Re: [RFC PATCH v2 01/19] dt-bindings: clock: Add VO subsystem clocks and update address requirements

[Krzysztof Kozlowski <krzk@kernel.org>]

On 24/12/2024 10:23, Michal Wilczynski wrote:
> 
> 
> On 12/24/24 09:53, Krzysztof Kozlowski wrote:
>> On Mon, Dec 23, 2024 at 12:50:59PM -0800, Stephen Boyd wrote:
>>> Quoting Michal Wilczynski (2024-12-23 04:55:35)
>>>> The T-Head TH1520 SoC’s AP clock controller now needs two address ranges
>>>> to manage both the Application Processor (AP) and Video Output (VO)
>>>> subsystem clocks. Update the device tree bindings to require two `reg`
>>>> entries, one for the AP clocks and one for the VO clocks.
>>>>
>>>> Additionally, introduce new VO subsystem clock constants in the header
>>>> file. These constants will be used by the driver to control VO-related
>>>> components such as display and graphics units.
>>>>
>>>> Signed-off-by: Michal Wilczynski <m.wilczynski@samsung.com>
>>>> ---
>>> [...]
>>>> diff --git a/Documentation/devicetree/bindings/clock/thead,th1520-clk-ap.yaml b/Documentation/devicetree/bindings/clock/thead,th1520-clk-ap.yaml
>>>> index 0129bd0ba4b3..f0df97a450ef 100644
>>>> --- a/Documentation/devicetree/bindings/clock/thead,th1520-clk-ap.yaml
>>>> +++ b/Documentation/devicetree/bindings/clock/thead,th1520-clk-ap.yaml
>>>> @@ -47,7 +54,9 @@ examples:
>>>>      #include <dt-bindings/clock/thead,th1520-clk-ap.h>
>>>>      clock-controller@ef010000 {
>>>>          compatible = "thead,th1520-clk-ap";
>>>> -        reg = <0xef010000 0x1000>;
>>>> +        reg = <0xef010000 0x1000>,
>>>> +              <0xff010000 0x1000>;
>>>
>>> I don't get it. Why not have two nodes and two devices? They have
>>> different register regions so likely they're different devices on the
>>> internal SoC bus. They may have the same input clks, but otherwise I
>>> don't see how they're the same node.
>>
>> That's a good point. Aren't here simply two different clock controllers?
> 
> Yeah there are two clock controllers, based on the review comments I was
> trying to re-use the driver, but the driver can also be re-used to serve
> multiple nodes and multiple compatible and .data properties, if that's
> fine with you that's how it will look like in v3.
Yeah, please drop my review tag and rework it to have two different
devices. Driver design should not influence DTS.

Best regards,
Krzysztof