degomme [Tue, 28 Mar 2017 00:05:53 +0000 (02:05 +0200)]
Mainly on 32 bits systems, target_disp can be seen as negative in one case.
This is due to the fact that in this case, the disp is actually an address, which is stored in a signed type.
We can't do much about that, and all MPI libraries I checked have this particular check disabled for this case.
degomme [Mon, 27 Mar 2017 21:19:51 +0000 (23:19 +0200)]
Add an smpi_is_shared call, and ignore memcpys in SMPI copy callback when buffers seem to be in shared memory.
For the global case this means we need to store the metadata as it was done in the local case
degomme [Sun, 26 Mar 2017 20:48:14 +0000 (22:48 +0200)]
support MPI_Op_commutative call, as it was already implemented internally
degomme [Sun, 26 Mar 2017 20:47:41 +0000 (22:47 +0200)]
activate some more tests
degomme [Fri, 24 Mar 2017 14:52:34 +0000 (15:52 +0100)]
Add MPI_Win_allocate, MPI_Win_create_dynamic, MPI_Win_attach, MPI_Win_detach, MPI_Win_set_info, MPI_Win_get_info.
Frederic Suter [Mon, 27 Mar 2017 18:27:59 +0000 (20:27 +0200)]
plug leaks
Frederic Suter [Mon, 27 Mar 2017 14:36:08 +0000 (16:36 +0200)]
finish to plug ptaskL07 leak
network and cpu models share the same lmm as the host model. the host
model frees the lmm, but both sub models have to set their local
reference to nullptr before the parent destructors are called to
prevent double free/delete.
Frederic Suter [Mon, 27 Mar 2017 14:17:27 +0000 (16:17 +0200)]
this test has been converted
Frederic Suter [Mon, 27 Mar 2017 12:16:05 +0000 (14:16 +0200)]
dumb me
Frederic Suter [Mon, 27 Mar 2017 12:11:51 +0000 (14:11 +0200)]
one more conversion
Frederic Suter [Mon, 27 Mar 2017 08:57:16 +0000 (10:57 +0200)]
make isend return a ref on the created Comm
Frederic Suter [Mon, 27 Mar 2017 08:16:07 +0000 (10:16 +0200)]
tesh conversion
add Mailbox::listen() and this_actor::isend() in the process
Frederic Suter [Sun, 26 Mar 2017 14:36:17 +0000 (16:36 +0200)]
sunday bug chase
Frederic Suter [Sun, 26 Mar 2017 14:03:54 +0000 (16:03 +0200)]
mostly cosmetics
Frederic Suter [Sun, 26 Mar 2017 13:31:59 +0000 (15:31 +0200)]
dict to map for storage model properties
Frederic Suter [Sun, 26 Mar 2017 13:01:22 +0000 (15:01 +0200)]
fix build?
Frederic Suter [Sun, 26 Mar 2017 12:47:35 +0000 (14:47 +0200)]
sunday smell chase
Frederic Suter [Sun, 26 Mar 2017 11:11:21 +0000 (13:11 +0200)]
conversion++
Frederic Suter [Sun, 26 Mar 2017 10:28:12 +0000 (12:28 +0200)]
one more teshsuite convertion
a couple extra s4u methods were needed
Frederic Suter [Fri, 24 Mar 2017 16:25:15 +0000 (17:25 +0100)]
better with that file ...
Frederic Suter [Fri, 24 Mar 2017 16:23:33 +0000 (17:23 +0100)]
TESH: msg to s4u - act 1
Frederic Suter [Fri, 24 Mar 2017 12:29:24 +0000 (13:29 +0100)]
bug-- and a bit less brain overload
Frederic Suter [Fri, 24 Mar 2017 12:28:35 +0000 (13:28 +0100)]
match new with delete in fact
Martin Quinson [Fri, 24 Mar 2017 12:14:45 +0000 (13:14 +0100)]
restore a check ensuring that this example is not used on an inadapted platform
Martin Quinson [Fri, 24 Mar 2017 10:25:29 +0000 (11:25 +0100)]
match free with mallocs
That's weird to put a free here, but having a new sg_size_t would be
even worse, wouldn't it?
