Re: [PATCH] cmpc_acpi: Added support for Classmate PC ACPI devices.

[Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com>]

[-- Attachment #1: Type: text/plain, Size: 2202 bytes --]

On Wed, Sep 30, 2009 at 12:27:48PM -0600, Bjorn Helgaas wrote:
> On Wednesday 30 September 2009 11:51:43 am Thadeu Lima de Souza Cascardo wrote:
> > On Wed, Sep 30, 2009 at 09:50:38AM -0600, Bjorn Helgaas wrote:
> > > On Monday 28 September 2009 07:38:00 pm Thadeu Lima de Souza Cascardo wrote:
> > > > This add supports for devices like keyboard, backlight, tablet and
> > > > accelerometer.
> > > 
> > > Something about the PNP IDs bothered me yesterday, but I couldn't
> > > figure out what.  You're using:
> > > 
> > >   ACCE0000
> > >   TBLT0000
> > >   IPML200
> > >   FnBT0000
> > > 
> > > Normally these OEM-specific ACPI drivers claim PNP IDs like ATK0100,
> > > TOS6200, FUJ02B1, etc., that are obviously OEM-specific.
> > > 
> > > But the ones you're claiming seem pretty generic, like things
> > > we could imagine appearing on machines other than the Classmate.
> > > 
> > > So I guess my question is, "Are these really generic?  If so, should
> > > this be split into separate, non-Classmate named drivers, such as
> > > 'accelerometer', 'tablet', etc.?"
> > > 
> > 
> > Although I agree that {ATK,TOS,FUJ,IBM}* give pretty much a certainty
> > these are OEM-specific, I think we'd need some evidence that
> > {ACCE,TBLT,IPML,FnBT}* are generic enough to grant them a non-classmate
> > driver.
> > 
> > What I mean is that we should push this driver forward and turn these
> > into a generic driver if we get some evidence these are not
> > Classmate-specific. And that will happen when we hit some other system
> > with these devices.
> 
> Yep, I agree with you that it's not worth splitting things up until
> we find these devices on more systems.  Just out of curiosity, where
> did you find specs for these devices?  If there's a public spec, it'd
> be nice to have a pointer to it in the driver.
> 
> Bjorn

There was no public spec. I used as reference some source code released
as GPL. This implementation used proc files as user interface. Other
references are the dumped tables, which I am using to implement what is
very likely a rfkill device. We are still doing some tests before adding
it to the driver.

Regards,
Cascardo.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 197 bytes --]