Re: [PATCH 2/5] dt-bindings: mfd: atmel,at91-usart: convert to json-schema

[Lee Jones <lee@xxxxxxxxxx>]

On Mon, 05 Sep 2022, Krzysztof Kozlowski wrote:

> On 05/09/2022 17:22, Sergiu.Moga@xxxxxxxxxxxxx wrote:
> > On 05.09.2022 17:37, Lee Jones wrote:
> >>
> >> On Fri, 19 Aug 2022, Sergiu.Moga@xxxxxxxxxxxxx wrote:
> >>
> >>> On 18.08.2022 11:39, Krzysztof Kozlowski wrote:
> >>>> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
> >>>>
> >>>> On 17/08/2022 10:55, Sergiu Moga wrote:
> >>>>> Convert at91 USART DT Binding for Atmel/Microchip SoCs to
> >>>>> json-schema format.
> >>>>>
> >>>>> Signed-off-by: Sergiu Moga <sergiu.moga@xxxxxxxxxxxxx>
> >>>>> ---
> >>>>>    .../bindings/mfd/atmel,at91-usart.yaml        | 190 ++++++++++++++++++
> >>>>>    .../devicetree/bindings/mfd/atmel-usart.txt   |  98 ---------
> >>>>>    2 files changed, 190 insertions(+), 98 deletions(-)
> >>>>>    create mode 100644 Documentation/devicetree/bindings/mfd/atmel,at91-usart.yaml
> >>>>>    delete mode 100644 Documentation/devicetree/bindings/mfd/atmel-usart.txt
> >>>>>
> >>>>> diff --git a/Documentation/devicetree/bindings/mfd/atmel,at91-usart.yaml b/Documentation/devicetree/bindings/mfd/atmel,at91-usart.yaml
> >>>>> new file mode 100644
> >>>>> index 000000000000..cf15d73fa1e8
> >>>>> --- /dev/null
> >>>>> +++ b/Documentation/devicetree/bindings/mfd/atmel,at91-usart.yaml
> >>>> One more thing - I think this should be in serial directory, not mfd,
> >>>> even though it includes SPI. MFD is just a Linux naming/wrapper device.
> >>>>
> >>>> Best regards,
> >>>> Krzysztof
> >>>
> >>> I would rather keep it in this directory, since its corresponding driver
> >>> is also in the mfd directory.
> >>
> >> Looks like a UART driver to me.
> >>
> >> Which MFD driver does this pertain to?
> >>
> > 
> > Hi,
> > 
> > It's this one: drivers/mfd/at91-usart.c[1]
> > 
> > 
> > [1] 
> > https://elixir.bootlin.com/linux/v6.0-rc4/source/drivers/mfd/at91-usart.c
> 
> Which is not a "real MFD driver" because it probes exactly one child
> (depending on the chosen serial protocol). Aren't MFD supposed to have
> more then one child?

It's a single piece of silicon which supports multiple functions.

There is no stipulation detailing simultaneous usage.

Still, happy to receive suggestions on implementing this differently.

-- 
Lee Jones [李琼斯]