Logo AND Algorithmique Numérique Distribuée

Public GIT Repository
simgrid.git
8 years agodocument last changes
Martin Quinson [Fri, 27 Jan 2012 21:16:39 +0000 (22:16 +0100)]
document last changes

8 years agoauto-set the amount of threads when the requested amount is negative
Martin Quinson [Fri, 27 Jan 2012 21:15:09 +0000 (22:15 +0100)]
auto-set the amount of threads when the requested amount is negative

8 years agoreally kill that file
Martin Quinson [Fri, 27 Jan 2012 21:04:37 +0000 (22:04 +0100)]
really kill that file

8 years agoRevert "Put PROCESSOR_COUNT in gras_config.h"
Martin Quinson [Fri, 27 Jan 2012 20:36:53 +0000 (21:36 +0100)]
Revert "Put PROCESSOR_COUNT in gras_config.h"

We need the runtime amount of cores, not the compile-time one.

This reverts commit 4ca11874986e1a694eb4183f8656ccac844cf3ce.

8 years agoRevert "Add a script to detect the number of core."
Martin Quinson [Fri, 27 Jan 2012 20:36:13 +0000 (21:36 +0100)]
Revert "Add a script to detect the number of core."

There is no use of detecting the amount of cores at compile time. We
need the amount of cores at runtime.

This reverts commit 39b99d7659c7f2ecd789c141a8fced9bf9684cff.

8 years agoadd a function xbt_os_get_numcores()
Martin Quinson [Fri, 27 Jan 2012 20:32:42 +0000 (21:32 +0100)]
add a function xbt_os_get_numcores()

8 years agodocument lastest changes
Martin Quinson [Fri, 27 Jan 2012 19:16:32 +0000 (20:16 +0100)]
document lastest changes

8 years agoenable SMPI by default (and rewrite the cmake doc)
Martin Quinson [Fri, 27 Jan 2012 19:14:40 +0000 (20:14 +0100)]
enable SMPI by default (and rewrite the cmake doc)

8 years agoadd a link to aespinosa's work on vagrant to get ruby working on Windows
Martin Quinson [Fri, 27 Jan 2012 17:37:49 +0000 (18:37 +0100)]
add a link to aespinosa's work on vagrant to get ruby working on Windows

8 years agoRegenerate timings of Lua tesh files
Christophe Thiéry [Thu, 26 Jan 2012 07:27:45 +0000 (08:27 +0100)]
Regenerate timings of Lua tesh files

8 years agoSIMIX refactoring: replace 'request' by a more precise term: 'simcall'
Christophe Thiéry [Wed, 25 Jan 2012 17:56:52 +0000 (18:56 +0100)]
SIMIX refactoring: replace 'request' by a more precise term: 'simcall'

All functions that make a system call (or simcall) are now simcall_*
(instead of SIMIX_req_*). They all interrupt the execution flow until
the kernel answers the simcall.
Let's hope that SIMIX will be easier to understand ;)

8 years agoRename SIMIX files
Christophe Thiéry [Wed, 25 Jan 2012 16:42:23 +0000 (17:42 +0100)]
Rename SIMIX files

8 years agoKeep an int parameter to SIMIX_context_get_nthreads
Christophe Thiéry [Wed, 25 Jan 2012 13:27:27 +0000 (14:27 +0100)]
Keep an int parameter to SIMIX_context_get_nthreads

8 years agoParmap: implement the POSIX synchronization mode (not working yet)
Christophe Thiéry [Wed, 25 Jan 2012 13:17:39 +0000 (14:17 +0100)]
Parmap: implement the POSIX synchronization mode (not working yet)

8 years agoDie if the context/synchro option is wrong
Christophe Thiéry [Wed, 25 Jan 2012 13:16:43 +0000 (14:16 +0100)]
Die if the context/synchro option is wrong

8 years agoSurf: create the parmap with the correct number of worker threads
Christophe Thiéry [Wed, 25 Jan 2012 13:16:19 +0000 (14:16 +0100)]
Surf: create the parmap with the correct number of worker threads

8 years agoAllow xbt_parmap_destroy to be called with a NULL parameter
Christophe Thiéry [Wed, 25 Jan 2012 13:11:47 +0000 (14:11 +0100)]
Allow xbt_parmap_destroy to be called with a NULL parameter

8 years agoRevert "Use xbt_config mechanism instead of hard coding value."
Christophe Thiéry [Wed, 25 Jan 2012 10:33:03 +0000 (11:33 +0100)]
Revert "Use xbt_config mechanism instead of hard coding value."

