Hi Somisetty,

On 27.10.2011 21:46, Daniel Wagner wrote:
>>>> +void connman_service_packet_stats(struct connman_service *service,
>>>> +   unsigned int *packets_transmitted)
>>>> +{
>>>> + struct connman_stats_data *data;
>>>> + data = &stats_get(service)->data;
>>>> +
>>>> + *packets_transmitted = (data->rx_packets + data->tx_packets);
>>>> +}
>>>> +
>>>
>>> I think you should add the appropriate notify callback and register
>>> session.c to the callbacks. This would adhere to the style we are using
>>> to inform about service.c changes.
>>>
>> Ok we shall make it as a notification.
> 
> Maybe if you are using the iptables code (Thomasz just added the match 
> support to
> it, which is needed for IDLETIMER) then the notify call back is propably not 
> needed
> anymore. 
> 
> So the first thing would be to find out how the iptables should look like for
> getting this use case working. Then we can figure how we can implement it in
> ConnMan. Thomasz is _the_ expert here :)

After thinking a bit more on this, I would like to propose following
steps. The whole traffic counting (idle timeout) is a difficult topic
and I have to work on this anyway.

So instead of you having to do this let me handle it. For implementing
the Priority flag you just can use a fixed timeout (e.g. using
g_timeout_add(..., 5)). Meanwhile I take care that the session code can
handle the idle timeout correctly with iptables.

Is this okay for you?

cheers,
daniel
_______________________________________________
connman mailing list
[email protected]
http://lists.connman.net/listinfo/connman

Reply via email to