Merge tag 'linux-can-fixes-for-5.4-20191113' of...

Merge tag 'linux-can-fixes-for-5.4-20191113' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can

Marc Kleine-Budde says:

====================
pull-request: can 2019-11-13

this is a pull request of 9 patches for net/master, hopefully for the v5.4
release cycle.

All nine patches are by Oleksij Rempel and fix locking and use-after-free bugs
in the j1939 stack found by the syzkaller syzbot.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/can/core.h

@@ -65,5 +65,6 @@ extern void can_rx_unregister(struct net *net, struct net_device *dev,
 			      void *data);
 
 extern int can_send(struct sk_buff *skb, int loop);
+void can_sock_destruct(struct sock *sk);
 
 #endif /* !_CAN_CORE_H */

net/can/af_can.c

@@ -86,11 +86,12 @@ static atomic_t skbcounter = ATOMIC_INIT(0);
 
 /* af_can socket functions */
 
-static void can_sock_destruct(struct sock *sk)
+void can_sock_destruct(struct sock *sk)
 {
 	skb_queue_purge(&sk->sk_receive_queue);
 	skb_queue_purge(&sk->sk_error_queue);
 }
+EXPORT_SYMBOL(can_sock_destruct);
 
 static const struct can_proto *can_get_proto(int protocol)
 {

net/can/j1939/main.c

@@ -51,6 +51,7 @@ static void j1939_can_recv(struct sk_buff *iskb, void *data)
 	if (!skb)
 		return;
 
+	j1939_priv_get(priv);
 	can_skb_set_owner(skb, iskb->sk);
 
 	/* get a pointer to the header of the skb
@@ -104,6 +105,7 @@ static void j1939_can_recv(struct sk_buff *iskb, void *data)
 	j1939_simple_recv(priv, skb);
 	j1939_sk_recv(priv, skb);
  done:
+	j1939_priv_put(priv);
 	kfree_skb(skb);
 }
 
@@ -150,6 +152,10 @@ static void __j1939_priv_release(struct kref *kref)
 
 	netdev_dbg(priv->ndev, "%s: 0x%p\n", __func__, priv);
 
+	WARN_ON_ONCE(!list_empty(&priv->active_session_list));
+	WARN_ON_ONCE(!list_empty(&priv->ecus));
+	WARN_ON_ONCE(!list_empty(&priv->j1939_socks));
+
 	dev_put(ndev);
 	kfree(priv);
 }
@@ -207,6 +213,9 @@ static inline struct j1939_priv *j1939_ndev_to_priv(struct net_device *ndev)
 {
 	struct can_ml_priv *can_ml_priv = ndev->ml_priv;
 
+	if (!can_ml_priv)
+		return NULL;
+
 	return can_ml_priv->j1939_priv;
 }
 

net/can/j1939/socket.c

@@ -78,7 +78,6 @@ static void j1939_jsk_add(struct j1939_priv *priv, struct j1939_sock *jsk)
 {
 	jsk->state |= J1939_SOCK_BOUND;
 	j1939_priv_get(priv);
-	jsk->priv = priv;
 
 	spin_lock_bh(&priv->j1939_socks_lock);
 	list_add_tail(&jsk->list, &priv->j1939_socks);
@@ -91,7 +90,6 @@ static void j1939_jsk_del(struct j1939_priv *priv, struct j1939_sock *jsk)
 	list_del_init(&jsk->list);
 	spin_unlock_bh(&priv->j1939_socks_lock);
 
-	jsk->priv = NULL;
 	j1939_priv_put(priv);
 	jsk->state &= ~J1939_SOCK_BOUND;
 }
@@ -349,6 +347,34 @@ void j1939_sk_recv(struct j1939_priv *priv, struct sk_buff *skb)
 	spin_unlock_bh(&priv->j1939_socks_lock);
 }
 
+static void j1939_sk_sock_destruct(struct sock *sk)
+{
+	struct j1939_sock *jsk = j1939_sk(sk);
+
+	/* This function will be call by the generic networking code, when then
+	 * the socket is ultimately closed (sk->sk_destruct).
+	 *
+	 * The race between
+	 * - processing a received CAN frame
+	 *   (can_receive -> j1939_can_recv)
+	 *   and accessing j1939_priv
+	 * ... and ...
+	 * - closing a socket
+	 *   (j1939_can_rx_unregister -> can_rx_unregister)
+	 *   and calling the final j1939_priv_put()
+	 *
+	 * is avoided by calling the final j1939_priv_put() from this
+	 * RCU deferred cleanup call.
+	 */
+	if (jsk->priv) {
+		j1939_priv_put(jsk->priv);
+		jsk->priv = NULL;
+	}
+
+	/* call generic CAN sock destruct */
+	can_sock_destruct(sk);
+}
+
 static int j1939_sk_init(struct sock *sk)
 {
 	struct j1939_sock *jsk = j1939_sk(sk);
@@ -371,6 +397,7 @@ static int j1939_sk_init(struct sock *sk)
 	atomic_set(&jsk->skb_pending, 0);
 	spin_lock_init(&jsk->sk_session_queue_lock);
 	INIT_LIST_HEAD(&jsk->sk_session_queue);
+	sk->sk_destruct = j1939_sk_sock_destruct;
 
 	return 0;
 }
@@ -443,6 +470,12 @@ static int j1939_sk_bind(struct socket *sock, struct sockaddr *uaddr, int len)
 		}
 
 		jsk->ifindex = addr->can_ifindex;
+
+		/* the corresponding j1939_priv_put() is called via
+		 * sk->sk_destruct, which points to j1939_sk_sock_destruct()
+		 */
+		j1939_priv_get(priv);
+		jsk->priv = priv;
 	}
 
