Re: gpe-handler / acpi on ASUSL2400D

[Sebastian Thias <post-OXl1b/uZKHizQB+pC5nmwQ@public.gmane.org>]

Am Mittwoch 26 März 2003 19:57 schrieben Sie:
> > From: Sebastian Thias [mailto:post-OXl1b/uZKHizQB+pC5nmwQ@public.gmane.org]
> > i'm a newbie on both acpi- and kernel-programming.
> > I've got an ASUSL2400D-laptop and kenel 2.5.65.
>
> Sounds like you're a quick learner. :)

Well, to be honest I started yesterday at noon. After a couple of dozens of
kernel-crashes things started to get working :-)

>
> > Upon pressing one of my laptops hotkeys, I receive an error
> > message about
> > a missing gpe-handler.
> > After fiddling a bit, I wrote a module that installs a
> > gpe-handler for 0x1D
> > (the one triggered by my hotkeys). The handler calls
> > "\_GPE._L1D" (using
> > acpi_evaluate_object ...). Only then can I receive notify-events for
> > "\_SB_.HOTK" and only then buttons like switching
> > LCD-Backlights off or
> > changing the brightness work.
> >
> > My question: Why do I have to install a gpe-handler by myself
> > if there is
> > a mehod "\_GPE._L1D" in dsdt. Is that the usual way to do such things?
>
> Is the button driver installed?

Yes. Powerbutton works. I compiled all drivers into the kernel. Lid state is 
also reported correctly, but I don't receive Lid events. If I try to install 
a notify-handler for the lid, I get returned, that there is already a handler 
present.

>
> The usual way is that you have a driver (e.g. for the lid button) that
> registers for notifications on the lid device. When the lid button is
> pressed, it generates a GPE, which the lid driver finds out about
> because the core calls the GPE method associated with the GPE (e.g.
> _L1D) and that method does a Notify() on the lid device. The lid device
> gets the notification and tells power policy via /proc/acpi/event.
>
> I'm not quite clear what is happening in your case. Can you elaborate
> further?
>
> Regards -- Andy

As I said before, the only "button-like" thing that is working, is the 
power-button. Any other button produces:
   "acpi_ev_gpe_dispatch[...] No handler present for GPE[00], disabling 
event."
(the gpe number is incorrect. The dispatch code always tells "00", regardless 
which gpe has really been triggered)

If I install my gpe-handler that calls _L1D "manually", I can also receive 
sleep-button events in /proc/acpi/events. The other hotkeys also work then 
(when a "\_SB_.HOTK" notify-handler is installed). Lid-events are still not
reported.
  If I interpret your answer correctly, it seems that the core does NOT call
the _L1D method. What could be the reason for this? Do you want to see
my dsdt?

Greetings
      Sebastian



-------------------------------------------------------
This SF.net email is sponsored by:
The Definitive IT and Networking Event. Be There!
NetWorld+Interop Las Vegas 2003 -- Register today!
http://ads.sourceforge.net/cgi-bin/redirect.pl?keyn0001en