[OLPC-devel] Why ACPI DSDT tables?

[David Woodhouse]

On Mon, 2006-07-10 at 13:00 -0400, Jim Gettys wrote:
> Unless there is someone signed up to do *all* the work involved in
> different kernel interfaces for battery, lid switch, etc, I think that
> David Zeuthen's observation that we may as well use ACPI for those
> standard interfaces wins.  Note that providing all those interfaces is
> a non-trivial amount of work.

No, it _is_ trivial -- /proc/apm provides the battery and AC info, and
stuff like the lid switch can be provided as keypresses.

We've been doing this for years -- it's what we did on the iPAQ, for
example.

It's also going to be _much_ easier just to do device drivers in the
kernel properly, instead of mucking about with ASL.

David didn't 'observe that we may as well use ACPI for those standard
interfaces'. He asked that we don't invent a _new_ kernel<->user
interface. I happen to think that this is an ideal time to come up with
an interface that all implementations (including APM, PMU and ACPI)
shall use -- but I'm not entirely adverse to just emulating /proc/apm
like everyone else does, either.

-- 
dwmw2