vSomeIP-Lib 3.5.6

[duartenfonseca]

Changes

• routing_manager_client

  • Forcing the ON_AVAILABLE log and send subscription of requested services even if availability was
    previously known.
    • Remove pending subscription entry right after calling on_subscription from application impl - race
      condition
    • Add missing mutex on request_debounce_timeout_ck and reset request_debounce_timer_running_ if
      requests_to_debounce_ is empty - race condition (issue: missing request)
    • lock registration_state_mutex_ before sending subscribes and unsubscribes - race condition
    • re-request service availability triggers from the routing host when a client loses the connection
      to a service provider

• routing_manager_impl

  • Convert pending_sd_offers_ from vector to set to avoid repeated service entries.

• routing_manager_stub

  • Send a config_command to share the _client hostname with the _target application. the
    config_command message is only sent after the routing_info is sent, thereby guaranteeing
    that the routing_info is processed first.
  • Send ADD_CLIENT command to all the client registered, without considering if they are
    requesting services from this client.

• routing_manager_base

  • Added a change to add_known_client so that the order of calling config_command and
    routing_info_command would not matter. This change only has impact when the clients adds itself
    as a known_client and was made to safeguard its hostname.

• endpoints

  • local comunication
    • Force endpoint to perform port assignment when local network interface becomes unavailable.
    • Check on local_tcp_client and local_uds_client if socket is open before sending.
    • Do not call shutdown_and_close_socket if socket state is connecting - race condition
    • A condition variable was implemented on local-tcp_client endpoint to notify the stop process
      to exit the 10ms wait if an EOF is received in receive_cbk, since the queue is cleared in this case.
  • Ensuring the lifetime of message_buffer_ptr on uds.
  • external communication
    • A state machine was created to cascade the stop -> init + start The goal was to ensure that
      the udp_server endpoint waits on both unicast and multicast stop, and after they are both stopped,
      and the flag restarting is set, the init and start can be called. (issue: Bad File Descriptor)
    • Syncronize the join and leave process of udp server multicast sockets. also (issue: in some situations
      after STR, vsomeip will not join the multicast group used by service discovery).
    • leave and join multicast address only starts when network is ready. (issue: join/leave multicast was
      causing CPU load, because this process starts when network is not ready).
    • Remove use of linger-with-timeout in tse, which fixes blocking socket operations
    • Ensure "shutdown_and_close" is called on "stop".
  • local and external communication
    • Endpoints would start every time find_or_create_local would execute, even if endpoint already existed.
    • Fix is_sending_ status
    • Increase mutex_ scope to include was_not_connected flag on connect (issue: REMOTE_ERROR on response)
    • cancel_and_connect_cbk was updated to only report the connection as successfully terminated when
      cancel_and_connect_cbk is called without error. Previously, it was reporting the connection as successfully
      terminated even when there was a connection error, preventing the endpoint from being.
      restarted.

• netlink

  • Removes the interface index when we get a RTM_DELLINK.
  • Added an option to consider the reset of sd_route_set_ when interface goes down. This is enabled by default.

• security

  • For each received response/notification, call is_client_allowed_to_offer to check whether the sending client
    is allowed to offer the service. This is done only for internal TCP communication; for UDS the check is done
    with the bound client.

• service discovery

  • Fix multicast threshold for initial events.
  • Stop find/offer debounce timers when the service discovery is stopped.
  • Added a mecaism on start of SD, to try rejooin the multicast address if any message was received by 1.1s
    (default timer) (issue: in some situations after STR, vsomeip will not join the multicast group used by
    service discovery).

• IO threads

  • To mitigate IO threads exit and stalled threads, a solution in which the io_context object processing is
    ran for determined duration was developed. This feature is disable by default and can be enable by
    configurations.

• configurations

  • Added configuration to enable io thread mitigation.
  • Enable configuration of client-specific debouncing.
  • Allow global configuration of maximum (additional) dispatchers and dispatch time.
  • Unspecify catch exception for boost exceptions.
  • Added configuration to consider the reset of sd_route_set_ when interface goes down. This is enabled by
    default.
  • Enable global configuration of request debounce time.
  • Support multiple IPsec activation files per connection.

• application_impl

  • Fix host assignment with non-local routing: add SO_REUSEADDR to the non-local routing host check

• runtime

  • Remove the get_mutex_ from runtime_impl - This mutex is not needed, as the static initialization of the
    shared_ptr is atomic. (issue: tombstone in application shutdown)

---

This discussion was created from the release vSomeIP-Lib 3.5.6.