After a Sunday debugging night I almost have the problem located (this does not mean I'll be able to solve it). Proper /dev/input/tablet/usb/x are being generated for each handled report, and input server add-ons are also being spawned for each one. From kernel land, WaitForReport at _ReadReport fails, but it was too late tonight and didn't came with any clue. The point it is, as I've completely copied MouseDevice, should this being happening too to usb mice? I'm going to do some tests, and need to check bug tracker, maybe someone already reported something similar. Regards Enrique