Martin Quinson [Tue, 23 May 2017 23:37:12 +0000 (01:37 +0200)]
silence a really supurious warning in cmake
Martin Quinson [Tue, 23 May 2017 23:35:08 +0000 (01:35 +0200)]
fix make dist
Martin Quinson [Tue, 23 May 2017 23:29:05 +0000 (01:29 +0200)]
cosmetics to please codacy
Martin Quinson [Tue, 23 May 2017 23:18:14 +0000 (01:18 +0200)]
sonar: also ignore that rule in hpp files
Martin Quinson [Tue, 23 May 2017 23:15:55 +0000 (01:15 +0200)]
plug a small memleak
Frederic Suter [Tue, 23 May 2017 23:14:13 +0000 (01:14 +0200)]
one level less in the last lib
Martin Quinson [Tue, 23 May 2017 23:05:48 +0000 (01:05 +0200)]
Java package names should not be uppercased
Martin Quinson [Tue, 23 May 2017 22:40:46 +0000 (00:40 +0200)]
remove an empty file
Martin Quinson [Tue, 23 May 2017 22:35:01 +0000 (00:35 +0200)]
remove an useless import in python
Martin Quinson [Tue, 23 May 2017 22:26:55 +0000 (00:26 +0200)]
sonar: dont' complain when we follow the advices of codacy
Martin Quinson [Tue, 23 May 2017 22:24:48 +0000 (00:24 +0200)]
hunt down some more short negation forms
Martin Quinson [Tue, 23 May 2017 22:11:19 +0000 (00:11 +0200)]
hunt down some more short negation forms
Martin Quinson [Tue, 23 May 2017 21:48:29 +0000 (23:48 +0200)]
kill all horizontal tabs
Martin Quinson [Tue, 23 May 2017 21:38:42 +0000 (23:38 +0200)]
remove (void) in C++
Martin Quinson [Tue, 23 May 2017 21:16:36 +0000 (23:16 +0200)]
add the codacy badge to the readme
Martin Quinson [Tue, 23 May 2017 21:11:57 +0000 (23:11 +0200)]
please codacy: use long form of negation in C++
and clang-format got annoying on this one
Frederic Suter [Tue, 23 May 2017 21:04:23 +0000 (23:04 +0200)]
a few more smells
degomme [Tue, 23 May 2017 19:41:34 +0000 (21:41 +0200)]
remove spurious non breaking spaces, grrr
Matthieu Volat [Tue, 23 May 2017 18:10:18 +0000 (20:10 +0200)]
Add support for flang and ifort Fortran compilers.
Require to properly set flags/libs in the wrappers.
Tested with flang under FreeBSD and ifort under Linux.
Matthieu Volat [Tue, 23 May 2017 17:42:18 +0000 (19:42 +0200)]
Only use -Wno-format-nonliteral if compiler is GCC or Clang.
Matthieu Volat [Tue, 23 May 2017 17:20:54 +0000 (19:20 +0200)]
Define worker_function more rigourously to xbt_main_func_t.
Matthieu Volat [Tue, 23 May 2017 17:19:05 +0000 (19:19 +0200)]
Handle some litteral string -> char* casts.
It's 100% legit in C, but icc seems to process it like C++, and
since simgrid is being ported to C++, well...
Matthieu Volat [Tue, 23 May 2017 17:15:16 +0000 (19:15 +0200)]
Use FindThreads to check for pthreading
It will prefer using threading as library if possible, and fallback
on -pthread only if necessary, which some compilers/systems
do not support anymore.
Matthieu Volat [Tue, 23 May 2017 17:12:20 +0000 (19:12 +0200)]
Flush output of energy tests.
That'll ensure in order output.
Martin Quinson [Tue, 23 May 2017 17:08:22 +0000 (19:08 +0200)]
useless cosmetics around NS3 long routes
Martin Quinson [Tue, 23 May 2017 17:04:53 +0000 (19:04 +0200)]
update the NS3 doc to be more explicit about long routes
Martin Quinson [Tue, 23 May 2017 16:19:56 +0000 (18:19 +0200)]
NS3: more informative messages in one case of misconfiguration
Martin Quinson [Tue, 23 May 2017 16:11:46 +0000 (18:11 +0200)]
ns3: preserve the rich informative type in a function call
degomme [Tue, 23 May 2017 16:35:02 +0000 (18:35 +0200)]
Add formatting to out fortran energy tests.
