Hi,
I'm using the latest of ofono and connman. I'm having an issue with a
PPP connection and async events. With the latest ofono code, I added
LE920A4 to plugins/telit.c. So it's using atmodem driver for gprs and
gprs-context. The modem is running in mode 0x1201, so below is the
mapping of aux and modem.
/dev/ttyUSB0 (telit) 255/255/255 [00] ==> (null) (null)
/dev/cdc-wdm0 (gobi) 255/255/255 [02] ==> (null) (null)
/dev/ttyUSB1 (telit) 255/0/0 [03] ==> (null) (null)
/dev/ttyUSB2 (telit) 255/0/0 [04] ==> (null) (null) -- for mode
0x1201, used for creating PPP connection
/dev/ttyUSB3 (telit) 255/0/0 [05] ==> (null) (null) -- for mode
0x1201, used as chat for AT commands
/dev/ttyUSB4 (telit) 255/0/0 [06] ==> (null) (null)
Initially after the PPP session is established, everything is fine. I
can ping using the ppp0 interface. But as soon as I get an async
event (+CREG, +CIEV, etc), ping stops working. If I leave ping
running, I'll eventually get a ppp_disconnect with Reason 5 and state
3. The events are logged on the aux channel, but I'm not sure if
they're coming on modem channel also.
This can happen without any associated NWDETACH (sometimes I'll just
see a +CREG after PPP is connected). In this case ppp_disconnect
calls ofono_gprs_context_deactivated followed by g_at_chat_resume. It
looks like at_chat_resume ends up calling io_disconnect because of the
below code.
if (g_at_io_get_channel(chat->io) == NULL) {
io_disconnect(chat);
return;
}
Is it expected that chat->io goes away because of the ppp_disconnect?
When this happens the connman ofono plugin still thinks there's a
network so it will repeatedly try to attach. But the attach fails as
it can't send AT commands anymore.
I've tried changing how events are handling while a PPP session is
active by changing settings similar to below -
g_at_chat_send(gcd->chat, "AT+CMER=1,0,0,0", none_prefix, NULL, NULL, NULL);
g_at_chat_send(gcd->chat, "AT+CGEREP=2,1", none_prefix, NULL, NULL, NULL);
g_at_chat_send(gcd->chat, "AT+CAOC=1", none_prefix, NULL, NULL, NULL);
The above changes are in gprs-context.c/at_cgdcont_cb right before
sending AT+CGDATA so I would expect them to suppress events on the
modem channel. But I still sometimes get a +CREG update which appears
to disrupt the PPP session. Any ideas?
BTW, I've trying using QMI on /dev/cdc-wdm0 using qmicli and
qmi-network without any luck. It just times out.
Thanks,
Daryl
_______________________________________________
ofono mailing list
[email protected]
https://lists.ofono.org/mailman/listinfo/ofono
