Re: [PATCH v6 1/3] dt-bindings: marvell: Document the AC5/AC5X compatibles

[Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>]

On 12/05/2022 03:20, Chris Packham wrote:
> 
> On 12/05/22 04:34, Krzysztof Kozlowski wrote:
>> On 11/05/2022 01:10, Chris Packham wrote:
>>> Describe the compatible properties for the Marvell Alleycat5/5X switches
>>> with integrated CPUs.
>>>
>>> Alleycat5:
>>> * 98DX2538: 24x1G + 2x10G + 2x10G Stack
>>> * 98DX2535: 24x1G + 4x1G Stack
>>> * 98DX2532: 8x1G + 2x10G + 2x1G Stack
>>> * 98DX2531: 8x1G + 4x1G Stack
>>> * 98DX2528: 24x1G + 2x10G + 2x10G Stack
>>> * 98DX2525: 24x1G + 4x1G Stack
>>> * 98DX2522: 8x1G + 2x10G + 2x1G Stack
>>> * 98DX2521: 8x1G + 4x1G Stack
>>> * 98DX2518: 24x1G + 2x10G + 2x10G Stack
>>> * 98DX2515: 24x1G + 4x1G Stack
>>> * 98DX2512: 8x1G + 2x10G + 2x1G Stack
>>> * 98DX2511: 8x1G + 4x1G Stack
>>>
>>> Alleycat5X:
>>> * 98DX3500: 24x1G + 6x25G
>>> * 98DX3501: 16x1G + 6x10G
>>> * 98DX3510: 48x1G + 6x25G
>>> * 98DX3520: 24x2.5G + 6x25G
>>> * 98DX3530: 48x2.5G + 6x25G
>>> * 98DX3540: 12x5G/6x10G + 6x25G
>>> * 98DX3550: 24x5G/12x10G + 6x25G
>>>
>>> Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
>>> ---
>>>
>>> Notes:
>>>      Changes in v6:
>>>      - New
>>>
>>>   .../bindings/arm/marvell/armada-98dx2530.yaml | 27 +++++++++++++++++++
>>>   1 file changed, 27 insertions(+)
>>>   create mode 100644 Documentation/devicetree/bindings/arm/marvell/armada-98dx2530.yaml
>>>
>>> diff --git a/Documentation/devicetree/bindings/arm/marvell/armada-98dx2530.yaml b/Documentation/devicetree/bindings/arm/marvell/armada-98dx2530.yaml
>>> new file mode 100644
>>> index 000000000000..6d9185baf0c5
>>> --- /dev/null
>>> +++ b/Documentation/devicetree/bindings/arm/marvell/armada-98dx2530.yaml
>>> @@ -0,0 +1,27 @@
>>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>>> +%YAML 1.2
>>> +---
>>> +$id: http://scanmail.trustwave.com/?c=20988&d=heX74s-dh8HSCAJmafRigZHOoyY0XQDl80QSCXWitw&u=http%3a%2f%2fdevicetree%2eorg%2fschemas%2farm%2fmarvell%2farmada-98dx2530%2eyaml%23
>>> +$schema: http://scanmail.trustwave.com/?c=20988&d=heX74s-dh8HSCAJmafRigZHOoyY0XQDl80oVWnOltA&u=http%3a%2f%2fdevicetree%2eorg%2fmeta-schemas%2fcore%2eyaml%23
>>> +
>>> +title: Marvell Alleycat5/5X Platforms
>>> +
>>> +maintainers:
>>> +  - Chris Packham <chris.packham@alliedtelesis.co.nz>
>>> +
>>> +properties:
>>> +  $nodename:
>>> +    const: '/'
>>> +  compatible:
>>> +    oneOf:
>>> +
>>> +      - description: Alleycat5 (98DX25xx)
>>> +        items:
>>> +          - const: marvell,ac5
>> This is confusing and does not look correct. The DTS calls AC5 a SoC and
>> you cannot have SoC alone. It's unusable without a SoM or board.
>>
>>> +
>>> +      - description: Alleycat5X (98DX35xx)
>>> +        items:
>>> +          - const: marvell,ac5x
>>> +          - const: marvell,ac5
>> This entry looks correct except ac5x once is called a SoC and once a
>> RD-AC5X board.
>>
>> It cannot be both. Probably you need third compatible, assuming AC5x is
>> a flavor of AC5.
> 
> Yeah it's a bit confusing
> 
> RD-AC5X-(bunch of extra numbers and letters) is the board I have.
> AC5X is a L3 switch chip with integrated CPU.
> AC5 is a L3 switch chip with integrated CPU.
> 
> Switch wise the AC5X and AC5 are quite different but the CPU block is 
> the same between the two.
> 
>>
>> items:
>>   - enum:
>>       - marvell,rd-ac5x
>>   - const: marvell,ac5x
>>   - const: marvell,ac5
> 
> I can go with that but I'm a little vague on what the requirements are. 
> I was trying to follow the armada-7k-8k.yaml as an example.
> 
> If I look at the cn9130-crb-A board it ends up with:
> 
>    compatible = "marvell,cn9130", "marvell,armada-ap807-quad", 
> "marvell,armada-ap807";
> 
> I know the ap807 has something to do with the vagaries of the cn9130 SoC 
> but isn't the "marvell,cn9130" still referring to the SoC. From what 
> you've said shouldn't there be a "marvell,cn9130-crb" somewhere in the mix?
> 
> Perhaps I've picked a bad example but the other dtbs I've poked at don't 
> have any board binding.

The CN9130 looks wrong the same way. They have cn9130.dtsi with "Marvell
Armada CN9130 SoC", so it is clearly a SoC. It has its own compatibles.
Then this DTSI is included in board DTSes. Till now everything is correct.

However the board DTS does not define its own compatible and re-uses SoC
compatible, so this is wrong.

It seems it was done like this inf commit 6a380172f171 ("dt-bindings:
marvell: Declare the CN913x SoC compatibles")
.

That commit even explains "There are three development boards based on
these SoCs:" but then fails to define these boards and instead later
everything uses SoC compatibles as board ones!

Anyone knowing Marvell HW/architecture could fix it?

Best regards,
Krzysztof