[v2,0/3] Enable JPEG Encoder on RK3566/RK3568

Hello,

the following series adds support for and enables one of the hardware
video encoders on the RK3566 and RK3568 line of SoCs by Rockchip,
initially just for the JPEG format in line with what the kernel supports.

The encoder block is separate from the Hantro decoder instance, as they
are in different power domains and have wildly different memory addresses
as well.

The encoder hardware seemingly also supports VP8 and H.264 in addition
to just JPEG, as is evident from both the downstream vendor stack and the
register listing in the TRM. The hantro driver in Linux, however, does
not yet support encoding these formats.

The first patch modifies the bindings with a new compatible, and adds
the ability to just have a vepu interrupt without a vdpu interrupt.

The second patch makes the actual driver changes to support this variant.

The third and final patch makes the necessary device tree changes for
the rk356x device tree file to add both the node for the encoder and
its MMU.

The series has been tested on a PINE64 Quartz64 Model A with an RK3566
SoC using GStreamer.

Below you'll also find an interdiff against V1, which falsely assumed
this encoder instance only did JPEG.

Regards,
Nicolas Frattaroli

Changes in v2:
 - rename compatible as it's not JPEG only
 - rename device tree nodes as it's not JPEG only
 - reword commits as it's not JPEG only
 - get rid of a whole bunch of redundant struct definitions, as, you
   guessed it, it's not JPEG only

Nicolas Frattaroli (3):
  dt-bindings: media: rockchip-vpu: Add RK3568 VEPU compatible
  media: hantro: Add support for RK356x encoder
  arm64: dts: rockchip: Add Hantro encoder node to rk356x

 .../bindings/media/rockchip-vpu.yaml          |  2 ++
 arch/arm64/boot/dts/rockchip/rk356x.dtsi      | 21 ++++++++++++++++
 drivers/staging/media/hantro/hantro_drv.c     |  1 +
 drivers/staging/media/hantro/hantro_hw.h      |  1 +
 .../staging/media/hantro/rockchip_vpu_hw.c    | 25 +++++++++++++++++++
 5 files changed, 50 insertions(+)

Interdiff against v1:

On Montag, 9. Mai 2022 09:25:23 CEST Krzysztof Kozlowski wrote:
> On 08/05/2022 22:25, Nicolas Frattaroli wrote:
> > The RK3568 and RK3566 have a Hantro VPU node solely dedicated to
> > encoding. This patch adds a compatible for it, and also allows
> > the bindings to only come with a vepu interrupt.
> > 
> > Signed-off-by: Nicolas Frattaroli <frattaroli.nicolas@gmail.com>
> > ---
> >  Documentation/devicetree/bindings/media/rockchip-vpu.yaml | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/Documentation/devicetree/bindings/media/rockchip-vpu.yaml b/Documentation/devicetree/bindings/media/rockchip-vpu.yaml
> > index bacb60a34989..4045f107ca4e 100644
> > --- a/Documentation/devicetree/bindings/media/rockchip-vpu.yaml
> > +++ b/Documentation/devicetree/bindings/media/rockchip-vpu.yaml
> > @@ -22,6 +22,7 @@ properties:
> >            - rockchip,rk3288-vpu
> >            - rockchip,rk3328-vpu
> >            - rockchip,rk3399-vpu
> > +          - rockchip,rk3568-vepu
> >            - rockchip,px30-vpu
> >        - items:
> >            - const: rockchip,rk3188-vpu
> > @@ -40,6 +41,7 @@ properties:
> >    interrupt-names:
> >      oneOf:
> >        - const: vdpu
> > +      - const: vepu
> 
> This should be enum (for both lines above) and you should add
> allOf:if:then with a constraints which variant can have which interrupts.
> 
> 
> Best regards,
> Krzysztof
> 

So something like this?

  interrupt-names:
    oneOf:
      - enum:
         - vdpu
         - vepu
      - items:
          - const: vepu
          - const: vdpu

What's the difference between a list of consts and an enum here?
I'm not very familiar with dt-schema, my apologies.

Also, since I don't know which of the other variants can have
the encoding interrupt and this wasn't brought up until now, I think
my solution will be to have a check for -vepu in the compatible and in
that case require that only the vepu interrupt is present, if that's
alright with you.

Regards,
Nicolas Frattaroli

On 09/05/2022 11:24, Nicolas Frattaroli wrote:
> On Montag, 9. Mai 2022 09:25:23 CEST Krzysztof Kozlowski wrote:
>> On 08/05/2022 22:25, Nicolas Frattaroli wrote:
>>> The RK3568 and RK3566 have a Hantro VPU node solely dedicated to
>>> encoding. This patch adds a compatible for it, and also allows
>>> the bindings to only come with a vepu interrupt.
>>>
>>> Signed-off-by: Nicolas Frattaroli <frattaroli.nicolas@gmail.com>
>>> ---
>>>  Documentation/devicetree/bindings/media/rockchip-vpu.yaml | 2 ++
>>>  1 file changed, 2 insertions(+)
>>>
>>> diff --git a/Documentation/devicetree/bindings/media/rockchip-vpu.yaml b/Documentation/devicetree/bindings/media/rockchip-vpu.yaml
>>> index bacb60a34989..4045f107ca4e 100644
>>> --- a/Documentation/devicetree/bindings/media/rockchip-vpu.yaml
>>> +++ b/Documentation/devicetree/bindings/media/rockchip-vpu.yaml
>>> @@ -22,6 +22,7 @@ properties:
>>>            - rockchip,rk3288-vpu
>>>            - rockchip,rk3328-vpu
>>>            - rockchip,rk3399-vpu
>>> +          - rockchip,rk3568-vepu
>>>            - rockchip,px30-vpu
>>>        - items:
>>>            - const: rockchip,rk3188-vpu
>>> @@ -40,6 +41,7 @@ properties:
>>>    interrupt-names:
>>>      oneOf:
>>>        - const: vdpu
>>> +      - const: vepu
>>
>> This should be enum (for both lines above) and you should add
>> allOf:if:then with a constraints which variant can have which interrupts.
>>
>>
>> Best regards,
>> Krzysztof
>>
> 
> So something like this?
> 
>   interrupt-names:
>     oneOf:
>       - enum:
>          - vdpu
>          - vepu
>       - items:
>           - const: vepu
>           - const: vdpu

Yes

> 
> What's the difference between a list of consts and an enum here?
> I'm not very familiar with dt-schema, my apologies.

The effect is the same, just oneOf is a bit more complicated way to
describe it.

> 
> Also, since I don't know which of the other variants can have
> the encoding interrupt and this wasn't brought up until now, I think
> my solution will be to have a check for -vepu in the compatible and in
> that case require that only the vepu interrupt is present, if that's
> alright with you.

If you meant by adding a "if" case for only rockchip,rk3568-vepu, it's ok.

Best regards,
Krzysztof