Re: [PATCH 00/12] riscv: Allwinner D1 platform support

[Conor.Dooley]

On 16/08/2022 03:42, Samuel Holland wrote:
> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
> 
> On 8/15/22 12:12 PM, Conor.Dooley@xxxxxxxxxxxxx wrote:
>> On 15/08/2022 08:05, Conor Dooley wrote:
>>> Hey Samuel,
>>> Apologies for the last email, the first 5 patches in this series
>>> didn't correctly land in my inbox and I skimmed & replied to the
>>> wrong cover letter... I blame it being pre 8 AM ;)
>>>
>>> On 15/08/2022 06:08, Samuel Holland wrote:
>>>> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
>>>>
>>>> This series adds the Kconfig/defconfig plumbing and devicetrees for a
>>>> range of Allwinner D1-based boards. Many features are already enabled,
>>>> including USB, Ethernet, and WiFi.
>>>>
>>>> The SoC devicetree uses bindings from the following series which have
>>>> not yet been merged:
>>>> - SRAM controller:
>>>>     https://lore.kernel.org/lkml/20220815041248.53268-1-samuel@xxxxxxxxxxxx/
>>>> - NVMEM cell bits property change:
>>>>     https://lore.kernel.org/lkml/20220814173656.11856-1-samuel@xxxxxxxxxxxx/
>>>> - In-package LDO regulators:
>>>>     https://lore.kernel.org/lkml/20220815043436.20170-1-samuel@xxxxxxxxxxxx/
>>>>
>>>> All three of these are required to set the correct I/O domain voltages
>>>> in the pin controller, which I would consider important to have in the
>>>> initial version of the devicetree.
>>>>
>>>> The SoC devicetree does contain one small hack to avoid a dependency on
>>>> the audio codec binding, since that is not ready yet: the codec node
>>>> uses a bare "simple-mfd", "syscon" compatible.
>>>
>>>
>>> I tried this series on top of both 6.0-rc1 & next, but I saw a shed
>>> load of errors from dtbs_check. I'll take a look at the rest of the
>>> series later on, but just FYI. This time I did at least read the
>>> right cover letter and actually applied your dependent series..
>>> /facepalm
>>>
>>> I have been trying to get riscv down to 0 warnings and would not be
>>> in favour of accepting patches that go in the other direction.
>>> (This patchset currently produces about 47 new warnings)
> 
> Only 4 unique ones :)
> 
> Thanks for checking this. I had done a couple of passes cleaning things up. The
> remaining warnings are the ones I was unsure how to fix.
> 
>> FWIW this is what I see, LMK if I am missing something.
>> allwinner/sun20i-d1-lichee-rv.dtb: i2c@2502000: Unevaluated properties are not allowed ('dma-names', 'dmas' were unexpected)
> 
> This one is new with dt-schema 2022.08.1 (I was previously on d1f3102e locally).
> I'm not sure if it is a bug in the binding or in the dt-schema tool. The
> warnings go away if I do:

Yup, there were some changes to handling of unevaluatedProperties
between .07 & .08
> 
> --- a/Documentation/devicetree/bindings/i2c/marvell,mv64xxx-i2c.yaml
> +++ b/Documentation/devicetree/bindings/i2c/marvell,mv64xxx-i2c.yaml
> @@ -72,6 +72,7 @@ required:
>     - interrupts
> 
>   allOf:
> +  - $ref: /schemas/dma/dma.yaml#
>     - $ref: /schemas/i2c/i2c-controller.yaml#
>     - if:
>         properties:
> 
> But I wouldn't expect this to be necessary, since dma.yaml has "select: true".
> Compare the this with the UARTs. Neither serial/snps-dw-apb-uart.yaml nor
> serial/serial.yaml references any DMA properties, yet there is no warning for those.
> 
>> allwinner/sun20i-d1-clockworkpi-v3.14.dtb: pmic@34: compatible: 'oneOf' conditional failed, one must be fixed:
>> allwinner/sun20i-d1-devterm-v3.14.dtb: pmic@34: compatible: 'oneOf' conditional failed, one must be fixed:
> 
> The new compatible is in linux-next. I can add it to the cover letter for v2:
> https://lore.kernel.org/linux-devicetree/20220810013430.27061-1-samuel@xxxxxxxxxxxx/
> https://git.kernel.org/lee/mfd/c/97ecda0f615e

Please do :)

> 
>> allwinner/sun20i-d1-lichee-rv.dtb: ethernet@4500000: Unevaluated properties are not allowed ('mdio', 'reset-names', 'resets' were unexpected)
> 
> This showed up with a dt-schema update in the last year or so. This may be a bug
> in the binding? allwinner,sun8i-a83t-emac.yaml picks up these three properties
> from snps,dwmac.yaml, but does not have them in its own "properties" block.
> (resets and reset-names are even in "required"!)
> 
>> allwinner/sun20i-d1-lichee-rv-86-panel-720p.dtb: ethernet@4500000: Unevaluated properties are not allowed ('phy-supply' was unexpected)
>> allwinner/sun20i-d1-dongshan-nezha-stu.dtb: ethernet@4500000: Unevaluated properties are not allowed ('phy-supply' was unexpected)
> 
> I think this one is just flat out missing from the binding. It is present in the
> other two Allwinner Ethernet MAC bindings, but not this one.

Cool. Maybe Rob Or Krzysztof can shed some light on the others.

Thanks,
Conor.