Hi Jiri, Benjamin,

quoted

So the solution consists in relying inconditionaly to the quirk MULTI_INPUT for
hid-multitouch. Before registering the input device in hid-input, we can test if
it has been populated, and if not, then we simply don't register it. In order to
prevent a regression for drivers that do not fill the input device, we need to
add an other quirk.

Hi everybody,

I am in progress of reviewing this patchset, and so far don't have any 
major obejctions; I expect to have the review finished this week, so 3.10 
is definitely a realistic target.

I agree, I think it looks sound in general.

Just to set my own expectations correctly -- Henrik, are you going to 
review this patchset as well, or there is no need for me to wait for your 
feedback?

No need to wait any longer ;-)

Cheers,
Henrik