p Test allreduce
$ $VALGRIND_NO_LEAK_CHECK ${bindir:=.}/../../../smpi_script/bin/smpirun -wrapper "${bindir:=.}/../../../bin/simgrid-mc" -map -hostfile ../hostfile_coll -platform ${platfdir:=.}/small_platform.xml -np 4 --log=xbt_cfg.thres:critical ${bindir:=.}/coll-allreduce-with-leaks --log=smpi_config.thres:warning --cfg=smpi/display-allocs:yes --cfg=smpi/simulate-computation:no --log=smpi_coll.thres:error --log=smpi_mpi.thres:error --log=smpi_pmpi.thres:error --cfg=smpi/list-leaks:10 --log=no_loc
-> [0.000000] [mc_safety/INFO] Check a safety property. Reduction is: dpor.
> [0.000000] [smpi/INFO] [rank 0] -> Tremblay
> [0.000000] [smpi/INFO] [rank 1] -> Tremblay
> [0.000000] [smpi/INFO] [rank 2] -> Tremblay
> [0.000000] [smpi/INFO] [rank 3] -> Tremblay
+> [0.000000] [mc_dfs/INFO] Start a DFS exploration. Reduction is: dpor.
> [0.000000] [smpi_utils/INFO] Probable memory leaks in your code: SMPI detected 8 unfreed MPI handles:
> [0.000000] [smpi_utils/WARNING] To get more information (location of allocations), compile your code with -trace-call-location flag of smpicc/f90
> [0.000000] [smpi_utils/INFO] 4 leaked handles of type MPI_Comm
> Largest allocation at once from a single process was 28 bytes, at coll-allreduce-with-leaks.c:28. It was called 1 times during the whole simulation.
> If this is too much, consider sharing allocations for computation buffers.
> This can be done automatically by setting --cfg=smpi/auto-shared-malloc-thresh to the minimum size wanted size (this can alter execution if data content is necessary)
->
+>
> [0.000000] [smpi_utils/INFO] Probable memory leaks in your code: SMPI detected 8 unfreed MPI handles:
> [0.000000] [smpi_utils/WARNING] To get more information (location of allocations), compile your code with -trace-call-location flag of smpicc/f90
> [0.000000] [smpi_utils/INFO] 4 leaked handles of type MPI_Comm
> Largest allocation at once from a single process was 28 bytes, at coll-allreduce-with-leaks.c:28. It was called 1 times during the whole simulation.
> If this is too much, consider sharing allocations for computation buffers.
> This can be done automatically by setting --cfg=smpi/auto-shared-malloc-thresh to the minimum size wanted size (this can alter execution if data content is necessary)
->
-> [0.000000] [mc_safety/INFO] No property violation found.
-> [0.000000] [mc_safety/INFO] 63 unique states visited; 16 backtracks (500 transition replays, 422 states visited overall)
+>
+> [0.000000] [mc_dfs/INFO] DFS exploration ended. 33 unique states visited; 7 backtracks (140 transition replays, 101 states visited overall)