[PATCH 2/5] dt-bindings: display: atmel: add optional output-mode property

Rob Herring robh at kernel.org
Mon Apr 16 14:22:45 UTC 2018


On Sat, Apr 14, 2018 at 8:54 AM, Peter Rosin <peda at axentia.se> wrote:
> On 2018-04-13 19:46, Rob Herring wrote:
>> On Mon, Apr 09, 2018 at 12:59:15PM +0200, Peter Rosin wrote:
>>> Useful for beating cases where an output mode selection heuristic
>>> fails.
>>>
>>> Signed-off-by: Peter Rosin <peda at axentia.se>
>>> ---
>>>  Documentation/devicetree/bindings/display/atmel/hlcdc-dc.txt | 4 ++++
>>>  1 file changed, 4 insertions(+)
>>>
>>> diff --git a/Documentation/devicetree/bindings/display/atmel/hlcdc-dc.txt b/Documentation/devicetree/bindings/display/atmel/hlcdc-dc.txt
>>> index 82f2acb3d374..dc478455b883 100644
>>> --- a/Documentation/devicetree/bindings/display/atmel/hlcdc-dc.txt
>>> +++ b/Documentation/devicetree/bindings/display/atmel/hlcdc-dc.txt
>>> @@ -10,6 +10,10 @@ Required properties:
>>>   - #address-cells: should be set to 1.
>>>   - #size-cells: should be set to 0.
>>>
>>> +Optional properties:
>>> + - output-mode: override any output mode selection hueristic and force a
>>> +   particular output mode. One of "rgb444", "rgb565", "rgb666" and "rgb888".
>>> +
>>
>> This needs to be generic, not just added to some random display
>> controller binding.
>>
>> It also belongs in the port or endpoint node as is done for camera
>> interfaces.
>
> Hmm, should I extend media/video-interfaces.txt with more bus types (or since
> I'm targeting parallel interfaces, perhaps the new bus types should be
> autodetected from other props?) or should a write a new binding similar to
> it?

Ultimately, I'd like it documented in one place. It can stay in
media/video-interfaces.txt for now, but we should probably move it or
portions of it out of media/. I don't have a good suggestion other
than up a level where people will find it easily.

> One question regarding bus-width, should it include hsync/vsync/de/clk?
> If yes, how to distinguish rgb565 with all those four from rgb666 with
> only de/clk (some panels do not need hsync/vsync)? 20 lines in both
> cases...

I'd say no, it is just the data lines.

> Or are rgb444/rgb565/rgb666/rgb888 already supported by the media video
> interface binding? That's not at all obvious to me.

Not sure, I'd guess not.

Rob


More information about the dri-devel mailing list