Re: [PATCH 3/3] dt-bindings: i2c: realtek,rtl9301-i2c: extend for RTL9310 support

[Krzysztof Kozlowski <krzk@kernel.org>]

On 01/07/2025 14:34, Jonas Jelonek wrote:
> Hi Krzysztof,
> 
> 
> On 01.07.2025 13:33, Krzysztof Kozlowski wrote:
>> On 01/07/2025 11:17, Jonas Jelonek wrote:
>>> This extends the dt-bindings for the I2C driver for RTL9300 to account
>> Please do not use "This commit/patch/change", but imperative mood. See
>> longer explanation here:
>> https://elixir.bootlin.com/linux/v5.17.1/source/Documentation/process/submitting-patches.rst#L95
> I'll fix this in v2.
>>> for the added support for RTL9310 series.
>>>
>>> A new property is added to explicitly set the SCL num/hardware instance
>>> of the controller that is used. In contrast to RTL9300 the driver needs
>>> to know that explicitly for RTL9310 because the SCL selection is now in
>>> a global register instead of a master-specific register.
>>>
>>> The regex for child-node address is adjusted to account for the fact
>>> that RTL9310 supports 12 instead of 8 SDA lines.
>>>
>>> A single generic compatible "realtek,rtl9310-i2c" is added. To best
>>> knowledge, all existing SoCs of RTL9310 series (RTL9311, RTL9312,
>>> RTL9313) have equal I2C capabilities thus don't need special treatment.
>>
>> You always need specific front compatible (and fallback if applicable).
>>
> Since I only have RTL9313 variant in my device, I'd be able to add
> 'realtek,rtl9313-i2c' as a verified compatible. For others, I do not 
> have a list
> of which variants actually exist, if there are more variants than just 
> RTL9311,
> RTL9312 and RTL9313. Should I add compatibles for those anyway or just for
> that variant I have?


You have some very odd wrapping of emails.

Anyway, you keep mentioning in multiple places rtl9311-9313, so that's
confusing. If you mention them, I would expect compatibles. They cannot
use rtl9310 compatible alone.

I don't mind skipping them, but then just don't mention any sort of
treatment for other devices. You add this and only this hardware, if you
do not want to follow the make-binding-complete principle (see writing
bindings).

> 

> 'realtek,rtl9301-i2c' seems to be such a fallback for RTL9300, should 
> the one
> for RTL9310 be 'realtek,rtl9311-i2c' or is 'realtek,rtl9310-i2c' fine?
> Just asking because this isn't obvious to me right now.
>>
>>> However, in the unlikely case of future differences with specific
>>> SoCs within this series, more can be added as needed.
>>>
>>> Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
>>> ---
>>>   .../bindings/i2c/realtek,rtl9301-i2c.yaml     | 33 ++++++++++++++++---
>>>   1 file changed, 29 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/Documentation/devicetree/bindings/i2c/realtek,rtl9301-i2c.yaml b/Documentation/devicetree/bindings/i2c/realtek,rtl9301-i2c.yaml
>>> index eddfd329c67b..3b32da3de2af 100644
>>> --- a/Documentation/devicetree/bindings/i2c/realtek,rtl9301-i2c.yaml
>>> +++ b/Documentation/devicetree/bindings/i2c/realtek,rtl9301-i2c.yaml
>>> @@ -10,9 +10,11 @@ maintainers:
>>>     - Chris Packham <chris.packham@alliedtelesis.co.nz>
>>>   
>>>   description:
>>> -  The RTL9300 SoC has two I2C controllers. Each of these has an SCL line (which
>>> -  if not-used for SCL can be a GPIO). There are 8 common SDA lines that can be
>>> -  assigned to either I2C controller.
>>> +  The RTL9300 SoCs have two I2C controllers. Each of these has an SCL line
>>> +  (which if not-used for SCL can be a GPIO). There are 8 common SDA lines
>>> +  that can be assigned to either I2C controller.
>>> +  The RTL9310 SoCs have equal capabilities but support 12 common SDA lines
>>> +  which can be assigned to either I2C controller.
>>>   
>>>   properties:
>>>     compatible:
>>> @@ -24,6 +26,7 @@ properties:
>>>                 - realtek,rtl9303-i2c
>>>             - const: realtek,rtl9301-i2c
>>>         - const: realtek,rtl9301-i2c
>>> +      - const: realtek,rtl9310-i2c
>>>   
>>>     reg:
>>>       description: Register offset and size this I2C controller.
>>> @@ -34,8 +37,18 @@ properties:
>>>     "#size-cells":
>>>       const: 0
>>>   
>>> +  scl-num:
>>
>> No, you do not get own instance IDs.
> Is that meant for the wording/naming of the property and/or its
> description or for the general idea of this property?

You do not get such property. We don't accept it, it's generic rule.
Nowhere in the kernel... unless this is a standard, generic property
(there is no vendor prefix), but I could not find it. If it is standard
property, where is it defined in dtschema or common bindings?

I don't get the need for this property and description does not help, so
just drop it.


Best regards,
Krzysztof