 Nathalie Furmento
|
9aee33b47d
test of new function starpu_init_task()
|
12 years ago |
|
Nathalie Furmento
|
14eddc11c7
New function starpu_init_task similar to starpu_insert_task without the submission of the task.
|
12 years ago |
 Samuel Thibault
|
edce30c7f0
port r11194 from 1.1: Also disable checking _head of renewed request lists
|
12 years ago |
|
Samuel Thibault
|
ea2ac2cc59
port r11191 from 1.1: Remove spurious frees
|
12 years ago |
|
Samuel Thibault
|
67fe6714cd
port r11189 from 1.1: Fix race on job ndeps: create_cg_task increments ndeps, it needs to hold the sync_mutex
|
12 years ago |
|
Samuel Thibault
|
55124f1005
port 11187 from 1.1: Fix livelock while running in valgrind: we really have to call sched_yield explicitly, rep nop is not enough
|
12 years ago |
|
Samuel Thibault
|
4c2ca351f2
port 11185 from 1.1: Fix names in documentation
|
12 years ago |
|
Samuel Thibault
|
489766f3ff
port r11183 from 1.1: document callback_arg_free and prologue_callback_arg_free
|
12 years ago |
|
Samuel Thibault
|
d74d3de4a0
port r11179 from 1.1: The starpufft caller can wait for the last task, the last task thus must not be detached
|
12 years ago |
|
Nathalie Furmento
|
0362a99418
mpi/src/starpu_mpi.c: use correct types to initialise mpi objects, and set the default handler to MPI_ERRORS_RETURN, the default MPI_ERRORS_ARE_FATAL exits the program and hence does not allow to see a backtrace
|
12 years ago |
|
Nathalie Furmento
|
05379546cf
mpi/tests/mpi_redux.c: use MPI_STATUS_IGNORE instead of NULL to indicate to MPI_Wait to ignore the status
|
12 years ago |
|
Nathalie Furmento
|
4d363ebb6b
src/core/sched_policy.c: do not display message scheduling policy not found when calling with STARPU_SCHED=help
|
12 years ago |
|
Nathalie Furmento
|
3e7427f3ae
mpi/tests/: use MPI_STATUS_IGNORE instead of NULL to indicate to MPI_Wait to ignore the status
|
12 years ago |
|
Samuel Thibault
|
3c7645aeff
port r11168 from 1.1: Also do not lie about spinlocks either, rather ignore accesses to busy_count. Also actually fix a missing check in perfmodel.c
|
12 years ago |
|
Samuel Thibault
|
886afdc84b
port 11166 from 1.1: entry is also allocated from reading files, also disable helgrind check there
|
12 years ago |
|
Nathalie Furmento
|
637d011bfc
src/common/thread.c: _STARPU_TRACE_MUTEX_LOCKED does not take argument
|
12 years ago |
|
Nathalie Furmento
|
1130a318ad
tests/main/insert_task.c: deal with ENODEV error
|
12 years ago |
|
Samuel Thibault
|
5bd1d28ab6
Fix worker initialization order: initialize fields before creating the first context.
|
12 years ago |
|
Samuel Thibault
|
c6d046885f
Put worker initialization in same order as structure. Add missing nsched_ctxs initialization
|
12 years ago |
|
Nathalie Furmento
|
a57110af5a
main/insert_task.c: check more testcasese
|
12 years ago |
|
Samuel Thibault
|
aea35461df
Document each racy spot normally reported by helgrind
|
12 years ago |
|
Samuel Thibault
|
0d1b0375b3
backport r11155 from 1.1: Disabling access to a variable locally does not actually seem to prevent helgrind from raising the race issues, on the contrary it prevents it from recording where the race comes from... Simply just disable checking these variables completely.
|
12 years ago |
|
Samuel Thibault
|
01294dcc77
Warn about using --enable-spinlock-check
|
12 years ago |
|
Samuel Thibault
|
4c97a56f4f
Add --enable-spinlock-check configure option to enable
|
12 years ago |
|
Nathalie Furmento
|
53d95c3f90
mpi/src/starpu_mpi_insert_task.c: set task::cl_arg_free to 1 when calling starpu_mpi_insert_task, it will clean the memory allocated to pack the codelet arguments
|
12 years ago |
|
Samuel Thibault
|
9de08d8da8
Make file/line handling for lock tracing compiled only when lock tracing is enabled instead of all the time
|
12 years ago |
|
Nathalie Furmento
|
461451be20
tests: indicate errorcheck/invalid_blocking_calls.c is expected to fail
|
12 years ago |
|
Samuel Thibault
|
02db02487b
fix memleak
|
12 years ago |
|
Samuel Thibault
|
b74c9583ff
Initialize buffer to be written for testing disk bandwidth, to avoid any page fault cost, and please valgrind
|
12 years ago |
|
Samuel Thibault
|
7e4f845927
Fix buffer overflow
|
12 years ago |
