Skip to content

Conversation

wzamazon
Copy link
Contributor

In allocate_state_shared(), state_region->base is used to
calculate peer->state when CPU atomics is not used.

Currently it is set only if accelerated btl is used,
but it is used for non-accelerated btl too, causing peer->state
to be set incorrectly.

This patch addressed the issue.

Signed-off-by: Wei Zhang [email protected]
(cherry picked from commit e7f9715)

In allocate_state_shared(), state_region->base is used to
calculate peer->state when CPU atomics is not used.

Currently it is set only if accelerated btl is used,
but it is used for non-accelerated btl too, causing peer->state
to be set incorrectly.

This patch addressed the issue.

Signed-off-by: Wei Zhang <[email protected]>
(cherry picked from commit e7f9715)
@wzamazon wzamazon changed the title osc/rdma: set state_region->base for non accelerated btl [v5.0.x] osc/rdma: set state_region->base for non accelerated btl Jun 10, 2022
@wzamazon wzamazon requested a review from awlauria June 10, 2022 13:34
@awlauria awlauria merged commit 7d0750a into open-mpi:v5.0.x Jun 10, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants