Using u-blox TOBY-R200, ModemManager not detecting already registered with network
Dan Williams
dcbw at redhat.com
Tue Feb 6 05:19:31 UTC 2018
On Mon, 2018-02-05 at 17:47 +0000, Matthew Starr wrote:
> > On Mon, 2018-02-05 at 15:45 +0000, Matthew Starr wrote:
> > > > On Thu, 2018-02-01 at 22:09 +0000, Matthew Starr wrote:
> > > > > I am using a TOBY-R200 which does not support USB Ethernet
> > > > > like
> > > > > TOBY-
> > > > > L2 models, but only supports AT interface over USB
> > > > > serial. When I
> > > > > use a NetworkManager profile to start a 3G/4G connection
> > > > > using
> > > > > ModemManager with the TOBY-R200, the registration step takes
> > > > > a
> > > > > long time (~37 seconds). The modem does eventually connect
> > > > > and
> > > > > everything work, but it takes too long for my use case. When
> > > > > I
> > > > > use a u-blox LISA
> > > > > U230 modem with the same software configuration, hardware
> > > > > platform, and SIM card, I get a connection quickly (~5
> > > > > second).
> > > > > When I put ModemManager in debug mode there is a difference
> > > > > in the
> > > > > connection between the two at the registration step as shown
> > > > > below. I have the same issue on ModemManager versions 1.4.14,
> > > > > 1.6.4, and 1.6.12.
> > > >
> > > > <snip>
> > > >
> > > > > On the TOBY R200, instead of realizing that the modem is
> > > > > already
> > > > > registered with the network, it runs AT+COPS=0 to reregister
> > > > > with
> > > > > the network. I confirmed that the modem registered with the
> > > > > network by checking the AT+CREG?, AT+CEREG?, and AT+CGREG?
> > > > > commands. Here is what ModemManager is outputting when
> >
> > connecting
> > > > > to the network:
> > > >
> > > > Do you see this line at all?
> > > >
> > > > Modem XXX: 3GPP Registration state changed (XXX -> registering)
> > > >
> > > > Basically, right after enabling the modem, ModemManager should
> > > > be
> > > > running initial registration checks. That will run a sequence
> > > > of
> > > > AT+CREG?, AT+CGREG?, and AT+CEREG? requests which should report
> > > > the
> > > > operator ID (eg MCC/MNC). This should all happen before the
> > > > connect, but perhaps the NM connect request is happening to
> > > > quickly.
> > >
> > > I see that line and waited until that line was output to command
> > > Network Manager to bring up the GSM connection. See log below
> > >
> > > >
> > > > Can you include a bit more of the ModemManager debug output
> > > > from
> >
> > the
> > > > time the modem is "enabling" until what you show below?
> >
> > Thanks; this is a bug in ModemManager but also odd behavior in the
> > modem. This could actually be triggered by your SIM though, given
> > the
> > description of the TOBY-R2 series response to the +COPS command in
> > ublox
> > document UBX-13002752 (R56). Anyway:
> >
> > (ttyACM1): <-- '<CR><LF>+CGREG:
> > 1,"xxxx","xxxxxxxx",6,"6C"<CR><LF><CR><LF>+CIEV:
> > 9,2<CR><LF><CR><LF>+CEREG: 4<CR><LF>'
> > (ttyACM0): --> 'AT+COPS=3,2;+COPS?<CR>'
> > (ttyACM0): <-- '<CR><LF>+COPS: 0<CR><LF><CR><LF>OK<CR><LF>'
> > modem_3gpp_load_operator_name(): loading Operator Name...
> > (ttyACM0): --> 'AT+COPS=3,0;+COPS?<CR>'
> > (ttyACM0): <-- '<CR><LF>+COPS: 0<CR><LF><CR><LF>OK<CR><LF>'
> >
> > Here MM is requesting the "set format only" COPS response, then
> > asking for
> > the current operator name and number. The modem replies with "0"
> > both times; usually devices are supposed to reply with the MCC/MNC
> > and
> > the operator name. Might be worth asking ublox why the device
> > responds
> > this way, but again it could be due to the SIM card.
> >
> > Anyway, the modem just isn't giving us any information about the
> > MCC/MNC and operator name, despite the fact that it's registered
> > with the
> > network and must already know these things. But MM currently looks
> > to
> > see if it has a valid operator MCC/MNC before short-circuiting the
> > automatic
> > registration AT+COPS=0 that you're running into.
> >
> > Could you try patch below and see if that makes things work better?
> > (Attached too just in case my mailer screws up the formatting...)
> >
> > Dan
> >
>
> What version of ModemManager is that patch for? I tried to apply it
> to version 1.6.12 (the newest version I have) and it failed on
> several hunks.
Was for git master, I'll do a backport to 1.6.12 for you tomorrow.
Dan
More information about the ModemManager-devel
mailing list