multi-touch-dev team mailing list archive

["Henrik Rydberg" <rydberg@xxxxxxxxxxx>]

On Thu, Jun 14, 2012 at 05:55:22PM +0200, Florian Echtler wrote:
> On 11.05.2012 14:20, Henrik Rydberg wrote:
> >>>Better yet, from the protocol details, it looks like it would be
> >>>straight-forward to write an MT driver directly.
> >>Can you suggest a kernel driver which I could use as base?
> >Sure, drivers/input/mouse/bcm5974.c seems like a candidate. It also
> >uses fixed endpoints directly.
> I've finally found a bit of time to work on the kernel driver. I
> have a basic character driver up and running, however, I've hit a
> bit of a snag regarding the input subsystem.

Great!

> AFAICT, all drivers including the bcm5974 example you mentioned are
> using interrupt endpoints. However, the Surface 2.0 has bulk
> endpoints which, consequently, require continuous polling. Does the
> input subsystem support that, or is a separate kernel thread
> required?

I see, it makes sense, given the large amount of data. You can
probably use input-polldev with a reasonable framerate.

Henrik