tcp: fix a possible lockdep splat in tcp_done()

syzbot found that if __inet_inherit_port() returns an error,
we call tcp_done() after inet_csk_prepare_forced_close(),
meaning the socket lock is no longer held.

We might fix this in a different way in net-next, but
for 5.4 it seems safer to relax the lockdep check.

Fixes: d983ea6f16b8 ("tcp: add rcu protection around tp->fastopen_rsk")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index b2ac4f0..42187a3 100644 (file)
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -3842,8 +3842,12 @@ void tcp_done(struct sock *sk)
 {
        struct request_sock *req;
 
-       req = rcu_dereference_protected(tcp_sk(sk)->fastopen_rsk,
-                                       lockdep_sock_is_held(sk));
+       /* We might be called with a new socket, after
+        * inet_csk_prepare_forced_close() has been called
+        * so we can not use lockdep_sock_is_held(sk)
+        */
+       req = rcu_dereference_protected(tcp_sk(sk)->fastopen_rsk, 1);
+
        if (sk->sk_state == TCP_SYN_SENT || sk->sk_state == TCP_SYN_RECV)
                TCP_INC_STATS(sock_net(sk), TCP_MIB_ATTEMPTFAILS);