[PATCH 2/3] dt-bindings: i2c: apple, i2c: allow multiple compatibles

Sven Peter sven at svenpeter.dev
Mon Nov 22 12:48:39 PST 2021



On Mon, Nov 22, 2021, at 18:59, Janne Grunau wrote:
> On 2021-11-21 21:22:47 +0100, Sven Peter wrote:
>> Hi,
>> 
>> On Sun, Nov 21, 2021, at 18:15, Janne Grunau wrote:
>> > The intention was to have a SoC-specific and base compatible string
>> > to allow forward compatibility and SoC specific quirks,
>> >
>> > Fixes: df7c4a8c1b47 ("dt-bindings: i2c: Add Apple I2C controller bindings")
>> > Signed-off-by: Janne Grunau <j at jannau.net>
>> > Cc: Mark Kettenis <kettenis at openbsd.org>
>> > ---
>> 
>> Yeah, this should've been "apple,t8103-i2c", "apple,i2c" all along :/
>> Given that we have no i2c nodes in the dts yet and that this binding was
>> only added for -rc1 I think it's fine to just drop "apple,t8103-i2c"
>> here instead of marking it as deprecated and keeping it around forever
>> if Mark Kettenis also agrees.
>> 
>> >  Documentation/devicetree/bindings/i2c/apple,i2c.yaml | 9 +++++----
>> >  1 file changed, 5 insertions(+), 4 deletions(-)
>> >
>> > diff --git a/Documentation/devicetree/bindings/i2c/apple,i2c.yaml 
>> > b/Documentation/devicetree/bindings/i2c/apple,i2c.yaml
>> > index 22fc8483256f..f1cb96c08212 100644
>> > --- a/Documentation/devicetree/bindings/i2c/apple,i2c.yaml
>> > +++ b/Documentation/devicetree/bindings/i2c/apple,i2c.yaml
>> > @@ -20,9 +20,10 @@ allOf:
>> > 
>> >  properties:
>> >    compatible:
>> > -    enum:
>> > -      - apple,t8103-i2c
>> > -      - apple,i2c
>> > +    items:
>> > +      - enum:
>> > +        - apple,t8103-i2c
>> > +      - const: apple,i2c
>> 
>> Nit: the enum makes sense once we add t6000-i2c but right now
>> 
>> properties:
>>   compatible:
>>     items:
>>       - const: apple,t8103-i2c
>>       - const: apple,i2c
>> 
>> also works and look a bit less weird.
>
> I split it from change which in addition added "apple,t6000-i2c" as 
> second enum value. I have no strong preference but the weirdness will 
> hopefully vanish soon.

I'm slightly in favor of not introducing the enum. You'd have to fix the indentation
if you wanted to introduce it anyway [1] (checkpatch there also seems to complain about
the commit hash for the Fixes tag but that looks like a spurious warning to me since
df7c4a8c1b47 is in Linus' tree [2]):

yamllint warnings/errors:
./Documentation/devicetree/bindings/i2c/apple,i2c.yaml:25:9: [warning] wrong indentation: expected 10 but found 8 (indentation)


Best,

Sven

[1] http://patchwork.ozlabs.org/project/devicetree-bindings/patch/20211121171545.27402-3-j@jannau.net/
[2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=df7c4a8c1b47



More information about the linux-arm-kernel mailing list