Re: [PATCH 01/14] dt-bindings: Document STM32MP1 Reset Clock Controller (RCC) bindings

[Gabriel FERNANDEZ]

On 02/07/2018 07:03 PM, Rob Herring wrote:
> On Mon, Feb 5, 2018 at 1:01 AM, Gabriel FERNANDEZ
> <gabriel.fernandez@xxxxxx>  wrote:
>> Hi Rob,
>>
>> Thanks for reviewing.
>>
>>
>> On 02/05/2018 07:09 AM, Rob Herring wrote:
>>> On Fri, Feb 02, 2018 at 03:03:29PM +0100,gabriel.fernandez@xxxxxx  wrote:
>>>> From: Gabriel Fernandez<gabriel.fernandez@xxxxxx>
>>>>
>>>> The RCC block is responsible of the management of the clock and reset
>>>> generation for the complete circuit.
>>>>
>>>> Signed-off-by: Gabriel Fernandez<gabriel.fernandez@xxxxxx>
>>>> ---
>>>>    .../devicetree/bindings/mfd/st,stm32-rcc.txt       | 85 ++++++++++++++++++++++
>>>>    1 file changed, 85 insertions(+)
>>>>    create mode 100644 Documentation/devicetree/bindings/mfd/st,stm32-rcc.txt
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/mfd/st,stm32-rcc.txt b/Documentation/devicetree/bindings/mfd/st,stm32-rcc.txt
>>>> new file mode 100644
>>>> index 0000000..28017a1
>>>> --- /dev/null
>>>> +++ b/Documentation/devicetree/bindings/mfd/st,stm32-rcc.txt
>>>> @@ -0,0 +1,85 @@
>>>> +STMicroelectronics STM32 Peripheral Reset Clock Controller
>>>> +==========================================================
>>>> +
>>>> +The RCC IP is both a reset and a clock controller.
>>>> +
>>>> +Please also refer to reset.txt for common reset controller binding usage.
>>>> +
>>>> +Please also refer to clock-bindings.txt for common clock controller
>>>> +binding usage.
>>>> +
>>>> +
>>>> +Required properties:
>>>> +- compatible: "simple-mfd", "syscon"
>>>> +- reg: should be register base and length as documented in the datasheet
>>>> +
>>>> +- Sub-nodes:
>>>> +  - compatible: "st,stm32mp1-rcc-clk"
>>>> +    - #clock-cells: 1, device nodes should specify the clock in their
>>>> +      "clocks" property, containing a phandle to the clock device node,
>>>> +      an index specifying the clock to use.
>>>> +
>>>> +  - compatible: "st,stm32mp1-rcc-rst"
>>>> +    - #reset-cells: Shall be 1
>>>> +
>>>> +Example:
>>>> +    rcc: rcc@50000000 {
>>>> +            compatible = "syscon", "simple-mfd";
>>>> +            reg = <0x50000000 0x1000>;
>>>> +
>>>> +            rcc_clk: rcc-clk@50000000 {
>>>> +                    #clock-cells = <1>;
>>>> +                    compatible = "st,stm32mp1-rcc-clk";
>>>> +            };
>>>> +
>>>> +            rcc_rst: rcc-reset@50000000 {
>>> You should not have the same unit-address twice.
>>>
>>> IMO, this should just be:
>>>
>>>        rcc: rcc@50000000 {
>>>                compatible = "st-stm32mp1-rcc";
>>>                reg = <0x50000000 0x1000>;
>>>                #clock-cells = <1>;
>>>                #reset-cells = <1>;
>>>        };
>>>
>>> There's no reason a node can't provide more than 1 function.
>> RCC is an dedicated IP for clocks and resets, but also for power
>> management (patches will be sent later)
> If there's additional functions, they should be part of the binding
> now, not later. bindings should not unnecessarily evolve.
Yes you're right i will do it.
>> Then i need to probe 3 drivers with same IP.
> Drivers and DT nodes don't have to be 1-1. A parent driver can create
> additional child devices.
>
> Also, looking at your existing bindings for RCC IP, it is done as I suggested.
>
>> It's also a way to avoid use of 'CLK_OF_DECLARE_DRIVER' and i need it to
>> probe the 3th driver.
> Sounds like a Linux problem, not a DT issue.
>
> Rob
RCC is an IP block wich exposed multiple functionalities (clock, reset, 
power management),
mfd should be the best solution to populate each functionality in 
natural way,
and to access to registers ?

Gabriel