tipc: fix flow control accounting for implicit connect
authorParthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Tue, 25 Sep 2018 16:21:58 +0000 (18:21 +0200)
committerDavid S. Miller <davem@davemloft.net>
Tue, 25 Sep 2018 17:47:37 +0000 (10:47 -0700)
In the case of implicit connect message with data > 1K, the flow
control accounting is incorrect. At this state, the socket does not
know the peer nodes capability and falls back to legacy flow control
by return 1, however the receiver of this message will perform the
new block accounting. This leads to a slack and eventually traffic
disturbance.

In this commit, we perform tipc_node_get_capabilities() at implicit
connect and perform accounting based on the peer's capability.

Signed-off-by: Parthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/tipc/socket.c

index 3f03ddd0e35b2f1b6acad1c788faee1976924b3b..b6f99b021d09b19a6fe0d6776d2c824a913cdacb 100644 (file)
@@ -1419,8 +1419,10 @@ static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dlen)
        /* Handle implicit connection setup */
        if (unlikely(dest)) {
                rc = __tipc_sendmsg(sock, m, dlen);
-               if (dlen && (dlen == rc))
+               if (dlen && dlen == rc) {
+                       tsk->peer_caps = tipc_node_get_capabilities(net, dnode);
                        tsk->snt_unacked = tsk_inc(tsk, dlen + msg_hdr_sz(hdr));
+               }
                return rc;
        }