This reverts commit c9c23fbcad8829230381279f08b72d22c82c7d20.

8 years agoAdd configuration auto for contexts/nthreads and use the PROCESSOR_COUNT.
navarro [Wed, 25 Jan 2012 10:36:00 +0000 (11:36 +0100)]
Add configuration auto for contexts/nthreads and use the PROCESSOR_COUNT.
Default number is 1.

8 years agoCompile tracing examples only if tracing is enabled.
navarro [Wed, 25 Jan 2012 10:27:13 +0000 (11:27 +0100)]
Compile tracing examples only if tracing is enabled.

8 years agoPut PROCESSOR_COUNT in gras_config.h
navarro [Wed, 25 Jan 2012 09:29:17 +0000 (10:29 +0100)]
Put PROCESSOR_COUNT in gras_config.h

8 years agoAdd a script to detect the number of core.
navarro [Tue, 24 Jan 2012 10:32:04 +0000 (11:32 +0100)]
Add a script to detect the number of core.

8 years agoDocument the new option --cfg=surf/nthreads
Christophe Thiéry [Tue, 24 Jan 2012 09:34:08 +0000 (10:34 +0100)]
Document the new option --cfg=surf/nthreads

8 years ago[trace] new example inspired by storage systems to trace user variables
Lucas Schnorr [Mon, 23 Jan 2012 15:57:31 +0000 (16:57 +0100)]
[trace] new example inspired by storage systems to trace user variables

8 years ago[trace] add new test, check tracing of g5k platform files
Lucas Schnorr [Mon, 23 Jan 2012 14:06:31 +0000 (15:06 +0100)]
[trace] add new test, check tracing of g5k platform files

8 years ago[trace] types are local, remove global dictionary
Lucas Schnorr [Mon, 23 Jan 2012 13:56:26 +0000 (14:56 +0100)]
[trace] types are local, remove global dictionary

8 years ago[trace] more testing in process migration tracing example
Lucas Schnorr [Mon, 23 Jan 2012 13:37:39 +0000 (14:37 +0100)]
[trace] more testing in process migration tracing example

8 years ago[trace] update tesh, remove timestamping check, add new test with (un)categorized...
Lucas Schnorr [Mon, 23 Jan 2012 13:05:40 +0000 (14:05 +0100)]
[trace] update tesh, remove timestamping check, add new test with (un)categorized tracing

8 years ago[trace] remove repeated parameter, plus add new test for tracing by MSG process
Lucas Schnorr [Mon, 23 Jan 2012 13:02:42 +0000 (14:02 +0100)]
[trace] remove repeated parameter, plus add new test for tracing by MSG process

8 years ago[trace] replace get_type by get_type_or_null when we are defining types
Lucas Schnorr [Mon, 23 Jan 2012 13:01:29 +0000 (14:01 +0100)]
[trace] replace get_type by get_type_or_null when we are defining types

8 years ago[trace] remove recursive get type support, search operations are local in the type...
Lucas Schnorr [Mon, 23 Jan 2012 13:01:01 +0000 (14:01 +0100)]
[trace] remove recursive get type support, search operations are local in the type hierarchy

8 years ago[trace] new example that will be used to test platform tracing
Lucas Schnorr [Mon, 23 Jan 2012 12:28:28 +0000 (13:28 +0100)]
[trace] new example that will be used to test platform tracing

8 years ago[trace] renaming source file to reflect its speciality
Lucas Schnorr [Mon, 23 Jan 2012 12:15:24 +0000 (13:15 +0100)]
[trace] renaming source file to reflect its speciality

8 years ago[trace] sanity checks
Lucas Schnorr [Mon, 23 Jan 2012 12:11:48 +0000 (13:11 +0100)]
[trace] sanity checks

8 years agoPlace MSG init before testing args for --help argument.
navarro [Mon, 23 Jan 2012 10:13:43 +0000 (11:13 +0100)]
Place MSG init before testing args for --help argument.

