Historial de Commits

Autor SHA1 Mensaje Fecha
  Nathalie Furmento 6cacff4980 mpi/tests/user_defined_datatype.c: make sure tags are different for each iteration hace 12 años
  Nathalie Furmento 8ef6b950f3 tests/loader.c: send SIGQUIT on timeout to generate core file hace 12 años
  Andra Hugo 5c1a97e5c7 fix resize for lp2 (compute correctly the bw && the latency) hace 12 años
  Samuel Thibault d95e1c8398 explain helgrind about some races we are fine with hace 12 años
  Samuel Thibault 95e0b19b3e Add some missing mutex/cond/rwlock_destroy hace 12 años
  Samuel Thibault 4b2844fb36 Also automatically release opencl memory on shutdown hace 12 años
  Andra Hugo a603573008 fix valgrind issue hace 12 años
  Samuel Thibault 32e124f6b6 revert spurious change in r9000, keep the suppression for now hace 12 años
  Samuel Thibault c44a90ff25 after r8987, mem chunks can now not have a data pointer hace 12 años
  Samuel Thibault add756f1af Note about racyness. We can't really add a suppression for this case, it would hide potential other races hace 12 años
  Samuel Thibault 0cd466b90a revert r8991: the busy_mutex must be taken after the spinlock, so we can not take the spinlock after the mutex. Reading an int really is atomic on all our targets anyway. hace 12 años
  Samuel Thibault 5e18100b8c more suppressions hace 12 años
  Samuel Thibault 78e6c17e27 Fix setting memory limit above 2GiB hace 12 años
  Samuel Thibault d3065fdc1b Make valgrind yield to other threads when busy-looping, otherwise we may get stuck hace 12 años
  Samuel Thibault a1db18eda4 count data as freed only after it is actually freed hace 12 años
  Samuel Thibault 5b3d87d1c3 Permit to disable sequential consistency for a given task. hace 12 años
  Samuel Thibault 450dc26dec more suppression hace 12 años
  Samuel Thibault bc6f8a0868 add more valgrind suppression counterpart hace 12 años
  Samuel Thibault bb2c29edc8 Fix race on a system where reading an int wouldn't be atomic. hace 12 años
  Samuel Thibault 0fe579e3ec suppress more race warnings about termination, which is not racy hace 12 años
  Samuel Thibault e95a4579fd Avoid taking the global tag lock after having taken some tag spinlock. Fixes race in main/tag_wait_api hace 12 años
  Samuel Thibault c8ff9b7dc6 Get size before freeing interfaces hace 12 años
  Samuel Thibault f24ad3cbab rework deallocating handle memory chunks: we do not actually need to take the burden of going through the whole mc_list, since we already have the mc pointer. This also permits to properly mark per-worker chunks as deallocated hace 12 años
  Samuel Thibault dc4867f7e8 tasks[i] are not actually tasks, but arrays of tasks :) hace 12 años
  Samuel Thibault b1ae105330 With the locking fixes from r8919, _starpu_request_mem_chunk_removal does not need to access the child handle any more hace 12 años
  Samuel Thibault 3ff66999cd Add missing task cleaning hace 12 años
  Nathalie Furmento 5ea90822ca mpi/examples/stencil/stencil5.c: reduce values when quick-check is enabled hace 12 años
  Andra Hugo 2ac86bbbe1 fix lp2: take latency & bw into account hace 12 años
  Nathalie Furmento e3df31128b mpi/src/starpu_mpi_collective.c: only allocate memory if needed hace 12 años
  Nathalie Furmento 32cfa5ab67 mpi: New configure option --enable-mpi-progression-hook to enable the hace 12 años