +0200)]
cosmetics
paul bedaride [Thu, 18 Oct 2012 11:17:50 +0000 (13:17 +0200)]
Ref Guide style update
Martin Quinson [Wed, 17 Oct 2012 21:25:06 +0000 (23:25 +0200)]
Cancel the comms of leaving processes instead of destroying them
This avoids segfaults when a process leaves the simulation before all
its asynchronous comms (included detached ones) end.
This may leak the canceled actions, but that's another bug, in
comm_cancel.
Arnaud Giersch [Wed, 17 Oct 2012 14:14:28 +0000 (16:14 +0200)]
Ignore the first two lines of output.
Martin Quinson [Wed, 17 Oct 2012 20:04:27 +0000 (22:04 +0200)]
Ensure that the mallocator are really inactive when running MC
- before, they were created during the preinit and then destroyed when
MC was activated.
- now, they are created inactive, and activated when the configuration
stops if MC is still not activated.
- this mandates some changes around the initialization mechanism in
the whole stack.
- one hint that it was a good idea is that before, _surf_init_status
were never reaching the value 2 (meaning "config now forbidden",
according to the preexisting asserts).
- Now this value is set from surf_routing when the first element of
platform is created. This is a horrible hack intermixing the
layers, but it was already accredited by the parser that
configuration could occur until the first <AS>, but no later.
- we now would need a proper way of increasing the init level of
simgrid.
I hope that this change did not break anything, but much more work
would be mandated to clean the init mess properly.
schnorr [Wed, 17 Oct 2012 13:58:38 +0000 (15:58 +0200)]
[trace] fix tesh'es to compare trace file considering all lines except the first two
schnorr [Wed, 17 Oct 2012 13:53:11 +0000 (15:53 +0200)]
[trace] output the command line used to generate the trace file
Arnaud Giersch [Wed, 17 Oct 2012 12:40:46 +0000 (14:40 +0200)]
Function MSG_get_host_number() is not deprecated anymore.
Acknowledge commit
b00e4794c073855ab40697e9ec648c6a67f013e9
(Doesn't see why MSG_get_host_number should be deprecated.)
Arnaud Giersch [Wed, 17 Oct 2012 11:12:45 +0000 (13:12 +0200)]
Cosmetics: remove spaces at EOL.
Arnaud Giersch [Wed, 17 Oct 2012 08:37:06 +0000 (10:37 +0200)]
Use function-like macros instead of inline functions for MC_is_active().
When using functions, Gcc apparently does not remove dead branches when
building without optimizations, and then complains at link time with:
Linking C executable ../../bin/gras_stub_generator
../../lib/libsimgrid.so.3.8: error: undefined reference to 'MC_ignore'
../../lib/libsimgrid.so.3.8: error: undefined reference to 'MC_do_the_modelcheck_for_real'
../../lib/libsimgrid.so.3.8: error: undefined reference to 'MC_process_clock_get'
../../lib/libsimgrid.so.3.8: error: undefined reference to 'MC_process_clock_add'
../../lib/libsimgrid.so.3.8: error: undefined reference to 'MC_memory_init'
../../lib/libsimgrid.so.3.8: error: undefined reference to 'MC_new_stack_area'
collect2: error: ld returned 1 exit status
This is if course when building with -Denable_model-checking=OFF.
Martin Quinson [Tue, 16 Oct 2012 20:58:18 +0000 (22:58 +0200)]
s/MC_IS_ENABLED/MC_is_active()/ to remove an unfriendly pitfall of the codebase
This fixes #14976 (thanks to LucasN for the report)
Martin Quinson [Tue, 16 Oct 2012 20:51:00 +0000 (22:51 +0200)]
cosmetics and improvement
Arnaud Giersch [Tue, 16 Oct 2012 19:31:57 +0000 (21:31 +0200)]
Distinguish outputs for mmalloc_test on 32 and 64bits architectures.
