Re: [PATCH v2 1/1] dt-bindings: interrupt-controller: fsl,ls-extirq: workaround wrong interrupt-map number

[Krzysztof Kozlowski <krzk@kernel.org>]

On 06/10/2024 15:30, Krzysztof Kozlowski wrote:
> On 04/10/2024 17:36, Frank Li wrote:
>> On Fri, Oct 04, 2024 at 08:43:23AM +0200, Krzysztof Kozlowski wrote:
>>> On Thu, Oct 03, 2024 at 05:43:15PM -0400, Frank Li wrote:
>>>> The driver(drivers/irqchip/irq-ls-extirq.c) have not use standard DT
>>>> function to parser interrupt-map. So it doesn't consider '#address-size'
>>>> in parent interrupt controller, such as GIC.
>>>>
>>>> When dt-binding verify interrupt-map, item data matrix is spitted at
>>>> incorrect position. So cause below warning:
>>>>
>>>> arch/arm64/boot/dts/freescale/fsl-ls1088a-qds.dtb: interrupt-controller@14:
>>>> interrupt-map: [[0, 0, 1, 0, 0, 4, 1, 0], [1, 0, 1, 4, 2, 0, 1, 0], ...
>>>> is too short
>>>>
>>>> Reduce minItems and maxItems to workaround this warning for
>>>> 'fsl,ls1088a-extirq', 'fsl,ls2080a-extirq' and fsl,lx2160a-extirq.
>>>> Other keep the same restriction.
>>>>
>>>> Signed-off-by: Frank Li <Frank.Li@nxp.com>
>>>> ---
>>>> Change from v1 to v2
>>>> - remove duplicate function in commit message
>>>> - only reduce miniItems for after 1088a chips
>>>> - maxItems change to 9. Otherwise report too long.
>>>> ---
>>>>  .../interrupt-controller/fsl,ls-extirq.yaml   | 27 +++++++++++++++++--
>>>>  1 file changed, 25 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/interrupt-controller/fsl,ls-extirq.yaml b/Documentation/devicetree/bindings/interrupt-controller/fsl,ls-extirq.yaml
>>>> index 199b34fdbefc4..1bfced6ed620c 100644
>>>> --- a/Documentation/devicetree/bindings/interrupt-controller/fsl,ls-extirq.yaml
>>>> +++ b/Documentation/devicetree/bindings/interrupt-controller/fsl,ls-extirq.yaml
>>>> @@ -82,14 +82,37 @@ allOf:
>>>>              enum:
>>>>                - fsl,ls1043a-extirq
>>>>                - fsl,ls1046a-extirq
>>>> +    then:
>>>> +      properties:
>>>> +        interrupt-map:
>>>> +          minItems: 12
>>>> +          maxItems: 12
>>>> +        interrupt-map-mask:
>>>> +          items:
>>>> +            - const: 0xf
>>>> +            - const: 0
>>>> +
>>>> +  - if:
>>>> +      properties:
>>>> +        compatible:
>>>> +          contains:
>>>> +            enum:
>>>>                - fsl,ls1088a-extirq
>>>>                - fsl,ls2080a-extirq
>>>>                - fsl,lx2160a-extirq
>>>> +# The driver(drivers/irqchip/irq-ls-extirq.c) have not use standard DT
>>>> +# function function to parser interrupt-map. So it doesn't consider
>>>
>>> Same issue as last time, double function.
>>>
>>> Please run scripts/checkpatch.pl and fix reported warnings. Then please
>>> run 'scripts/checkpatch.pl --strict' and (probably) fix more warnings.
>>> Some warnings can be ignored, especially from --strict run, but the code
>>> here looks like it needs a fix. Feel free to get in touch if the warning
>>> is not clear.
>>
>> Thanks, I forget add --strict this time.
>>
>>>
>>>
>>>> +# '#address-size' in parent interrupt controller, such as GIC.
>>>> +#
>>>> +# When dt-binding verify interrupt-map, item data matrix is spitted at
>>>> +# incorrect position. Reduce minItems and maxItems to workaround this
>>>> +# problem.
>>>> +
>>>>      then:
>>>>        properties:
>>>>          interrupt-map:
>>>> -          minItems: 12
>>>> -          maxItems: 12
>>>> +          minItems: 8
>>>> +          maxItems: 9
>>>
>>> Are you sure it works? I see 12 items in fsl-ls1088a.dtsi.
>>
>> interrupt-map =
>>    <0 0 &gic GIC_SPI 0 IRQ_TYPE_LEVEL_HIGH>,
>>    <1 0 &gic GIC_SPI 1 IRQ_TYPE_LEVEL_HIGH>,
>>    ...
>>    <11 0 &gic GIC_SPI 11 IRQ_TYPE_LEVEL_HIGH>;
>>
>> Total 12*6 = 72 data.
>>
>> Normal each row should be 6 data.
>>
>> but when GIC have #address-size, <2>,  dt-schemal split at at (6+2=8).
>>
>> "interrupt-map: [[0, 0, 1, 0, 0, 4, 1, 0], [1, 0, 1, 4, 2, 0, 1, 0]"
>>
>> So 72/8 = 9, I just realize it can divide to whole number.  so minItems
>> can be set 9 also.
> 
> I read it three times but I cannot parse it.
> 
> I cannot translate above interrupt-map to anything meaningful, so it
> looks to me that not only address-cells is ignored but entire format is
> different from what the spec asks.
> 
> This is not some ancient 15 year old code, but was added in 2019.
> 
> You should not add fake constraints to a valid property, because what if
> other system implements this correctly?

I read now Rob's response: dropping constraints is also a way.

Best regards,
Krzysztof