On 08/22/2013 06:45 AM, Rob Herring wrote: > On 08/21/2013 03:54 AM, Mark Rutland wrote: ... >> We should fix the binding documents to make clear which bindings require >> their reg properties at fixed indexes, and those which allow for >> arbitrarily ordered named reg entries. Named reg entries are really >> useful for blocks with optional components, and given we have drivers >> using them, they're already mandatory for some bindings. > > No, we should fix the bindings that have arbitrarily ordered reg > properties. There is no obvious reg examples of this that I see from a > quick scan of dts files. > > There is this questionable use of "empty" for interrupt-names: > > arch/arm/boot/dts/imx23.dtsi: > interrupts = <0 14 20 0 > 13 13 13 13>; > interrupt-names = "empty", "ssp0", "ssp1", "empty", > "gpmi0", "gpmi1", "gpmi2", "gpmi3"; > Well, there's been an assertion that bindings that define an ordered interrupts property shouldn't have a interrupt-names property, and bindings that define interrupt-names entries explicitly don't require a specific order of entries in the interrupts property. I assume from your statement, you don't agree, and think that interrupts must always be in a specific order, even if interrupt-names exists? If so, that'd make interrupt-names at least a bit useless and at worst horribly misleading, since people will assume it defines the order and that the order can be arbitrary. -- 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
