prov/efa: Queue txes when handshake is enforced but not made #10115
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
darrylabbate
approved these changes
Jun 25, 2024
shijin-aws
force-pushed
the
queue_txes_before_handshake
branch
from
29c3a6b to
e413982
Compare
|
Latest push fixed unit test failure and added new unit tests to cover the changes |
shijin-aws
force-pushed
the
queue_txes_before_handshake
branch
2 times, most recently
from
69c0ec3 to
e311af6
Compare
|
@shijin-aws PR had the ucx rdm_tagged_peek test failure which was removed from #10124. If you rebase and re-push it will fix it. |
This function performs duplicate iteration over ope_queued_list. This is an oversight from 3cfc0bb, which merged the queued_rnr_list and the queued_ctrl_list. This patch fixes this bug by making the function only flush these two kinds of queued opes. Signed-off-by: Shi Jin <sjina@amazon.com>
Signed-off-by: Shi Jin <sjina@amazon.com>
…dshake is made Signed-off-by: Shi Jin <sjina@amazon.com>
Introduce efa_rdm_txe_enforce_handshake() function to handle the handshake triggering and the txe queueing. Signed-off-by: Shi Jin <sjina@amazon.com>
With a single ope_queued_list, we can remove queued_entry from the list equipotentially by checking if ope has any queued flags. Signed-off-by: Shi Jin <sjina@amazon.com>
When efa_rdm_ope_post_read has error, we should still continue after writing the txe/rxe error. When ctsdata pkt post returns FI_EAGAIN, we should still continue instead of break because opes may come from different eps. Signed-off-by: Shi Jin <sjina@amazon.com>
shijin-aws
force-pushed
the
queue_txes_before_handshake
branch
from
ec584b2 to
d56053d
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently, when a handshake is required but not made for a request to a given peer, we always return EAGAIN, which caused nasty user experience during the startup stage. This PR contains a series of commits that allows queuing the requests (when it's not for inject), and return 0 in such situation.