Martin Quinson [Wed, 13 Jan 2016 22:24:18 +0000 (23:24 +0100)]
port NS3 to simgrid::s4u::Host
Gabriel Corona [Wed, 13 Jan 2016 13:43:24 +0000 (14:43 +0100)]
Merge simgrid::Host into simgrid::s4u::Host
Martin Quinson [Wed, 13 Jan 2016 14:31:51 +0000 (15:31 +0100)]
document the update mechanisms
Martin Quinson [Wed, 13 Jan 2016 14:28:05 +0000 (15:28 +0100)]
more documentation in surf::Routing
Martin Quinson [Wed, 13 Jan 2016 13:44:37 +0000 (14:44 +0100)]
port NS3 to the new signals in simgrid::Host
Gabriel Corona [Wed, 13 Jan 2016 11:16:04 +0000 (12:16 +0100)]
Move the signals from simgrid::surf::Host to simgrid::Host
Fix issues of dependencies between extensions in destruction.
Gabriel Corona [Wed, 13 Jan 2016 10:24:21 +0000 (11:24 +0100)]
Public simgrid::xbt:signal<F> class
It is either an alias for boost::signals2::signal<F> or a wrapper
around sigc::signal<R, Args...>. We are going to need it in order to
exposer the signals to users and move them in simgrid::Host.
Martin Quinson [Wed, 13 Jan 2016 10:11:54 +0000 (11:11 +0100)]
[appveyor] disable the tests that are broken by the perl version to get the jarfile
Martin Quinson [Wed, 13 Jan 2016 08:18:27 +0000 (09:18 +0100)]
[cmake] cosmetics: mark 2 variables as advanced
Martin Quinson [Tue, 12 Jan 2016 16:14:08 +0000 (17:14 +0100)]
[cmake/java] copy all native libs in one shot to fix parallel build
Gabriel Corona [Tue, 12 Jan 2016 14:43:08 +0000 (15:43 +0100)]
Remove surf_host_model_execute_parallel_task()
Gabriel Corona [Tue, 12 Jan 2016 13:46:22 +0000 (14:46 +0100)]
Remove surf_host_* functions
They are just C wrappers around the C++ API.
Gabriel Corona [Tue, 12 Jan 2016 13:38:50 +0000 (14:38 +0100)]
Fix stupidly broken indexing in SIMIX_process_parallel_execute()
Martin Quinson [Tue, 12 Jan 2016 13:39:35 +0000 (14:39 +0100)]
re-move content into separate callback functions (and empty the destructor)
Gabriel Corona [Tue, 12 Jan 2016 12:58:41 +0000 (13:58 +0100)]
Add missing XBT_ALWAYS_INLINE
"Yes, you can be XBT_ALWAYS_INLINE too."
Gabriel Corona [Tue, 12 Jan 2016 12:58:16 +0000 (13:58 +0100)]
Fix aliased template parameter name
This is purely stylistic.
Gabriel Corona [Tue, 12 Jan 2016 11:24:12 +0000 (12:24 +0100)]
Revert "remove a C-like pimple"
This reverts commit
71c619779ed661b50ce589403a0f4df7ab0865de.
This commits generates a lot of segfaults.
Martin Quinson [Mon, 11 Jan 2016 23:44:21 +0000 (00:44 +0100)]
sg_functions should be wrapping the C++ public API, not internals
Martin Quinson [Mon, 11 Jan 2016 23:53:14 +0000 (00:53 +0100)]
Make sg_* functions callable from simdag
There was an issue when they were doing a simcall, but now, every
simdag simulation is supposed to run in maestro context.
Martin Quinson [Mon, 11 Jan 2016 22:14:25 +0000 (23:14 +0100)]
make MC compile with -Wmissing-declaration
Martin Quinson [Mon, 11 Jan 2016 21:49:42 +0000 (22:49 +0100)]
kill an unused static function
Martin Quinson [Mon, 11 Jan 2016 21:37:37 +0000 (22:37 +0100)]
Accept the fact that NS3 timings may vary with the version
Don't take the exact timing into account in the tesh file.
