Re: [RFC PATCH 07/16] dt-bindings: pinctrl: ralink: add new compatible strings

[Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx>]

On 02/03/2023 10:17, Arınç ÜNAL wrote:
> On 2.03.2023 11:28, Krzysztof Kozlowski wrote:
>> On 01/03/2023 09:15, Arınç ÜNAL wrote:
>>> On 1.03.2023 05:44, Rob Herring wrote:
>>>> On Tue, Feb 28, 2023 at 07:46:36PM +0300, Arınç ÜNAL wrote:
>>>>> On 27/02/2023 20:33, Rob Herring wrote:
>>>>>> On Wed, Feb 22, 2023 at 09:39:23PM +0300, arinc9.unal@xxxxxxxxx wrote:
>>>>>>> From: Arınç ÜNAL <arinc.unal@xxxxxxxxxx>
>>>>>>>
>>>>>>> Add the ralink,rt2880-pinmux compatible string. It had been removed from
>>>>>>> the driver which broke the ABI.
>>>>>>>
>>>>>>> Add the mediatek compatible strings. Change the compatible string on the
>>>>>>> examples with the mediatek compatible strings.
>>>>>>>
>>>>>>> Signed-off-by: Arınç ÜNAL <arinc.unal@xxxxxxxxxx>
>>>>>>> ---
>>>>>>>     .../devicetree/bindings/pinctrl/ralink,mt7620-pinctrl.yaml | 7 +++++--
>>>>>>>     .../devicetree/bindings/pinctrl/ralink,mt7621-pinctrl.yaml | 7 +++++--
>>>>>>>     .../devicetree/bindings/pinctrl/ralink,rt2880-pinctrl.yaml | 7 +++++--
>>>>>>>     .../devicetree/bindings/pinctrl/ralink,rt305x-pinctrl.yaml | 7 +++++--
>>>>>>>     .../devicetree/bindings/pinctrl/ralink,rt3883-pinctrl.yaml | 7 +++++--
>>>>>>>     5 files changed, 25 insertions(+), 10 deletions(-)
>>>>>>>
>>>>>>> diff --git a/Documentation/devicetree/bindings/pinctrl/ralink,mt7620-pinctrl.yaml b/Documentation/devicetree/bindings/pinctrl/ralink,mt7620-pinctrl.yaml
>>>>>>> index 1e63ea34146a..531b5f616c3d 100644
>>>>>>> --- a/Documentation/devicetree/bindings/pinctrl/ralink,mt7620-pinctrl.yaml
>>>>>>> +++ b/Documentation/devicetree/bindings/pinctrl/ralink,mt7620-pinctrl.yaml
>>>>>>> @@ -17,7 +17,10 @@ description:
>>>>>>>     properties:
>>>>>>>       compatible:
>>>>>>> -    const: ralink,mt7620-pinctrl
>>>>>>> +    enum:
>>>>>>> +      - mediatek,mt7620-pinctrl
>>>>>>> +      - ralink,mt7620-pinctrl
>>>>>>
>>>>>> We don't update compatible strings based on acquistions nor marketing
>>>>>> whims. If you want to use 'mediatek' for new things, then fine.
>>>>>
>>>>> Understood. Only the SoCs with rtXXXX were rebranded, the mtXXXX SoCs share
>>>>> the same architecture from Ralink, so they were incorrectly called Ralink
>>>>> SoCs.
>>>>>
>>>>> I can remove the new strings from Ralink SoCs and add them only for MediaTek
>>>>> SoCs. Or you could make an exception for this one, regarding the situation.
>>>>> Whatever you think is best.
>>>>
>>>> I'm not in a position to make an exception as I know little about this
>>>> platform. Carrying both strings is a NAK. Either you (and everyone using
>>>> these platforms) care about the ABI and are stuck with the "wrong"
>>>> string. In the end, they are just unique identifiers. Or you don't care
>>>> and break the ABI and rename everything. If you do that, do just that in
>>>> your patches and make it crystal clear in the commit msg that is your
>>>> intention and why that is okay.
>>>
>>> Ralink had their MIPS SoCs pre-acquisition, RT2880, etc. MediaTek
>>> introduced new SoCs post-acquisition, MT7620, MT7621, MT7628, and
>>> MT7688, utilising the same platform from Ralink, sharing the same
>>> architecture code, pinctrl core driver, etc.
>>>
>>> I don't intend to break the ABI at all. On the contrary, I fix it where
>>> possible.
>>>
>>> If I understand correctly, from this conversation and what Krzysztof
>>> said, all strings must be kept on the schemas so I can do what I said on
>>> the composed mail. Only match the pin muxing information on the strings
>>> that won't match multiple pin muxing information from other schemas.
>>>
>>> This way we don't break the ABI, introduce new compatible strings while
>>> keeping the remaining ones, and make schemas match correctly.
>>>
>>> Let me know if this is acceptable to you.
>>
>> If by "introduce new compatible strings" you mean duplicate compatibles
>> to fix the ralink->mediatek, then you ignored entire email from Rob -
>> this and previous. We don't do this. Leave them as is.
>>
>> If you meant something else, explain more...
> 
> Let me put them in a group to better explain.
> 
> ## Fix ABI
> 
> ralink,rt2880-pinmux was there before, it was removed which broke the 
> ABI. I'm reintroducing it to fix it.
> 
> ## New strings to be able to split bindings
> 
> New strings are needed for MT7628/MT7688 and some RT SoCs to be able to 
> properly document the pin muxing information.

Then ok.

> 
> ## Incorrect naming
> 
> MT7620, MT7621, MT7628, and MT7688 SoCs are incorrectly called Ralink, 
> introduce new ralink->mediatek compatible strings to address it.

So this part was addressed by Rob - we don't do it, because it does not
matter. Ralink is now Mediatek, thus there is no conflict and no issues
with different vendor used.

> 
> ## Exception for RT SoCs to be called MediaTek
> 
> This is where I was asking an exception to be made. Rob told us here 
> they know little about the platform so I explained it.
> 
> MediaTek acquired Ralink and then introduced new MediaTek SoCs utilising 
> the same platform from Ralink.
> 
> Anyway, now that I look at this again, it makes sense to me as well not 
> to rename the Ralink SoCs. I'll call the RT SoCs Ralink on the kconfig, 
> pinctrl driver,

These are separate. We did not comment on how you call Linux drivers.
The mail thread was only about:

> and dt-binding schemas on my next version.

Best regards,
Krzysztof