Re: [PATCH v2] spi: dt-bindings: jcore,spi: convert spi-jcore to dtschema

From: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
To: Kousik Sanagavarapu <five231003@gmail.com>
Cc: linux-kernel@vger.kernel.org, devicetree@vger.kernel.org,
	linux-spi@vger.kernel.org, Rob Herring <robh@kernel.org>,
	Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
	Mark Brown <broonie@kernel.org>,
	Shuah Khan <skhan@linuxfoundation.org>,
	Javier Carrasco <javier.carrasco.cruz@gmail.com>,
	Daniel Baluta <daniel.baluta@nxp.com>
Subject: Re: [PATCH v2] spi: dt-bindings: jcore,spi: convert spi-jcore to dtschema
Date: Fri, 22 Mar 2024 07:49:57 +0100	[thread overview]
Message-ID: <6552bcb8-e046-4882-91da-1094fff3d239@linaro.org> (raw)
In-Reply-To: <5dd3237f-e0a2-4214-a63f-233e89a26b8d@linaro.org>

On 22/03/2024 07:34, Krzysztof Kozlowski wrote:
> On 22/03/2024 07:23, Kousik Sanagavarapu wrote:
>> On Fri, 22 Mar 2024, 11:33 Krzysztof Kozlowski, <
>> krzysztof.kozlowski@linaro.org> wrote:
>>
>>> On 21/03/2024 19:02, Kousik Sanagavarapu wrote:
>>>
>>>> +  spi-max-frequency:
>>>> +    $ref: /schemas/types.yaml#/definitions/uint32
>>>
>>> No, drop. From which other SPI binding did you take it? I asked you to
>>> look at existing code.
>>>
>>
>> Without this, "make dt_binding_check" would break though, right at the
>> position in the example where "spi-max-frequency" is used.  That was
>> also the reason why additionalProperties was set to true in the last
>> iteration, but after reading the doc more carefully I realized that was
>> wrong after you pointed it out.
>>
>> I followed along bindings/spi/nvidia,tegra114-spi.yaml.
> 
> OK, you are right, the property is used here in controller node, however
> Linux driver never parsed it. It was never used, so I propose to drop it
> from the binding and example. You can mention in commit msg that
> spi-max-frequency was not documented thus you drop it from the example.
> 
> DTS should be fixed as well. I'll send a patch for it.

Cc Daniel,

BTW, J2 core is rather odd platform to work on... Even cross compiling
and building that DTB is tricky. If I failed, I have doubts that you
tested the DTS with your binding.

This applies to all GSoC or some Linux Mentorship programs: I suggest to
choose for conversion bindings with more users and bigger possible
impact. So first I would look at ARM64 and ARMv7 platforms. We still
have around 1000 and 3500 unique warnings about undocumented compatibles
for ARM64 defconfig and ARM multi_v7! That's the platforms you should
choose.

Not SuperH, ARC, or whatever with only one DTS which is difficult to
build for regular developer.

Best regards,
Krzysztof