Re: ANN: apple motion sensor reverse engineered
On Fri, 2005-08-19 at 13:21 +0200, Johannes Berg wrote:
> Hi,
>
> I have successfully reverse engineered the AMS using a variety of python
> scripts, otool, libiberty (for C++ name demangling) and a bit of time.
> Most useful is an old script I have that uses graphviz' dot to make a
> flow graph out of asm.
>
> The OSX driver is remarkably well written. But maybe all OSX drivers
> are, I don't know :)
Haha... some are, and some are very bad :) Probably depends who wrote it
at Apple...
> What I found is documented here:
>
> http://johannes.sipsolutions.net/PowerBook/Apple_Motion_Sensor_Specification
>
> Some things like how to set the sensitivity are missing, but in normal
> operation the OSX driver do this.
>
> I have written a proof of concept driver that creates a HID device which
> fully works, but I suppose I cannot publish it since I disassembled the
> OSX driver. OTOH, it should be trivial to write a driver from this :)
Hrm... I'm not sure there is such a problem with disassembling, but
then, I'm no lawyer. Cleanroom is surely better. Hopefully, somebody
will pick up from your doc soon enough.
> Have fun. Maybe I'll see a real driver that can even handle interrupts
> (I have no idea how to get the interrupt GPIO stuff from the PROM/OF)
> when I get back online after the weekend? ;)
You can look at how the sound driver does.
Ben.
Reply to: