Millian Poquet [Fri, 16 Nov 2018 15:11:10 +0000 (16:11 +0100)]
Millian Poquet [Thu, 15 Nov 2018 18:15:17 +0000 (19:15 +0100)]
[misc] fix deprecation msg for Actor::setProperty
Augustin Degomme [Thu, 15 Nov 2018 10:02:33 +0000 (11:02 +0100)]
handle MPI_ANY_SOURCE in recv replay
Arnaud Giersch [Thu, 15 Nov 2018 08:56:19 +0000 (09:56 +0100)]
Macro is always defined (by cmakedefine01).
Arnaud Giersch [Thu, 15 Nov 2018 08:54:18 +0000 (09:54 +0100)]
Add a comment explaining the "ifndef".
Arnaud Giersch [Thu, 15 Nov 2018 08:35:53 +0000 (09:35 +0100)]
Include seems useless.
Arnaud Giersch [Thu, 15 Nov 2018 08:02:28 +0000 (09:02 +0100)]
SIMGRID_HAVE_MC is already defined in <simgrid/config.h>.
Arnaud Giersch [Thu, 15 Nov 2018 07:54:07 +0000 (08:54 +0100)]
Fix comments.
Martin Quinson [Thu, 15 Nov 2018 06:34:24 +0000 (07:34 +0100)]
docs improvements
- some words about trace replay, mostly stolen from tuto. Ways more should be said.
- A warning about remainings of previous cmake configuration
- A word on VM in the 10000ft view of S4U
Martin Quinson [Mon, 12 Nov 2018 05:14:01 +0000 (06:14 +0100)]
fix clang build
Martin Quinson [Mon, 12 Nov 2018 04:12:37 +0000 (05:12 +0100)]
small doc improvement
Martin Quinson [Mon, 12 Nov 2018 03:49:39 +0000 (04:49 +0100)]
Context creation: Fix one OOP design issue
Martin Quinson [Sun, 11 Nov 2018 05:17:17 +0000 (06:17 +0100)]
improve a comment
Martin Quinson [Sun, 11 Nov 2018 04:39:11 +0000 (05:39 +0100)]
Factorize code between the calling sites of the ActorImpl ctor
Martin Quinson [Sun, 11 Nov 2018 03:23:03 +0000 (04:23 +0100)]
tiny factorizations and doc improvements
Martin Quinson [Sun, 11 Nov 2018 02:54:30 +0000 (03:54 +0100)]
remove the argc/argv version of simcall_process_create
Martin Quinson [Sun, 11 Nov 2018 01:49:31 +0000 (02:49 +0100)]
MSG: kill the MSG_process_create_from_stdfunc() pimple
Martin Quinson [Sun, 11 Nov 2018 00:31:57 +0000 (01:31 +0100)]
java: stop using MSG_process_create_from_stdfunc
Use simcall_process_create directly instead. This code would be better
if smx_actor were a real class.
Martin Quinson [Sat, 10 Nov 2018 23:54:05 +0000 (00:54 +0100)]
java process creation: this was already binded by java_main_jprocess()
Martin Quinson [Sat, 10 Nov 2018 23:12:30 +0000 (00:12 +0100)]
eclipse config file: there is some Java in this project
Martin Quinson [Sat, 10 Nov 2018 21:33:19 +0000 (22:33 +0100)]
doc: say that set_receiver() adds a silly ref to the actor
I tried hard to fix it instead, but I don't manage for now.
Martin Quinson [Tue, 6 Nov 2018 06:19:42 +0000 (07:19 +0100)]
Apple renamed all their OSes
https://www.engadget.com/2016/06/13/os-x-is-now-macos/
Martin Quinson [Sun, 4 Nov 2018 11:54:44 +0000 (12:54 +0100)]
please sonar: kill an unused type and reduce the visibility of another one
Arnaud Giersch [Wed, 7 Nov 2018 21:11:04 +0000 (22:11 +0100)]
Fix distcheck.
Arnaud Giersch [Fri, 2 Nov 2018 20:21:47 +0000 (21:21 +0100)]
Cosmetics.
