[Freedreno] RFC: DSI host capabilities (was: [PATCH RFC 03/10] drm/panel: Add LGD panel driver for Sony Xperia XZ3)

[Dmitry Baryshkov]

On 05/07/2023 16:29, Maxime Ripard wrote:
> On Wed, Jul 05, 2023 at 03:05:33PM +0200, Neil Armstrong wrote:
>> On 05/07/2023 14:04, Maxime Ripard wrote:
>>> Hi,
>>>
>>> On Tue, May 30, 2023 at 03:36:04PM +0300, Dmitry Baryshkov wrote:
>>>> On 30/05/2023 15:15, AngeloGioacchino Del Regno wrote:
>>>>> Il 30/05/23 13:44, Dmitry Baryshkov ha scritto:
>>>>>> On Tue, 30 May 2023 at 10:24, Neil Armstrong
>>>>>> <neil.armstrong at linaro.org> wrote:
>>>>>>>
>>>>>>> Hi Marijn, Dmitry, Caleb, Jessica,
>>>>>>>
>>>>>>> On 29/05/2023 23:11, Marijn Suijten wrote:
>>>>>>>> On 2023-05-22 04:16:20, Dmitry Baryshkov wrote:
>>>>>>>> <snip>
>>>>>>>>>> +   if (ctx->dsi->dsc) {
>>>>>>>>>
>>>>>>>>> dsi->dsc is always set, thus this condition can be dropped.
>>>>>>>>
>>>>>>>> I want to leave room for possibly running the panel without DSC (at a
>>>>>>>> lower resolution/refresh rate, or at higher power consumption if there
>>>>>>>> is enough BW) by not assigning the pointer, if we get access to panel
>>>>>>>> documentation: probably one of the magic commands sent in this driver
>>>>>>>> controls it but we don't know which.
>>>>>>>
>>>>>>> I'd like to investigate if DSC should perhaps only be enabled if we
>>>>>>> run non certain platforms/socs ?
>>>>>>>
>>>>>>> I mean, we don't know if the controller supports DSC and those
>>>>>>> particular
>>>>>>> DSC parameters so we should probably start adding something like :
>>>>>>>
>>>>>>> static drm_dsc_config dsc_params_qcom = {}
>>>>>>>
>>>>>>> static const struct of_device_id panel_of_dsc_params[] = {
>>>>>>>            { .compatible = "qcom,sm8150", , .data = &dsc_params_qcom },
>>>>>>>            { .compatible = "qcom,sm8250", , .data = &dsc_params_qcom },
>>>>>>>            { .compatible = "qcom,sm8350", , .data = &dsc_params_qcom },
>>>>>>>            { .compatible = "qcom,sm8450", , .data = &dsc_params_qcom },
>>>>>>> };
>>>>>>
>>>>>> I think this would damage the reusability of the drivers. The panel
>>>>>> driver does not actually care if the SoC is SM8350, sunxi-something or
>>>>>> RCar.
>>>>>> Instead it cares about host capabilities.
>>>>>>
>>>>>> I think instead we should extend mipi_dsi_host:
>>>>>>
>>>>>> #define MIPI_DSI_HOST_MODE_VIDEO BIT(0)
>>>>>> #define MIPI_DSI_HOST_MODE_CMD  BIT(1)
>>>>>> #define MIPI_DSI_HOST_VIDEO_SUPPORTS_COMMANDS BIT(2)
>>>>>> // FIXME: do we need to provide additional caps here ?
>>>>>>
>>>>>> #define MIPI_DSI_DSC_1_1 BIT(0)
>>>>>> #define MIPI_DSI_DSC_1_2 BIT(1)
>>>>>> #define MIPI_DSI_DSC_NATIVE_422 BIT(2)
>>>>>> #define MIPI_DSI_DSC_NATIVE_420 BIT(3)
>>>>>> #define MIPI_DSI_DSC_FRAC_BPP BIT(4)
>>>>>> // etc.
>>>>>>
>>>>>> struct mipi_dsi_host {
>>>>>>     // new fields only
>>>>>>      unsigned long mode_flags;
>>>>>>      unsigned long dsc_flags;
>>>>>> };
>>>>>>
>>>>>> Then the panel driver can adapt itself to the host capabilities and
>>>>>> (possibly) select one of the internally supported DSC profiles.
>>>>>>
>>>>>
>>>>> I completely agree about extending mipi_dsi_host, other SoCs could reuse
>>>>> that and
>>>>> support for DSC panels would become a lot cleaner.
>>>>
>>>> Sounds good. I will wait for one or two more days (to get the possible
>>>> feedback on fields/flags/etc) and post an RFC patch to dri-devel.
>>>
>>> I just came across that discussion, and couldn't find those patches, did
>>> you ever send them?

No, I got sidetracked by other issues.

>>>
>>> Either way, I'm not really sure it's a good idea to multiply the
>>> capabilities flags of the DSI host, and we should just stick to the
>>> spec. If the spec says that we have to support DSC while video is
>>> output, then that's what the panels should expect.
>>
>> Except some panels supports DSC & non-DSC, Video and Command mode, and
>> all that is runtime configurable. How do you handle that ?
> 
> In this case, most of the constraints are going to be on the encoder
> still so it should be the one driving it. The panel will only care about
> which mode has been selected, but it shouldn't be the one driving it,
> and thus we still don't really need to expose the host capabilities.

This is an interesting perspective. This means that we can and actually 
have to extend the drm_display_mode with the DSI data and compression 
information.

For example, the panel that supports all four types for the 1080p should 
export several modes:

1920x1080-command
1920x1080-command-DSC
1920x1080-video
1920x1080-video-DSC

where video/command and DSC are some kinds of flags and/or information 
in the drm_display_mode? Ideally DSC also has several sub-flags, which 
denote what kind of configuration is supported by the DSC sink (e.g. 
bpp, yuv, etc).

Another option would be to get this handled via the bus format 
negotiation, but that sounds like worse idea to me.

> This is very much like HDMI: the encoder knows what the monitor is
> capable of, will take a decision based on its capabilities and the
> monitor's and will then let the monitor know. But the monitor never
> knows what the encoder is truly capable of, nor will it enforce
> something.
> 
> Maxime

-- 
With best wishes
Dmitry