Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <b3c7db03-6614-47d9-a9e0-a8e51c836d86@linaro.org>
Date:   Wed, 10 May 2023 09:06:11 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.11.0
Subject: Re: [PATCH v3 2/5] dt-bindings: clocks: atmel,at91rm9200-pmc: convert
 to yaml
Content-Language: en-US
To:     Claudiu.Beznea@microchip.com, mturquette@baylibre.com,
        sboyd@kernel.org, robh+dt@kernel.org,
        krzysztof.kozlowski+dt@linaro.org, Nicolas.Ferre@microchip.com,
        alexandre.belloni@bootlin.com
Cc:     linux-clk@vger.kernel.org, devicetree@vger.kernel.org,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org
References: <20230509052757.539274-1-claudiu.beznea@microchip.com>
 <20230509052757.539274-3-claudiu.beznea@microchip.com>
 <e463eb68-3ea0-5230-76fd-4a2ee66bf397@linaro.org>
 <773d0d90-29c7-b1bd-bd16-898b435eafb6@microchip.com>
From:   Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
In-Reply-To: <773d0d90-29c7-b1bd-bd16-898b435eafb6@microchip.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 10/05/2023 09:00, Claudiu.Beznea@microchip.com wrote:
> On 09.05.2023 09:25, Krzysztof Kozlowski wrote:
>> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
>>
>> On 09/05/2023 07:27, Claudiu Beznea wrote:
>>> Convert Atmel PMC documentation to yaml. Along with it clock names
>>> were adapted according to the current available device trees as
>>> different controller versions accept different clocks (some of them
>>> have 3 clocks as input, some has 2 clocks as inputs and some with 2
>>> input clocks uses different clock names).
>>>
>>
>> Thank you for your patch. There is something to discuss/improve.
>>
>>> +title: Atmel Power Management Controller (PMC)
>>> +
>>> +maintainers:
>>> +  - Claudiu Beznea <claudiu.beznea@microchip.com>
>>> +
>>> +description:
>>> +  The power management controller optimizes power consumption by controlling all
>>> +  system and user peripheral clocks. The PMC enables/disables the clock inputs
>>> +  to many of the peripherals and to the processor.
>>> +
>>> +properties:
>>> +  compatible:
>>> +    oneOf:
>>> +      - items:
>>> +          - enum:
>>> +              - atmel,at91sam9g15-pmc
>>> +              - atmel,at91sam9g20-pmc
>>> +              - atmel,at91sam9g25-pmc
>>> +              - atmel,at91sam9g35-pmc
>>> +              - atmel,at91sam9x25-pmc
>>> +              - atmel,at91sam9x35-pmc
>>> +          - enum:
>>> +              - atmel,at91sam9260-pmc
>>> +              - atmel,at91sam9x5-pmc
>>
>> I missed it last time - why you have two enums? We never talked about
>> this. It's usually wrong... are you sure this is real hardware:
>> atmel,at91sam9g20-pmc, atmel,at91sam9260-pmc
>> ?
> 
> I have 2 enums because there are some hardware covered by:
> "vendor-name,hardware-v1-pmc", "syscon" and some covered by:
> "vendor-name,hardware-v2-pmc", "vendor-name,hardware-v1-pmc", "syscon".

The enum does not say this. At all.

So again, answer, do not ignore:
is this valid setup:
atmel,at91sam9g20-pmc, atmel,at91sam9260-pmc
?

> 
> Many AT91 device trees compatibles were written in this way. Thus when new
> versions of the same IP has been introduced the drivers were not
> necessarily updated but the compatibles in device trees were updated e.g.
> with "vendor-name,hardware-v2-pmc" (the full compatible becoming
> "vendor-name,hardware-v2-pmc", "vendor-name,hardware-v1-pmc", "syscon") and
> let the drivers fall back to already in driver supported compatible
> "vendor-name,hardware-v1-pmc", "syscon". In general v2 comes with new
> features in addition to v1.
> 
> That way they AT91 ensures the ABI properties of DT and thus when the
> drivers were finally updated with the new features of the
> "vendor-name,hardware-v2-pmc" DT remained in place.
> 
> Please let me know if these could be handled better in YAML.


enum + const + syscon, like every binding that type does in all
bindings. Don't invent some new syntax.

Best regards,
Krzysztof