Re: [PATCH v3 1/5] media: dt-bindings: add non-pixel property in iris schema

[Krzysztof Kozlowski <krzk@kernel.org>]

On 02/07/2025 13:32, Vikash Garodia wrote:
> 
> On 7/2/2025 4:43 PM, Krzysztof Kozlowski wrote:
>> On 27/06/2025 17:48, Vikash Garodia wrote:
>>> Existing definition limits the IOVA to an addressable range of 4GiB, and
>>> even within that range, some of the space is used by IO registers,
>>> thereby limiting the available IOVA to even lesser. Video hardware is
>>> designed to emit different stream-ID for pixel and non-pixel buffers,
>>> thereby introduce a non-pixel sub node to handle non-pixel stream-ID.
>>>
>>> With this, both iris and non-pixel device can have IOVA range of 0-4GiB
>>> individually. Certain video usecases like higher video concurrency needs
>>> IOVA higher than 4GiB.
>>>
>>> Add reference to the reserve-memory schema, which defines reserved IOVA
>>
>> No. That schema is always selected. This makes no sense at all.
> I could not get this, are you suggesting to drop this reference ?

What does the schema says?

  7 title: /reserved-memory Child Node Common


Is this the binding for reserved-memory node? Not sure, your subject
does not have proper prefix, but diff suggested that not.

Maybe I missed something.


>>
>>> regions that are *excluded* from addressable range. Video hardware
>>> generates different stream IDs based on the predefined range of IOVA
>>> addresses. Thereby IOVA addresses for firmware and data buffers need to
>>> be non overlapping. For ex. 0x0-0x25800000 address range is reserved for
>>> firmware stream-ID, while non-pixel (bitstream) stream-ID can be
>>> generated by hardware only when bitstream buffers IOVA address is from
>>> 0x25800000-0xe0000000.
>>> Non-pixel stream-ID can now be part of the new sub-node, hence iommus in
>>> iris node can have either 1 entry for pixel stream-id or 2 entries for
>>> pixel and non-pixel stream-ids.
>>>
>>> Signed-off-by: Vikash Garodia <quic_vgarodia@quicinc.com>
>>> ---
>>>  .../bindings/media/qcom,sm8550-iris.yaml           | 40 ++++++++++++++++++++--
>>>  1 file changed, 38 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/Documentation/devicetree/bindings/media/qcom,sm8550-iris.yaml b/Documentation/devicetree/bindings/media/qcom,sm8550-iris.yaml
>>> index c79bf2101812d83b99704f38b7348a9f728dff44..4dda2c9ca1293baa7aee3b9ee10aff38d280fe05 100644
>>> --- a/Documentation/devicetree/bindings/media/qcom,sm8550-iris.yaml
>>> +++ b/Documentation/devicetree/bindings/media/qcom,sm8550-iris.yaml
>>> @@ -65,10 +65,31 @@ properties:
>>>        - const: core
>>>  
>>>    iommus:
>>> +    minItems: 1
>>>      maxItems: 2
>>
>> No, why hardware suddenly has different amount?
> Its not about hardware started to have a new stream-ID. You can look for the
> description in the commit which explains the need for a new device and hence the
> split of stream-IDs in iris device OR non-pixel device.

But this is not a new device! This is sm8550. Existing device.

>>
>>>  
>>>    dma-coherent: true
>>>  
>>> +  non-pixel:
>>
>> Why EXISTING hardware grows?
> Same here, the commit describes the limitation of existing design and also
> explains the need for having the non-pixel device. Its not that the hardware is
> growing here, rather the hardware stream-IDs are utilized differently to get
> higher device addressable range.

You are not doing this for a new device. There is no new device here at
all. Nowhere here is a new device.

Changes for a new device COME TOGETHER with the new device.

What you are doing here is changing existing hardware without any
explanation why.

>>
>>> +    type: object
>>> +    additionalProperties: false
>>> +
>>> +    description:
>>> +      Non pixel context bank is needed when video hardware have distinct iommus
>>> +      for non pixel buffers. Non pixel buffers are mainly compressed and
>>> +      internal buffers.
>>> +
>>> +    properties:
>>> +      iommus:
>>> +        maxItems: 1
>>> +
>>> +      memory-region:
>>> +        maxItems: 1
>>> +
>>> +    required:
>>> +      - iommus
>>> +      - memory-region
>>> +
>>>    operating-points-v2: true
>>>  
>>>    opp-table:
>>> @@ -86,6 +107,7 @@ required:
>>>  
>>>  allOf:
>>>    - $ref: qcom,venus-common.yaml#
>>> +  - $ref: /schemas/reserved-memory/reserved-memory.yaml
>>
>> This makes no sense. how is this device a reserved memory?
> Again, explained the "excluded" portion from IOVA part in commit description.
> For such excluded region, reserved memory would be needed. I have followed the
> adsp example in the reserved-memory schema[1], its same for iris.
> 
> [1]
> https://github.com/devicetree-org/dt-schema/blame/main/dtschema/schemas/reserved-memory/reserved-memory.yaml

Read the title there.


>>
>>>    - if:
>>>        properties:
>>>          compatible:
>>> @@ -117,6 +139,16 @@ examples:
>>>      #include <dt-bindings/power/qcom-rpmpd.h>
>>>      #include <dt-bindings/power/qcom,rpmhpd.h>
>>>  
>>> +    reserved-memory {
>>> +      #address-cells = <2>;
>>> +      #size-cells = <2>;
>>
>> Why do you need this?
> Was planning to drop this, as the reserved-memory region have it defined.
>>
>>> +
>>> +      iris_resv: reservation-iris {
>>
>> Mixing MMIO and non-MMIO is not the way to go. This is also not relevant
>> here. Don't embed other things into your binding example.
>>
>>
>>> +        iommu-addresses = <&iris_non_pixel 0x0 0x0 0x0 0x25800000>,
>>> +                          <&iris_non_pixel 0x0 0xe0000000 0x0 0x20000000>;
>>> +      };
>>> +    };
>>> +
>>>      video-codec@aa00000 {
>>>          compatible = "qcom,sm8550-iris";
>>>          reg = <0x0aa00000 0xf0000>;
>>> @@ -144,12 +176,16 @@ examples:
>>>          resets = <&gcc GCC_VIDEO_AXI0_CLK_ARES>;
>>>          reset-names = "bus";
>>>  
>>> -        iommus = <&apps_smmu 0x1940 0x0000>,
>>> -                 <&apps_smmu 0x1947 0x0000>;
>>> +        iommus = <&apps_smmu 0x1947 0x0000>;
>>
>> Why did the device or hardware change? Nothing explains in commit msg
>> what is wrong with existing device and existing binding.
> Same query here, the commit well explains the limitation with existing device
> and how adding a new sub node mitigates the situation.

I read it and still do not get what is wrong with existing device. Which
hardware emits different stream-ID? How does it affect users? How can I
reproduce the problem?

Remember, that you are now affecting ABI.

Best regards,
Krzysztof