tcp/dccp: allow a connection when sk_max_ack_backlog is zero

[ Upstream commit 3479c75 ]

If the backlog of listen() is set to zero, sk_acceptq_is_full() allows
one connection to be made, but inet_csk_reqsk_queue_is_full() does not.
When the net.ipv4.tcp_syncookies is zero, inet_csk_reqsk_queue_is_full()
will cause an immediate drop before the sk_acceptq_is_full() check in
tcp_conn_request(), resulting in no connection can be made.

This patch tries to keep consistent with 64a1465 ("[NET]: Revert
incorrect accept queue backlog changes.").

Link: https://lore.kernel.org/netdev/[email protected]/
Fixes: ef547f2 ("tcp: remove max_qlen_log")
Signed-off-by: Zhongqiu Duan <[email protected]>
Reviewed-by: Kuniyuki Iwashima <[email protected]>
Reviewed-by: Jason Xing <[email protected]>
Reviewed-by: Eric Dumazet <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
Signed-off-by: Sasha Levin <[email protected]>

Author: pipchu

include/net/inet_connection_sock.h

@@ -282,7 +282,7 @@ static inline int inet_csk_reqsk_queue_len(const struct sock *sk)
 
 static inline int inet_csk_reqsk_queue_is_full(const struct sock *sk)
 {
-	return inet_csk_reqsk_queue_len(sk) >= READ_ONCE(sk->sk_max_ack_backlog);
+	return inet_csk_reqsk_queue_len(sk) > READ_ONCE(sk->sk_max_ack_backlog);
 }
 
 bool inet_csk_reqsk_queue_drop(struct sock *sk, struct request_sock *req);