Sessions new #46
Open
Sessions new #46
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov> Signed-off-by: Howard Pritchard <howardp@lanl.gov>
the way fboxes works has issues for the sessions implementation, in particular tthe session finalize approach. what happens without this temporary fix is that if there is not some fully shcnronizing call prior to calling session_finalize, there are cases where a process may be probing its fast mailboxes for processes that are tearing down theses fboxes. That results in segfauls and sigbus problems. The fast box mechanism will need to be supplemented with some kind of shutdown mechanism that will tell the owner of the fboxes when its okay to actually tear them down. IN the interest of making progress using the sessions prototype with applications, shut down the fbox process for the prototype and return to coming up with a real fix at a later date. relates to #3 drop use of MPI_Flag what we're reading at the forum now for Sessions proposal has ditched MPI_Flags. Now using info object to MPI_Session_init to specify thread support level desired. Signed-off-by: Howard Pritchard <hppritcha@gmail.com>
and so much more Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <hppritcha@gmail.com>
remove extraneous method add patch to bypass excid exchange some fixes to excid bypass path sessions/fortran: add missing redefine for profiling interfaces sessions: move ompi_mpi_instance_cleanup_pml ahead in cleanup stack to avoid melt-downs in ompi_win_finalize, etc. under certain cases. fix attr subsys and ucx compile errors swat uninitialized variables fixes many test failures when not using --enable-debug and application calling MPI_Comm_create_errhandler, etc. sessions: put MPI_Intercomm_merge back to orig don't use MPI_Comm_create_from_group since its not supported by all PMLs. Signed-off-by: Howard Pritchard <howardp@lanl.gov>
minor patchups sessions: rebase over ULFM attributes patch up to adjust for MPI_FT key sessions: remove pmix glue functions for PMIx group operations. Now calling pmix directly sessions: adjustment for ucx initialization since it makes calls to hwloc at this point which needs PMIx and hence rte_init to have been called. sessions: rebase over ULFM OB1 changes sessions: rebase over ULFM portals adjustments sessions: rebase over ULFM minor ucx fixup sessions: rebase over ULFM changes to FT code ucx: improve configury to handle case of HAVE_UCP_ATTR_MEMORY_TYPES not defined remove redundeant check for gni provier remove some commented out print statements Signed-off-by: Howard Pritchard <hppritcha@gmail.com>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Turns out that the predefined datatypes are initialized statically, so the first time through calling ompi_datatype_finalize the magic cookie gets cleared if code is compiled in debug mode. This results in a second call to ompi_datatype_finalize to blow up with an assert since the cookie had been cleared in the first pass. Add a new OBJ_DESTRUCT variant which does not check/clear cookie even when the code is compiled in --enable-debug mode. Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
this patch fixes problems encountered trying to handle session_init following a previous finalize that shut down the bml. Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
ompi_mmpi_instance_unit_basic_count will drop to zero with the next call to ompi_mpi_instance_release. This patch may need a better solution. Signed-off-by: Howard Pritchard <howardp@lanl.gov>
this leads to meltdowns in a new session is inited as the predefined datatypes have been destructed. Signed-off-by: Howard Pritchard <hppritcha@gmail.com>
The fix to not destruct the predefined datatypes seem to have addressed the problem. This commit removes the hack patch. Signed-off-by: Howard Pritchard <howardp@lanl.gov>
per the MPI 4.0 standard related to #37 Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
this didn't make it into the MPI 4.0 standard related to #52 Signed-off-by: Howard Pritchard <howardp@lanl.gov>
forum decided attrs had enough problems already and shouldn't be extended to other mpi objects, so sessions attrs are not part of the MPI 4.0 standard related to #47 Signed-off-by: Howard Pritchard <howardp@lanl.gov>
processing to match with current master. Signed-off-by: Howard Pritchard <howardp@lanl.gov>
remove sessions (instance) attribute type a this is not part of the MPI 4.0 standard. Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
check related to #49 Signed-off-by: Howard Pritchard <howardp@lanl.gov>
this was not part of the MPI 4.0 standard related to #54 Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
not part of the MPI 4 standard so removing related to #54 Signed-off-by: Howard Pritchard <howardp@lanl.gov>
and some more windows from group stuff too. None of these made it into the sessions API for MPI 4.0 and its very likely they will be added in a MPI 5.0 release. Other approaches to topology constructs outside of sessions proposal are being explored. Related to #58 Related to #54 Signed-off-by: Howard Pritchard <howardp@lanl.gov>
for ULFM communicators. Related to #43 Signed-off-by: Howard Pritchard <howardp@lanl.gov>
comm_create_from_group and intercomm_create_from_groups related to #56 Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: tomhers <tom.herschberg@gmail.com>
Signed-off-by: tomhers <tom.herschberg@gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.