Martin Quinson [Mon, 11 Jan 2016 21:37:10 +0000 (22:37 +0100)]
be more robust to strange failures when handling exceptions
Martin Quinson [Mon, 11 Jan 2016 21:08:35 +0000 (22:08 +0100)]
surf::Host don't need no netcard
Martin Quinson [Mon, 11 Jan 2016 21:06:21 +0000 (22:06 +0100)]
kill unused sg_host_msg_destroy()
Martin Quinson [Mon, 11 Jan 2016 16:48:04 +0000 (17:48 +0100)]
[energy] sanitize what's a simcall and what's not; inline some calls
Gabriel Corona [Mon, 11 Jan 2016 16:23:44 +0000 (17:23 +0100)]
[simix] Fix unitialized thread_ in ThreadContext
Gabriel Corona [Mon, 11 Jan 2016 15:15:20 +0000 (16:15 +0100)]
Rewrite the energy plugin using the extension mechanism
Gabriel Corona [Mon, 11 Jan 2016 12:52:44 +0000 (13:52 +0100)]
[energy] C++-ification, remove nested xbt_dynar
Gabriel Corona [Mon, 11 Jan 2016 12:15:17 +0000 (13:15 +0100)]
Add missing energy.h
Gabriel Corona [Mon, 11 Jan 2016 11:55:40 +0000 (12:55 +0100)]
Fix an #include
Gabriel Corona [Mon, 11 Jan 2016 10:56:07 +0000 (11:56 +0100)]
Remove SIMIX wrappers
Gabriel Corona [Mon, 11 Jan 2016 10:52:04 +0000 (11:52 +0100)]
Remove SIMIX_host_get_current_power_peak() and SIMIX_host_get_power_peak_at()
They are mere wrappers around the same surf_ calls.
Gabriel Corona [Mon, 11 Jan 2016 10:40:58 +0000 (11:40 +0100)]
Remove SIMIX_host_get_properties() and SIMIX_host_get_process_list()
Martin Quinson [Mon, 11 Jan 2016 10:43:06 +0000 (11:43 +0100)]
remove completely the energy concern from the core
- It is moved to the plugin only.
- In particular, there is nothing in the core headers anymore
- SMPI still defines some energy methods
- the command-line config still needs to know the init method
Also, I'm sorry, but I renamed simgrid::Host->(p_cpu/p_netcard) to
pimpl_cpu/pimpl_netcard. I should have separated the rename, sorry.
Martin Quinson [Sun, 10 Jan 2016 15:06:40 +0000 (16:06 +0100)]
[appveyor] Perl 5.20 fails also. Try v5.18
Martin Quinson [Sun, 10 Jan 2016 10:02:51 +0000 (11:02 +0100)]
[appveyor] try to alternate our Perl provider
I hope that it will workaround https://github.com/toddr/IPC-Run/issues/16
Martin Quinson [Sun, 10 Jan 2016 09:42:16 +0000 (10:42 +0100)]
Do the Right Thing for the host_(get/set)_pstate
- The prefered public interface is in simgrid::Host
- the getter directly calls pimpl_cpu->getPState()
- the setter simcalls to the pimpl, without diverting to simix
- (this will be automatically forwarded to Java with SWIG one day)
- The C public interface is sg_host_(get/set)_pstate
- MSG_host_(get/set)_pstate is #defined to this for compatibility
- SMPI adapts the interface in smpi_host_(get/set)_pstate to make
the receiver host implicit.
I don't see the point but don't want to mess with SMPI yet.
One day, the whole of SimGrid will be organized this way. One day.
Martin Quinson [Sun, 10 Jan 2016 00:32:20 +0000 (01:32 +0100)]
mark simix::Context as public to please windows
Martin Quinson [Sun, 10 Jan 2016 00:06:55 +0000 (01:06 +0100)]
kill two more useless calls
Martin Quinson [Sun, 10 Jan 2016 00:05:13 +0000 (01:05 +0100)]
kill dead code
Martin Quinson [Sun, 10 Jan 2016 00:01:35 +0000 (01:01 +0100)]
[cmake/java] only recompile the javadoc if the documentation is enabled
This prevents the jarfile to be rebuilt each time you hit the "build"
button, dramatically speeding things up <3
Martin Quinson [Sat, 9 Jan 2016 23:54:40 +0000 (00:54 +0100)]
make it easier to mark execution paths as DEADCODE, and mark one of them :)
Martin Quinson [Sat, 9 Jan 2016 23:35:51 +0000 (00:35 +0100)]
C++-ifies a wrapper function, and kill it
This path will be long :(
Martin Quinson [Sat, 9 Jan 2016 23:26:22 +0000 (00:26 +0100)]
bypass simix in simcall to surf::Cpu::setPState
Martin Quinson [Sat, 9 Jan 2016 23:12:54 +0000 (00:12 +0100)]
kill 2 useless calls
Martin Quinson [Sat, 9 Jan 2016 22:58:09 +0000 (23:58 +0100)]
compile with -Wmissing-declarations to catch more errors
Martin Quinson [Sat, 9 Jan 2016 22:19:46 +0000 (23:19 +0100)]
Cosmetics: rename host::on to host::turnOn
Martin Quinson [Sat, 9 Jan 2016 22:08:20 +0000 (23:08 +0100)]
kill useless code
Martin Quinson [Fri, 8 Jan 2016 21:12:51 +0000 (22:12 +0100)]
don't free VM's netcard: is not a wrapper anymore but the PM one
Martin Quinson [Fri, 8 Jan 2016 20:37:59 +0000 (21:37 +0100)]
kill a now unused function
Gabriel Corona [Fri, 8 Jan 2016 12:45:36 +0000 (13:45 +0100)]
Bypass the C binding layer (surf_action and surf_host)
Gabriel Corona [Tue, 5 Jan 2016 15:08:52 +0000 (16:08 +0100)]
[simix] Call directly the C++ SURF API instead of calling the C bindings
Gabriel Corona [Thu, 7 Jan 2016 14:47:30 +0000 (15:47 +0100)]
Make XBT_TEST_UNIT() work in C++
testall.c expects the unit test functions (XBT_TEST_UNIT()) to use the
C linkage "C" so we have to define them so. Before this commit, they
had to be manually placed in a extern "C" {} block.