Frederic Suter [Fri, 24 Mar 2017 10:44:52 +0000 (11:44 +0100)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
Frederic Suter [Fri, 24 Mar 2017 10:44:43 +0000 (11:44 +0100)]
plug leak
Frederic Suter [Fri, 24 Mar 2017 10:44:16 +0000 (11:44 +0100)]
improve that mess a bit
Frederic Suter [Fri, 24 Mar 2017 07:44:30 +0000 (08:44 +0100)]
no need to test if hard coded value is in a range
Martin Quinson [Thu, 23 Mar 2017 23:30:43 +0000 (00:30 +0100)]
pet sonar: kill a 'using namespace'
Martin Quinson [Thu, 23 Mar 2017 23:15:34 +0000 (00:15 +0100)]
sonar: ignore missing namespaces in examples
Martin Quinson [Thu, 23 Mar 2017 22:28:16 +0000 (23:28 +0100)]
petting sonar as every night before bed
Martin Quinson [Thu, 23 Mar 2017 15:58:59 +0000 (16:58 +0100)]
fix case of a header file
Martin Quinson [Thu, 23 Mar 2017 15:58:30 +0000 (16:58 +0100)]
Add s4u::Mutex to the doc
Frederic Suter [Thu, 23 Mar 2017 12:39:36 +0000 (13:39 +0100)]
no need to call cimix everytime for that
Frederic Suter [Thu, 23 Mar 2017 12:27:56 +0000 (13:27 +0100)]
surf_fopen to ifsopen + no need to reimplement streambuffers
Frederic Suter [Thu, 23 Mar 2017 11:55:02 +0000 (12:55 +0100)]
32-bit requires smthg longer than size_t
Frederic Suter [Thu, 23 Mar 2017 11:22:14 +0000 (12:22 +0100)]
start to mess up with storage
+ Storage size cannot change, do not ask surf all the time
+ replace internal dict by a map
+ use ifstream and c++ string tools
Martin Quinson [Thu, 23 Mar 2017 08:55:46 +0000 (09:55 +0100)]
Improve the doc on how to release
One day, I should read it BEFORE releasing to save time :(
Martin Quinson [Thu, 23 Mar 2017 08:04:49 +0000 (09:04 +0100)]
Initialize the 3.16 release cycle
Martin Quinson [Wed, 22 Mar 2017 16:50:52 +0000 (17:50 +0100)]
release v3.15
Martin Quinson [Wed, 22 Mar 2017 16:25:38 +0000 (17:25 +0100)]
doxygen fixups, again
Martin Quinson [Wed, 22 Mar 2017 16:09:08 +0000 (17:09 +0100)]
doxygen fixups
I kill the reference guides from the archive to save half of the
archive size. Seriously, I'm getting tired of doxygen.
degomme [Wed, 22 Mar 2017 14:47:05 +0000 (15:47 +0100)]
SMPI copyright bump before release
degomme [Wed, 22 Mar 2017 14:41:52 +0000 (15:41 +0100)]
Changelog update
degomme [Wed, 22 Mar 2017 14:40:30 +0000 (15:40 +0100)]
Add mpi_get_accumulate call. And MPI_NO_OP operation
Martin Quinson [Wed, 22 Mar 2017 14:37:31 +0000 (15:37 +0100)]
we will release today
Frederic Suter [Wed, 22 Mar 2017 14:30:19 +0000 (15:30 +0100)]
fix links
Frederic Suter [Wed, 22 Mar 2017 14:22:54 +0000 (15:22 +0100)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
Frederic Suter [Wed, 22 Mar 2017 14:19:59 +0000 (15:19 +0100)]
pre-release update
Martin Quinson [Wed, 22 Mar 2017 14:19:12 +0000 (15:19 +0100)]
cleanups on NEWS and ChangeLog
Frederic Suter [Wed, 22 Mar 2017 11:42:43 +0000 (12:42 +0100)]
smells -= a lot
Frederic Suter [Wed, 22 Mar 2017 11:06:55 +0000 (12:06 +0100)]
bugs and smells
Frederic Suter [Wed, 22 Mar 2017 10:28:54 +0000 (11:28 +0100)]
fix example (bummer)
chase smells
Frederic Suter [Wed, 22 Mar 2017 09:30:45 +0000 (10:30 +0100)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
Frederic Suter [Wed, 22 Mar 2017 09:29:58 +0000 (10:29 +0100)]
add actor-kill S4U example
rename launching to actor-create
degomme [Wed, 22 Mar 2017 09:03:31 +0000 (10:03 +0100)]
smells --
degomme [Wed, 22 Mar 2017 09:03:10 +0000 (10:03 +0100)]
Better handling of lock modes. Should avoid some rare deadlocks.
