yanglimingcn commented on code in PR #3145:
URL: https://github.com/apache/brpc/pull/3145#discussion_r2508530349
##########
src/brpc/rdma/rdma_endpoint.cpp:
##########
@@ -972,9 +1001,9 @@ ssize_t RdmaEndpoint::HandleCompletion(ibv_wc& wc) {
// Update window
uint32_t wnd_thresh = _local_window_capacity / 8;
- if (_window_size.fetch_add(acks, butil::memory_order_relaxed) >=
wnd_thresh
+ if (_remote_rq_window_size.fetch_add(acks,
butil::memory_order_relaxed) >= wnd_thresh
Review Comment:
This part receives an update on the rq queue size from Recv, so it would be
more appropriate to call this variable _rq_window_size, which corresponds to
_sq_window_size mentioned above.
##########
src/brpc/rdma/rdma_endpoint.cpp:
##########
@@ -972,9 +1001,9 @@ ssize_t RdmaEndpoint::HandleCompletion(ibv_wc& wc) {
// Update window
uint32_t wnd_thresh = _local_window_capacity / 8;
- if (_window_size.fetch_add(acks, butil::memory_order_relaxed) >=
wnd_thresh
+ if (_remote_rq_window_size.fetch_add(acks,
butil::memory_order_relaxed) >= wnd_thresh
Review Comment:
The meaning of `_remote_rq_window_size` is not entirely clear. On the
receiving end, `SendAck` means informing the sending end that data reception is
complete.
The sending end can now clear `snd_buffer`. The original logic assumed that
clearing `snd_buffer` would free up space for `sqe`, but in reality, even if
`snd_buffer` is cleared, space may not necessarily be freed up for `sqe`,
causing the problem.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