Martin Quinson [Wed, 7 Nov 2018 19:51:31 +0000 (20:51 +0100)]
Merge pull request #309 from mpoquet/generate-pkg-config
Provide a pkg-config file for SimGrid
Millian Poquet [Wed, 7 Nov 2018 18:09:02 +0000 (19:09 +0100)]
[cmake] generate/install a pkg-config simgrid.pc
Augustin Degomme [Tue, 6 Nov 2018 17:06:05 +0000 (18:06 +0100)]
fix gcc 9 warnings. Have to check if the std::move removal is harmful
Martin Quinson [Sun, 4 Nov 2018 04:33:09 +0000 (05:33 +0100)]
these archaic callbacks are not used anymore, so kill them
Martin Quinson [Sun, 4 Nov 2018 04:04:55 +0000 (05:04 +0100)]
simplify MSG process creation/destruction
Use a proper callback to cleanup things, and somehow, we don't need
anything special anymore when creating a MSG process, so stop making
things complicated for no reason.
Martin Quinson [Sun, 4 Nov 2018 03:50:03 +0000 (04:50 +0100)]
Note that this easy task is actually impssible as is
Martin Quinson [Sat, 27 Oct 2018 14:31:49 +0000 (16:31 +0200)]
Deprecate the C API of parallel_execute
Martin Quinson [Sun, 4 Nov 2018 00:45:07 +0000 (01:45 +0100)]
plug a memleak
Augustin Degomme [Sat, 3 Nov 2018 14:57:58 +0000 (14:57 +0000)]
try to have mc work with freebsd 12 and flang
Martin Quinson [Fri, 2 Nov 2018 21:49:24 +0000 (22:49 +0100)]
finish objectifying the backtraces
Arnaud Giersch [Fri, 2 Nov 2018 17:01:20 +0000 (18:01 +0100)]
Use $() instead of backticks and remove useless use of grep.
Arnaud Giersch [Fri, 2 Nov 2018 16:48:27 +0000 (17:48 +0100)]
Copy constructor and assignment operators are implicitely deleted anyway (dixit clang 8).
Arnaud Giersch [Thu, 1 Nov 2018 20:18:23 +0000 (21:18 +0100)]
Use libunwind-generic instead of specific libunwind-PLAT.
It's the way to go according to libunwind-ptrace(3), and libunwind-generic
seems to point to the correct libunwind-PLAT anyway.
Martin Quinson [Mon, 29 Oct 2018 20:01:15 +0000 (21:01 +0100)]
Fix build without stacktrace_backtrace
Martin Quinson [Mon, 29 Oct 2018 15:51:45 +0000 (16:51 +0100)]
replace our own crude code with Boost.stacktrace
Augustin Degomme [Mon, 29 Oct 2018 12:25:59 +0000 (13:25 +0100)]
add libboost_stacktrace_backtrace to excludes
Martin Quinson [Sun, 28 Oct 2018 23:13:08 +0000 (00:13 +0100)]
provide a backtrace implementation that uses boost.stacktrace
Martin Quinson [Sun, 28 Oct 2018 21:30:38 +0000 (22:30 +0100)]
Hide the backtrace implementation in a private pimpl
This opens the path to an implementation with Boost.Stacktrace when found
Martin Quinson [Sun, 28 Oct 2018 20:58:57 +0000 (21:58 +0100)]
BuildSimGrid.sh: Try to install the lib if possible
Martin Quinson [Sun, 28 Oct 2018 16:40:51 +0000 (17:40 +0100)]
cmake: rewrite the detection of boost optional components, look for stacktrace
Arnaud Giersch [Sun, 28 Oct 2018 22:00:16 +0000 (23:00 +0100)]
Workaround bug in Intel-C.cmake.
See:
https://gitlab.kitware.com/cmake/cmake/issues/18166
https://gitlab.kitware.com/cmake/cmake/commit/
735e69f4458850cc335710af36d1382b891cbb5e
Arnaud Giersch [Sun, 28 Oct 2018 20:20:04 +0000 (21:20 +0100)]
Avoid bleeding edge ‘for’ loop initial declarations.
I suspect Travis to fail because cmake doesn't use -std=c11 with try_run.
Arnaud Giersch [Sun, 28 Oct 2018 20:10:16 +0000 (21:10 +0100)]
Kill obsolete config test programs.
Arnaud Giersch [Sat, 27 Oct 2018 19:43:55 +0000 (21:43 +0200)]
Remove obsolete exclusions.
Arnaud Giersch [Fri, 26 Oct 2018 12:38:40 +0000 (14:38 +0200)]
Udpate .mailmap.
