Re: [PATCH] riscv: dts: sifive unmatched: Expose the FU740 core supply regulator.

From: Krzysztof Kozlowski
Date: Tue Nov 16 2021 - 06:08:17 EST


On 16/11/2021 10:53, Heiko Stübner wrote:
> Hi Vincent,
>
> Am Dienstag, 16. November 2021, 01:52:59 CET schrieb Vincent Pelletier:
>> Provides monitoring of core voltage and current:
>> tps544b20-i2c-0-1e
>> Adapter: i2c-ocores
>> vout1: 906.00 mV
>> temp1: -40.0°C (high = +125.0°C, crit = +150.0°C)
>> iout1: 5.06 A (max = +20.00 A, crit max = +26.00 A)
>>
>> Signed-off-by: Vincent Pelletier <plr.vincent@xxxxxxxxx>
>>
>> --
>> Note for review: this patch has one warning from checkpatch.pl:
>> WARNING: DT compatible string "tps544b20" appears un-documented -- check ./Documentation/devicetree/bindings/
>> #32: FILE: arch/riscv/boot/dts/sifive/hifive-unmatched-a00.dts:55:
>> + compatible = "tps544b20";
>> This chip is handled by the existing pmbus module, and there is indeed no
>> matching entry in Documentation/devicetree/bindings/hwmon/pmbus. I am not
>> especially knowledgeable about this chip, I only know it is used by this
>> board, so I am not sure I can do the best job in putting such a file
>> together.
>> If needed I can git it a try.
>
> Devicetree bindings are supposed to be stable into the future, so an actually
> reviewed binding is quite necessary ;-) .
>
> In the case of your tps544b20 it should also be pretty easy to do, as
>
> Documentation/devicetree/bindings/hwmon/pmbus/ti,ucd90320.yaml
>
> is probably a pretty good match to what you need in terms of Yaml notation.
> Just need to replace the naming in your copy and drop in the correct
> description from
>
> https://www.ti.com/lit/ds/symlink/tps544b20.pdf?ts=1637055780278
>
> and you have a working binding.
>
> Then just add another patch to your series that mimics
>
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8a36e38d8b0fbb92609e837a67f919202ec7ec51
>
> and include the relevant maintainers that scripts/get_maintainer.pl will
> give you, and you're all set :-)
>

Hi Heiko,

In current form the bindings would be close to trivial and we actually
do not know how proper bindings would look like (the device is not
trivial). Therefore based on Rob's recent comments - better to have
trivial schema than nothing - I sent a patch adding them to trivial-devices:

https://lore.kernel.org/linux-devicetree/20211116110207.68494-1-krzysztof.kozlowski@xxxxxxxxxxxxx/T/#u


Best regards,
Krzysztof