Re: [PATCH v2 6/6] dt-bindings: dma: Convert Qualcomm BAM DMA binding to json format

From: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
To: Kuldeep Singh <singh.kuldeep87k@gmail.com>,
	Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Cc: Rob Herring <robh+dt@kernel.org>,
	Krzysztof Kozlowski <krzk+dt@kernel.org>,
	Andy Gross <agross@kernel.org>,
	Bjorn Andersson <bjorn.andersson@linaro.org>,
	Vinod Koul <vkoul@kernel.org>,
	linux-kernel@vger.kernel.org, devicetree@vger.kernel.org,
	linux-arm-msm@vger.kernel.org, dmaengine@vger.kernel.org
Subject: Re: [PATCH v2 6/6] dt-bindings: dma: Convert Qualcomm BAM DMA binding to json format
Date: Tue, 12 Apr 2022 08:43:59 +0200	[thread overview]
Message-ID: <8ff07720-3c52-99e6-8046-501f4ae28518@linaro.org> (raw)
In-Reply-To: <20220412061953.GA95928@9a2d8922b8f1>

On 12/04/2022 08:19, Kuldeep Singh wrote:
> On Mon, Apr 11, 2022 at 01:38:41PM +0200, Krzysztof Kozlowski wrote:
>> On 11/04/2022 12:58, Kuldeep Singh wrote:
>>>> This is something new and it seems only one SoC defines it (not even one
>>>> BAM version). I wonder whether this is actually correct or this
>>>> particular version of BAM is slightly different. Maybe someone could
>>>> clarify it, but if no - looks ok.
>>>
>>> Yes, sdm845.dtsi uses 4 entries and rest 1.
>>
>> Yes, I know. This does not solve my wonder.
>>
>>>
>>>>
>>>>> +
>>>>> +  num-channels:
>>>>> +    $ref: /schemas/types.yaml#/definitions/uint32
>>>>> +    description:
>>>>> +      Indicates supported number of DMA channels in a remotely controlled bam.
>>>>> +
>>>>> +  qcom,controlled-remotely:
>>>>> +    $ref: /schemas/types.yaml#/definitions/flag
>>>>
>>>> type: boolean
>>>
>>> Boolean comes under flag in types.yaml
>>>
>>> definitions:
>>>   flag:
>>>     oneOf:
>>>       - type: boolean
>>>         const: true
>>>       - type: 'null'
>>>
>>> I have seen other boolean properties(spi-cpol, spi-cpha and bunch of
>>> others) using type flag. I think we should keep flag here.
>>
>> type:boolean is just shorter and example-schema recommends it. If you
>> want to base on something (as a template, pattern) then the
>> example-schema is the source, the preferred one.
> 
> I had seen other spec using flag and that's why kept same here.

I understand, you mentioned it before. However other spec is not the
example-schema...

> Which example schema are you talking about?

There is only one example-schema.
$ find ./linux -name 'example-schema*'

>>>> clocks, clock-names, qcom-ee - these are required according to old bindings.
>>>
>>> I missed qcom,ee. Will add in v3.
>>>
>>> For clocks and clock-names , there are two platforms(msm8996.dtsi,
>>> sdm845.dtsi) where these properties are missing. And I don't want to add
>>> some random values. Shall I skip them here? and let board owners add
>>> them later.
>>
>> These are required, so the SoC DTSI should be fixed. Not with random
>> clocks but something proper. :)
> 
> Yes absolutely :-)
> I have kept Srinivas in copy, who sent initial support for both the
> dtsi. Probably he can confirm provided his email doesn't bounce.
> 
> Anyway Krzysztof, can you confirm the same as you have been actively
> contributing to Qcom peripherals. I will add credit in follow-up
> submission.

Honestly not now, because I don't have access to related datasheets (I
am working on this). You can though try to look at original (vendor)
sources:
https://git.codelinaro.org/clo/la/kernel/msm-4.19 (sdm845)
https://git.codelinaro.org/clo/la/kernel/msm-3.18 (msm8996)

Best regards,
Krzysztof