8 years agoChange cluster sharing policy for FULLDUPLEX by default.
navarro [Mon, 23 Jan 2012 10:06:54 +0000 (11:06 +0100)]
Change cluster sharing policy for FULLDUPLEX by default.
Update tesh files.

8 years agoRemove unused variables and add missing prototypes.
navarro [Mon, 23 Jan 2012 09:51:41 +0000 (10:51 +0100)]
Remove unused variables and add missing prototypes.

8 years agoUse of busy wait and not posix for te moment.
unknown [Mon, 23 Jan 2012 09:09:31 +0000 (10:09 +0100)]
Use of busy wait and not posix for te moment.
On windows we don't have futex so we must use busy wait.

8 years agoUpdate factor for model LegrandVelho.
navarro [Mon, 23 Jan 2012 09:04:45 +0000 (10:04 +0100)]
Update factor for model LegrandVelho.
Update times from tesh.

8 years ago[trace] adding combinations of tracing parameters to smpirun
Lucas Schnorr [Sun, 22 Jan 2012 23:43:47 +0000 (00:43 +0100)]
[trace] adding combinations of tracing parameters to smpirun

8 years ago[trace] be more strict about errors
Lucas Schnorr [Sun, 22 Jan 2012 23:34:52 +0000 (00:34 +0100)]
[trace] be more strict about errors

8 years ago[trace] fixing some memory leaks in smpi instrumentation code
Lucas Schnorr [Sun, 22 Jan 2012 23:22:07 +0000 (00:22 +0100)]
[trace] fixing some memory leaks in smpi instrumentation code

8 years ago[trace] re-writing some tracing functions
Lucas Schnorr [Sun, 22 Jan 2012 23:15:22 +0000 (00:15 +0100)]
[trace] re-writing some tracing functions

8 years ago[trace] very simple program that creates, executes and destroy a task
Lucas Schnorr [Sun, 22 Jan 2012 17:00:12 +0000 (18:00 +0100)]
[trace] very simple program that creates, executes and destroy a task

8 years ago[trace] function now longer exist
Lucas Schnorr [Sun, 22 Jan 2012 16:58:49 +0000 (17:58 +0100)]
[trace] function now longer exist

8 years ago[trace] some checks in tracing functions
Lucas Schnorr [Sun, 22 Jan 2012 16:58:28 +0000 (17:58 +0100)]
[trace] some checks in tracing functions

8 years ago[trace] remove the container from its parent
Lucas Schnorr [Sun, 22 Jan 2012 16:55:34 +0000 (17:55 +0100)]
[trace] remove the container from its parent

8 years ago[trace] in SMPI, follow the same start/end interface used in MSG
Lucas Schnorr [Sun, 22 Jan 2012 16:41:38 +0000 (17:41 +0100)]
[trace] in SMPI, follow the same start/end interface used in MSG

8 years ago[trace] moving traced smpi example to example/smpi/tracing/, plus new simple program
Lucas Schnorr [Sun, 22 Jan 2012 16:33:39 +0000 (17:33 +0100)]
[trace] moving traced smpi example to example/smpi/tracing/, plus new simple program

8 years ago[trace] update test output
Lucas Schnorr [Sat, 21 Jan 2012 11:02:19 +0000 (12:02 +0100)]
[trace] update test output

8 years ago[trace] very small memory leak fix in this example
Lucas Schnorr [Fri, 20 Jan 2012 22:27:00 +0000 (23:27 +0100)]
[trace] very small memory leak fix in this example

