Re: GSM dongle fails to connect

[Dan Williams <dcbw redhat com>]

On Thu, 2012-11-08 at 17:16 +0100, Aleksander Morgado wrote:
> On 08/11/12 16:52, Radu Rendec wrote:
> > On Thu, 2012-11-08 at 15:49 +0100, Aleksander Morgado wrote:
> >> There are different ways of detecting which port is to be used for
> >> control and which for data. In some huawei modems we base the logic on
> >> the GETPORTMODE result.
> > 
> > I see - it's much more complex than I thought :)
> > 
> >> Just reply with the ModemManager debug log attached, that's ok.
> > 
> > Ok, I attached 2 sets of log captures. The first set consits of only the
> > file 3g-modem-messages.log.gz. It's the first capture, which I made 10
> > days ago and it's actually extracted from /var/log/messages - I didn't
> > redirect the daemons output to any file.
> > 
> > The second set of captures is made a few minutes ago and contains
> > daemons output redirected to files.
> > 
> > Please let me know if there's any other information or test that would
> > be helpful for debugging.
> > 
> > Simple crazy idea: maybe the pppd timeout is too low? I noticed it takes
> > only 5 seconds of LCP timeouts before the connection is aborted.
> > 
> > Would it be helpful if I try to install the old Fedora 16 (the
> > distribution that used to work with the same 3g modem/sim) on another
> > system, test it again and make another set of captures if it works?
> > 
> 
> Your issue is exactly the one I linked in a previous mail. Your
> GETPORTMODE reply says:
> 
>  ^GETPORTMODE: TYPE: WCDMA: huawei,NDIS:0,PCUI:1,GPS:2
> 
> NDIS is reported but no MDM port. Once MM grabs the ports, it uses the
> same for control and data, which is not ok:
>   tty/ttyUSB1 primary
>   tty/ttyUSB0 secondary
>   tty/ttyUSB1 data
> 
> If you are able to recompile ModemManager yourself, please try with the
> '06-huawei-simple' branch from upstream git.

Branch looks fine to me, but I don't have a device that mixes up NDIS
and MDM.  So at least it doesn't regress on non-screwed-up Huawei
devices :)

Dan