Re: Initial state of modems
- From: Dan Williams <dcbw redhat com>
- To: Aleksander Morgado <aleksander lanedo com>
- Cc: Networkmanager List <networkmanager-list gnome org>
- Subject: Re: Initial state of modems
- Date: Tue, 15 Nov 2011 12:45:52 -0600
On Tue, 2011-11-15 at 19:33 +0100, Aleksander Morgado wrote:
> > > i use a Fasttrack xTend FXT009 with an external power supply.
> > > i use MM on an embedded device and the modem is connected over usb.
> > > When i have an active connection with the modem, remove the power from
> > > the embedded device (so MM has no chance to cleanup anything) and then i
> > > restart the embedded device (and MM also restarts), MM can not handle
> > > the modem because the modem is still in connection mode.
> > > i talked to aleksander on irc and a simple solution is to send "+++"
> > > before any other at-command.
> > > any problems expected with this solution?
> >
> > It depends on the modem. '+++' is one of the ways to break out of
> > "data" mode and into "AT" mode, but it's only one of a few. Others
> > include dropping DTR for a while or "flashing" the port, and some of
> > this depends on the settings you send in the initialization string. One
> > question I've got though is if you're removing power from the modem,
> > does NM get a port removed signal from udev? That's the expected
> > sequence here. If you remove the device, the kernel should notice that,
> > send a remove event to MM (through udev) and MM should clean up the
> > device anyway. Is that not happening?
> >
>
> The case here is the opposite. ModemManager (or the whole PC) gets
> rebooted; while the modem itself is always powered on (and connected).
> Therefore, this issue is only applicable to modems with an external
> power supply, like the Fastrack and Cinterion/Siemens ones (at least).
> If sending '+++' works for this case, we could extend those plugins with
> a new custom init command just doing that.
>
> Although it seems to me that the modems should be able to detect that
> the link with the DTE is down, and drop the connection automatically in
> that case.
Yeah, though implementation is dicey. AT&Dx is one way to do this but
of course then we have to twiddle serial port bits and stuff too.
Closing the port should also cause a hangup and force the modem to drop
the data stream, but not all do that for various reasons. Adding +++ to
probing isn't a bad idea, and will be even better when we are just
probing for AT compatibility instead of the whole 6-level probing
scheme.
Dan
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]