8 years ago[trace] removing the tracing per task (where each task was a container)
Lucas Schnorr [Fri, 20 Jan 2012 16:18:42 +0000 (17:18 +0100)]
[trace] removing the tracing per task (where each task was a container)

the --cfg=tracing/msg/task:1 parameter is no longer supported

reason:
- not used

8 years ago[trace] when destroying a container, remove its references from allContainers
Lucas Schnorr [Fri, 20 Jan 2012 15:25:19 +0000 (16:25 +0100)]
[trace] when destroying a container, remove its references from allContainers

8 years ago[trace] remove info messages from process migration test, update its tesh
Lucas Schnorr [Fri, 20 Jan 2012 15:24:44 +0000 (16:24 +0100)]
[trace] remove info messages from process migration test, update its tesh

8 years ago[trace] removing memory leaks from the process migration tracing example
Lucas Schnorr [Fri, 20 Jan 2012 15:05:04 +0000 (16:05 +0100)]
[trace] removing memory leaks from the process migration tracing example

8 years ago[trace] when a process migrates, we need to clean-up its container properly (includin...
Lucas Schnorr [Fri, 20 Jan 2012 15:02:10 +0000 (16:02 +0100)]
[trace] when a process migrates, we need to clean-up its container properly (including from its parent)

8 years ago[trace] when a process ends, remove its container from the parent as well
Lucas Schnorr [Fri, 20 Jan 2012 14:39:51 +0000 (15:39 +0100)]
[trace] when a process ends, remove its container from the parent as well

8 years ago[trace] debug messages for recursive destruction of containers and types
Lucas Schnorr [Fri, 20 Jan 2012 14:38:34 +0000 (15:38 +0100)]
[trace] debug messages for recursive destruction of containers and types

8 years ago[trace] avoid return or break within a xbt_dict_foreach (it causes a memory leak)
Lucas Schnorr [Fri, 20 Jan 2012 14:37:54 +0000 (15:37 +0100)]
[trace] avoid return or break within a xbt_dict_foreach (it causes a memory leak)

8 years ago[trace] separate initialization function from setting the root container
Lucas Schnorr [Fri, 20 Jan 2012 14:14:51 +0000 (15:14 +0100)]
[trace] separate initialization function from setting the root container

8 years ago[trace] function to remove a child from its parent container
Lucas Schnorr [Fri, 20 Jan 2012 14:13:36 +0000 (15:13 +0100)]
[trace] function to remove a child from its parent container

8 years ago[trace] avoid comparing the tracing logs and the generated trace files
Lucas Schnorr [Fri, 20 Jan 2012 10:35:22 +0000 (11:35 +0100)]
[trace] avoid comparing the tracing logs and the generated trace files

details:
- the issues raised by these tests were not related to the tracing system

8 years ago[trace] cleaner code for tracing examples
Lucas Schnorr [Fri, 20 Jan 2012 10:31:29 +0000 (11:31 +0100)]
[trace] cleaner code for tracing examples

8 years agowe knew we won't explain how to install emacs here (shorten the URL)
Martin Quinson [Sun, 22 Jan 2012 15:18:25 +0000 (16:18 +0100)]
we knew we won't explain how to install emacs here (shorten the URL)

8 years agoSurf: also parallelize update_actions_state() of models
Christophe Thiéry [Fri, 20 Jan 2012 16:15:46 +0000 (17:15 +0100)]
Surf: also parallelize update_actions_state() of models

8 years agoAdd an option --cfg=surf/nthreads:N to update Surf models in parallel
Christophe Thiéry [Fri, 20 Jan 2012 14:39:58 +0000 (15:39 +0100)]
Add an option --cfg=surf/nthreads:N to update Surf models in parallel

8 years agoSurf: allow to call share_resources in parallel (never enabled yet)
Christophe Thiéry [Fri, 20 Jan 2012 13:47:48 +0000 (14:47 +0100)]
Surf: allow to call share_resources in parallel (never enabled yet)

8 years agoParmap: add a value XBT_PARMAP_DEFAULT that decides the synchro mode for you
Christophe Thiéry [Fri, 20 Jan 2012 13:47:19 +0000 (14:47 +0100)]
Parmap: add a value XBT_PARMAP_DEFAULT that decides the synchro mode for you

8 years agoRemove old MSG_set_channel_number from examples.
navarro [Thu, 19 Jan 2012 19:34:23 +0000 (20:34 +0100)]
Remove old MSG_set_channel_number from examples.

8 years agoFix SIMIX_post_io. For now SIMIX_req_file_read makes a Surf sleep action.
Christophe Thiéry [Thu, 19 Jan 2012 16:10:30 +0000 (17:10 +0100)]
Fix SIMIX_post_io. For now SIMIX_req_file_read makes a Surf sleep action.

8 years agoStarting simix mecanism for io.
navarro [Thu, 19 Jan 2012 15:31:41 +0000 (16:31 +0100)]
Starting simix mecanism for io.

8 years agoA for was a bit too much here. It should be readable now.
Christophe Thiéry [Thu, 19 Jan 2012 14:22:31 +0000 (15:22 +0100)]
A for was a bit too much here. It should be readable now.

8 years agoUse %zu to format a sizeof value.
Arnaud Giersch [Wed, 18 Jan 2012 15:47:12 +0000 (16:47 +0100)]
Use %zu to format a sizeof value.

8 years agoDon't turn warnings about clobbered variables into errors on older gcc.
Arnaud Giersch [Wed, 18 Jan 2012 14:21:39 +0000 (15:21 +0100)]
Don't turn warnings about clobbered variables into errors on older gcc.

Rationale: given warnings seem bogus.

8 years agoEnable warnings iff enable_compile_warnings=ON.
Arnaud Giersch [Wed, 18 Jan 2012 14:19:29 +0000 (15:19 +0100)]
Enable warnings iff enable_compile_warnings=ON.

8 years agoInline test_all() in masterslave_mailbox.
Arnaud Giersch [Thu, 19 Jan 2012 09:45:43 +0000 (10:45 +0100)]
Inline test_all() in masterslave_mailbox.

8 years agoAdd mm.c to fragile sources
navarro [Thu, 19 Jan 2012 09:28:11 +0000 (10:28 +0100)]
Add mm.c to fragile sources

8 years agoCompletely remove channel mecanism with tesh tested examples.
navarro [Tue, 17 Jan 2012 16:47:48 +0000 (17:47 +0100)]
Completely remove channel mecanism with tesh tested examples.

8 years agoPut XBT_WARN when try to use channel and not alias system.
navarro [Tue, 17 Jan 2012 16:03:18 +0000 (17:03 +0100)]
Put XBT_WARN when try to use channel and not alias system.
Give the right function with alias.

8 years agoAlso display some information when procname is NULL
navarro [Tue, 17 Jan 2012 14:45:59 +0000 (15:45 +0100)]
Also display some information when procname is NULL

8 years agoRevert "Fix segfault when using XBT_... and when process have no name."
navarro [Tue, 17 Jan 2012 14:08:14 +0000 (15:08 +0100)]
Revert "Fix segfault when using XBT_... and when process have no name."
Use of commit 1eac1757ad2fe043df17ac10262648504c226907 Don't crash with NULL procname.

This reverts commit 9c0bd4e1db324336f4bbbfbe6e5d909a27e54d68.

8 years agoDon't crash with NULL procname.
Arnaud Giersch [Tue, 17 Jan 2012 14:02:05 +0000 (15:02 +0100)]
Don't crash with NULL procname.

8 years agoAFAICS, there is no problem with e.msg.
Arnaud Giersch [Mon, 16 Jan 2012 17:07:29 +0000 (18:07 +0100)]
AFAICS, there is no problem with e.msg.