 	/* set default transmit pgn */
@@ -560,8 +593,8 @@ static int j1939_sk_release(struct socket *sock)
 	if (!sk)
 		return 0;
 
-	jsk = j1939_sk(sk);
 	lock_sock(sk);
+	jsk = j1939_sk(sk);
 
 	if (jsk->state & J1939_SOCK_BOUND) {
 		struct j1939_priv *priv = jsk->priv;
@@ -1059,51 +1092,72 @@ static int j1939_sk_sendmsg(struct socket *sock, struct msghdr *msg,
 {
 	struct sock *sk = sock->sk;
 	struct j1939_sock *jsk = j1939_sk(sk);
-	struct j1939_priv *priv = jsk->priv;
+	struct j1939_priv *priv;
 	int ifindex;
 	int ret;
 
+	lock_sock(sock->sk);
 	/* various socket state tests */
-	if (!(jsk->state & J1939_SOCK_BOUND))
-		return -EBADFD;
+	if (!(jsk->state & J1939_SOCK_BOUND)) {
+		ret = -EBADFD;
+		goto sendmsg_done;
+	}
 
+	priv = jsk->priv;
 	ifindex = jsk->ifindex;
 
-	if (!jsk->addr.src_name && jsk->addr.sa == J1939_NO_ADDR)
+	if (!jsk->addr.src_name && jsk->addr.sa == J1939_NO_ADDR) {
 		/* no source address assigned yet */
-		return -EBADFD;
+		ret = -EBADFD;
+		goto sendmsg_done;
+	}
 
 	/* deal with provided destination address info */
 	if (msg->msg_name) {
 		struct sockaddr_can *addr = msg->msg_name;
 
-		if (msg->msg_namelen < J1939_MIN_NAMELEN)
-			return -EINVAL;
+		if (msg->msg_namelen < J1939_MIN_NAMELEN) {
+			ret = -EINVAL;
+			goto sendmsg_done;
+		}
 
-		if (addr->can_family != AF_CAN)
-			return -EINVAL;
+		if (addr->can_family != AF_CAN) {
+			ret = -EINVAL;
+			goto sendmsg_done;
+		}
 
-		if (addr->can_ifindex && addr->can_ifindex != ifindex)
-			return -EBADFD;
+		if (addr->can_ifindex && addr->can_ifindex != ifindex) {
+			ret = -EBADFD;
+			goto sendmsg_done;
+		}
 
 		if (j1939_pgn_is_valid(addr->can_addr.j1939.pgn) &&
-		    !j1939_pgn_is_clean_pdu(addr->can_addr.j1939.pgn))
-			return -EINVAL;
+		    !j1939_pgn_is_clean_pdu(addr->can_addr.j1939.pgn)) {
+			ret = -EINVAL;
+			goto sendmsg_done;
+		}
 
 		if (!addr->can_addr.j1939.name &&
 		    addr->can_addr.j1939.addr == J1939_NO_ADDR &&
-		    !sock_flag(sk, SOCK_BROADCAST))
+		    !sock_flag(sk, SOCK_BROADCAST)) {
 			/* broadcast, but SO_BROADCAST not set */
-			return -EACCES;
+			ret = -EACCES;
+			goto sendmsg_done;
+		}
 	} else {
 		if (!jsk->addr.dst_name && jsk->addr.da == J1939_NO_ADDR &&
-		    !sock_flag(sk, SOCK_BROADCAST))
+		    !sock_flag(sk, SOCK_BROADCAST)) {
 			/* broadcast, but SO_BROADCAST not set */
-			return -EACCES;
+			ret = -EACCES;
+			goto sendmsg_done;
+		}
 	}
 
 	ret = j1939_sk_send_loop(priv, sk, msg, size);
 
+sendmsg_done:
+	release_sock(sock->sk);
+
 	return ret;
 }
 

net/can/j1939/transport.c

@@ -255,6 +255,7 @@ static void __j1939_session_drop(struct j1939_session *session)
 		return;
 
 	j1939_sock_pending_del(session->sk);
+	sock_put(session->sk);
 }
 
 static void j1939_session_destroy(struct j1939_session *session)
@@ -266,6 +267,9 @@ static void j1939_session_destroy(struct j1939_session *session)
 
 	netdev_dbg(session->priv->ndev, "%s: 0x%p\n", __func__, session);
 
