On Wed, Aug 30, 2017 at 04:58:29PM +0200, Andrew Lunn wrote: > > > > > Your example shows there's GPIO phandle *and* specifier. > > > > > > > > Would "GPIO specifier" be enough here? > > > > > > No, specifier is the cells following GPIO (or any other) phandle. > > > > So this should be "GPIO phandle and specifier of ...", is that correct? > > > > I have found very few (< 4) occurrences of this language in (lots of) > > '-gpios' > > property descriptions under Documentation/devicetree/bindings/. Is this a > > new > > requirement? > > Sometimes it is just easier to refer to another document: > > GPIO, as defined in Documentation/devicetree/binding/gpio/gpio.txt
Yes, and what I care about here is how many GPIOs, direction and active state. IOW, worry about the information necessary to validate a specific instance is correct. And hopefully someday we'll have a format parseable to do that checking, and all the free form text will be gone. Rob