can: j1939: prevent deadlock by moving j1939_sk_errqueue()

In the Linux kernel, the following vulnerability has been resolved:

can: j1939: prevent deadlock by moving j1939_sk_errqueue()

This commit addresses a deadlock situation that can occur in certain
scenarios, such as when running data TP/ETP transfer and subscribing to
the error queue while receiving a net down event. The deadlock involves
locks in the following order:

3
j1939_session_list_lock -> active_session_list_lock
j1939_session_activate
...
j1939_sk_queue_activate_next -> sk_session_queue_lock
...
j1939_xtp_rx_eoma_one

2
j1939_sk_queue_drop_all -> sk_session_queue_lock
...
j1939_sk_netdev_event_netdown -> j1939_socks_lock
j1939_netdev_notify

1
j1939_sk_errqueue -> j1939_socks_lock
__j1939_session_cancel -> active_session_list_lock
j1939_tp_rxtimer

CPU0 CPU1
---- ----
lock(&priv->active_session_list_lock);
lock(&jsk->sk_session_queue_lock);
lock(&priv->active_session_list_lock);
lock(&priv->j1939_socks_lock);

The solution implemented in this commit is to move the
j1939_sk_errqueue() call out of the active_session_list_lock context,
thus preventing the deadlock situation.