This is meant to ease comparison with reference, as free format may differ with various compilers.
And output is also nicer.
Martin Quinson [Tue, 23 May 2017 16:01:54 +0000 (18:01 +0200)]
NS3: be explicit when ignoring long routes
Martin Quinson [Tue, 23 May 2017 15:44:46 +0000 (17:44 +0200)]
give some hints about platforms that can be used in NS3
Martin Quinson [Mon, 22 May 2017 23:22:57 +0000 (01:22 +0200)]
more user-friendly error message
Explicit direclty the 3 potentially missing libraries instead of
forcing them to configure thrice to discover them.
Martin Quinson [Mon, 22 May 2017 23:03:34 +0000 (01:03 +0200)]
smpi_replay: make it easier to override some actions
There is even an example.
Frederic Suter [Mon, 22 May 2017 21:39:58 +0000 (23:39 +0200)]
list recent changes
Frederic Suter [Mon, 22 May 2017 21:39:09 +0000 (23:39 +0200)]
remove unvalidated Bconnection attribute
Frederic Suter [Mon, 22 May 2017 21:38:03 +0000 (23:38 +0200)]
few smells less in MC
Frederic Suter [Mon, 22 May 2017 21:37:18 +0000 (23:37 +0200)]
now useless that app-pmm is dead
Martin Quinson [Mon, 22 May 2017 21:23:53 +0000 (23:23 +0200)]
Kill a stupid example
A tiled matrix multiplication is a good idea, but doing the
multiplication for real in MSG is not so much
Martin Quinson [Mon, 22 May 2017 21:12:53 +0000 (23:12 +0200)]
Make attribute 'model' of <storage_type> optional: for future usage
Martin Quinson [Mon, 22 May 2017 20:14:04 +0000 (22:14 +0200)]
Remove attribute content_type of <storage_type>: was never used
Martin Quinson [Mon, 22 May 2017 19:49:22 +0000 (21:49 +0200)]
Make it easier to work with the XML parser
We ask flex to generate C code, but compile it with g++ so it breaks.
So we fix the C code with a bunch of sed. Who would think of
generating C++ code directly?
Martin Quinson [Mon, 22 May 2017 18:52:36 +0000 (20:52 +0200)]
sonar: v3.16 is out now
Martin Quinson [Mon, 22 May 2017 17:34:45 +0000 (19:34 +0200)]
Convert our README file to markdown
This is only to get all those nice little badges on github and gitlab
[no ci]
Martin Quinson [Mon, 22 May 2017 17:29:36 +0000 (19:29 +0200)]
sonar: disable dupplication in teshsuite too
Martin Quinson [Mon, 22 May 2017 15:09:05 +0000 (17:09 +0200)]
try to please travis
Frederic Suter [Mon, 22 May 2017 20:17:34 +0000 (22:17 +0200)]
no more content_type
Frederic Suter [Mon, 22 May 2017 19:38:18 +0000 (21:38 +0200)]
a vector of keyval structs is a map
Martin Quinson [Mon, 22 May 2017 13:10:08 +0000 (15:10 +0200)]
travis: previous cosmetics commit made it fail in any case
Martin Quinson [Mon, 22 May 2017 13:01:07 +0000 (15:01 +0200)]
tiny updates to the doc of huge pages
Martin Quinson [Mon, 22 May 2017 12:38:39 +0000 (14:38 +0200)]
Merge pull request #168 from simgrid/smpi_huge_pages
Add support for huge pages in SMPI
Martin Quinson [Mon, 22 May 2017 12:36:22 +0000 (14:36 +0200)]
we want to execute ctest on PR on travis
Tom Cornebize [Mon, 22 May 2017 09:09:11 +0000 (11:09 +0200)]
Add a comment and extend the doc.
Tom Cornebize [Mon, 22 May 2017 08:06:16 +0000 (10:06 +0200)]
Add documentation for smpi/shared-malloc-hugepage and SMPI_PARTIAL_SHARED_MALLOC.
