prov/efa: Make DGRAM provider use new av_entry struct #10706
+390
−27
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
a-szegel
force-pushed
the
modify-efa-dgram-av-path
branch
from
c686ee3 to
d7a715b
Compare
|
This needs unit testing, but it should be ready to review. |
struct efa_ep_addr is used in both the DGRAM and RDM providers, so the structure definition should be in a common file, and not an RDM only file. Signed-off-by: Seth Zegelstein <[email protected]>
Update the DGRAM provider to use the new efa_base_av_entry structure. This change splits the AV logic for the DGRAM and the RDM providers. Splitting the av logic for DGRAM makes DGRAM's AV logic simpler and faster in a few ways. 1. This change removes the efa_con struct from the av_entry because the DGRAM protocol does not maintain an idea of a connection to the endpoint receiving the messages. It does not matter if the remote endpoint we were previously communicating with exits, and a new QP opens with the same AH (NIC), and QPN (Endpoint). As long as someone is there to post recv buffers for our messages, DGRAM provider is happy. 2. This changes gets rid of the current/prev av reverse lookup map (ahn/qpn -> av_entry), and creates a single reverse look up map because the dgram provider does not care about old connections. Signed-off-by: Seth Zegelstein <[email protected]>
a-szegel
force-pushed
the
modify-efa-dgram-av-path
branch
from
d7a715b to
c3d1cd4
Compare
|
A high level comment is that I want to get rid of |
|
I am wondering if it is cleaner to decouple rdm_peer with efa_conn. As far as I read the code you only access peer from coon during the av insert and remove path. You don't need rdm peer for reverse look up. It should be much less changes if you do in this way |
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.
Update the DGRAM provider to use the new efa_base_av_entry structure. This change splits the AV logic for the DGRAM and the RDM providers. Splitting the av logic for DGRAM makes DGRAM's AV logic simpler and faster in a few ways.
This change removes the efa_con struct from the av_entry because the DGRAM protocol does not maintain an idea of a connection to the endpoint receiving the messages. It does not matter if the remote endpoint we were previously communicating with exits, and a new QP opens with the same AH (NIC), and QPN (Endpoint). As long as someone is there to post recv buffers for our messages, DGRAM provider is happy.
This changes gets rid of the current/prev av reverse lookup map (ahn/qpn -> av_entry), and creates a single reverse look up map because the dgram provider does not care about old connections.