Martin Quinson [Tue, 16 Oct 2012 20:04:22 +0000 (22:04 +0200)]
document that we really need xbt_os_thread_atfork (for tesh+mmalloc)
Martin Quinson [Tue, 16 Oct 2012 16:18:27 +0000 (18:18 +0200)]
improve the mmalloc tesh to ensure that use after free don't ruin mmalloc
Martin Quinson [Tue, 16 Oct 2012 16:11:50 +0000 (18:11 +0200)]
Update the mmalloc tesh file
- We now write the offset in hexa
- exact returned pointers changed a bit with last mmalloc rewamping,
but they still seem reasonable
Martin Quinson [Tue, 16 Oct 2012 16:10:33 +0000 (18:10 +0200)]
display offsets in hexa (counting in decimal sometimes difficult)
Martin Quinson [Tue, 16 Oct 2012 16:06:31 +0000 (18:06 +0200)]
Do not store any metadata where the user (was) legitimate to write
Before, free fragments were chained within the user area, leading to
obscure breakdown when the user was mean enough to write into the
block after a free(). Now, these data are stored as swag directly into
the mdp->heapinfo that is where we store our metadata.
We have one extra complication due to the fact that heapinfo must be
reallocated when we mmap more memory. When this happens, we have to
update all swag hooks manually to apply the offset.
Yup, this went that bad :-/
Arnaud Giersch [Tue, 16 Oct 2012 16:09:44 +0000 (18:09 +0200)]
Use double quotes around variable expansion to protect spaces.
Arnaud Giersch [Tue, 16 Oct 2012 12:46:57 +0000 (14:46 +0200)]
Make working directory really clean between builds.
Arnaud Giersch [Tue, 16 Oct 2012 12:46:57 +0000 (14:46 +0200)]
Regenerate tesh files for chord examples.
Martin Quinson [Tue, 16 Oct 2012 00:12:07 +0000 (02:12 +0200)]
sg_cmdline is a public dynar containing all command-line arguments we got
Martin Quinson [Mon, 15 Oct 2012 23:23:50 +0000 (01:23 +0200)]
cosmetics
Martin Quinson [Mon, 15 Oct 2012 22:52:43 +0000 (00:52 +0200)]
ensures that runall is never called on empty set
Not sure of when it can happen, but it does, forcing the context
factories to be defensive. This was causing issues in java coroutines
as this factory was not defensive enough. So avoiding future issues
seems interesting to me.
Martin Quinson [Mon, 15 Oct 2012 21:57:55 +0000 (23:57 +0200)]
search libunwind with cmake and die at config time if not found
Martin Quinson [Sun, 14 Oct 2012 19:57:17 +0000 (21:57 +0200)]
ouch, I've lost an eye on that
Martin Quinson [Thu, 11 Oct 2012 07:18:24 +0000 (09:18 +0200)]
try to get /some/ info from the cdash by not intentionaly breaking the tesh files
Martin Quinson [Thu, 11 Oct 2012 00:45:57 +0000 (02:45 +0200)]
more cruft to ignore
Lucas Nussbaum [Sun, 14 Oct 2012 10:56:09 +0000 (12:56 +0200)]
chord: avoid duplicate call of MSG_comm_test()
Commit
caffead (exactly one year ago!) introduced a duplicate call to
MSG_comm_test(): MSG_comm_test() could return true on line 336, and
then be called again there. That caused a crash since the communication
was already finalized in the first MSG_comm_test() call.
suter [Fri, 12 Oct 2012 23:01:17 +0000 (01:01 +0200)]
list bug fixed in SD_simulate
suter [Fri, 12 Oct 2012 22:57:03 +0000 (00:57 +0200)]
fix an issue that occurs when SD_simulate is called with a positive
value (simulate for X seconds). The simulation does not always stop
after the expected duration.
Cause: The "less than" test is sometimes tricky with doubles.
Fix: Replace by "difference smaller than 0.00001 (the maxmin precision).
+ indent
+ one more verbose message
navarro [Fri, 12 Oct 2012 12:52:49 +0000 (14:52 +0200)]
Compilation with warning flags
navarro [Fri, 12 Oct 2012 12:42:40 +0000 (14:42 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
navarro [Fri, 12 Oct 2012 12:42:32 +0000 (14:42 +0200)]
Finish the tutorial for new model and api
Marion Guthmuller [Fri, 12 Oct 2012 11:41:03 +0000 (13:41 +0200)]
model-checker : display malloc backtrace according to address
navarro [Fri, 12 Oct 2012 10:06:01 +0000 (12:06 +0200)]
Oups rename function
navarro [Fri, 12 Oct 2012 09:48:51 +0000 (11:48 +0200)]
Add first steps to explain How to add a new model in Simgrid