[LU-13049] LNet: Handle shutdown properly

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.14.0
• Severity: 3
• Assignee: Amir Shehata (Inactive)
• Reporter: Amir Shehata (Inactive)

Description

There are two code paths which can lookup a peer while the system is shutting down:

1. net_shutdown_lndnet()>lnet_peer_tables_cleanup()>lnet_peer_table_del_rtrs_locked()->lnet_del_route()
2. lnet_mt_event_handler()>lnet_handle_recovery_reply()>lnet_find_peer_ni_locked()

In both of these cases the_lnet.ln_state might be shutting down. Currently lnet_get_peer_ni_locked() asserts on ln_stat == LNET_STATE_RUNNING.

This should be handled gracefully. If the state isn't running, then it should return NULL. Callers of the function handle NULL return code.