Skip to content

mpi-rma: embiggen second level mpi rma support infrastructure #13288

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

Merged
merged 1 commit into from
Jun 3, 2025

Conversation

hppritcha
Copy link
Member

The top level API files generated by the bindings generation code and the lower level OSC frameworks have been embiggened previously, but the middle layer within OMPI has not been till this patch.

There's a present for the RMA WG in this commit in the form of a comment. Something for them to work on.

Related to #13151

The top level API files generated by the bindings generation code
and the lower level OSC frameworks have been embiggened previously,
but the middle layer within OMPI has not been till this patch.

There's a present for the RMA WG in this commit in the form of
a comment.  Something for them to work on.

Related to open-mpi#13151

Signed-off-by: Howard Pritchard <[email protected]>
@hppritcha hppritcha requested a review from devreal June 3, 2025 15:15
Copy link
Contributor

@devreal devreal left a comment

Choose a reason for hiding this comment

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

Thanks @hppritcha for flagging this. We'll discuss this in the RMA WG 👍

Comment on lines +223 to +228
/*
* No this isn't really right but the MPI Forum RMA WG was spleeping during
* Big count proposal reading and didn't put in something for Table 12.1 of the
* MPI 5.0 standard for embiggened disp_unit, so here we go with int * in accordance
* with that table.
*/
Copy link
Contributor

Choose a reason for hiding this comment

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

@hppritcha hppritcha merged commit 2f3bfd9 into open-mpi:main Jun 3, 2025
15 checks passed
@jeffhammond
Copy link
Contributor

I don't know why you use ptrdiff_t for disp_unit when it's either 1 or the result of sizeof(element) and the return type of sizeof() is size_t in C.

I would change this back and make disp_unit greater than INT_MAX an error, because there is absolutely no use case for this.

mpiwg-rma/rma-issues#31

@hppritcha
Copy link
Member Author

With big count disp_unit argument is of type MPI_Aint for the "_c" routines. So that's the reason why we're using ptrdiff_t now.

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