Frederic Suter [Mon, 22 May 2017 08:03:37 +0000 (10:03 +0200)]
bummer
Frederic Suter [Mon, 22 May 2017 07:32:54 +0000 (09:32 +0200)]
start reducing smells in MC code
Tom Cornebize [Mon, 22 May 2017 07:43:27 +0000 (09:43 +0200)]
Clean the code.
Martin Quinson [Sun, 21 May 2017 16:37:22 +0000 (18:37 +0200)]
kill an unreachable statement (thanks sonar)
Martin Quinson [Sun, 21 May 2017 15:32:51 +0000 (17:32 +0200)]
instr: kill an unused structure field to please clang
Martin Quinson [Sun, 21 May 2017 00:16:36 +0000 (02:16 +0200)]
cosmetics
Martin Quinson [Sun, 21 May 2017 15:03:13 +0000 (17:03 +0200)]
Merge branch 'master' of github.com:simgrid/simgrid
Martin Quinson [Sun, 21 May 2017 14:58:46 +0000 (16:58 +0200)]
Disable uninteresting events in TI mode
Only 3 events must be logged in TI mode, but if we don't disable them,
chunks of Pajé trace ends up in the file supposed to be TI formated.
Martin Quinson [Sun, 21 May 2017 14:58:11 +0000 (16:58 +0200)]
loudly fail when asked to replay a non-existing file
Martin Quinson [Sun, 21 May 2017 00:04:54 +0000 (02:04 +0200)]
drop module strbuff. We don't need it anymore.
Killing XBT (and reduce our maintainance burden) was a major
motivation in the switch to C++.
Martin Quinson [Sat, 20 May 2017 23:59:07 +0000 (01:59 +0200)]
remove a broken feature: backtraces in log format
Martin Quinson [Sat, 20 May 2017 23:50:56 +0000 (01:50 +0200)]
Drop unused functions: xbt_str_varsubst(), xbt_strbuff_varsubst()
Martin Quinson [Sat, 20 May 2017 23:28:07 +0000 (01:28 +0200)]
sonar: try to reduce verbosity of the coverage
Martin Quinson [Sat, 20 May 2017 23:23:05 +0000 (01:23 +0200)]
Remove an empty file (thx, sonar)
Martin Quinson [Sat, 20 May 2017 22:06:08 +0000 (00:06 +0200)]
travis: activate MC on linux so that sonar checks it (fix #158)
Martin Quinson [Sat, 20 May 2017 21:22:45 +0000 (23:22 +0200)]
restore (and comment) a test for a feature
Martin Quinson [Sat, 20 May 2017 21:22:26 +0000 (23:22 +0200)]
reassure sonar: reinterpret_cast are the way to go in this file
Martin Quinson [Fri, 19 May 2017 23:32:47 +0000 (01:32 +0200)]
instr: merge the TI output in the converted code (WIP)
The TI output is a pimple added at some point to trick the TRACING
infrastructure into generating a Time Independent trace of the
executed MPI application, so that this trace can later be replayed
with smpi_replay.
This was done in pure C, with a large amount of pointers to functions.
This cannot be cleanly ported to C++ without a large amount of
factories so that we can subclass each class of event and pick the
right one at runtime.
Instead, this was converted in a ugly way, with a global defining the
type of output that we want today, and each print() function checking
that global to adapt its code.
The bug that I'm chasing is still there, as the replay does not look
for the file that was actually generated, but it's getting better
already.
We will soon rewrite the whole tracing infrastructure with many signals
that are fired at the places that we want to observe. Then, both the
TRACING and the TI modules will register functions in these signals to
be informed of what's going on in the simulation. But they will not
share any code anymore.
Soon...
