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: <8e98b31a-ef4d-a8bd-32f3-e6f496f2b138@linaro.org>
Date:   Thu, 11 May 2023 13:10:35 +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 v1 3/3] dt-bindings: ASoC: Add ESS ES9218P codec bindings
Content-Language: en-US
To:     Aidan MacDonald <aidanmacdonald.0x0@gmail.com>
Cc:     lgirdwood@gmail.com, broonie@kernel.org, robh+dt@kernel.org,
        krzysztof.kozlowski+dt@linaro.org, conor+dt@kernel.org,
        perex@perex.cz, tiwai@suse.com, alsa-devel@alsa-project.org,
        devicetree@vger.kernel.org, linux-kernel@vger.kernel.org
References: <20230510112349.939991-1-aidanmacdonald.0x0@gmail.com>
 <20230510112349.939991-3-aidanmacdonald.0x0@gmail.com>
 <7b3a37e8-0210-c539-5b5b-bf8e587707ea@linaro.org>
 <7Zga7RBqLNcaG5ylTIY4Qn7CUdE1IJsW@localhost>
From:   Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
In-Reply-To: <7Zga7RBqLNcaG5ylTIY4Qn7CUdE1IJsW@localhost>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 11/05/2023 12:15, Aidan MacDonald wrote:
> 
> Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> writes:
> 
>> On 10/05/2023 13:23, Aidan MacDonald wrote:
>>> ...
>>> +  ess,max-clock-div:
>>> +    $ref: /schemas/types.yaml#/definitions/uint32
>>> +    description:
>>> +      Sets the maximum MCLK divider for generating the internal CLK.
>>> +      CLK must be at least 20x the I2C bus speed or I2C transactions
>>> +      will fail. The maximum divider should be chosen to ensure that
>>> +      CLK will not fall below the limit.
>>> +    enum:
>>> +      - 1
>>> +      - 2
>>> +      - 4
>>> +      - 8
>>> +    default: 1
>>
>> Why do you need to customize it per board?
>>
> 
> There's no generic API to read or change the bus speed (SCL frequency)
> for I2C adapters, so it's impossible to calculate a limit on the MCLK
> divider automatically.
> 
> Defaulting to 1 is always safe, but means wasting power at lower sample
> rates. If you know what the bus speed is you can use a higher divider
> limit to save power, and it has to be done at the board/firmware level
> because that's the only place where the bus speed is known.

If I understand correctly, you only miss a way to get bus_freq_hz from
i2c_timings to calculate the divider by yourself? This looks like Linux
limitation, so we shouldn't push it into DT, but rather fix Linux. The
I2C bus rate is known, the MCLK rate as well, so divider is possible to
deduce.

I am actually surprised that I2C core does not store the timings
anywhere and each bus host has to deal with it on its own.

Best regards,
Krzysztof