Modem wakeup...

andrzej zaborowski balrogg at
Tue Jul 24 15:07:06 CEST 2007

On 24/07/07, Nils Faerber <nils.faerber at> wrote:
> Arnaud Patard (Rtp) schrieb:
> > Nils Faerber <nils.faerber at> writes:
> >> Arnaud Patard (Rtp) schrieb:
> >>> Nils Faerber <nils.faerber at> writes:
> >>> I've no idea on how the modem is working too but at least I can see that
> >>> there's a irq handler for a "modem wakeup" interrupt. Unfortunately, I
> >>> don't see where the gpio is configured to do interrupts.
> >>>
> >>> Can you try adding :
> >>> s3c2410_gpio_cfgpin(GTA01_IRQ_MODEM, S3C2410_GPF1_EINT1);
> >>>
> >>> before set_irq_type(GTA01_IRQ_MODEM, IRQT_RISING); in
> >>> mach-gta01:gta01_machine_init() ?
> >> Something is very strangenow...
> >> If I add the above command the NEO does not boot-up anymore, "kernel
> >> panic, unable to mount rootfs, tried jffs2".
> > oh :(
> No problem, I know that this is development stuff and not final wisdom ;)
> >> I tried several times, both on V3 and V4.
> >> After I removed the command again everything returned to normal, device
> >> boots.
> >> What side effect could
> >>      s3c2410_gpio_cfgpin(GTA01_IRQ_MODEM, S3C2410_GPF1_EINT1);
> >> possibly have concerning the NAND Flash?
> >
> > too bad. My fault. sorry :(
> Really no problem!
> I rather try broken stuff than nothing at all. So also wrong tries are
> greatly appreciated! And besides... also failures serve a purpose: You
> learn how not to do it ;)
> > This should have been:
> > s3c2410_gpio_cfgpin(S3C2410_GPF1, S3C2410_GPF1_EINT1);

I would bet on either EINT0/GPF0 or EINT1/GPF1 to be the modem interrupt signal.

I don't know if the modem needs to be configured into some "sleep"
mode to start driving that signal. Maybe some command like AT+CMUX is
used for that, but it's unlikely. None of the TI-specific commands
that are printed by AT+CLAC seems to be related to this either (I
don't have the documentation, but I found that there is a publicly
available documentation for a modem called "Enfora Enabler III" that
uses many of the same non-standard commands).

If we can't work that out, it's always possible to just configure the
modem so that it doesn't print any unsolicited responses other than
the new call/new SMS indication ones, and set uart0 as a wake-up event
source. It would be interesting to see how the other smartphones (Treo
or HTC) do that, perhaps by running their ROMs in an emulator,
shouldn't be very hard.


More information about the openmoko-kernel mailing list