net: fix a lockdep splat

We have for each socket :

One spinlock (sk_slock.slock)
One rwlock (sk_callback_lock)

Possible scenarios are :

(A) (this is used in net/sunrpc/xprtsock.c)
read_lock(&sk->sk_callback_lock) (without blocking BH)
<BH>
spin_lock(&sk->sk_slock.slock);
...
read_lock(&sk->sk_callback_lock);
...

(B)
write_lock_bh(&sk->sk_callback_lock)
stuff
write_unlock_bh(&sk->sk_callback_lock)

(C)
spin_lock_bh(&sk->sk_slock)
...
write_lock_bh(&sk->sk_callback_lock)
stuff
write_unlock_bh(&sk->sk_callback_lock)
spin_unlock_bh(&sk->sk_slock)

This (C) case conflicts with (A) :

CPU1 [A]                         CPU2 [C]
read_lock(callback_lock)
<BH>                             spin_lock_bh(slock)
<wait to spin_lock(slock)>
                                 <wait to write_lock_bh(callback_lock)>

We have one problematic (C) use case in inet_csk_listen_stop() :

local_bh_disable();
bh_lock_sock(child); // spin_lock_bh(&sk->sk_slock)
WARN_ON(sock_owned_by_user(child));
...
sock_orphan(child); // write_lock_bh(&sk->sk_callback_lock)

lockdep is not happy with this, as reported by Tetsuo Handa

It seems only way to deal with this is to use read_lock_bh(callbacklock)
everywhere.

Thanks to Jarek for pointing a bug in my first attempt and suggesting
this solution.

Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Tested-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Jarek Poplawski <jarkao2@gmail.com>
Tested-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

4 files changed, 8 insertions, 8 deletions

diff --git a/net/rds/tcp_connect.c b/net/rds/tcp_connect.c
index c397524c039..c519939e8da 100644
--- a/net/rds/tcp_connect.c
+++ b/net/rds/tcp_connect.c
@@ -43,7 +43,7 @@ void rds_tcp_state_change(struct sock *sk)
 	struct rds_connection *conn;
 	struct rds_tcp_connection *tc;
 
-	read_lock(&sk->sk_callback_lock);
+	read_lock_bh(&sk->sk_callback_lock);
 	conn = sk->sk_user_data;
 	if (conn == NULL) {
 		state_change = sk->sk_state_change;
@@ -68,7 +68,7 @@ void rds_tcp_state_change(struct sock *sk)
 			break;
 	}
 out:
-	read_unlock(&sk->sk_callback_lock);
+	read_unlock_bh(&sk->sk_callback_lock);
 	state_change(sk);
 }
 
diff --git a/net/rds/tcp_listen.c b/net/rds/tcp_listen.c
index 975183fe695..27844f231d1 100644
--- a/net/rds/tcp_listen.c
+++ b/net/rds/tcp_listen.c
@@ -114,7 +114,7 @@ void rds_tcp_listen_data_ready(struct sock *sk, int bytes)
 
 	rdsdebug("listen data ready sk %p\n", sk);
 
-	read_lock(&sk->sk_callback_lock);
+	read_lock_bh(&sk->sk_callback_lock);
 	ready = sk->sk_user_data;
 	if (ready == NULL) { /* check for teardown race */
 		ready = sk->sk_data_ready;
@@ -131,7 +131,7 @@ void rds_tcp_listen_data_ready(struct sock *sk, int bytes)
 		queue_work(rds_wq, &rds_tcp_listen_work);
 
 out:
-	read_unlock(&sk->sk_callback_lock);
+	read_unlock_bh(&sk->sk_callback_lock);
 	ready(sk, bytes);
 }
 
diff --git a/net/rds/tcp_recv.c b/net/rds/tcp_recv.c
index 1aba6878fa5..e4379740410 100644
--- a/net/rds/tcp_recv.c
+++ b/net/rds/tcp_recv.c
@@ -324,7 +324,7 @@ void rds_tcp_data_ready(struct sock *sk, int bytes)
 
 	rdsdebug("data ready sk %p bytes %d\n", sk, bytes);
 
-	read_lock(&sk->sk_callback_lock);
+	read_lock_bh(&sk->sk_callback_lock);
 	conn = sk->sk_user_data;
 	if (conn == NULL) { /* check for teardown race */
 		ready = sk->sk_data_ready;
@@ -338,7 +338,7 @@ void rds_tcp_data_ready(struct sock *sk, int bytes)
 	if (rds_tcp_read_sock(conn, GFP_ATOMIC, KM_SOFTIRQ0) == -ENOMEM)
 		queue_delayed_work(rds_wq, &conn->c_recv_w, 0);
 out:
-	read_unlock(&sk->sk_callback_lock);
+	read_unlock_bh(&sk->sk_callback_lock);
 	ready(sk, bytes);
 }
 
diff --git a/net/rds/tcp_send.c b/net/rds/tcp_send.c
index a28b895ff0d..2f012a07d94 100644
--- a/net/rds/tcp_send.c
+++ b/net/rds/tcp_send.c
@@ -224,7 +224,7 @@ void rds_tcp_write_space(struct sock *sk)
 	struct rds_connection *conn;
 	struct rds_tcp_connection *tc;
 
-	read_lock(&sk->sk_callback_lock);
+	read_lock_bh(&sk->sk_callback_lock);
 	conn = sk->sk_user_data;
 	if (conn == NULL) {
 		write_space = sk->sk_write_space;
@@ -244,7 +244,7 @@ void rds_tcp_write_space(struct sock *sk)
 		queue_delayed_work(rds_wq, &conn->c_send_w, 0);
 
 out:
-	read_unlock(&sk->sk_callback_lock);
+	read_unlock_bh(&sk->sk_callback_lock);
 
 	/*
 	 * write_space is only called when data leaves tcp's send queue if