Martin Quinson [Sun, 28 Oct 2018 08:34:44 +0000 (09:34 +0100)]
travis: this format is hard to read and I'm stupid. The perfect match
Martin Quinson [Sun, 28 Oct 2018 08:25:49 +0000 (09:25 +0100)]
travis: Display full cmake logs on error
Martin Quinson [Sat, 27 Oct 2018 20:22:21 +0000 (22:22 +0200)]
Hide the content of a simgrid::xbt::Backtrace and inline a function
Martin Quinson [Sat, 27 Oct 2018 20:06:53 +0000 (22:06 +0200)]
Simplify how we check for C++11 and friends
Martin Quinson [Sat, 27 Oct 2018 20:03:16 +0000 (22:03 +0200)]
FindSimGrid: Make sure our headers are compiled with C++11 at least
Martin Quinson [Sat, 27 Oct 2018 20:02:36 +0000 (22:02 +0200)]
try to please sonar on this ugly code
Martin Quinson [Sat, 27 Oct 2018 19:18:38 +0000 (21:18 +0200)]
backtrace: factorize the function headers between the portability layers
Martin Quinson [Sat, 27 Oct 2018 17:44:47 +0000 (19:44 +0200)]
docs: fix all examples of git clone to use https
Martin Quinson [Sat, 27 Oct 2018 17:09:18 +0000 (19:09 +0200)]
backtrace: fix various portability issues I just introduced
I'm really a bummer
Martin Quinson [Sat, 27 Oct 2018 16:55:53 +0000 (18:55 +0200)]
these portability bits should be useless
Martin Quinson [Sat, 27 Oct 2018 16:47:04 +0000 (18:47 +0200)]
inline two C files that were included (gosh)
Martin Quinson [Sat, 27 Oct 2018 16:18:49 +0000 (18:18 +0200)]
backtrace: kill two unused functions
Martin Quinson [Sat, 27 Oct 2018 16:01:47 +0000 (18:01 +0200)]
inline an include file. Not used from C anymore
Martin Quinson [Sat, 27 Oct 2018 15:20:25 +0000 (17:20 +0200)]
cosmetics around backtraces
- Use a type hiding the vector
- Remove a deprecated function that nobody uses
This is the first step toward the rewriting of the backtraces. My
motivation is that our ugly hack with addr2line is not working anymore
on my Debian while boost::stacktrace entered testing. I plan to
convert to that new system.
Martin Quinson [Sat, 27 Oct 2018 15:01:25 +0000 (17:01 +0200)]
remove an obsolete file
Martin Quinson [Sat, 27 Oct 2018 14:32:57 +0000 (16:32 +0200)]
Revert "[smpi] use maps on actor to use one buffer for each"
This reverts commit
4b1773e713af47a4917278519ab87386d1de48c5.
I fail to understand the problem that this commit wants to solve, and
it introduces many issues. Amongst which, a memory exhaustion on 32bit.
Martin Quinson [Sat, 27 Oct 2018 14:32:33 +0000 (16:32 +0200)]
Revert "Fix multiple memleaks in the replay of multiple SMPI instances"
This reverts commit
cfa37467445963796bfdafdec9bad8ec8566a60a.
Martin Quinson [Sat, 27 Oct 2018 14:32:17 +0000 (16:32 +0200)]
ignore more examples' binaries
Martin Quinson [Fri, 26 Oct 2018 16:31:00 +0000 (18:31 +0200)]
cosmetics, sorry for the noise
Martin Quinson [Fri, 26 Oct 2018 16:21:25 +0000 (18:21 +0200)]
Tuto algo: say how to specify where SimGrid is installed
Martin Quinson [Fri, 26 Oct 2018 16:14:19 +0000 (18:14 +0200)]
MPI tuto: use the https version of git repos
Millian Poquet [Thu, 25 Oct 2018 17:59:44 +0000 (19:59 +0200)]
[examples] smpi-replay-mm: free -> xbt_free
Millian Poquet [Thu, 25 Oct 2018 15:42:25 +0000 (17:42 +0200)]
Merge branch 'smpi-replay-mm-robustness'
Millian Poquet [Thu, 25 Oct 2018 15:37:42 +0000 (17:37 +0200)]
[examples] smpi-replay-mm: sleep(1) at job end
This is done to check that the job ends after the ranks
without huge modifications to the output log analysis
Millian Poquet [Thu, 25 Oct 2018 15:25:05 +0000 (17:25 +0200)]
[examples] smpi-replay-mm: fix output sort
Millian Poquet [Thu, 25 Oct 2018 15:12:56 +0000 (17:12 +0200)]
[examples] smpi-replay-mm: align actors' name
Martin Quinson [Wed, 24 Oct 2018 22:25:16 +0000 (00:25 +0200)]
docs: Fix broken links, bummer
Martin Quinson [Wed, 24 Oct 2018 22:11:30 +0000 (00:11 +0200)]
update the releasing procedure to align with the new website on framagit
Martin Quinson [Wed, 24 Oct 2018 22:11:01 +0000 (00:11 +0200)]
docker smpi tuto: we need libboost-all-dev for replay.cpp
Martin Quinson [Wed, 24 Oct 2018 22:04:48 +0000 (00:04 +0200)]
docs: user-friendly URLs
Martin Quinson [Wed, 24 Oct 2018 14:00:42 +0000 (16:00 +0200)]
Fix multiple memleaks in the replay of multiple SMPI instances
Martin Quinson [Wed, 24 Oct 2018 08:21:14 +0000 (10:21 +0200)]
travis: try to use the addon to install homebrew packages
Martin Quinson [Tue, 23 Oct 2018 20:18:47 +0000 (22:18 +0200)]
fix some doxygen errors
Martin Quinson [Mon, 22 Oct 2018 23:08:28 +0000 (01:08 +0200)]
docs: Rewrite the intro of platform description
Millian Poquet [Tue, 23 Oct 2018 13:31:05 +0000 (15:31 +0200)]
[examples] smpi-replay-mmd: hack--, memleak++
This commit removes a memory hack in the smpi-replay-mmd example.