Martin Quinson [Fri, 8 Jan 2016 00:21:24 +0000 (01:21 +0100)]
woopsie, sorry
Martin Quinson [Thu, 7 Jan 2016 23:23:25 +0000 (00:23 +0100)]
RoutingEdgeWrapper is probably the dumbest wrapper I've ever seen. Kill it
Martin Quinson [Thu, 7 Jan 2016 23:17:12 +0000 (00:17 +0100)]
[jenkins] that was a bad idea
Martin Quinson [Thu, 7 Jan 2016 22:56:49 +0000 (23:56 +0100)]
[jenkins] don't be too demanding for osx/freebsd slaves
Martin Quinson [Thu, 7 Jan 2016 22:22:45 +0000 (23:22 +0100)]
[jenkins] typo
Martin Quinson [Thu, 7 Jan 2016 22:06:53 +0000 (23:06 +0100)]
Merge branch 'master' of github.com:mquinson/simgrid
Martin Quinson [Thu, 7 Jan 2016 22:06:01 +0000 (23:06 +0100)]
[jenkins] remove the right files from /tmp
Martin Quinson [Thu, 7 Jan 2016 22:04:08 +0000 (23:04 +0100)]
[jenkins] also cleanup the files forgotten in the tree by java errors
Martin Quinson [Thu, 7 Jan 2016 21:57:10 +0000 (22:57 +0100)]
[jenkins] preliminary checks to save time when we are building a new slave
Martin Quinson [Thu, 7 Jan 2016 21:55:31 +0000 (22:55 +0100)]
[jenkins] more info on the available disk space
Martin Quinson [Thu, 7 Jan 2016 21:52:53 +0000 (22:52 +0100)]
Merge pull request #45 from mpoquet/master
Patch candidate for issue 15
Millian Poquet [Thu, 7 Jan 2016 21:24:05 +0000 (22:24 +0100)]
Warning removal: switched the next_pid variable in MSG_process_create_with_environment from int to unsigned int
Martin Quinson [Thu, 7 Jan 2016 20:29:28 +0000 (21:29 +0100)]
better leaking than segfaulting
Millian Poquet [Thu, 7 Jan 2016 20:02:46 +0000 (21:02 +0100)]
Removal of TRACE_msg_process_end
Martin Quinson [Thu, 7 Jan 2016 20:02:08 +0000 (21:02 +0100)]
untangle resource::state (ON/OFF) and VM::state (STARTED/MIGRATING/etc)
Millian Poquet [Thu, 7 Jan 2016 18:46:16 +0000 (19:46 +0100)]
Fix TRACE_msg_process_destroy: PJ_container_get_or_null is used instead of PJ_container_get to avoid crash on double destruction of a TRACE process (or just if the corresponding process was not created). Header updated since the host was not used in this function. The function TRACE_msg_process_end now calls TRACE_msg_process_destroy.
Millian Poquet [Thu, 7 Jan 2016 18:14:08 +0000 (19:14 +0100)]
In MSG_process_create_with_environment, inversion of the SIMIX process creation and the TRACE process creation
Millian Poquet [Thu, 7 Jan 2016 17:47:18 +0000 (18:47 +0100)]
Revert "A candidate patch to Github issue #15. Three problems related to TRACE might be corrected by this patch."
This reverts commit
58a9c67822f4640964044ec485426d090bcb123f.
Martin Quinson [Thu, 7 Jan 2016 13:52:21 +0000 (14:52 +0100)]
give simgrid::Host a getState() method, and use it
Martin Quinson [Thu, 7 Jan 2016 10:04:50 +0000 (11:04 +0100)]
shallow cleanups in net_cste: overide markers and kill useless code
Martin Quinson [Thu, 7 Jan 2016 10:02:01 +0000 (11:02 +0100)]
don't commit my personal eclipse settings
Martin Quinson [Wed, 6 Jan 2016 21:38:58 +0000 (22:38 +0100)]
kill methods overriding super to the same content
Martin Quinson [Wed, 6 Jan 2016 21:32:35 +0000 (22:32 +0100)]
arguable cosmetics
Martin Quinson [Wed, 6 Jan 2016 20:58:09 +0000 (21:58 +0100)]
kill an eclipse warning about unused static functions
Martin Quinson [Wed, 6 Jan 2016 20:51:46 +0000 (21:51 +0100)]
kill a useless obscure function
Martin Quinson [Wed, 6 Jan 2016 20:46:49 +0000 (21:46 +0100)]
cosmetics in comments and error messages
Martin Quinson [Wed, 6 Jan 2016 20:45:13 +0000 (21:45 +0100)]
remove a C-like pimple
Martin Quinson [Wed, 6 Jan 2016 20:37:45 +0000 (21:37 +0100)]
plug a memleak
Martin Quinson [Wed, 6 Jan 2016 17:47:23 +0000 (18:47 +0100)]
strange, tesh ignores these extra arguments
Martin Quinson [Wed, 6 Jan 2016 17:15:44 +0000 (18:15 +0100)]
speed up the tests a bit
Martin Quinson [Wed, 6 Jan 2016 17:14:31 +0000 (18:14 +0100)]
fix a compilation error when NS3 is activated
degomme [Wed, 6 Jan 2016 10:31:35 +0000 (11:31 +0100)]
only cleanup previous build when starting a new one. Jenkins needs some data in these folders to post-process results.
Martin Quinson [Wed, 6 Jan 2016 09:10:41 +0000 (10:10 +0100)]
NetworkCM02: various cosmetics
Martin Quinson [Wed, 6 Jan 2016 08:43:30 +0000 (09:43 +0100)]
kill an unused function prototype
Martin Quinson [Tue, 5 Jan 2016 21:22:40 +0000 (22:22 +0100)]
make it compile with -fpermissive even on travis
Martin Quinson [Tue, 5 Jan 2016 21:01:43 +0000 (22:01 +0100)]
[energy] react sainly when the host speed is 0
(fix #44)
Martin Quinson [Tue, 5 Jan 2016 20:23:16 +0000 (21:23 +0100)]
kill deprecated code
Martin Quinson [Tue, 5 Jan 2016 20:22:49 +0000 (21:22 +0100)]
Adapt lua to the latest code changes
Gabriel Corona [Tue, 5 Jan 2016 13:21:18 +0000 (14:21 +0100)]
[simix] Move other simcall to the Host class
Gabriel Corona [Tue, 5 Jan 2016 13:04:00 +0000 (14:04 +0100)]
[simix] Make host->{on,off}() do the simcall directly
Gabriel Corona [Tue, 5 Jan 2016 09:25:56 +0000 (10:25 +0100)]
[simix] Comments and fast path for simgrid::simix::run_kernel
degomme [Tue, 5 Jan 2016 14:27:58 +0000 (15:27 +0100)]
add some missing prototypes
degomme [Tue, 5 Jan 2016 14:11:15 +0000 (15:11 +0100)]
-Wno-unused-local-typedefs was only added post clang 3.3, so it breaks our freebsd builder. Make this flag gcc only
degomme [Tue, 5 Jan 2016 13:57:19 +0000 (14:57 +0100)]
avoid some boost warning/errors on windows
degomme [Tue, 5 Jan 2016 13:55:00 +0000 (14:55 +0100)]
weirdly, DEPENDS clause seems to be ignored sometimes when using make -j1 (as in ci on some nodes)
Gabriel Corona [Mon, 4 Jan 2016 12:07:44 +0000 (13:07 +0100)]
[simix] Remove lots of simcalls, use run_kernel instead
We should be able to most of the "Func" and "Proc" simcalls without
handler. Most getters can be done directly in the application process
without kernel synchronization.