-
Notifications
You must be signed in to change notification settings - Fork 887
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Cleanup handling of OPAL_PREFIX vs PMIX_INSTALL_PREFIX #4052
Conversation
@bwbarrett This replaces the prior PR on the matter, and represents a rollup of all the discussion and proposed changes. Sorry for the confusion. |
The IBM CI (PGI Compiler) build failed! Please review the log, linked below. Gist: https://gist.github.com/40e76ee2ddf9bc2b93c53c656874df75 |
I think the pgi failure might have been a file system issue. Let's retry. |
The IBM CI (PGI Compiler) build failed! Please review the log, linked below. Gist: https://gist.github.com/548f9d030235992401b4dbc9551a3cc3 |
Humm the PGI compiler thing might be real. It looks like the |
Hmmm...let me check. I updated asm in PMIx master, but I bet the changes weren't ported to the branch prior to making this update. |
Here is a capture of the configure output from *** Assembler
checking dependency style of pgcc... pgcc
checking for perl... /usr/bin/perl
checking for __atomic builtin atomics... no
checking for processor support of __atomic builtin atomic compare-and-swap on 128-bit values... no
checking for __atomic builtin atomic compare-and-swap on 128-bit values with -mcx16 flag... no
checking for __sync builtin atomics... no
checking for 64-bit __sync builtin atomics... no
checking for processor support of __sync builtin atomic compare-and-swap on 128-bit values... no
checking for __sync builtin atomic compare-and-swap on 128-bit values with -mcx16 flag... no
configure: error: __sync builtin atomics requested but not found. |
I just checked and the atomics on the PMIx v2.0 branch matches the PMIx master, so it looks like something from OPAL didn't come over? |
I didn't find any substantive change in OPAL atomics that was missing from PMIx master. However, the opal_config_asm.m4 did change - notably, it removed that error line! So I'm guessing that is the "issue" here. I'm updating PMIx master and will bring it across to the PMIx 2.0 branch so it can land here. |
Wait a minute - we have PR #3757 waiting which adds that error so we abort, and folks indicate they want that behavior. So I'm guessing that PMIx is actually up-to-date minus the recent renaming of the opal_atomic_init stuff (which we'll pickup in the next round when something substantive changes in the atomics). So whatever it is, it looks like there actually is something broken in the PGI-based atomics |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Until we resolve the PMIx+PGI issue I'll put a hold on this. It's otherwise fine.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Now that we sorted out the PGI CI issue, I think this is good to go.
Or it is not needed anymore? |
We should probably revert 71da0fc from master to be consistent - we want to handle the OPAL_PREFIX case for the internal PMIx code, but leave anything else alone. |
This reverts commit 71da0fc. (per open-mpi#4052).
This reverts commit 71da0fc. (per open-mpi#4052). Refs: open-mpi#3980 Signed-off-by: Artem Polyakov <[email protected]>
Signed-off-by: Ralph Castain <[email protected]>
…hes between location directives for OPAL and PMIx. Provide a more helpful error message and error out if we find a mismatch. If any OPAL values are set and the PMIx equivalent is not, then transfer it. Do not clear PMIX_INSTALL_PREFIX from the daemon's launch environment Fixes #3980 Closes #4007 Refs #3985 Signed-off-by: Ralph Castain <[email protected]> (cherry picked from commit a239b4c)
Signed-off-by: Ralph Castain <[email protected]> (cherry picked from commit 53c9270)
Signed-off-by: Ralph Castain <[email protected]>
Signed-off-by: Ralph Castain <[email protected]>
MTT results:
|
Do a better job of detecting mismatches between location directives for OPAL and PMIx. Provide a more helpful error message and error out if we find a mismatch. If any OPAL values are set and the PMIx equivalent is not, then transfer it.
Update to track PMIx v2.0.1rc
Fixes #3980
Closes #4007
Closes #3985
Signed-off-by: Ralph Castain [email protected]