Ideally, user apps should not need to do such hacks so it has been
removed.
smpi_replay_run changes argc/argv but does not deallocate the lost
memory, which results in a memory leak.
Millian Poquet [Mon, 22 Oct 2018 13:51:20 +0000 (15:51 +0200)]
[examples] smpi-replay-mmd: (ugly) memory clean-up
Martin Quinson [Fri, 19 Oct 2018 05:55:48 +0000 (07:55 +0200)]
Merge branch 'fix-multi-smpi-buffer-double-free' into 'master'
Fix multi smpi buffer double free
See merge request simgrid/simgrid!3
Martin Quinson [Thu, 18 Oct 2018 21:04:22 +0000 (23:04 +0200)]
TODO--
MERCIER Michael [Thu, 18 Oct 2018 15:55:15 +0000 (17:55 +0200)]
[smpi] use maps on actor to use one buffer for each
Martin Quinson [Wed, 17 Oct 2018 16:11:12 +0000 (18:11 +0200)]
Sanitize the API of this_actor::parallel_execute()
- Use std::vector instead of C arrays (old API remains but is not
documented -- it should be properly deprecated)
- The flop_amounts and comm_amounts arrays are not automatically freed
by the internal functions, and should be properly cleaned by their
creator.
- EXCEPTION: in ptask model, sequential exec and regular comms don't
have a real caller, so the internal function still has to free
these arrays... Sick Sad World.
- The proper solution would be to have the only copy of these arrays
in the Action instead of having it in s4u.
- But for now, Actions start as soon as created. So if you want to
init them without starting, you have to have the data in s4u and
only create the implementation side when you start the stuff.
- That should obviously be fixed :) First step in that direction
would be to have the constructor of each action NOT register the
action in the LMM, but have an Action::start() in charge of this.
For each subclass of Action.
Martin Quinson [Wed, 17 Oct 2018 16:09:14 +0000 (18:09 +0200)]
Hide ugly things from the doc
Martin Quinson [Wed, 17 Oct 2018 08:56:42 +0000 (10:56 +0200)]
docs: document <config>
Martin Quinson [Tue, 16 Oct 2018 20:40:22 +0000 (22:40 +0200)]
docs: fix broken links
Martin Quinson [Tue, 16 Oct 2018 20:11:43 +0000 (22:11 +0200)]
docs: new section on M&S hints
Martin Quinson [Tue, 16 Oct 2018 09:29:43 +0000 (11:29 +0200)]
remove my config files from the git. Not sure it ever helped anyone
Martin Quinson [Tue, 16 Oct 2018 09:28:15 +0000 (11:28 +0200)]
another thing that SMPI cannot do yet
Martin Quinson [Mon, 15 Oct 2018 19:42:48 +0000 (21:42 +0200)]
COPYING: move our homepage to framagit
Millian Poquet [Tue, 16 Oct 2018 16:34:16 +0000 (18:34 +0200)]
[examples] smpi-replay-mmd: static + cleanup
Millian Poquet [Tue, 16 Oct 2018 15:35:52 +0000 (17:35 +0200)]
[doc] SMPI_app_instance_register: improve doxygen
Millian Poquet [Tue, 16 Oct 2018 15:26:53 +0000 (17:26 +0200)]
[examples] smpi-replay-mmd: sayonara MSG