[LU-15076] Add locking to ksocknal_find_timed_out_conn() for safe ksnc_tx_queue list processing

Tue, 23 Aug 2022 19:24:49 +0000

GFP happened in this peice of code from ksocknal_find_timed_out_conn():

 if ((!list_empty(&conn->ksnc_tx_queue) ||
                     conn->ksnc_sock->sk->sk_wmem_queued != 0) &&
                    ktime_get_seconds() >= conn->ksnc_tx_deadline) {
                        /* Timed out messages queued for sending or
                         * buffered in the socket's send buffer */
                        ksocknal_conn_addref(conn);
                        list_for_each_entry(tx, &conn->ksnc_tx_queue,
                                            tx_list)
                                tx->tx_hstatus = <----GFP here
                                        LNET_MSG_STATUS_LOCAL_TIMEOUT; \{code}


It looks like ksnc_tx_queue processing requires some additional locking.
For instance, like it done in ksocknal_write_callback():

spin_lock_bh(&sched->kss_lock);
...
if (!conn->ksnc_tx_scheduled && /* not being progressed */
!list_empty(&conn->ksnc_tx_queue)) { /* packets to send */
list_add_tail(&conn->ksnc_tx_list, &sched->kss_tx_conns);
...
spin_unlock_bh(&sched->kss_lock);

Whamcloud Community JIRA

[LU-15076] Add locking to ksocknal_find_timed_out_conn() for safe ksnc_tx_queue list processing