skmsg: Avoid lock_sock() in sk_psock_backlog()

We do not have to lock the sock to avoid losing sk_socket,
instead we can purge all the ingress queues when we close
the socket. Sending or receiving packets after orphaning
socket makes no sense.

We do purge these queues when psock refcnt reaches zero but
here we want to purge them explicitly in sock_map_close().
There are also some nasty race conditions on testing bit
SK_PSOCK_TX_ENABLED and queuing/canceling the psock work,
we can expand psock->ingress_lock a bit to protect them too.

As noticed by John, we still have to lock the psock->work,
because the same work item could be running concurrently on
different CPU's.
Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20210331023237.41094-5-xiyou.wangcong@gmail.com

include/linux/skmsg.h

@@ -99,6 +99,7 @@ struct sk_psock {
 	void (*saved_write_space)(struct sock *sk);
 	void (*saved_data_ready)(struct sock *sk);
 	struct proto			*sk_proto;
+	struct mutex			work_mutex;
 	struct sk_psock_work_state	work_state;
 	struct work_struct		work;
 	union {
@@ -347,6 +348,7 @@ static inline void sk_psock_report_error(struct sk_psock *psock, int err)
 }
 
 struct sk_psock *sk_psock_init(struct sock *sk, int node);
+void sk_psock_stop(struct sk_psock *psock, bool wait);
 
 #if IS_ENABLED(CONFIG_BPF_STREAM_PARSER)
 int sk_psock_init_strp(struct sock *sk, struct sk_psock *psock);

net/core/skmsg.c

@@ -497,7 +497,7 @@ static int sk_psock_handle_skb(struct sk_psock *psock, struct sk_buff *skb,
 	if (!ingress) {
 		if (!sock_writeable(psock->sk))
 			return -EAGAIN;
-		return skb_send_sock_locked(psock->sk, skb, off, len);
+		return skb_send_sock(psock->sk, skb, off, len);
 	}
 	return sk_psock_skb_ingress(psock, skb);
 }
@@ -511,8 +511,7 @@ static void sk_psock_backlog(struct work_struct *work)
 	u32 len, off;
 	int ret;
 
-	/* Lock sock to avoid losing sk_socket during loop. */
-	lock_sock(psock->sk);
+	mutex_lock(&psock->work_mutex);
 	if (state->skb) {
 		skb = state->skb;
 		len = state->len;
@@ -529,7 +528,7 @@ static void sk_psock_backlog(struct work_struct *work)
 		skb_bpf_redirect_clear(skb);
 		do {
 			ret = -EIO;
-			if (likely(psock->sk->sk_socket))
+			if (!sock_flag(psock->sk, SOCK_DEAD))
 				ret = sk_psock_handle_skb(psock, skb, off,
 							  len, ingress);
 			if (ret <= 0) {
@@ -553,7 +552,7 @@ static void sk_psock_backlog(struct work_struct *work)
 			kfree_skb(skb);
 	}
 end:
-	release_sock(psock->sk);
+	mutex_unlock(&psock->work_mutex);
 }
 
 struct sk_psock *sk_psock_init(struct sock *sk, int node)
@@ -591,6 +590,7 @@ struct sk_psock *sk_psock_init(struct sock *sk, int node)
 	spin_lock_init(&psock->link_lock);
 
 	INIT_WORK(&psock->work, sk_psock_backlog);
+	mutex_init(&psock->work_mutex);
 	INIT_LIST_HEAD(&psock->ingress_msg);
 	spin_lock_init(&psock->ingress_lock);
 	skb_queue_head_init(&psock->ingress_skb);
@@ -631,7 +631,7 @@ static void __sk_psock_purge_ingress_msg(struct sk_psock *psock)
 	}
 }
 
-static void sk_psock_zap_ingress(struct sk_psock *psock)
+static void __sk_psock_zap_ingress(struct sk_psock *psock)
 {
 	struct sk_buff *skb;
 
@@ -639,9 +639,7 @@ static void sk_psock_zap_ingress(struct sk_psock *psock)
 		skb_bpf_redirect_clear(skb);
 		kfree_skb(skb);
 	}
-	spin_lock_bh(&psock->ingress_lock);
 	__sk_psock_purge_ingress_msg(psock);
-	spin_unlock_bh(&psock->ingress_lock);
 }
 
 static void sk_psock_link_destroy(struct sk_psock *psock)
@@ -654,6 +652,18 @@ static void sk_psock_link_destroy(struct sk_psock *psock)
 	}
 }
 