Frederic Suter [Wed, 22 Mar 2017 08:54:43 +0000 (09:54 +0100)]
yet another example conversion
Frederic Suter [Wed, 22 Mar 2017 08:21:17 +0000 (09:21 +0100)]
vain attempt to chase last leak in ptaskL07
the destructors of CpuModel and NetworkModel do not seem to be called
at all
save one traversal while at it
Martin Quinson [Wed, 22 Mar 2017 01:51:10 +0000 (02:51 +0100)]
another round for sonar
Martin Quinson [Wed, 22 Mar 2017 01:44:28 +0000 (02:44 +0100)]
cosmetics: fix the 'work' of clang-format
Martin Quinson [Wed, 22 Mar 2017 01:42:40 +0000 (02:42 +0100)]
please sonar
Martin Quinson [Wed, 22 Mar 2017 01:29:29 +0000 (02:29 +0100)]
plug two memleaks
Martin Quinson [Wed, 22 Mar 2017 00:38:05 +0000 (01:38 +0100)]
Implement natively VM_getVMByName. Java VMs are now freed
It's amusing to think that no java VM were ever freed before...
Martin Quinson [Tue, 21 Mar 2017 23:36:50 +0000 (00:36 +0100)]
Allow maestro to do simcalls in the Java world too
The detection of whether we are in maestro was broken in the
JavaContext, leading to an incredible mayhem in my code...
Gosh, that JNI code is the ultimate masochist experience.
Martin Quinson [Tue, 21 Mar 2017 22:37:57 +0000 (23:37 +0100)]
Implement VM::all() as a native, and use it in tests
This is to kill the array of all VMs in the java world, that retains
references to the VMs that thus cannot be cleaned. One method to go
and I'm good with it.
Martin Quinson [Tue, 21 Mar 2017 22:28:12 +0000 (23:28 +0100)]
s4u::VM: make getState() public, and add a state for destroyed VMs
Martin Quinson [Tue, 21 Mar 2017 22:24:11 +0000 (23:24 +0100)]
cosmetics
Martin Quinson [Tue, 21 Mar 2017 20:16:02 +0000 (21:16 +0100)]
java: rename an internal method
Martin Quinson [Tue, 21 Mar 2017 15:58:41 +0000 (16:58 +0100)]
java examples: make it easier to run them from the cmdline or from eclipse
Frederic Suter [Tue, 21 Mar 2017 14:01:12 +0000 (15:01 +0100)]
plug (massive) leak with ptaskL07
Martin Quinson [Tue, 21 Mar 2017 13:49:18 +0000 (14:49 +0100)]
make it easier to run that example from eclipse directly
Martin Quinson [Tue, 21 Mar 2017 13:33:40 +0000 (14:33 +0100)]
Merge branch 'master' of scm.gforge.inria.fr:/gitroot/simgrid/simgrid
Martin Quinson [Tue, 21 Mar 2017 12:52:33 +0000 (13:52 +0100)]
that malloc is really useless
Martin Quinson [Tue, 21 Mar 2017 12:33:24 +0000 (13:33 +0100)]
jtask: remove the ability to receive from a given host
This is disabled in the underlying code since a long time
Frederic Suter [Tue, 21 Mar 2017 12:35:35 +0000 (13:35 +0100)]
Merge branch 'master' of git+ssh://scm.gforge.inria.fr//gitroot/simgrid/simgrid
Frederic Suter [Tue, 21 Mar 2017 12:35:20 +0000 (13:35 +0100)]
update documentation with new examples
degomme [Tue, 21 Mar 2017 12:28:23 +0000 (13:28 +0100)]
make this test a bit shorter
degomme [Tue, 21 Mar 2017 12:28:13 +0000 (13:28 +0100)]
Fix stupid tag collision bug. RMA tag is used to order consecutive MPI_Accumulate calls.
