Martin Quinson [Sat, 26 Mar 2022 23:22:25 +0000 (00:22 +0100)]
MBI: typo
Martin Quinson [Sat, 26 Mar 2022 23:08:44 +0000 (00:08 +0100)]
MBI mark the test requiring the detection of local variables as failing
Martin Quinson [Sat, 26 Mar 2022 22:59:49 +0000 (23:59 +0100)]
MBI: buffer on stack is a programming error, not a MPI error
Martin Quinson [Sat, 26 Mar 2022 22:59:06 +0000 (23:59 +0100)]
MBI test: disply the right buffer on error
Martin Quinson [Sat, 26 Mar 2022 10:01:16 +0000 (11:01 +0100)]
Resynch MBI generators with upstream
CollP2PMessageRaceGenerator.py: Replace source of second recv for message race.
(+ whitespace cleanups)
Martin Quinson [Sat, 26 Mar 2022 09:58:00 +0000 (10:58 +0100)]
Reactivate the last MBI generator now that it produces tests that compile
Arnaud Giersch [Sat, 26 Mar 2022 13:28:48 +0000 (14:28 +0100)]
Fix dead links.
Not sure what to do with the remainging links to
pda.gforge.inria.fr / lists.gforge.inria.fr / gforge.inria.fr/tracker
[ci-skip]
Arnaud Giersch [Fri, 25 Mar 2022 21:51:22 +0000 (22:51 +0100)]
Cosmetics: don't talk about ActorIDTrait if it's not required.
Arnaud Giersch [Fri, 25 Mar 2022 21:33:31 +0000 (22:33 +0100)]
Slightly improve message on deadlock.
Arnaud Giersch [Fri, 25 Mar 2022 21:21:07 +0000 (22:21 +0100)]
Remove unused member.
Arnaud Giersch [Fri, 25 Mar 2022 15:43:53 +0000 (16:43 +0100)]
[sonar] Don't mix public and private members (mc::State).
Arnaud Giersch [Fri, 25 Mar 2022 15:36:47 +0000 (16:36 +0100)]
[sonar] Don't mix public and private members (mc::ModelChecker).
Arnaud Giersch [Fri, 25 Mar 2022 13:35:13 +0000 (14:35 +0100)]
[sonar] Don't mix public and private members (resource::DiskImpl).
Martin Quinson [Fri, 25 Mar 2022 20:23:13 +0000 (21:23 +0100)]
Expose signals Activity::on_suspend and Activity::on_resume in s4u
Earlier, they were only accessible from ActivityImpl
Martin Quinson [Fri, 25 Mar 2022 20:00:40 +0000 (21:00 +0100)]
kill a signal that never worked anyway
Martin Quinson [Fri, 25 Mar 2022 14:53:10 +0000 (15:53 +0100)]
Yet another tiny improvement to the release process
Arnaud Giersch [Fri, 25 Mar 2022 08:36:50 +0000 (09:36 +0100)]
Make it possible to have an empty profile callback.
Arnaud Giersch [Fri, 25 Mar 2022 08:26:53 +0000 (09:26 +0100)]
Sonar is right: a std:vector is just fine here.
Jean-Edouard BOULANGER [Fri, 25 Mar 2022 07:26:51 +0000 (08:26 +0100)]
Add remaining Comm bindings and examples
Arnaud Giersch [Thu, 24 Mar 2022 15:47:57 +0000 (16:47 +0100)]
Make NOW a static member of EngineImpl.
Arnaud Giersch [Thu, 24 Mar 2022 15:33:18 +0000 (16:33 +0100)]
Make maxpid a static member of ActorIDTrait.
Arnaud Giersch [Thu, 24 Mar 2022 13:16:17 +0000 (14:16 +0100)]
Use a std::unique_ptr, and avoid explicit new/delete.
Arnaud Giersch [Wed, 23 Mar 2022 21:09:20 +0000 (22:09 +0100)]
Save more std::string copies.
Arnaud Giersch [Wed, 23 Mar 2022 14:53:40 +0000 (15:53 +0100)]
Avoid unnecessary copy of std::string parameter.
Martin Quinson [Thu, 24 Mar 2022 19:41:09 +0000 (20:41 +0100)]
kill trailing spaces and tabs in MBI files
Martin Quinson [Thu, 24 Mar 2022 19:38:59 +0000 (20:38 +0100)]
Sync MBI generators with upstream modifications
Martin Quinson [Thu, 24 Mar 2022 17:17:30 +0000 (18:17 +0100)]
Remove a reference to gforge from the README
Martin Quinson [Wed, 23 Mar 2022 15:46:01 +0000 (16:46 +0100)]
Revalidate the tesh outputs after cosmetics to the output in mc_dfs
Martin Quinson [Wed, 23 Mar 2022 15:44:46 +0000 (16:44 +0100)]
Fix the dependency of inter-comm transitions
Saying that COMM are indep with non-COMM transitions is good, but this
shouldnt mess up with the computation between COMM transitions...
