Accellerometer ioctl

[Andy Green]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:

> More questions: What about the resume stuff? IIRC the idea was that the 
> accellerometers could be programmed to wake up on certain gestures^wevents -- 
> is that still possible and if so, how?

We didn't do any work on it yet, IIRC one of the motion sensors is a
viable wakeup source and the other isn't.  Mainly we wanted to make sure
we can turn them off so they don't eat power in suspend.

There is a threshold feature in the devices that will make an interrupt
if the force in any axis exceeds some limit, and it can interrupt on a
"tap", that's it.  Anything more you would need to wake and wait to see
if it was a gesture, although you would be dead to it for a second or
two while you woke up.  Really to do that properly you would need a
little MPU managing the sensors in suspend and doing the gesture
detection and we don't have it.

>> If you hit trouble or have questions please don't hesitate.  If you
>> think the driver should do some manipulation on the data first to cut
>> the report rate to userspace, that's really interesting too.
> 
> Well, properly dealing with accellerometers is a complex thing. I have added 
> an entry to the GSoC 2008 list of ideas. At the end of the day, you want 
> gestures, not data. Towards this we will probably need to adopt something 
> like tslib does.

Better that we do it on the kernel side.  But I didn't really understand
what it was we were doing with the data in userland yet.  So far we just
pass it through pretty raw.

- -Andy
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iD8DBQFH1VtPOjLpvpq7dMoRAjNaAJ9fG2NkFNsX77QwhV5I1vfiX66t2QCggx6Z
iRfRmlkoNlp+WJkhcxFTqu0=
=5tsI
-----END PGP SIGNATURE-----