On 21/07/2024 11:45, Laurent Pinchart wrote: > Hi Krzysztof, > > On Sun, Jul 21, 2024 at 11:23:12AM +0200, Krzysztof Kozlowski wrote: >> On 19/07/2024 22:39, Laurent Pinchart wrote: >>> + maxItems: 1 >>> + >>> + interrupts: >>> + maxItems: 1 >>> + >>> + vdd-supply: true >>> + >>> + gpio-controller: true >>> + >>> + '#gpio-cells': >>> + const: 2 >>> + >>> + gpio-reserved-ranges: true >>> + >>> + "#pwm-cells": >>> + const: 3 >>> + >>> +required: >>> + - compatible >>> + - reg >>> + - gpio-controller >>> + - "#gpio-cells" >>> + - "#pwm-cells" >>> + >>> +allOf: >>> + - if: >>> + properties: >>> + compatible: >>> + contains: >>> + const: adi,adp5585-01 >>> + then: >>> + properties: >>> + gpio-reserved-ranges: false >>> + else: >>> + properties: >>> + gpio-reserved-ranges: >>> + items: >>> + - const: 5 >>> + - const: 1 >> >> Why reserved ranges are fixed? If they pins are *always* not accessible, >> then these are not GPIOs. This really looks incorrect. > > It's model-dependent. The ADP5585 has 11 pins that can be used as GPIOs. > They are named GPIO 1 to GPIO 11 in the datasheet. The -01 variant uses > the pin associated with GPIO 6 for a different purpose, so GPIO 6 is not > usable. That maps to index 5 as GPIO numbers in DT bindings are 0-based. > I've decided to handle that as a reserved GPIO range to keep the GPIO 7 > to GPIO 11 indices the same across all ADP5585 variants. Ah, I missed the fact that gpio-reserved-ranges are not required, so some of variants can just skip it. It's fine. Best regards, Krzysztof
