[Linux-stm32] [PATCH RFC 1/6] dt-bindings: dma: qcom, gpi: Retire passing the protocol ID

[Konrad Dybcio]

On 7/31/25 1:46 AM, Rob Herring wrote:
> On Wed, Jul 30, 2025 at 11:33:28AM +0200, Konrad Dybcio wrote:
>> From: Konrad Dybcio <konrad.dybcio at oss.qualcomm.com>
>>
>> This is a software construct that has no business being expressed in
>> dt-bindings. Drivers can be constructed to retrieve the protocol ID at
>> runtime or hardcode them per protocol.
>>
>> Remove it, as a pre-requisite for further simplifying the GENI
>> bindings.
>>
>> Signed-off-by: Konrad Dybcio <konrad.dybcio at oss.qualcomm.com>
>> ---
>>  Documentation/devicetree/bindings/dma/qcom,gpi.yaml | 5 ++---
>>  1 file changed, 2 insertions(+), 3 deletions(-)
>>
>> diff --git a/Documentation/devicetree/bindings/dma/qcom,gpi.yaml b/Documentation/devicetree/bindings/dma/qcom,gpi.yaml
>> index bbe4da2a11054f0d272017ddf5d5f7e47cf7a443..745613b93b210afd38946030f7477e91e08c907a 100644
>> --- a/Documentation/devicetree/bindings/dma/qcom,gpi.yaml
>> +++ b/Documentation/devicetree/bindings/dma/qcom,gpi.yaml
>> @@ -61,14 +61,13 @@ properties:
>>      maxItems: 13
>>  
>>    "#dma-cells":
>> -    const: 3
>> +    const: 2
> 
> I think you need to keep 3 and note it is deprecated. Does an existing 
> kernel support this being 2 already. If not, ABI break...

Hm, I thought this is a case of relaxing the requirements, although I
suppose some software might have had a "if n_cells != 3" (and not < 3)
check

Konrad