+	WARN_ON_ONCE(!list_empty(&session->sk_session_queue_entry));
+	WARN_ON_ONCE(!list_empty(&session->active_session_list_entry));
+
 	skb_queue_purge(&session->skb_queue);
 	__j1939_session_drop(session);
 	j1939_priv_put(session->priv);
@@ -1042,12 +1046,13 @@ j1939_session_deactivate_activate_next(struct j1939_session *session)
 		j1939_sk_queue_activate_next(session);
 }
 
-static void j1939_session_cancel(struct j1939_session *session,
+static void __j1939_session_cancel(struct j1939_session *session,
 				 enum j1939_xtp_abort err)
 {
 	struct j1939_priv *priv = session->priv;
 
 	WARN_ON_ONCE(!err);
+	lockdep_assert_held(&session->priv->active_session_list_lock);
 
 	session->err = j1939_xtp_abort_to_errno(priv, err);
 	/* do not send aborts on incoming broadcasts */
@@ -1062,6 +1067,20 @@ static void j1939_session_cancel(struct j1939_session *session,
 		j1939_sk_send_loop_abort(session->sk, session->err);
 }
 
+static void j1939_session_cancel(struct j1939_session *session,
+				 enum j1939_xtp_abort err)
+{
+	j1939_session_list_lock(session->priv);
+
+	if (session->state >= J1939_SESSION_ACTIVE &&
+	    session->state < J1939_SESSION_WAITING_ABORT) {
+		j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS);
+		__j1939_session_cancel(session, err);
+	}
+
+	j1939_session_list_unlock(session->priv);
+}
+
 static enum hrtimer_restart j1939_tp_txtimer(struct hrtimer *hrtimer)
 {
 	struct j1939_session *session =
@@ -1108,8 +1127,6 @@ static enum hrtimer_restart j1939_tp_txtimer(struct hrtimer *hrtimer)
 		netdev_alert(priv->ndev, "%s: 0x%p: tx aborted with unknown reason: %i\n",
 			     __func__, session, ret);
 		if (session->skcb.addr.type != J1939_SIMPLE) {
-			j1939_tp_set_rxtimeout(session,
-					       J1939_XTP_ABORT_TIMEOUT_MS);
 			j1939_session_cancel(session, J1939_XTP_ABORT_OTHER);
 		} else {
 			session->err = ret;
@@ -1169,7 +1186,7 @@ static enum hrtimer_restart j1939_tp_rxtimer(struct hrtimer *hrtimer)
 			hrtimer_start(&session->rxtimer,
 				      ms_to_ktime(J1939_XTP_ABORT_TIMEOUT_MS),
 				      HRTIMER_MODE_REL_SOFT);
-			j1939_session_cancel(session, J1939_XTP_ABORT_TIMEOUT);
+			__j1939_session_cancel(session, J1939_XTP_ABORT_TIMEOUT);
 		}
 		j1939_session_list_unlock(session->priv);
 	}
@@ -1375,7 +1392,6 @@ j1939_xtp_rx_cts_one(struct j1939_session *session, struct sk_buff *skb)
 
  out_session_cancel:
 	j1939_session_timers_cancel(session);
-	j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS);
 	j1939_session_cancel(session, err);
 }
 
@@ -1572,7 +1588,6 @@ static int j1939_xtp_rx_rts_session_active(struct j1939_session *session,
 
 		/* RTS on active session */
 		j1939_session_timers_cancel(session);
-		j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS);
 		j1939_session_cancel(session, J1939_XTP_ABORT_BUSY);
 	}
 
@@ -1583,7 +1598,6 @@ static int j1939_xtp_rx_rts_session_active(struct j1939_session *session,
 			     session->last_cmd);
 
 		j1939_session_timers_cancel(session);
-		j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS);
 		j1939_session_cancel(session, J1939_XTP_ABORT_BUSY);
 
 		return -EBUSY;
@@ -1785,7 +1799,6 @@ static void j1939_xtp_rx_dat_one(struct j1939_session *session,
 
  out_session_cancel:
 	j1939_session_timers_cancel(session);
-	j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS);
 	j1939_session_cancel(session, J1939_XTP_ABORT_FAULT);
 	j1939_session_put(session);
 }
@@ -1866,6 +1879,7 @@ struct j1939_session *j1939_tp_send(struct j1939_priv *priv,
 		return ERR_PTR(-ENOMEM);
 
 	/* skb is recounted in j1939_session_new() */
+	sock_hold(skb->sk);
 	session->sk = skb->sk;
 	session->transmission = true;
 	session->pkt.total = (size + 6) / 7;
@@ -2028,7 +2042,11 @@ int j1939_cancel_active_session(struct j1939_priv *priv, struct sock *sk)
 				 &priv->active_session_list,
 				 active_session_list_entry) {
 		if (!sk || sk == session->sk) {
-			j1939_session_timers_cancel(session);
+			if (hrtimer_try_to_cancel(&session->txtimer) == 1)
+				j1939_session_put(session);
+			if (hrtimer_try_to_cancel(&session->rxtimer) == 1)
+				j1939_session_put(session);
+
 			session->err = ESHUTDOWN;
 			j1939_session_deactivate_locked(session);
 		}