On May 1, 2011, at 3:52 PM, Grant Erickson wrote: > On Apr 29, 2011, at 9:51 AM, Grant Erickson wrote: >> I am investigating all sort of odd state issues with connman and my user >> interface / application and the latest oddity is that when I issue: >> >> net.connman.Service.Disconnect >> >> and then: >> >> net.connman.Manager.SetProperty OfflineMode 1 >> >> I end up in a state in which no services are connected, yet connman still >> reports that State=online and ConnectedTechnologies = [ wifi ], despite the >> fact that there are no connected services or technologies. Is this normal, >> expected, as-designed behavior? > > I've done some additional instrumentation and debugging of this and the > high-level cause is that on connection to a service, > __connman_notifier_connect is called twice whereas on the service disconnect > path, __connman_notifier_disconnect is called only once, leading to the > __connman_notifier_count_connected inconsistency. > > I've not fully narrowed in to the root cause for this; however, > __connman_service_indicate_state seems to be called twice with new_state 4 > (ready) and type 1 (IPv4), leading to the two calls to > __connman_notifier_connect. The state transitions, incorrectly (?), seem to > be: > > association -> configuration -> ready -> online -> ready > > It seems to me that __connman_notifier_connect should only be called for the > ONLINE state, not the READY state. What's the design philosophy for calling > it from the READY state or is that a defect?
FWIW, I looked at whether there are any behavioral deltas between 0.71 and 0.73. As far as this issue is concerned, the behavior is the same. Still investigating the odd state transition behavior. Best, Grant _______________________________________________ connman mailing list [email protected] http://lists.connman.net/listinfo/connman
