casaroli commented on PR #22089: URL: https://github.com/apache/nuttx/pull/22089#issuecomment-7043758553
We ran your cases. You are right. With this PR, when we send 1 MiB to our own address, the read-ahead of the echo server takes the IOBs down to the throttle, the server waits in send() and stops reading, and the transfer stops. With a default limit per connection, as you suggest, all five cases pass, and four local transfers at the same time pass too. We will change this PR to that limit. First we fix a bug that the limit shows: after a sender waits on the send limit, the next send() calls fail with ENOMEM. This also happens on master with NET_SEND_BUFSIZE=8192. The bad data node to node is a separate bug: after 8 retransmissions the sender frees the data of a write buffer but keeps the connection open, and the next data goes out with the old sequence number. We will fix it in another PR. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
