On Tue, Jul 24, 2018 at 12:10 PM Paul Burton <paul.burton@xxxxxxxx> wrote: > > The binding for the cascoda,ca8210 IEEE 802.15.4 (6LoWPAN) device > includes an extclock-gpio property which does not contain a gpio-list, > but is instead an integer representing a pin of the device itself. This > falls foul of the gpios_property check, for example: > > DTC arch/mips/boot/dts/img/pistachio_marduk.dtb > arch/mips/boot/dts/img/pistachio_marduk.dtb: Warning (gpios_property): > /spi@18100f00/sixlowpan@4: Missing property '#gpio-cells' in node > /clk@18144000 or bad phandle (referred from extclock-gpio[0]) > > Extend the checking for false-positives in prop_is_gpio() to detect this > case in addition to the existing nr-gpio case. The false-positive cases > are described by an array including a compatible string & property name. > A NULL compatible string indicates that the property may be present in > any node, otherwise the property is only allowed in a node compatible > with the given string. This allows us to whitelist the extclock-gpio > property for the cascoda,ca8210 device without allowing it anywhere > else. IMO the binding should be fixed. It wasn't reviewed and there are no dts files using it. I see several issues with it. Rob
