[RFC] Probing ttys by default

Aleksander Morgado aleksander at aleksander.es
Fri Mar 21 04:33:55 PDT 2014


On Fri, Mar 21, 2014 at 12:11 PM, Lars Knudsen <larsgk at gmail.com> wrote:
> First, can someone with a deeper knowledge of how the modemmanager probes
> tell me (for a near term solution) if it's possible to report back some
> magic keyword to the initial "AT" that will let the modemmanager release a
> device instantly?
>
> Our problem is that we need to release very soon - and we want to support
> Chromebook and Linux (in general) with our education platform (
> www.empirikit.com ) - preferably without hacking our device to report as
> something else - but we might be forced to to give an acceptable user
> experience.


Hum... hard to tell how you can do that without needing to change MM.

The probing logic is more or less explained here:
http://www.freedesktop.org/software/ModemManager/api/1.2.0/ch02s02.html

Basically, if MM decides that needs to probe the device, it will start
with AT probing (sends "AT" several times) to see if the port is an
AT-capable port. If you don't want to get the port detected as AT, you
can just ignore the "AT" command and not reply anything. Note that if
you e.g. reply "ERROR" to that, the port will anyway get flagged as
AT-capable. After the AT probing fails, it will try QCDM probing,
sending a QCDM probe to see if the port replies correctly to it. If
nothing is replied to the probe, the port will finally be discarded.
This whole sequence may take up several seconds, as for each AT and
QCDM probe we retry a couple of times. BTW, all this is for MM 1.x,
the logic in MM 0.6.x is a bit different.

Which kind of TTY are you exposing? Is it a read-only tty (e.g. a
usual GPS NMEA port) or a read-write one?

-- 
Aleksander
https://aleksander.es


More information about the ModemManager-devel mailing list