ofi/common: fix code that broke sessions #12870
Merged
Conversation
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
With sessions initialization model (section 11.3 of MPI 4 standard) MPI may be initialized and finalized any number of times. This patch refactors code that was assuming a one shot init/finalize sequence for initializing Open MPI and its MCA param space The underlying problem with the replaced code was that when an app calls MPI_Session_finalize and there are no more sessions active, the MCA param space is destroyed. So if one does not build Open MPI to use dynamically load components, and a component is using static variables in a way that assumes the MCA param space is always preserved if a static variable is set to some value, then things break if a subsequent MPI_Session_init is invoked. Related to open-mpi#12869 Signed-off-by: Howard Pritchard <[email protected]>
hppritcha
force-pushed
the
fix_common_ofi_for_sessions
branch
from
a9a334d to
2f4ba1b
Compare
hppritcha
requested review from
jsquyres,
a team,
cpshereda and
edgargabriel
and removed request for
a team,
jsquyres and
cpshereda
edgargabriel
approved these changes
Oct 23, 2024
| } | ||
|
|
||
| if (0 > accelerator_rank_index) { | ||
|
|
||
| param = mca_base_var_find("opal", "opal_common", "ofi", "accelerator_rank"); |
There was a problem hiding this comment.
I think this looks good, there is a case that could be made that this should not be an ofi parameter, but an mca parameter of the accelerator framework (for a later PR)
There was a problem hiding this comment.
Ah good point. I was just keeping things the way they were. I've no objection to moving it in the end to accel framework.
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.
With sessions initialization model (section 11.3 of MPI 4 standard) MPI may be initialized and finalized any number of times.
This patch refactors code that was assuming a one shot init/finalize sequence for initializing Open MPI and its MCA param space
The underlying problem with the replaced code was that when an app calls MPI_Session_finalize and there are no more sessions active, the MCA param space is destroyed. So if one does not build Open MPI to use dynamically load components, and a component is using static variables in a way that assumes the MCA param space is always preserved if a static variable is set to some value, then things break if a subsequent MPI_Session_init is invoked.
Related to #12869