RE: [PATCH v2 1/2] regulator: act8945a: add regulator driver for ACT8945A

[Yang, Wenyou]

Hi Peter,

> -----Original Message-----
> From: Peter Korsgaard [mailto:jacmet@xxxxxxxxx] On Behalf Of Peter Korsgaard
> Sent: 2016年1月14日 19:43
> To: Yang, Wenyou <Wenyou.Yang@xxxxxxxxx>
> Cc: Liam Girdwood <lgirdwood@xxxxxxxxx>; Mark Brown <broonie@xxxxxxxxxx>;
> Rob Herring <robh+dt@xxxxxxxxxx>; Pawel Moll <pawel.moll@xxxxxxx>; Mark
> Rutland <mark.rutland@xxxxxxx>; Ian Campbell <ijc+devicetree@xxxxxxxxxxxxxx>;
> Kumar Gala <galak@xxxxxxxxxxxxxx>; Krzysztof Kozlowski
> <k.kozlowski@xxxxxxxxxxx>; devicetree@xxxxxxxxxxxxxxx; Ferre, Nicolas
> <Nicolas.FERRE@xxxxxxxxx>; linux-kernel@xxxxxxxxxxxxxxx; Javier Martinez
> Canillas <javier@xxxxxxxxxxxx>; Lee Jones <lee.jones@xxxxxxxxxx>; linux-arm-
> kernel@xxxxxxxxxxxxxxxxxxx
> Subject: Re: [PATCH v2 1/2] regulator: act8945a: add regulator driver for
> ACT8945A
> 
> >>>>> "Yang," == Yang, Wenyou <Wenyou.Yang@xxxxxxxxx> writes:
> 
> Hi,
> 
> >>
>  >>> Isn't the regulator part of the act8945a identical to act8865? Can't we just
> use  >>> the existing act8865-regulator.c driver? (E.G. support 8865 variant in the
> mfd  >>> driver, but only register the regulator sub device?)  >>  >> > Yes, the
> regulator part of the act8945a is identical to act8865.
>  >>
>  >> > But the act8865-regulator driver is a struct i2c_driver, and the mfd  > sub-
>   >> device driver is a platform_driver driver, it is not easy to use  > it.
>  >>
>  >> But the mfd driver could support the act8865 variant as well (by only
> registering  >> the regulator mfd cell).
> 
>  > I still don't understand.
> 
>  > The MFD sub device is registered as a platform device, the existing  > act8865
> driver is registered as an i2c_driver. How do they match? :)
> 
>  > Could you point out which exiting driver for me to reference.  Thank you!
> 
> What I'm saying is that it isn't nice that we end up with both act8865-regulator.c
> and act8954-regulator.c doing 95% the same. This can either be fixed by:
> 
> - Adding act8865 support to your mfd driver (but only registering the
>   regulator subdev for act8865) and changing the existing
>   act8865-regulator driver to work as a mfd subdev (platform device)
> 
> - Extending the act8865-regulator driver to register as a platform
>   driver in ADDITION to a i2c driver so it can be used with mfd. The
>   driver already uses regmap so it should only be a matter of adding the
>   pdev probe/remove functions.

After think twice, I think this will complicates the act8865-regulator driver.

Moreover, after following up the Mark's advice, a lot redundant code is removed from act8945a-regulater driver. It is very simple now.

So I think it is better to keep a separate act8945a-regulator driver.

Thank you for detailed description.


Best Regards,
Wenyou Yang