Skip to content

Conversation

devreal
Copy link
Contributor

@devreal devreal commented Jan 19, 2024

All but the udreg rcache module did not properly initialize the lock member in the mca_rcache_base_module_t parent, which triggers an assert on Mac OSX when trying to take that lock.

Also squashes some format string warnings (separate commit) in the gpusm rcache module. Clang-14 on Mac OSX complains about a mismatch between void* and char*.

Copy link
Member

@edgargabriel edgargabriel left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM.
I am wondering whether this could explain the issue that I saw with an application testcase last week, will rerun it once this is merged

@devreal devreal force-pushed the rcache_base_lock_init branch from cc03173 to 2e570a8 Compare January 19, 2024 15:22
@devreal
Copy link
Contributor Author

devreal commented Jan 19, 2024

I added the destructor calls and removed the warning fixes. I will move them into a separate PR.

Copy link
Member

@bosilca bosilca left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't see a valid reason to make to an object because:

  1. You don't use the regcache module as a reference counted object;
  2. No other framework is using the modules as derived-objects.

I would create a base_init function and let the different components call it directly.

@devreal devreal force-pushed the rcache_base_lock_init branch 2 times, most recently from 86dcd02 to 5f4b211 Compare January 19, 2024 21:14
@devreal
Copy link
Contributor Author

devreal commented Jan 19, 2024

@bosilca Makes sense. I made them inline functions but am happy to put them into base. Not sure what the right way to go here was.

@bosilca
Copy link
Member

bosilca commented Jan 19, 2024

It's a *_base_* function it should go into base (such that it is easy to know where it is located).

All but the udreg rcache module did not properly initialize the `lock`
member in the mca_rcache_base_module_t parent, which triggers
an assert on Mac OSX when trying to take that lock.

Signed-off-by: Joseph Schuchart <schuchart@icl.utk.edu>
@devreal devreal force-pushed the rcache_base_lock_init branch from 5f4b211 to f81fc3e Compare January 19, 2024 22:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants