Fix exploding idle ctxs array

[zhongchen530]

Fixes #13060 by ensuring strict matching between the requested option and option of the reused context.

[github-actions]

Hello! The Git Commit Checker CI bot found a few problems with this PR:

24c764b: fix exploding idle ctxs array

• check_signed_off: does not contain a valid Signed-off-by line

Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks!

[github-actions]

Hello! The Git Commit Checker CI bot found a few problems with this PR:

6917230: fix exploding idle ctxs array Signed-off-by: John ...

• check_signed_off: does not contain a valid Signed-off-by line

Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks!

[github-actions]

Hello! The Git Commit Checker CI bot found a few problems with this PR:

6917230: fix exploding idle ctxs array Signed-off-by: John ...

• check_signed_off: does not contain a valid Signed-off-by line

Please fix these problems and, if necessary, force-push new commits back up to the PR branch. Thanks!

[devreal]

This looks good to me but maybe someone more familiar with the oshmem component should take a look (@janjust)

[bosilca]

I don't understand how this fixes the issue as you replace a bitwise and (so any bits could match) with a strict matching equality. The & would trigger in all the cases where == is true, and more.

I reread the discussion in #13060 and I now think this is a correct patch but not for the reason claimed in the PR log, but because it ensures a strict matching on the context options. Additionally it also covers the case of selecting contexts with no options, which is a plus. I suggest you change the log to make this clear.

[zhongchen530]

I don't understand how this fixes the issue as you replace a bitwise and (so any bits could match) with a strict matching equality. The & would trigger in all the cases where == is true, and more.

I reread the discussion in #13060 and I now think this is a correct patch but not for the reason claimed in the PR log, but because it ensures a strict matching on the context options. Additionally it also covers the case of selecting contexts with no options, which is a plus. I suggest you change the log to make this clear.

I apologize for the delayed reply. By PR log, do you mean my commit message or the description of this PR?

[jsquyres]

By PR log, do you mean my commit message or the description of this PR?

I'm going to guess that @bosilca meant both: the git commit message and the PR description body.

[zhongchen530]

By PR log, do you mean my commit message or the description of this PR?

I'm going to guess that @bosilca meant both: the git commit message and the PR description body.

I have made changes to both. Please let me know if it still needs changes. Will this PR eventually get merged?

[jsquyres]

I have made changes to both. Please let me know if it still needs changes.

@bosilca Can you comment?

Will this PR eventually get merged?

That's the goal!

[janjust]

@zhongchen530 Please cherry-pick to v5.0.x