Re: [PATCH v5 bpf-next 07/11] tcp: Migrate TCP_NEW_SYN_RECV requests at receiving the final ACK.

From:   Martin KaFai Lau <redacted>
Date:   Fri, 14 May 2021 18:13:05 -0700

On Mon, May 10, 2021 at 12:44:29PM +0900, Kuniyuki Iwashima wrote:

quoted

diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index e690d1cff36e..fe666dc5c621 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c

@@ -1075,10 +1075,38 @@ struct sock *inet_csk_complete_hashdance(struct sock *sk, struct sock *child,
 	if (own_req) {
 		inet_csk_reqsk_queue_drop(sk, req);
 		reqsk_queue_removed(&inet_csk(sk)->icsk_accept_queue, req);

In the migration case 'sk != req->rsk_listener', is sk the right
one to pass in the above two functions?

Good catch, 'sk' should be 'req->rsk_listener' here.
Thank you!

quoted

-		if (inet_csk_reqsk_queue_add(sk, req, child))
+
+		if (sk != req->rsk_listener) {
+			/* another listening sk has been selected,
+			 * migrate the req to it.
+			 */
+			struct request_sock *nreq;
+
+			/* hold a refcnt for the nreq->rsk_listener
+			 * which is assigned in reqsk_clone()
+			 */
+			sock_hold(sk);
+			nreq = reqsk_clone(req, sk);
+			if (!nreq) {
+				inet_child_forget(sk, req, child);
+				goto child_put;
+			}
+
+			refcount_set(&nreq->rsk_refcnt, 1);
+			if (inet_csk_reqsk_queue_add(sk, nreq, child)) {
+				reqsk_migrate_reset(req);
+				reqsk_put(req);
+				return child;
+			}
+
+			reqsk_migrate_reset(nreq);
+			__reqsk_free(nreq);
+		} else if (inet_csk_reqsk_queue_add(sk, req, child)) {
 			return child;
+		}
 	}
 	/* Too bad, another child took ownership of the request, undo. */
+child_put:
 	bh_unlock_sock(child);
 	sock_put(child);
 	return NULL;