Arnaud Giersch [Thu, 10 Jan 2019 10:33:09 +0000 (11:33 +0100)]
Cosmetics: trailing withespaces.
Arnaud Giersch [Thu, 10 Jan 2019 10:31:30 +0000 (11:31 +0100)]
Avoid redefining built-in "list".
Arnaud Giersch [Thu, 10 Jan 2019 10:29:59 +0000 (11:29 +0100)]
Change string statement into comment.
Arnaud Giersch [Thu, 10 Jan 2019 10:20:21 +0000 (11:20 +0100)]
Now that it run on jenkins, disable sonar analysis on travis.
Arnaud Giersch [Thu, 10 Jan 2019 09:15:25 +0000 (10:15 +0100)]
[sonar] Merge exclusions.
Arnaud Giersch [Thu, 10 Jan 2019 08:59:12 +0000 (09:59 +0100)]
[sonar] Rename method (don't use "main").
Arnaud Giersch [Wed, 9 Jan 2019 20:34:29 +0000 (21:34 +0100)]
[sonar] Misc issues in bindings/lua.
Augustin Degomme [Thu, 10 Jan 2019 11:09:45 +0000 (12:09 +0100)]
bad formatting of suppression file, missing a mandatory comment line
Augustin Degomme [Thu, 10 Jan 2019 09:43:07 +0000 (10:43 +0100)]
tell valgrind to ignore cruft inside python3.. Let it for now try to find the remaining issues in simgrid
Augustin Degomme [Thu, 10 Jan 2019 09:31:02 +0000 (10:31 +0100)]
disable python for sanitizers
Augustin Degomme [Wed, 9 Jan 2019 14:43:55 +0000 (15:43 +0100)]
adopt single file for cobertura as well
Augustin Degomme [Wed, 9 Jan 2019 14:41:11 +0000 (15:41 +0100)]
typo --
Augustin Degomme [Wed, 9 Jan 2019 14:40:39 +0000 (15:40 +0100)]
generate only one file for all python reports, as sonarcloud wants it.
Arnaud Giersch [Wed, 9 Jan 2019 11:57:49 +0000 (12:57 +0100)]
Redo "Like java, python breaks mmalloc with FreeBSD+MC. Disable it."
Check option 'enable_python' *after* its declaration.
This fixes commit
882355d6bdf10129df36e89e4a0a5182959600ff.
Augustin Degomme [Wed, 9 Jan 2019 10:48:30 +0000 (11:48 +0100)]
add python coverage
May crash with some examples, such as actor-create..
Arnaud Giersch [Wed, 9 Jan 2019 09:17:52 +0000 (10:17 +0100)]
Field SwappedContext::stack_ became private, use get_stack().
Arnaud Giersch [Wed, 9 Jan 2019 08:51:55 +0000 (09:51 +0100)]
Like java, python breaks mmalloc with FreeBSD+MC. Disable it.
Arnaud Giersch [Wed, 9 Jan 2019 08:51:55 +0000 (09:51 +0100)]
[sonar] Fix never executed "break" statements.
Arnaud Giersch [Wed, 9 Jan 2019 08:51:55 +0000 (09:51 +0100)]
[sonar] Use the "&" operator with function names.
Arnaud Giersch [Wed, 9 Jan 2019 08:51:55 +0000 (09:51 +0100)]
Happy new year.
Martin Quinson [Wed, 9 Jan 2019 09:39:28 +0000 (10:39 +0100)]
more pythonic example
Martin Quinson [Wed, 9 Jan 2019 07:36:11 +0000 (08:36 +0100)]
sonar: don't exclude java from coverage
Martin Quinson [Wed, 9 Jan 2019 07:21:03 +0000 (08:21 +0100)]
sonar: disable a weird rule
Martin Quinson [Wed, 9 Jan 2019 07:18:16 +0000 (08:18 +0100)]
Fix MC builds
Martin Quinson [Tue, 8 Jan 2019 22:32:19 +0000 (23:32 +0100)]
travis is on Xenial now
Martin Quinson [Tue, 8 Jan 2019 22:25:58 +0000 (23:25 +0100)]
Missing files
Martin Quinson [Tue, 8 Jan 2019 22:09:57 +0000 (23:09 +0100)]
doc cosmetics, and cleanup the API
more functions in the API, don't help users but mess them up
- In python:
- remove this_actor.kill(pid)
- bind Actor.by_pid(pid)
- remove this_actor.kill_all(), we have Actor.kill_all()
- In C++:
- deprecate s4u::Actor::kill(pid)
- Various docstrings improvements
Martin Quinson [Tue, 8 Jan 2019 21:21:49 +0000 (22:21 +0100)]
sonar: ignore XML and python files from coverage
Martin Quinson [Tue, 8 Jan 2019 18:30:11 +0000 (19:30 +0100)]
new python example: actor-kill
Martin Quinson [Tue, 8 Jan 2019 17:13:55 +0000 (18:13 +0100)]
obey sonar and make SwappedContext::stack_ private
Martin Quinson [Tue, 8 Jan 2019 14:30:40 +0000 (15:30 +0100)]
Don't force on a factory that is not always avail
Martin Quinson [Tue, 8 Jan 2019 13:30:51 +0000 (14:30 +0100)]
tiny doc improvement
Augustin Degomme [Tue, 8 Jan 2019 16:23:49 +0000 (17:23 +0100)]
Don't get tied to a version, this one needed an upgrade
Augustin Degomme [Tue, 8 Jan 2019 15:48:46 +0000 (16:48 +0100)]
protect against entering the loop when no match is found
Augustin Degomme [Tue, 8 Jan 2019 14:02:04 +0000 (15:02 +0100)]
add sonarqube to the coverage build
Augustin Degomme [Tue, 8 Jan 2019 14:01:23 +0000 (15:01 +0100)]
don't wrash with 0 sized files
jacoco is failing as of now: probably we don't close the jvm properly, and the filling of the file is never triggered
Martin Quinson [Tue, 8 Jan 2019 07:51:01 +0000 (08:51 +0100)]
s/HAVE_SENDFILE/SG_HAVE_SENDFILE/
Now that the python-bindings load the private headers too (to get
Context::StopRequest() in Context.hpp), we get some naming conflict.
Interestingly, that's only visible on some architectures, not globally.
Martin Quinson [Tue, 8 Jan 2019 07:44:41 +0000 (08:44 +0100)]
move header template near to its destination
Martin Quinson [Tue, 8 Jan 2019 07:34:12 +0000 (08:34 +0100)]
ci: make target got renamed
Martin Quinson [Tue, 8 Jan 2019 07:21:45 +0000 (08:21 +0100)]
tiny doc improvement
Martin Quinson [Tue, 8 Jan 2019 07:20:12 +0000 (08:20 +0100)]
include pybind11.h first, as we should
Martin Quinson [Tue, 8 Jan 2019 07:16:56 +0000 (08:16 +0100)]
remove any virtuals from Context::StopRequest
Some clang versions fail otherwise:
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/8.2.1/../../../../include/c++/8.2.1/bits/exception_ptr.h:172:9: error: destructor called on non-final 'simgrid::kernel::context::Context::StopRequest' that has virtual functions but non-virtual destructor [-Werror,-Wdelete-non-virtual-dtor]
{ static_cast<_Ex*>(__x)->~_Ex(); }
^
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/8.2.1/../../../../include/c++/8.2.1/bits/exception_ptr.h:185:21: note: in instantiation of function template specialization 'std::__exception_ptr::__dest_thunk<simgrid::kernel::context::Context::StopRequest>' requested here
__exception_ptr::__dest_thunk<_Ex>);
^
/builds/workspace/SimGrid/build_mode/Debug/node/simgrid-manjaro/build/SimGrid-3.21.90/src/simix/ActorImpl.cpp:507:35: note: in instantiation of function template specialization 'std::make_exception_ptr<simgrid::kernel::context::Context::StopRequest>' requested here
actor->throw_exception(std::make_exception_ptr(simgrid::kernel::context::Context::StopRequest("host failed")));
^
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/8.2.1/../../../../include/c++/8.2.1/bits/exception_ptr.h:172:34: note: qualify call to silence this warning
{ static_cast<_Ex*>(__x)->~_Ex(); }
^
1 error generated.
Martin Quinson [Tue, 8 Jan 2019 06:06:03 +0000 (07:06 +0100)]
autopep8 our python examples
Martin Quinson [Tue, 8 Jan 2019 05:41:48 +0000 (06:41 +0100)]
Merge branch 'master' of framagit.org:simgrid/simgrid
Martin Quinson [Tue, 8 Jan 2019 05:40:32 +0000 (06:40 +0100)]
Also test is_suspended in the s4u example
Martin Quinson [Tue, 8 Jan 2019 05:37:28 +0000 (06:37 +0100)]
python: test Actor.is_suspended() and introduce this_actor.error
Martin Quinson [Tue, 8 Jan 2019 05:32:35 +0000 (06:32 +0100)]
Various doc improvements
Also kill meaningless functions:
- Actor::yield() was not even implemented, only declared
- this_actor::is_suspended() can only return false, because suspended
actors don't get executed.