The 454th message's tag was causing a collision with the BARRIER tag, causing the wrong message to be used.
So switch to a decrement instead, and start from a value below all of the ones we use (users can't use <0 values, it's only for internals)
Martin Quinson [Tue, 21 Mar 2017 12:17:49 +0000 (13:17 +0100)]
jtask: reduce the code duplication
Martin Quinson [Tue, 21 Mar 2017 12:17:19 +0000 (13:17 +0100)]
sanitizes the msg_task fields default value
Frederic Suter [Tue, 21 Mar 2017 12:03:38 +0000 (13:03 +0100)]
partially address a bug in Java initialization
SimGrid flag on command line were consumed at C level but stayed in the
original Java String[] args. This could mess users' args[i] if SG flags
were not put at the end of the command line.
This patch removes the SG flags from the Java arguments. However, the
number of arguments REMAINS UNCHANGED.
It is then UNSAFE to test if args.length is greater than the number of
YOUR OWN ARGUMENTS. It might be if you have --log or --cfg flags in
the command line. BTW, it already was ...
Martin Quinson [Tue, 21 Mar 2017 11:14:53 +0000 (12:14 +0100)]
code cleanups in jtask
Some checks are really easier when done on the Java side
Martin Quinson [Tue, 21 Mar 2017 10:36:23 +0000 (11:36 +0100)]
allow to use this test as a crash test
Martin Quinson [Tue, 21 Mar 2017 10:32:54 +0000 (11:32 +0100)]
java example simplification and cosmetics
Martin Quinson [Tue, 21 Mar 2017 10:26:30 +0000 (11:26 +0100)]
java example simplification + reindent + cosmetics
Martin Quinson [Tue, 21 Mar 2017 10:19:38 +0000 (11:19 +0100)]
java: plug a memleak
Martin Quinson [Tue, 21 Mar 2017 09:18:54 +0000 (10:18 +0100)]
increase a bit the size of this java test (+cosmetics)
Still under the second
Martin Quinson [Tue, 21 Mar 2017 08:49:21 +0000 (09:49 +0100)]
guess what? I hate our java crufty code
Frederic Suter [Tue, 21 Mar 2017 08:40:41 +0000 (09:40 +0100)]
add suspend/resume to S4U
update the actor-migration example to perfectly match the
process-migration example in C/MSG
Martin Quinson [Tue, 21 Mar 2017 07:50:03 +0000 (08:50 +0100)]
more cleanups to jprocess
- Don't catch error conditions that cannot occure
- Be more explicit on error conditions that may remain
- cosmetics
Martin Quinson [Tue, 21 Mar 2017 07:47:38 +0000 (08:47 +0100)]
initialize all fields of the xbt_ex_t class
Martin Quinson [Tue, 21 Mar 2017 06:25:22 +0000 (07:25 +0100)]
java: inline a function and don't leak a global ref
Frederic Suter [Mon, 20 Mar 2017 21:56:06 +0000 (22:56 +0100)]
attempt to extend S4U to migrate actors
probably need a cleaning/documentation pass
lack suspend/resume to mimic C example
Martin Quinson [Mon, 20 Mar 2017 21:46:24 +0000 (22:46 +0100)]
jprocess: more cruft removed
Martin Quinson [Mon, 20 Mar 2017 21:41:16 +0000 (22:41 +0100)]
further the jprocess creation by not testing conditions that cannot happen
The host and name was tested to not be null in the Java world before
arriving here
Martin Quinson [Mon, 20 Mar 2017 21:32:17 +0000 (22:32 +0100)]
deperlify the Java process creation: no need to pass the hostname where we can use the host
Martin Quinson [Mon, 20 Mar 2017 21:18:53 +0000 (22:18 +0100)]
java: code simplification: no need for an extra boolean here
Martin Quinson [Mon, 20 Mar 2017 21:02:39 +0000 (22:02 +0100)]
java: well, use a closure if you really want to