Martin Quinson [Tue, 22 Mar 2022 20:55:09 +0000 (21:55 +0100)]
tiny fix in release process
Martin Quinson [Tue, 22 Mar 2022 20:54:50 +0000 (21:54 +0100)]
MC: be explicit when reaching an execution end (it's important when reading user's output)
Martin Quinson [Tue, 22 Mar 2022 20:53:22 +0000 (21:53 +0100)]
Comm transitions are INDEP with non-comm transitions
Arnaud Giersch [Tue, 22 Mar 2022 14:55:32 +0000 (15:55 +0100)]
More XBT_PUBLIC for mingw.
[ci-skip]
Arnaud Giersch [Tue, 22 Mar 2022 13:52:36 +0000 (14:52 +0100)]
A few Sonar smells.
Arnaud Giersch [Tue, 22 Mar 2022 13:05:43 +0000 (14:05 +0100)]
Wrong find&replace.
Arnaud Giersch [Tue, 22 Mar 2022 12:54:44 +0000 (13:54 +0100)]
Misc Python smells from codefactor.io.
* unnecesary semicolon
* unnecessary "pass" statement.
* unnecessary "else" after "return".
* unnecessary "global"
* specify exception type
* testing for None should use the 'is' operator
* don't redefine built-ins
* unused statements (import, assignment)
Arnaud Giersch [Tue, 22 Mar 2022 12:54:44 +0000 (13:54 +0100)]
Unnecessary newlines.
[ci-skip]
Arnaud Giersch [Tue, 22 Mar 2022 12:54:44 +0000 (13:54 +0100)]
Define macro XBT_ATTRIB_DEPRECATED_v336.
Arnaud Giersch [Tue, 22 Mar 2022 12:54:44 +0000 (13:54 +0100)]
Remove deprecated features for next release (3.32).
FIXME: NetZoneImpl::on_route_creation is not removed.
Arnaud Giersch [Tue, 22 Mar 2022 12:54:44 +0000 (13:54 +0100)]
Start the dev cycle toward v3.32.
Martin Quinson [Tue, 22 Mar 2022 11:52:24 +0000 (12:52 +0100)]
Fix the release date in the release notes
Martin Quinson [Tue, 22 Mar 2022 10:38:44 +0000 (11:38 +0100)]
Release v3.31
Martin Quinson [Tue, 22 Mar 2022 10:33:55 +0000 (11:33 +0100)]
Update tuto MC to the latest tool output
Bruno Donassolo [Tue, 22 Mar 2022 10:05:43 +0000 (11:05 +0100)]
Add test for issue105
Martin Quinson [Tue, 22 Mar 2022 09:26:08 +0000 (10:26 +0100)]
Fix make distcheck, once again
Martin Quinson [Tue, 22 Mar 2022 08:49:46 +0000 (09:49 +0100)]
Complete the documentation of v3.31
Augustin Degomme [Mon, 21 Mar 2022 20:18:49 +0000 (21:18 +0100)]
pedantic is now implied by -analyze
Augustin Degomme [Mon, 21 Mar 2022 20:17:27 +0000 (21:17 +0100)]
Changelog update
Augustin Degomme [Mon, 21 Mar 2022 20:17:09 +0000 (21:17 +0100)]
-analyze implies pedantic mode
Augustin Degomme [Mon, 21 Mar 2022 19:40:29 +0000 (20:40 +0100)]
give a (small and random) value of smpi/ois and smpi/os flags for mpich3 tests. Rationale is that it can speedup some tests a lot, as without it and async_small_thres, a lot a small messages can be sent at the same time in a loop, piling in the system and causing a lot of lag when lmm needs to handle all of these messages at once
Arnaud Giersch [Sat, 19 Mar 2022 22:20:40 +0000 (23:20 +0100)]
Returning a value from simcall_blocking's lambda is misleading.
The return value is in fact obtained from observer->get_result().
Arnaud Giersch [Fri, 18 Mar 2022 13:26:23 +0000 (14:26 +0100)]
Cosmetics: coding-style.
