GSM modem doesn't power off when Neo does

Shawn Rutledge shawn.t.rutledge at
Tue Feb 19 23:35:29 CET 2008

On Feb 19, 2008 2:59 PM, andrzej zaborowski <balrogg at> wrote:
> That's a bit of a separate issue.  The current issue is that when you
> power-off the device it doesn't power off completely (not when you
> shut down gsmd, which may never have been running, or there may be
> more than one running when there's multiplexing).  gsmd is not

So that means, for example when you are using GPRS a second gsmd is
running, because it's multiplexed?  And then it is shut down when
you're done with GPRS?  What are the other use cases for multiplexing,
if any?

Anyway I suspect maybe gsmd is not killed, but exits by request in that case.

> requesting powering up of the modem, it assumes the modem is up the
> moment it starts, so it should also not deal with powering it down.

Seems to me it would be a good design for gsmd to have all the
responsibility to control the radio power.  That's hardware
abstraction: the problem is that there is one way to power on the
radio and another way to power it off, right?  well what if we put
gsmd in charge of it, make sure to finish debugging gsmd so it never
crashes, and provide an interface whereby any app can request gsmd to
set the power state.  (dbus or otherwise)  Oh but gsmd is supposed to
be portable... well so is the kernel, but then you need to have some
compile-time flags to enable workarounds for certain kinds of broken

Or the abstraction could even be pushed down into the kernel, so that
when you frob the /sys/whatever to turn off the radio it sends that
POFF command, and when you try to turn it on, it does it the other
way.  That way as long as the kernel is shut down normally rather than
crashing, it will still work.

More information about the gsmd-devel mailing list