On 02/06/13 01:06, Dan Williams wrote:
>>> Yeah, full debug output would be good then, since if MM sees that the
>>> > > device is not actually a modem, after probing it completely, it should
>>> > > never touch the device again unless the device goes away and re-appears.
>> > 
>> > Here comes the normal and the debug log. You are right, that
>> > modem-manager does stop probing after some time, but this is after
>> > 55s, not 10s. The process does not respawn (PID stays the same),
>> > but does the AT+GCAP probe nine times with a distance between four
>> > and ten seconds.
>> > 
>> > If there is a way to cut down the 55s more in the direction of 10s
>> > or 20s, this would be very helpful.
> Ok, the issue here is that after the first probe fails, MM should be
> smarter with plugins that require explicit probing (Iridium, Via CBP7).
> First, if the modem never replied with AT commands to anything, then
> there's no point in the other two plugins trying to probe.  Second,
> these two plugins don't care about QCDM, so there's no point in having
> their probes also try QCDM.

That is exactly what we do in MM git master now. AT probing failures are
detected early and we mark the ports as not being AT, so no further AT
commands are sent even if plugins want them. And also, plugins specify
whether they want QCDM probing or not.

-- 
Aleksander
_______________________________________________
networkmanager-list mailing list
[email protected]
https://mail.gnome.org/mailman/listinfo/networkmanager-list

Reply via email to