[ci-skip]
Martin Quinson [Mon, 21 Mar 2022 00:05:13 +0000 (01:05 +0100)]
Revalidate the tesh of a MC test that has now less combinatorial explosion
This is the result of the optimization introduced in
c40f6bc2301d8b7039f7ced0cd40b2bf6e3879ed
Augustin Degomme [Sun, 20 Mar 2022 23:36:15 +0000 (00:36 +0100)]
remove spurious barrier calls in comm creation and win creation.
standard allows them but they are not mandatory I think (and slow down mc)
Martin Quinson [Sun, 20 Mar 2022 23:20:44 +0000 (00:20 +0100)]
Resynchronize simgrid's MBI adapter from MBI project
Martin Quinson [Sun, 20 Mar 2022 23:11:43 +0000 (00:11 +0100)]
Add the remaining MBI generators
Martin Quinson [Sun, 20 Mar 2022 08:40:57 +0000 (09:40 +0100)]
small sonar fixes
Martin Quinson [Sun, 20 Mar 2022 08:24:25 +0000 (09:24 +0100)]
also remove the master example from example syncho-mutex.py
Augustin Degomme [Sun, 20 Mar 2022 23:15:01 +0000 (00:15 +0100)]
Have MPI collectives display their root and MPI_Op to handle mismatches lazily
Augustin Degomme [Sun, 20 Mar 2022 20:41:55 +0000 (21:41 +0100)]
add check for collectives, using check_collectives_ordering utility.
maybe I missed some ..
Augustin Degomme [Sun, 20 Mar 2022 20:41:08 +0000 (21:41 +0100)]
activate pedantic mode for MBI
Augustin Degomme [Sun, 20 Mar 2022 20:40:53 +0000 (21:40 +0100)]
make pedantic not the default...
Augustin Degomme [Sun, 20 Mar 2022 20:40:34 +0000 (21:40 +0100)]
SMPI: add utility to keep the order of collective calls performed by each process MPI and validate that all processes do the same.
It's not activated by default, and needs --cfg=smpi/pedantic:true option, as it may store too much data in memory for now
For each comm we maintain a vector of encountered collective calls
Each process stores the amount of calls performed in each communicator
At each new one we compare the amount of calls to the size of the corresponding vector
if we are the first process, add the call to the list
if not, compare its name to the one at the corresponding place in the vector, and cry for help if it's not the same.
Kudos mquinson and MBI for the idea.
Augustin Degomme [Sun, 20 Mar 2022 13:01:45 +0000 (14:01 +0100)]
add finalizing state for smpi actor:
if finalization barrier is used, tag is now different, in order not to match against other MPI_Barrier that may be in the MPI code
Martin Quinson [Sun, 20 Mar 2022 00:20:33 +0000 (01:20 +0100)]
Fail loudly when someone tries to use Actor::join() in MC (not implemented)
Martin Quinson [Sun, 20 Mar 2022 00:12:31 +0000 (01:12 +0100)]
Don't have an extra actor in s4u-synchro-mutex to create the others
In MC, it could happen that the master actor terminates, leaves and
frees the shared memory before the end of the other actors.
Martin Quinson [Sat, 19 Mar 2022 23:34:14 +0000 (00:34 +0100)]
MC: be more cautionous around MC_process_clock, the actor may not be there in case of bugs
Martin Quinson [Sat, 19 Mar 2022 21:34:29 +0000 (22:34 +0100)]
Cosmetics
Martin Quinson [Sat, 19 Mar 2022 20:09:48 +0000 (21:09 +0100)]
MBI: reduce a bit the amount of supurious warnings in generated code
Augustin Degomme [Sun, 20 Mar 2022 00:13:19 +0000 (01:13 +0100)]
MPI Win: add check if we try to delete a locked or opened window.
todo:
- refcount for wins,
- check if we need a separate counter for open/fence (going negative is hacky)
Martin Quinson [Sat, 19 Mar 2022 19:34:13 +0000 (20:34 +0100)]
Fix gcc builds :(
Martin Quinson [Sat, 19 Mar 2022 19:25:19 +0000 (20:25 +0100)]
make a single barrier per RMA window and share it through broadcast
Martin Quinson [Sat, 19 Mar 2022 19:12:48 +0000 (20:12 +0100)]
Give users a way to debug their barriers
Augustin Degomme [Sat, 19 Mar 2022 16:39:20 +0000 (17:39 +0100)]
use correct type
Augustin Degomme [Sat, 19 Mar 2022 16:37:34 +0000 (17:37 +0100)]
SMPI: add check for buffer size in windows creation (thx MBI)
Martin Quinson [Fri, 18 Mar 2022 22:12:46 +0000 (23:12 +0100)]
Finish to document the BMF options [no-ci]
Martin Quinson [Fri, 18 Mar 2022 21:45:47 +0000 (22:45 +0100)]
MBI: don't add the path of the generator in the generated codes
Martin Quinson [Fri, 18 Mar 2022 20:23:48 +0000 (21:23 +0100)]
Specify actor in the MC traces
Bruno Donassolo [Fri, 18 Mar 2022 21:26:28 +0000 (22:26 +0100)]
Fix doc: add bmf/precision
Bruno Donassolo [Fri, 18 Mar 2022 17:44:16 +0000 (18:44 +0100)]
Changelog
Bruno Donassolo [Fri, 18 Mar 2022 17:42:51 +0000 (18:42 +0100)]
Revalidate output for maxmin_bench tests.
