> These are just the issues I would like to implement drivers as standard I2C > driver model =) Attaching driver for one chip twice is ugly hack! While I'm not arguing the merits of using the standard I2C driver model, it won't actually help in this case since we would still need a structure representing shared state accessible by both the DVB and V4L2 subsystems. And that struct will need to be referenced counted, which is exactly what hybrid_tuner_request_state() does. In short, what you're proposing doesn't have any relevance to this case - it just moves the problem to some other mechanism for sharing private state between two drivers and having to reference count it. And at least in this case it's done the same way for all the tuner drivers (as opposed to different tuners re-inventing their own mechanism for sharing state between the different consumers). If you ever get around to implementing support for a hybrid device (where you actually have to worry about how both digital and analog share a single tuner), you'll appreciate some of these challenges and why what was done was not as bad/crazy as you might think. If anything, this little regression is yet another point of evidence that innocent refactoring and "cleanup" of existing code without really understanding what it does and/or performing significant testing can leave everybody worse off than if the well-intentioned committer had done nothing at all. Devin -- Devin J. Heitmueller - Kernel Labs http://www.kernellabs.com -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
