On Thu, Sep 19, 2013 at 05:42:45PM +0100, Mark Brown wrote: > On Wed, Sep 18, 2013 at 04:38:28PM +0100, Mark Rutland wrote: > > On Mon, Sep 16, 2013 at 04:19:53PM +0100, Guenter Roeck wrote: > > > > I think what you are saying is that describing a generic connector via > > > devicetree is not acceptable, even though it _does_ describe hardware. > > > I would have to describe a specific connector for a specific hardware > > > instead, which in turn would need its own driver. Is that correct ? > > > Regardless of how the connector is described, the block of hardware it > > connects to will have to be described, and some description of the > > connector will be necessary (either in the node for the block, or by > > phandle to a node for the connector). I agree that having a combined IP > > block + connector driver for each permutation is not good. > > Many of the things described only have passive components attached, or > things that otherwise don't need drivers - things like power inputs or > headphone connectors, they're mainly providing information to allow > userspace to behave differently (eg, display a charging indicator in the > UI). It's not 100% true but by and by large if detection is being done > using a GPIO it's probably something like that. Correct. However, gpio based 'detect' pins typically need debounce support which is not directly available through the gpio userspace API. I tried to add that earlier, but was told to use extcon instead as it provides the necessary infrastructure. Now it almost looks like I can not use it either because the required devicetree bindings may be considered unacceptable. Guenter -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html