Simply copy and paste output since it's hard to manually validate it
Bruno Donassolo [Fri, 18 Mar 2022 17:30:41 +0000 (18:30 +0100)]
Kill set_concurrency_share
Bruno Donassolo [Fri, 18 Mar 2022 15:47:29 +0000 (16:47 +0100)]
Kill expand_add
Bruno Donassolo [Fri, 18 Mar 2022 15:42:52 +0000 (16:42 +0100)]
Unify expand and expand_add
A single function to represent the use of some resource by an action.
Fixes https://framagit.org/simgrid/simgrid/-/issues/105
Bruno Donassolo [Fri, 18 Mar 2022 10:27:02 +0000 (11:27 +0100)]
Fix ctest for no Eigen builds
Bruno Donassolo [Fri, 18 Mar 2022 09:32:06 +0000 (10:32 +0100)]
Remove bmf host model. Add it as an option.
Remove --cfg=host/model:ptask_BMF.
Use --cfg=host/model:ptask_L07 --cfg=host/solver:bmf instead.
More consistent with other models (CPU, disk, etc)
Bruno Donassolo [Fri, 18 Mar 2022 09:00:55 +0000 (10:00 +0100)]
BMF: undo weight adding.
It doesn't seem necessary while we're not sure to the semantics gave to
fair_sharing with priorities
Bruno Donassolo [Thu, 17 Mar 2022 10:47:24 +0000 (11:47 +0100)]
Network: add solver as an option
Select a test which doesn't use crosstraffic and the timing is the same
for both models (maxmin and bmf)
Bruno Donassolo [Thu, 17 Mar 2022 10:44:09 +0000 (11:44 +0100)]
BMF: hopefully fix the fair_sharing with priority
Go back where we were supposed to be 1 week ago
Bruno Donassolo [Wed, 16 Mar 2022 18:04:40 +0000 (19:04 +0100)]
Disk: add solver as an option
Select one random test which timing should be the same between maxmin
and bmf
Bruno Donassolo [Wed, 16 Mar 2022 17:53:46 +0000 (18:53 +0100)]
System: methods -3 protected, +3 private
Bruno Donassolo [Wed, 16 Mar 2022 17:42:32 +0000 (18:42 +0100)]
Fix distcheck
Bruno Donassolo [Wed, 16 Mar 2022 17:36:27 +0000 (18:36 +0100)]
BMF: remove warning since the solver can be used by any model
Bruno Donassolo [Wed, 16 Mar 2022 17:21:24 +0000 (18:21 +0100)]
BMF: fix cloud-capping tests.
Bruno Donassolo [Wed, 16 Mar 2022 14:59:25 +0000 (15:59 +0100)]
BMF: Fix bug with threads
Our fair sharing is related to maxA_ not A_
Bruno Donassolo [Mon, 14 Mar 2022 14:52:03 +0000 (15:52 +0100)]
Avoid negative rate
Negatives rates doesn't make sense in real life.
In this case, set fair_sharing as if the resource were saturated
Bruno Donassolo [Wed, 16 Mar 2022 11:15:38 +0000 (12:15 +0100)]
Move common part for selective_update to System class.
And 1 more private variable in System \o/
Bruno Donassolo [Wed, 16 Mar 2022 10:10:56 +0000 (11:10 +0100)]
Rename update_modified_set to update_modified_cnst_set.
It's the modified_constraint_set we're updating...
Avoid confusion with modified_set_
Bruno Donassolo [Wed, 16 Mar 2022 09:30:33 +0000 (10:30 +0100)]
System: fix modified_set_ when selective_update is active
Using update_modified_set only on the first constraint of the variable
is valid only if the variable is active. Otherwise, it can make the
modified_set_ incomplete.
Bruno Donassolo [Tue, 15 Mar 2022 17:26:02 +0000 (18:26 +0100)]
Move maxmin solver from system
New class for maxmin solver:: lmm::MaxMin.
A tiny step towards better encapsulation, still far from good though.