+void sk_psock_stop(struct sk_psock *psock, bool wait)
+{
+	spin_lock_bh(&psock->ingress_lock);
+	sk_psock_clear_state(psock, SK_PSOCK_TX_ENABLED);
+	sk_psock_cork_free(psock);
+	__sk_psock_zap_ingress(psock);
+	spin_unlock_bh(&psock->ingress_lock);
+
+	if (wait)
+		cancel_work_sync(&psock->work);
+}
+
 static void sk_psock_done_strp(struct sk_psock *psock);
 
 static void sk_psock_destroy_deferred(struct work_struct *gc)
@@ -665,12 +675,12 @@ static void sk_psock_destroy_deferred(struct work_struct *gc)
 	sk_psock_done_strp(psock);
 
 	cancel_work_sync(&psock->work);
+	mutex_destroy(&psock->work_mutex);
 
 	psock_progs_drop(&psock->progs);
 
 	sk_psock_link_destroy(psock);
 	sk_psock_cork_free(psock);
-	sk_psock_zap_ingress(psock);
 
 	if (psock->sk_redir)
 		sock_put(psock->sk_redir);
@@ -688,8 +698,7 @@ static void sk_psock_destroy(struct rcu_head *rcu)
 
 void sk_psock_drop(struct sock *sk, struct sk_psock *psock)
 {
-	sk_psock_cork_free(psock);
-	sk_psock_zap_ingress(psock);
+	sk_psock_stop(psock, false);
 
 	write_lock_bh(&sk->sk_callback_lock);
 	sk_psock_restore_proto(sk, psock);
@@ -699,7 +708,6 @@ void sk_psock_drop(struct sock *sk, struct sk_psock *psock)
 	else if (psock->progs.stream_verdict)
 		sk_psock_stop_verdict(sk, psock);
 	write_unlock_bh(&sk->sk_callback_lock);
-	sk_psock_clear_state(psock, SK_PSOCK_TX_ENABLED);
 
 	call_rcu(&psock->rcu, sk_psock_destroy);
 }
@@ -770,14 +778,20 @@ static void sk_psock_skb_redirect(struct sk_buff *skb)
 	 * error that caused the pipe to break. We can't send a packet on
 	 * a socket that is in this state so we drop the skb.
 	 */
-	if (!psock_other || sock_flag(sk_other, SOCK_DEAD) ||
-	    !sk_psock_test_state(psock_other, SK_PSOCK_TX_ENABLED)) {
+	if (!psock_other || sock_flag(sk_other, SOCK_DEAD)) {
+		kfree_skb(skb);
+		return;
+	}
+	spin_lock_bh(&psock_other->ingress_lock);
+	if (!sk_psock_test_state(psock_other, SK_PSOCK_TX_ENABLED)) {
+		spin_unlock_bh(&psock_other->ingress_lock);
 		kfree_skb(skb);
 		return;
 	}
 
 	skb_queue_tail(&psock_other->ingress_skb, skb);
 	schedule_work(&psock_other->work);
+	spin_unlock_bh(&psock_other->ingress_lock);
 }
 
 static void sk_psock_tls_verdict_apply(struct sk_buff *skb, struct sock *sk, int verdict)
@@ -845,8 +859,12 @@ static void sk_psock_verdict_apply(struct sk_psock *psock,
 			err = sk_psock_skb_ingress_self(psock, skb);
 		}
 		if (err < 0) {
-			skb_queue_tail(&psock->ingress_skb, skb);
-			schedule_work(&psock->work);
+			spin_lock_bh(&psock->ingress_lock);
+			if (sk_psock_test_state(psock, SK_PSOCK_TX_ENABLED)) {
+				skb_queue_tail(&psock->ingress_skb, skb);
+				schedule_work(&psock->work);
+			}
+			spin_unlock_bh(&psock->ingress_lock);
 		}
 		break;
 	case __SK_REDIRECT:

net/core/sock_map.c

@@ -1540,6 +1540,7 @@ void sock_map_close(struct sock *sk, long timeout)
 	saved_close = psock->saved_close;
 	sock_map_remove_links(sk, psock);
 	rcu_read_unlock();
+	sk_psock_stop(psock, true);
 	release_sock(sk);
 	saved_close(sk, timeout);
 }