Martin Quinson [Tue, 8 Jan 2019 05:00:45 +0000 (06:00 +0100)]
rename make target to ease tab-completion
Martin Quinson [Tue, 8 Jan 2019 04:15:03 +0000 (05:15 +0100)]
Also make registered actors killable in python
Martin Quinson [Tue, 8 Jan 2019 04:04:50 +0000 (05:04 +0100)]
only use pybind11 when registering actors
Mixing with the underlying C API only makes things harder
Martin Quinson [Tue, 8 Jan 2019 03:57:18 +0000 (04:57 +0100)]
New python example: actor-daemon
Martin Quinson [Tue, 8 Jan 2019 03:43:16 +0000 (04:43 +0100)]
allow python actors to be killed
Only works for directly created actors for now, not for the registered
ones. I need to do the same kind of things for register_actors, now.
The funny thing is that I didn't had to clean the contexts for that.
Arnaud Giersch [Mon, 7 Jan 2019 16:59:26 +0000 (17:59 +0100)]
Don't turn gcc's -Wmaybe-uninitialized into an error.
It is sometimes bogus. For example:
src/include/xbt/parmap.hpp:225:19: error: may be used uninitialized in this function [-Werror=maybe-uninitialized]
return boost::none;
^
Arnaud Giersch [Mon, 7 Jan 2019 15:50:40 +0000 (16:50 +0100)]
Fix build with ASan (s/from/this/).
Arnaud Giersch [Mon, 7 Jan 2019 14:17:16 +0000 (15:17 +0100)]
Thread factory is always valid.
Arnaud Giersch [Mon, 7 Jan 2019 09:44:56 +0000 (10:44 +0100)]
Anonymize unused variable.
Arnaud Giersch [Mon, 7 Jan 2019 09:36:53 +0000 (10:36 +0100)]
Remove trailing whitespaces.
Arnaud Giersch [Mon, 7 Jan 2019 09:28:23 +0000 (10:28 +0100)]
Update copyright lines with new year.
Martin Quinson [Mon, 7 Jan 2019 11:23:16 +0000 (12:23 +0100)]
do not break sequential execution while fixing parallel execution
Martin Quinson [Mon, 7 Jan 2019 10:34:57 +0000 (11:34 +0100)]
Fix a race condition
When saving the context of maestro, we must ensure that we are
actually thread 0. Otherwise, we may end up saving the context of a
minion (a working thread).
Martin Quinson [Mon, 7 Jan 2019 09:01:39 +0000 (10:01 +0100)]
doc: fix copy/paste error
Martin Quinson [Sun, 6 Jan 2019 16:33:23 +0000 (17:33 +0100)]
please sonar, add 'explicit'
Martin Quinson [Sun, 6 Jan 2019 14:08:10 +0000 (15:08 +0100)]
factorize where seq_runall and par_runall save maestro context
Martin Quinson [Sun, 6 Jan 2019 13:21:53 +0000 (14:21 +0100)]
change static fields in SwappedCtx into regular fields of the factory
and other cleanups
Martin Quinson [Sun, 6 Jan 2019 12:05:26 +0000 (13:05 +0100)]
Kill the now useless ParallelRawContext and ParallelBoostContext
They were a perfect duplication of ParallelUContext :(
Martin Quinson [Sun, 6 Jan 2019 11:58:03 +0000 (12:58 +0100)]
merge ParallelUContext into SwappedContext, that can now run parallel or sequential
Martin Quinson [Sun, 6 Jan 2019 11:09:02 +0000 (12:09 +0100)]
Cosmetics in contexts
- Kill useless code
- Move a method to superclass
- Fix clang builds
Martin Quinson [Sun, 6 Jan 2019 11:06:03 +0000 (12:06 +0100)]
move a static field of SwappedContext into SwappedContextFactory
Martin Quinson [Sun, 6 Jan 2019 10:54:40 +0000 (11:54 +0100)]
Introduce a SwappedContextFactory to further reduce code duplication, working WIP
The code seems to work (I've run many tests locally), but it's far
from being clean yet. See this as a checkpoint commit.
Martin Quinson [Sat, 5 Jan 2019 22:28:31 +0000 (23:28 +0100)]
context: rewrite some comments, and useless cosmetics
Martin Quinson [Sat, 5 Jan 2019 22:12:40 +0000 (23:12 +0100)]
Move Parallel{Boost,Raw,U}Context::{initialize,finalize} to SwappedContext
Martin Quinson [Sat, 5 Jan 2019 21:43:11 +0000 (22:43 +0100)]
further cleanups in contexts
- Kill SerialRawContext and use SwappedContext
- Centralize everything related to the stack pointer in SwappedContext
Previously, alloc/dealloc was done in functions in Context.cpp while
the fields were duplicated in all SwappedContext childs
Martin Quinson [Sat, 5 Jan 2019 19:02:13 +0000 (20:02 +0100)]
Change a subclass into a superclass around contexts
SerialUContext was a subclass of UContext. SwappedContext does the
same, as a super class of UContext.
