[PATCH v2 1/3] regulator: mt6360: Add OF match table

[Matthias Brugger]

Hi Matti,

On 11/01/2021 11:32, Vaittinen, Matti wrote:
> 
> Hello Matthias & All,
> 
> On Mon, 2021-01-11 at 11:08 +0100, Matthias Brugger wrote:
>>
>> On 11/01/2021 03:18, gene_chen(陳俊宇) wrote:
>>> [ Internal Use - External ]
>>>
>>
>> Please don't top-post in the future.
>>
>>> Hi Matthias,
>>>
>>> I discussed OF match table with Mark in previous mail in our PATCH
>>> v3,
>>> MFD should just instantiate the platform device.
>>
>> Did you ever test that? Which MFD driver should instantiate the
>> device?
>> I had a look at mt6360-core.c (the obvious one) but I don't see any
>> reference to
>> the regulator [1]. What am I missing?
>>
>>>> Mark Brown <broonie at kernel.org> 於 2020年8月20日 週四 下午7:45寫道：
>>>>> This device only exists in the context of a single parent
>>>>> device, there
>>>>> should be no need for a compatible string here - this is just a
>>>>> detail
>>>>> of how Linux does things.  The MFD should just instntiate the
>>>>> platform
>>>>> device.
>>>> Trying to autoload module without of_id_table will cause run-time 
>>>> error:
>>>> ueventd: LoadWithAliases was unable to load
>>>> of:NregulatorT(null)Cmediatek,mt6360-regulator
>>>
>>> You shouldn't have this described in the device tree at all, like I
>>> say
>>> the MFD should just instantiate the platform device.
>>>
>>
>> Well from my understanding the regulator has a device-tree entry [2],
>> so it
>> needs to match against a device-tree node. My understanding is, that
>> you need a
>> the devicetree node to describe the regulators provided by the
>> device. TBH I'm a
>> bit puzzled about the comment from Mark here. How does another DT
>> node be able
>> to reference a regulator if this is not described in DT? I think we
>> need a DT
>> node here and the matching in the regulator and MFD driver to get the
>> regulator
>> loaded via udev.
> 
> Others are better to answer - but as I spotted this from my inbox I'll
> give my 2 cents :) 
> 
> This can be done W/O regulators having own compatible. Please see
> following:
> 
> drivers/mfd/rohm-bd718x7.c
> drivers/regulator/bd718x7-regulator.c
> Documentation/devicetree/bindings/mfd/rohm,bd71847-pmic.yaml
> Documentation/devicetree/bindings/regulator/rohm,bd71847-regulator.yaml
> 
> as example.

Thanks for the links. Yes that's' a way to go, but...

> 
> The device matching can be done via platform_device_id table.
> I think the MODULE_ALIAS() is needed for module matching - but I don't
> remember without further code browsing.
> 
> Anyways, the BD71837/47/50 DT entry without compatible for
> clk/regulators do probe and load the sub-devices.
> 
> As a "tradeoff" subdevices must retrieve the DT node from the parent
> device.
> 
> For my uneducated eyes the DT binding for regulators should be changed.
> Compatible should not be required and the example node should be moved
> to MFD binding document in the MFD node. But that's just my view on
> this - not willing to push this to any direction!
> 

Generally speaking, DT bindings are stable and can't be changed in a
none-compatible way. We could argue if there is any user of the compatible out
there (probably there isn't any, as the code seems to not even be working).
Apart from that all other cells in the MFD driver are defined as OF_MFD_CELL
[1]. Actually I suspect that mt6360_ldo should represent mt6360-regulator.

Regards,
Matthias

[1] https://elixir.bootlin.com/linux/latest/source/drivers/mfd/mt6360-core.c#L294

> Best Regards
> 	Matti Vaittinen
> 
> --
> Matti Vaittinen, Linux device drivers
> ROHM Semiconductors, Finland
> SWDC
> Kiviharjunlenkki 1E
> 90220 OULU
> FINLAND
> 
> ~~~ "I don't think so," said Rene Descartes. Just then he vanished ~~~
> 
> Simon says - in Latin please.
> "non cogito me" dixit Rene Descarte, deinde evanescavit
> 
> (Thanks for the translation Simon)
>