dw_hdmi is showing wrong colour after commit 7cd70656d1285b79("drm/bridge: display-connector: implement bus fmts callbacks")
Neil Armstrong
narmstrong at baylibre.com
Fri Jan 14 08:23:51 UTC 2022
Hi,
On 13/01/2022 21:01, Fabio Estevam wrote:
> Hi Biju,
>
> On Thu, Jan 13, 2022 at 2:45 PM Biju Das <biju.das.jz at bp.renesas.com> wrote:
>>
>> Hi All,
>>
>> RZ/G2{H, M, N} SoC has dw_hdmi IP and it was working ok(colour) till the commit
>> 7cd70656d1285b79("drm/bridge: display-connector: implement bus fmts callbacks").
>>
>> After this patch, the screen becomes greenish(may be it is setting it into YUV format??).
>>
>> By checking the code, previously it used to call get_input_fmt callback and set colour as RGB24.
>>
>> After this commit, it calls get_output_fmt_callbck and returns 3 outputformats(YUV16, YUV24 and RGB24)
>> And get_input_fmt callback, I see the outputformat as YUV16 instead of RGB24.
>>
>> Not sure, I am the only one seeing this issue with dw_HDMI driver.
This patch was introduced to maintain the bridge color format negotiation after using DRM_BRIDGE_ATTACH_NO_CONNECTOR,
but it seems it behaves incorrectly if the first bridge doesn't implement the negotiation callbacks.
Let me check the code to see how to fix that.
>
> I have tested linux-next 20220112 on a imx6q-sabresd board, which shows:
>
> dwhdmi-imx 120000.hdmi: Detected HDMI TX controller v1.30a with HDCP
> (DWC HDMI 3D TX PHY)
>
> The colors are shown correctly here.
>
The imx doesn't use DRM_BRIDGE_ATTACH_NO_CONNECTOR so the negotiation fails and use the RGB fallback input & output format.
Anyway thanks for testing
Neil
More information about the dri-devel
mailing list