Frederic Suter [Fri, 19 May 2017 12:32:02 +0000 (14:32 +0200)]
try to please clang 5.0
and remove useless exit in failing test
Frederic Suter [Fri, 19 May 2017 12:02:29 +0000 (14:02 +0200)]
couple of minor smells
Martin Quinson [Fri, 19 May 2017 11:34:52 +0000 (13:34 +0200)]
Merge branch 'master' of github.com:simgrid/simgrid
Martin Quinson [Fri, 19 May 2017 09:11:14 +0000 (11:11 +0200)]
Merge pull request #167 from simgrid/smpi_execute_public
Add smpi_execute_public
Tom Cornebize [Fri, 19 May 2017 08:10:52 +0000 (10:10 +0200)]
Renaming: smpi_execute_public → smpi_execute_benched
Frederic Suter [Fri, 19 May 2017 08:03:07 +0000 (10:03 +0200)]
bugs, smells and cosmetics of the day
Tom Cornebize [Thu, 18 May 2017 17:00:39 +0000 (19:00 +0200)]
Fix compilation for systems without MAP_HUGETLB.
Tom Cornebize [Thu, 18 May 2017 16:11:55 +0000 (18:11 +0200)]
Fix compilation warnings.
Frederic Suter [Thu, 18 May 2017 16:04:26 +0000 (18:04 +0200)]
attempt to get rid of all const_cast (take 2)
Frederic Suter [Thu, 18 May 2017 15:54:41 +0000 (17:54 +0200)]
attempt to get rid of all const_cast
Tom Cornebize [Thu, 18 May 2017 14:35:56 +0000 (16:35 +0200)]
Code style.
Tom Cornebize [Thu, 18 May 2017 14:06:26 +0000 (16:06 +0200)]
Code style.
Tom Cornebize [Tue, 9 May 2017 12:10:04 +0000 (14:10 +0200)]
Add support for huge pages in shared malloc.
Tom Cornebize [Sat, 13 May 2017 13:04:24 +0000 (15:04 +0200)]
Add function smpi_execute_public.
Frederic Suter [Wed, 17 May 2017 14:31:44 +0000 (16:31 +0200)]
another bunch of sonar treats
degomme [Wed, 17 May 2017 13:33:04 +0000 (15:33 +0200)]
Ignore pseudo-leaks when we use dlopen in smpi.
In a perfect world we should dlclose everything properly, but this leads to issues with some other atexit handles.
The xbt_postexit is always executed last, and closes the logs. If the SMPI code used XBT_LOG features, it is then needed until the last moment.
So we would need to store the handle at the XBT level instead of the SMPI one... And we don't really want that.
So we just hide these non-leaks from valgrind.
Frederic Suter [Wed, 17 May 2017 10:55:47 +0000 (12:55 +0200)]
please sonar, less free in c++
Frederic Suter [Wed, 17 May 2017 09:56:19 +0000 (11:56 +0200)]
a bit of cleaning in chainsend
Frederic Suter [Wed, 17 May 2017 06:28:42 +0000 (08:28 +0200)]
don't create dependencies if both root and end are not connected to any task
This happens when none of the actual jobs described in the DAX file
consumes and produces any file. In that case, root has no successor
and a dependency to end is created but end has no predecessor and a
dependency from root is created ... that already exists.
Martin Quinson [Tue, 16 May 2017 21:04:53 +0000 (23:04 +0200)]
code simplification in CMake. Hopefully no functional change
Martin Quinson [Tue, 16 May 2017 20:52:59 +0000 (22:52 +0200)]
More help to our fellow mac users
Martin Quinson [Tue, 16 May 2017 18:19:26 +0000 (20:19 +0200)]
plug some memleaks in the new unit test
Martin Quinson [Mon, 15 May 2017 08:55:25 +0000 (10:55 +0200)]
please sonar: nothing before the includes
Frederic Suter [Mon, 15 May 2017 15:01:17 +0000 (17:01 +0200)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
Frederic Suter [Mon, 15 May 2017 14:53:23 +0000 (16:53 +0200)]
mostly cosmetics
Martin Quinson [Mon, 15 May 2017 08:43:10 +0000 (10:43 +0200)]
LOOPAFTER is a better name for profile periodicity
Frederic Suter [Mon, 15 May 2017 07:07:50 +0000 (09:07 +0200)]
another bunch of sonar treats
Frederic Suter [Mon, 15 May 2017 06:43:41 +0000 (08:43 +0200)]
please sonar
Martin Quinson [Sun, 14 May 2017 22:22:48 +0000 (00:22 +0200)]
tmgr: cosmetics