Re: [PATCH v4 1/4] dt-bindings: mfd: Add TI TPS6594 PMIC

[Krzysztof Kozlowski]

On 28/03/2023 16:20, Julien Panis wrote:
> 
> 
> On 3/28/23 13:21, Krzysztof Kozlowski wrote:
>> On 28/03/2023 12:45, Julien Panis wrote:
>>>
>>> On 3/28/23 08:51, Krzysztof Kozlowski wrote:
>>>> On 27/03/2023 17:40, Julien Panis wrote:
>>>>> TPS6594 is a Power Management IC which provides regulators and others
>>>>> features like GPIOs, RTC, watchdog, ESMs (Error Signal Monitor), and
>>>>> PFSM (Pre-configurable Finite State Machine) managing the state of the
>>>>> device.
>>>>> TPS6594 is the super-set device while TPS6593 and LP8764X are derivatives.
>>>>>
>>>>> Signed-off-by: Julien Panis <jpanis@xxxxxxxxxxxx>
>>>>> ---
>>>>>    .../devicetree/bindings/mfd/ti,tps6594.yaml   | 231 ++++++++++++++++++
>>>>>    1 file changed, 231 insertions(+)
>>>>>    create mode 100644 Documentation/devicetree/bindings/mfd/ti,tps6594.yaml
>>>>>
>>>>> diff --git a/Documentation/devicetree/bindings/mfd/ti,tps6594.yaml b/Documentation/devicetree/bindings/mfd/ti,tps6594.yaml
>>>>> new file mode 100644
>>>>> index 000000000000..4498e6361b34
>>>>> --- /dev/null
>>>>> +++ b/Documentation/devicetree/bindings/mfd/ti,tps6594.yaml
>>>>> @@ -0,0 +1,231 @@
>>>>> +# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause
>>>>> +%YAML 1.2
>>>>> +---
>>>>> +$id: http://devicetree.org/schemas/mfd/ti,tps6594.yaml#
>>>>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>>>>> +
>>>>> +title: TI TPS6594 Power Management Integrated Circuit
>>>>> +
>>>>> +maintainers:
>>>>> +  - Julien Panis <jpanis@xxxxxxxxxxxx>
>>>>> +
>>>>> +description:
>>>>> +  TPS6594 is a Power Management IC which provides regulators and others
>>>>> +  features like GPIOs, RTC, watchdog, ESMs (Error Signal Monitor), and
>>>>> +  PFSM (Pre-configurable Finite State Machine) managing the state of the device.
>>>>> +  TPS6594 is the super-set device while TPS6593 and LP8764X are derivatives.
>>>> LP8764X? Compatible says LP8764.
>>>>
>>>>> +
>>>>> +properties:
>>>>> +  compatible:
>>>>> +    enum:
>>>>> +      - ti,lp8764
>>>> It's confusing. If x was wildcard, didn't you remove part of model name?
>>> OK, I will remove 'X' from model name in v5.
>> There is no x in compatible. What is (are) the model name(s)?
> 
> Basically, the model names are:
> - lp8764-q1 ('x' came from an error in internal documentation)

Ah, ok, that explains.

> - tps6593-q1
> - tps6594-q1
> 
> But...for these 3 pmics, there are 19 PN provided
> by TI. For instance: tps659413f, tps659411f, ...
> Each PN gives information about non-volatile memory
> settings and firmware. So, there can be several PN
> for a given model name.
> 
> Which solution is recommended in such situation ?
> 
> IMO, using the 3 'xxxxxx-q1' names for compatible
> strings makes sense (that's what already exists in
> linux for others TI pmics).
> But if you think that full PN names should be specified
> to differentiate internal FW/NVM configurations, I will do
> that and we will get 19 compatible strings.

I think it is fine as long as their programming model is exactly the same.

> 
>>
>>>>
>>>>> +      - ti,tps6593
>>>>> +      - ti,tps6594
>> (...)
>>
>>>>> +
>>>>> +  rtc:
>>>>> +    type: object
>>>>> +    description: RTC provided by this controller.
>>>>> +    $ref: /schemas/rtc/rtc.yaml#
>>>> I doubt that you can have here any RTC and any watchdog (below). This
>>>> should be specific binding instead. Or list of compatibles if you have 3
>>>> or more possible bindings.
>>>>
>>>> Additionally, judging by your DTS you do not have any resources in rtc
>>>> and watchdog, so these should not be nodes by themself in such case.
>>> It seems that I can't figure out what you and Rob mean by saying that
>>> "binding must be complete" and that "RTC and watchdog may or may not
>>> need binding changes".
>>> What does "specific binding" mean ?
>> Specific means not loose, not generic, precise with some accurate
>> properties.
>>
>>> Should we add some specific property
>>> for RTC/WDG provided by the PMIC ?
>> You know ask me to know what is in your device. I don't know. You should
>> know.
>>
>>> Should we write another yaml for both
>>> of them ?
>> Depends. Pretty often yes, but think what do you want to put there?
> 
> There's nothing to put there actually.
> 
>>
>>> Why shouldn't they use the generic rtc/watchdog yaml ?
>> There are no properties in these nodes, so you do not need nodes. Or if
>> you have properties then you need specific binding, not generic one.
> 
> We do not need nodes indeed, since we do not have properties to put in them.

OK. Then please remove the watchdog and rtc nodes. We asked about them
to make binding complete, because often they need additional data (e.g.
pins or interrupts). In this case the expectation will be to add
rtc.yaml and watchdog.yaml refs to main device node (top level) if you
ever need their properties (like start-year).

Best regards,
Krzysztof