Re: [PATCH 08/14] dt-bindings: dma: dma350: Document interrupt-names

[Krzysztof Kozlowski <krzk@kernel.org>]

On 24/08/2025 11:49, Jisheng Zhang wrote:
> On Sat, Aug 23, 2025 at 06:09:22PM +0200, Krzysztof Kozlowski wrote:
>> On 23/08/2025 17:40, Jisheng Zhang wrote:
>>> Currently, the dma350 driver assumes all channels are available to
>>> linux, this may not be true on some platforms, so it's possible no
>>> irq(s) for the unavailable channel(s). What's more, the available
>>> channels may not be continuous. To handle this case, we'd better
>>> get the irq of each channel by name.
>>
>> You did not solve the actual problem - binding still lists the
>> interrupts in specific order.
>>
>>>
>>> Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
>>> ---
>>>  Documentation/devicetree/bindings/dma/arm,dma-350.yaml | 5 +++++
>>>  1 file changed, 5 insertions(+)
>>>
>>> diff --git a/Documentation/devicetree/bindings/dma/arm,dma-350.yaml b/Documentation/devicetree/bindings/dma/arm,dma-350.yaml
>>> index 429f682f15d8..94752516e51a 100644
>>> --- a/Documentation/devicetree/bindings/dma/arm,dma-350.yaml
>>> +++ b/Documentation/devicetree/bindings/dma/arm,dma-350.yaml
>>> @@ -32,6 +32,10 @@ properties:
>>>        - description: Channel 6 interrupt
>>>        - description: Channel 7 interrupt
>>>  
>>> +  interrupt-names:
>>> +    minItems: 1
>>> +    maxItems: 8
>>
>> You need to list the items.
> 
> I found in current dt-bindings, not all doc list the items. So is it
> changed now?

Close to impossible... :) But even if you found 1% of bindings with
mistake, please kindly take 99% of bindings as the example. Not 1%.

Which bindings were these with undefined names?

> 
>>
>>
>>> +
>>>    "#dma-cells":
>>>      const: 1
>>>      description: The cell is the trigger input number
>>> @@ -40,5 +44,6 @@ required:
>>>    - compatible
>>>    - reg
>>>    - interrupts
>>> +  - interrupt-names
>>
>> That's ABI break, so no.
> 
> If there's no users of arm-dma350 in upstream so far, is ABI break
> allowed? The reason is simple: to simplify the driver to parse
> the irq.

You can try to make your case - see writing bindings. But what about all
out of tree users? All other open source projects? All other kernels? I
really do not ask about anything new here - that's a policy since long time.


Best regards,
Krzysztof