On Wed, 11 Apr 2012, Dmitry Torokhov wrote:

quoted

quoted

quoted

This patch adds driver for Elan I2C touchpad. These protocol of 
HID-I2C was defined by Microsoft. The kernel driver would use the 
multi-touch protocol format B.

If this device truly supports HID protocol shouldn't we define i2c-hid
transport, similar to usbhid?

I merely follow defined by Microsoft.
It was called "HID OVER I2C" by Microsoft.
(http://msdn.microsoft.com/en-us/library/windows/hardware/hh825917(v=vs.85).
aspx) But I agree to change from hid i2c to i2chid.
Do you have any better opinions?

I am not sure we are talking about same thing... I do not really care
whether it is named "hid i2c" or "i2chid". What I meant if the device
really speaks HID protocol, except that the transport is not USB or
bluetooth but I2C, then instead of writing a custom driver for this
particular part we need to add a new HID transport into
drivers/hid/i2c-hid that is similar to drivers/hid/usbhid/ and reuse the
rest of HID framework along with hid- multitouch.c which handles
HID-compliant multitouch devices.

I am CCing Jiri (HID maintainer) for additional input.

Yes, please.

In case you'd be looking into a bit less bloated example of how transport 
driver for HID subsystem looks like, see the Bluetooth one in 
net/bluetooth/hidp/ (and we don't currently have any other examples than 
these two :) ).

I'd be happy to merge it once you convert to to proper HID transport 
driver.

Could the event processing part then be integrated into hid-multitouch?

Thanks,

-- 
Jiri Kosina
SUSE Labs