The idea is to make SwappedContext the ancestor of BoostContext and
RawContext too, to factorize as much code as possible.
Martin Quinson [Sat, 5 Jan 2019 16:10:34 +0000 (17:10 +0100)]
make SIMIX_context_self() useless and use Context::self() instead
Martin Quinson [Sat, 5 Jan 2019 14:33:22 +0000 (15:33 +0100)]
move ContextFactory::self() into Context::self() and tidy it up
Martin Quinson [Sat, 5 Jan 2019 02:29:33 +0000 (03:29 +0100)]
Please sonar
- Add 'explicit' where needed
- Don't specify the return type of lambdas
- Verify that this != that before everything in the operator=
- Remove ';' in the comments to not fool the commented code detection mechanism
Martin Quinson [Sat, 5 Jan 2019 02:18:25 +0000 (03:18 +0100)]
reimplement this test in cpp to use standard threads
Martin Quinson [Sat, 5 Jan 2019 02:04:53 +0000 (03:04 +0100)]
make this test more robust to cruft that may remain on disk
Martin Quinson [Sat, 5 Jan 2019 01:28:55 +0000 (02:28 +0100)]
make xbt_os_thread_self() useless, and kill it
Also remove every parts of xbt_os_thread_t that I managed to make useless
Martin Quinson [Sat, 5 Jan 2019 01:21:19 +0000 (02:21 +0100)]
kill the unused xbt_os_thread_self_name()
Martin Quinson [Sat, 5 Jan 2019 01:11:22 +0000 (02:11 +0100)]
kill the now useless xbt_os_thread_{get,set}_extra_data()
Martin Quinson [Sat, 5 Jan 2019 01:02:31 +0000 (02:02 +0100)]
unify how threaded and parallelisable context factories find context_self
Martin Quinson [Sat, 5 Jan 2019 00:20:57 +0000 (01:20 +0100)]
Inline SIMIX_context_get_current()
Martin Quinson [Sat, 5 Jan 2019 00:08:40 +0000 (01:08 +0100)]
another small simplification now that threads are usable everywhere
Martin Quinson [Fri, 4 Jan 2019 23:45:27 +0000 (00:45 +0100)]
Further code simplifications now that the Thread context backend is always available
Martin Quinson [Fri, 4 Jan 2019 23:18:36 +0000 (00:18 +0100)]
remove unecessary imports to please sonar
Martin Quinson [Fri, 4 Jan 2019 23:09:51 +0000 (00:09 +0100)]
Fix make distcheck
Martin Quinson [Fri, 4 Jan 2019 23:06:02 +0000 (00:06 +0100)]
kill xbt_os_sem_t and the required cmake-detection cruft
Martin Quinson [Fri, 4 Jan 2019 22:50:59 +0000 (23:50 +0100)]
Introduce a portable xbt::OsSemaphore using C++11 threads
Martin Quinson [Fri, 4 Jan 2019 21:41:00 +0000 (22:41 +0100)]
tiny simplification: xbt_os_thread_bind is a noop when not HAVE_PTHREAD_SETAFFINITY
Martin Quinson [Fri, 4 Jan 2019 21:12:45 +0000 (22:12 +0100)]
track some remaining CamelCase deep inside the context code
Martin Quinson [Fri, 4 Jan 2019 21:05:49 +0000 (22:05 +0100)]
some useless s/process/actor/ in the source code
Martin Quinson [Fri, 4 Jan 2019 21:03:18 +0000 (22:03 +0100)]
sonar: ignore another rule as we don't have real security concerns
Martin Quinson [Fri, 4 Jan 2019 16:43:22 +0000 (17:43 +0100)]
use strncpy instead of strcpy to please sonar
Martin Quinson [Fri, 4 Jan 2019 13:51:45 +0000 (14:51 +0100)]
travis: don't run the tests before the sonar analysis
We exeed the 40mn timeframe on travis with those tests.
Too bad, we gonna lose the coverage information...
Martin Quinson [Fri, 4 Jan 2019 13:43:57 +0000 (14:43 +0100)]
py: try to fix MC builds
Martin Quinson [Fri, 4 Jan 2019 10:58:35 +0000 (11:58 +0100)]
sonar: we do have a process to remove deprecated code
Martin Quinson [Fri, 4 Jan 2019 10:13:42 +0000 (11:13 +0100)]
typo in comments
Martin Quinson [Fri, 4 Jan 2019 10:12:47 +0000 (11:12 +0100)]
cmake: mark some variables as advanced to hide them
Martin Quinson [Thu, 3 Jan 2019 22:23:38 +0000 (23:23 +0100)]
Merge branch 'master' of github.com:simgrid/simgrid