Re: [PATCH v5 08/12] dt-bindings: mediatek: Change the binding for mmsys clocks

[Matthias Brugger <mbrugger@suse.com>]

On 05/07/2019 03:35, CK Hu wrote:
> Hi, Uli:
> 
> On Thu, 2019-07-04 at 17:33 +0200, Ulrich Hecht wrote:
>>> On July 4, 2019 at 11:08 AM Matthias Brugger <matthias.bgg@gmail.com> wrote:
>>> You are right, it took far too long for me to respond with a new version of the
>>> series. The problem I face is, that I use my mt8173 based chromebook for
>>> testing. It needs some downstream patches and broke somewhere between my last
>>> email and a few month ago.
>>
>> If that Chromebook is an Acer R13 and you need a working kernel, you may want to have a look at https://github.com/uli/kernel/tree/elm-working-5.2 .
> 
> Thanks for your sample code, and your implementation is different than
> Matthias' version. In your version, mmsys is a single device which has
> clock function and display function, the clock function is placed in
> clock driver folder and display function is placed in drm driver folder.
> In Matthias' version, clock function is a sub device of mmsys. I've no
> idea of which one is better. I would get more information to make better
> decision.
> 

From the discussion we had here and the last comments from Stephen I thin what
we should do is:
- create a platform driver for the mmsys clock part, in the drivers/clk/mediatek/...
- the DRM driver creates a platform device which will probe the clock driver.

This way you won't need to change any compatible.

You will have to re-read the discussions in the different versions of this series.
My first approach was to usa a mfd driver for mmsys, which was rejected.
The last approach was to create a DTS sub node, also rejected :)

Regards,
Matthias

> Regards,
> CK
> 
>>
>> CU
>> Uli
>>
>> _______________________________________________
>> Linux-mediatek mailing list
>> Linux-mediatek@lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/linux-mediatek
> 
> 
>