[PATCH] add support for ACPI IBM hotkey events

David Zeuthen david at fubar.dk
Mon Mar 26 12:58:04 PDT 2007


On Mon, 2007-03-26 at 12:13 +0200, Danny Kukawka wrote:
> On Freitag, 19. Januar 2007, Danny Kukawka wrote:
> > On Friday 19 January 2007 04:35, Matthew Garrett wrote:
> > > On Sun, Jan 14, 2007 at 07:10:08PM +0100, Danny Kukawka wrote:
> > > > Since we already propagate these events for normal ACPI events via HAL
> > > > (as we report the related buttons as devices) and also provide such
> > > > events for Toshiba machines, I would see them also for IBM until this
> > > > _all_ will go in the future through the input layer.
> > >
> > > I believe that this is queued for 2.6.21. There's no real point in
> > > adding it in order to just remove it again, especially if it encourages
> > > people to depend on these event names.
> >
> > Yes, but this is not there atm. What I would see as minimum that we get the
> > sleep/s2disk(hibernate) events as we already have from normal ACPI events.
> > I don't see why we should not emit signals for them.
> >
> > Btw. nobody needs to depend on this signals, and there is also no need that
> > HAL should depend on kernel 2.6.21, where 2.6.20 is not release currently.
> 
> I checked this with 2.6.21rc5 and could not see that this now go trough the 
> input layer. The events are not available in an other way than via ACPI 
> (except sleep and power button), or do I miss something?
> 
> IMO we should apply the patch to 0.5.9 if there is nothing in 2.6.21.

Some thoughts

 - what will happen when a future kernel release includes support for
   these buttons via the input layer? My guess is that we'll get
   duplicate events... 

   I guess having this functionality available as a build time feature
   will help mitigate that; e.g. we punt the problem to distributors to
   rebuild a new hal package when they rebase to a kernel that has this
   support. That's not ideal though.

   (and the Fedora folks, like davej, likes to provide very invasive
   kernel updates for stable Fedora releases (say, going from 2.6.18
   to 2.6.19) so I guess it means closer collaboration between people
   managing a distro. In reality I'm unsure we would get this right
   in Fedora, e.g. whether me and davej (or Chuck) will manage to
   get the HAL package rebuilt when a newer kernel is available.
   But that's another thing. Dave, your thoughts?)

   As such we should probably leave the the build time option to
   "off" so distro pkg maintainers at least have to make a conscious
   decision to support this.)

   This means we need a patch for Toshiba buttons too - to make it a
   build-time option. Both should default to "off". I'd turn both
   on in Fedora and keep an eye on what's going into Fedora. Other
   distro package maintainers would have to do the same.

 - like with toshiba, it should probably include an fdi file to add
   the 'button' capability to computer

 - Just emit all events on computer

     David




More information about the hal mailing list