tcp: annotate lockless access to sk->sk_err

tcp_poll() reads sk->sk_err without socket lock held/owned.

We should used READ_ONCE() here, and update writers
to use WRITE_ONCE().

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

5 files changed, 13 insertions, 12 deletions

diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 288693981b00..01569de651b6 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -589,7 +589,8 @@ __poll_t tcp_poll(struct file *file, struct socket *sock, poll_table *wait)
 	}
 	/* This barrier is coupled with smp_wmb() in tcp_reset() */
 	smp_rmb();
-	if (sk->sk_err || !skb_queue_empty_lockless(&sk->sk_error_queue))
+	if (READ_ONCE(sk->sk_err) ||
+	    !skb_queue_empty_lockless(&sk->sk_error_queue))
 		mask |= EPOLLERR;
 
 	return mask;
@@ -3094,7 +3095,7 @@ int tcp_disconnect(struct sock *sk, int flags)
 	if (old_state == TCP_LISTEN) {
 		inet_csk_listen_stop(sk);
 	} else if (unlikely(tp->repair)) {
-		sk->sk_err = ECONNABORTED;
+		WRITE_ONCE(sk->sk_err, ECONNABORTED);
 	} else if (tcp_need_reset(old_state) ||
 		   (tp->snd_nxt != tp->write_seq &&
 		    (1 << old_state) & (TCPF_CLOSING | TCPF_LAST_ACK))) {
@@ -3102,9 +3103,9 @@ int tcp_disconnect(struct sock *sk, int flags)
 		 * states
 		 */
 		tcp_send_active_reset(sk, gfp_any());
-		sk->sk_err = ECONNRESET;
+		WRITE_ONCE(sk->sk_err, ECONNRESET);
 	} else if (old_state == TCP_SYN_SENT)
-		sk->sk_err = ECONNRESET;
+		WRITE_ONCE(sk->sk_err, ECONNRESET);
 
 	tcp_clear_xmit_timers(sk);
 	__skb_queue_purge(&sk->sk_receive_queue);
@@ -4692,7 +4693,7 @@ int tcp_abort(struct sock *sk, int err)
 	bh_lock_sock(sk);
 
 	if (!sock_flag(sk, SOCK_DEAD)) {
-		sk->sk_err = err;
+		WRITE_ONCE(sk->sk_err, err);
 		/* This barrier is coupled with smp_rmb() in tcp_poll() */
 		smp_wmb();
 		sk_error_report(sk);
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 8b5b6ca6617d..754ddbe0577f 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -4322,15 +4322,15 @@ void tcp_reset(struct sock *sk, struct sk_buff *skb)
 	/* We want the right error as BSD sees it (and indeed as we do). */
 	switch (sk->sk_state) {
 	case TCP_SYN_SENT:
-		sk->sk_err = ECONNREFUSED;
+		WRITE_ONCE(sk->sk_err, ECONNREFUSED);
 		break;
 	case TCP_CLOSE_WAIT:
-		sk->sk_err = EPIPE;
+		WRITE_ONCE(sk->sk_err, EPIPE);
 		break;
 	case TCP_CLOSE:
 		return;
 	default:
-		sk->sk_err = ECONNRESET;
+		WRITE_ONCE(sk->sk_err, ECONNRESET);
 	}
 	/* This barrier is coupled with smp_rmb() in tcp_poll() */
 	smp_wmb();
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 4f6894469b62..89daa6b953ff 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -596,7 +596,7 @@ int tcp_v4_err(struct sk_buff *skb, u32 info)
 		ip_icmp_error(sk, skb, err, th->dest, info, (u8 *)th);
 
 		if (!sock_owned_by_user(sk)) {
-			sk->sk_err = err;
+			WRITE_ONCE(sk->sk_err, err);
 
 			sk_error_report(sk);
 
@@ -625,7 +625,7 @@ int tcp_v4_err(struct sk_buff *skb, u32 info)
 
 	inet = inet_sk(sk);
 	if (!sock_owned_by_user(sk) && inet->recverr) {
-		sk->sk_err = err;
+		WRITE_ONCE(sk->sk_err, err);
 		sk_error_report(sk);
 	} else	{ /* Only an error on timeout */
 		WRITE_ONCE(sk->sk_err_soft, err);
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 71d01cf3c13e..f7e00d90a730 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -3699,7 +3699,7 @@ static void tcp_connect_init(struct sock *sk)
 	tp->rx_opt.rcv_wscale = rcv_wscale;
 	tp->rcv_ssthresh = tp->rcv_wnd;
 
-	sk->sk_err = 0;
+	WRITE_ONCE(sk->sk_err, 0);
 	sock_reset_flag(sk, SOCK_DONE);
 	tp->snd_wnd = 0;
 	tcp_init_wl(tp, 0);
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index 8823e2182713..b839c2f91292 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -67,7 +67,7 @@ u32 tcp_clamp_probe0_to_user_timeout(const struct sock *sk, u32 when)
 
 static void tcp_write_err(struct sock *sk)
 {
-	sk->sk_err = READ_ONCE(sk->sk_err_soft) ? : ETIMEDOUT;
+	WRITE_ONCE(sk->sk_err, READ_ONCE(sk->sk_err_soft) ? : ETIMEDOUT);
 	sk_error_report(sk);
 
 	tcp_write_queue_purge(sk);