8 years agoCosmetics: fix indentation.
Arnaud Giersch [Tue, 10 Jan 2012 13:27:58 +0000 (14:27 +0100)]
Cosmetics: fix indentation.

8 years agoFix segfault when using XBT_... and when process have no name.
navarro [Tue, 17 Jan 2012 14:01:45 +0000 (15:01 +0100)]
Fix segfault when using XBT_... and when process have no name.

8 years agoImplement SIMIX_process_from_PID
navarro [Tue, 17 Jan 2012 13:47:41 +0000 (14:47 +0100)]
Implement SIMIX_process_from_PID

8 years agoFactorize functions for smpi factor.
navarro [Tue, 17 Jan 2012 12:51:10 +0000 (13:51 +0100)]
Factorize functions for smpi factor.

8 years agoFix some memory leak.
navarro [Tue, 17 Jan 2012 11:31:24 +0000 (12:31 +0100)]
Fix some memory leak.

8 years agoUpdate changelog for smpi factor.
navarro [Tue, 17 Jan 2012 09:54:20 +0000 (10:54 +0100)]
Update changelog for smpi factor.

8 years agoAdd a platform with tag config and explain how to change smpi factor.
navarro [Mon, 16 Jan 2012 17:56:24 +0000 (18:56 +0100)]
Add a platform with tag config and explain how to change smpi factor.

8 years agoOffer the possibility to change smpi bandwidth and latency factor into tag config...
navarro [Mon, 16 Jan 2012 17:42:55 +0000 (18:42 +0100)]
Offer the possibility to change smpi bandwidth and latency factor into tag config or by command line.
Use --cfg=smpi/bw_factor="thresholh:return_value;threshold:return_value;..."
or --cfg=smpi/lat_factor="thresholh:return_value;threshold:return_value;...

8 years agoPut xbt_cfg_register for smpi into surf_config.c file
navarro [Mon, 16 Jan 2012 16:37:09 +0000 (17:37 +0100)]
Put xbt_cfg_register for smpi into surf_config.c file

8 years agoMerge branch 'master' of scm.gforge.inria.fr:/gitroot/simgrid/simgrid
Martin Quinson [Mon, 16 Jan 2012 15:32:50 +0000 (16:32 +0100)]
Merge branch 'master' of scm.gforge.inria.fr:/gitroot/simgrid/simgrid

8 years agodocument what I've done, and note that a written item is not finished yet
Martin Quinson [Mon, 16 Jan 2012 15:32:31 +0000 (16:32 +0100)]
document what I've done, and note that a written item is not finished yet

8 years agoLua chord: simplify the code by reusing tasks since it now works
Christophe Thiéry [Mon, 16 Jan 2012 11:18:45 +0000 (12:18 +0100)]
Lua chord: simplify the code by reusing tasks since it now works

8 years agoAdd a test for Chord in Lua
Christophe Thiéry [Mon, 16 Jan 2012 11:07:10 +0000 (12:07 +0100)]
Add a test for Chord in Lua

8 years agoMSG_task_dsend: don't apply a default function if cleanup is NULL
Christophe Thiéry [Mon, 16 Jan 2012 10:47:16 +0000 (11:47 +0100)]
MSG_task_dsend: don't apply a default function if cleanup is NULL

NULL now means that no cleanup function will be called by SIMIX.
This is useful if the task is garbage collected (as in Lua).

8 years agoLua: close the Lua world of a process being destroyed
Christophe Thiéry [Fri, 13 Jan 2012 16:04:03 +0000 (17:04 +0100)]
Lua: close the Lua world of a process being destroyed

Also make sure MSG_clean is called only once (by maestro).

8 years agoFix a compilation error
Christophe Thiéry [Fri, 13 Jan 2012 15:59:02 +0000 (16:59 +0100)]
Fix a compilation error

8 years agoLua: forgot the timeout argument of simgrid.comm.send
Christophe Thiéry [Fri, 13 Jan 2012 15:22:52 +0000 (16:22 +0100)]
Lua: forgot the timeout argument of simgrid.comm.send