OMPI: Notified RMA ops Public API#9
Open
joe-explr wants to merge 76 commits intodevreal:notified-rmafrom
Open
Conversation
|
Hello! The Git Commit Checker CI bot found a few problems with this PR: 8d7ea3b: Public APis to makefile
0b6b3c3: Public APIs for put_withnotify. get_with_notify
Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks! |
joe-explr
force-pushed
the
notified-rma-sm
branch
3 times, most recently
from
5a9c010 to
f79b94b
Compare
|
Hello! The Git Commit Checker CI bot found a few problems with this PR: 8ba738e: Edits
Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks! |
joe-explr
force-pushed
the
notified-rma-sm
branch
from
8ba738e to
80fef14
Compare
devreal
approved these changes
Nov 20, 2025
|
Hello! The Git Commit Checker CI bot found a few problems with this PR: dff9ea5: Public APIs for:
Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks! |
joe-explr
force-pushed
the
notified-rma-sm
branch
from
dff9ea5 to
6519bd0
Compare
joe-explr
commented
Dec 5, 2025
joe-explr
commented
Dec 5, 2025
Address sanitizer helps us catch memory bugs even if they don't manifest into faults right away. The instrumention incurs some overhead so this is run on a reduced set of mpi4py runs. Also tests `ompi_info` and `mpicc`. Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
The PMIx_Fence_nb function can return PMIX_OPERATION_SUCCEEDED to indicate that the function was executed atomically and the callback function will therefore not be called. The PMIx Standard lists a few reasons why this can happen, but the point here was to fix usage to properly handle that possibility. Signed-off-by: Ralph Castain <rhc@pmix.org>
Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
The variable will go out of scope and ASAN flags this in ompi_info. Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
without this patch I keep seeing this compiler warning when building with newer gcc's.
shmem_put_nb.c:230:6: warning: no previous prototype for ‘shmemx_alltoall_global_nb’ [-Wmissing-prototypes]
void shmemx_alltoall_global_nb(void *dest,
^~~~~~~~~~~~~~~~~~~~~~~~~
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
This commit adds notification support to the OSC SM component by implementing the put_with_notify, get_with_notify, rput_with_notify, and rget_with_notify functions. These functions perform the same operations as their non-notify counterparts but also increment notification counters after the data transfer completes. The changes include: - Added function pointer types for notify variants in osc.h - Added function prototypes in osc_sm.h - Implemented the notify functions in osc_sm_comm.c - Updated the module template to register the new functions - Removed TODO comments that have been addressed Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
put_with_notify get_with_notify Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
put_with_notify
get_with_notify
Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
…for a single and multi rank window. Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
joe-explr
force-pushed
the
notified-rma-sm
branch
from
4e90525 to
f0ee69f
Compare
…has_main pmix/prrte:advance shas
…uce_topo coll/acoll: Fix sbuf handling in reduce_topo
when fetching PMIX_GROUP_LOCAL_CID values from pmix server. Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Signed-off-by: Shachar Hasson <shasson@nvidia.com> Signed-off-by: Thomas Vegas <tvegas@nvidia.com>
Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
Signed-off-by: Orion Poplawski <orion@nwra.com>
open-mpi#13721) Signed-off-by: Orion Poplawski <orion@nwra.com>
It indicates if any accelerator support has been build in, and if we need all the frameworks dealing with such devices. Dont build the smcuda BTL, coll accelerator, SMSC accelerator, rcache gpusm and rgpusm without accelerators support. Signed-off-by: George Bosilca <gbosilca@nvidia.com>
PMIX returns success and does not set the output to NULL while looking for an optional key. Thus, to prevent segfaults we need to set the output value to a known value before. Signed-off-by: George Bosilca <gbosilca@nvidia.com>
Signed-off-by: George Bosilca <gbosilca@nvidia.com>
devreal
reviewed
Mar 6, 2026
ompi/mca/osc/osc.h
Outdated
| ompi_osc_base_module_free_fn_t osc_free; | ||
|
|
||
| ompi_osc_base_module_put_fn_t osc_put; | ||
| ompi_osc_base_module_put_notify_fn_t osc_put_notify; |
Owner
There was a problem hiding this comment.
Maybe we should move these to the end of the struct. That avoids creating gaps in the struct that the other osc components need to fill (for now). It's not just osc/ubcl, it's also osc/ucx, osc/rdma, and osc/portals
OMPI/RTE: Modify job name printing to use thread local storage
…thout_accelerator Avoiding building support for any accelerator-based modules without accelerators
Signed-off-by: Joseph Antony <jajoseph.antony18@gmail.com>
…eph.antony18@gmail.com>
… <jajoseph.antony18@gmail.com>
The newly added code to support shared memory queries in osc/rdma had the check for MPI_PROC_NULL too late, which caused an out-of-bounds access into the peer array. Signed-off-by: Joseph Schuchart <joseph.schuchart@stonybrook.edu>
Fix brace initialization
Drop __opal_attribute_always_inline__ for mca_part_persist_start
…sions Comm create from group: improve debug statements
…roc-null osc/rdma: Fix handling of MPI_PROC_NULL in shared_query
Signed-off-by: Matthew Whitlock <mwhitlo@sandia.gov>
…ailed pml/ob1: Return MPI_ERR_PROC_FAILED on unmatched I(m)probes when appropriate
remove btl-smcuda,rcache-gpusm,rcache-rgpusm from the list of components that have to be compiled as dsos at all times. The components do not contain any references/function calls to a GPU software stack anymore, everything is based off the accelerator framework APIs. Signed-off-by: Edgar Gabriel <Edgar.Gabriel@amd.com>
…smcuda opal/mca: remove smcuda from dso list
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
16 participants
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.
Summary of Changes: