On Fri, May 22, 2026 at 02:32:03PM +0200, Nicolas Frattaroli wrote:
> In anticipation of userspace being able to explicitly select supported
> sink formats, add handling of the YCBCR444 sink format. The AUTO path
> does not choose this format, but with explicit format selection added to
> the driver, it becomes a possibility.
>
> Check for both source and sink support of YCBCR444 in
> intel_dp_sink_format_valid.
>
> Acked-by: Daniel Stone
> Signed-off-by: Nicolas Frattaroli
> ---
> drivers/gpu/drm/i915/display/intel_dp.c | 21 +
> 1 file changed, 21 insertions(+)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_dp.c
> b/drivers/gpu/drm/i915/display/intel_dp.c
> index 1920d2f02666..143ed85224be 100644
> --- a/drivers/gpu/drm/i915/display/intel_dp.c
> +++ b/drivers/gpu/drm/i915/display/intel_dp.c
> @@ -1344,6 +1344,20 @@ intel_dp_mode_valid_downstream(struct intel_connector
> *connector,
>8, sink_format, true);
> }
>
> +static bool
> +intel_dp_can_ycbcr444(struct intel_dp *intel_dp)
> +{
> + if (source_can_output(intel_dp, INTEL_OUTPUT_FORMAT_YCBCR444) &&
> + !drm_dp_is_branch(intel_dp->dpcd))
> + return true;
> +
> + if (source_can_output(intel_dp, INTEL_OUTPUT_FORMAT_RGB) &&
> + dfp_can_convert_from_rgb(intel_dp, INTEL_OUTPUT_FORMAT_YCBCR444))
IIRC my previous conclusion was that we don't want the dfp convert
stuff here, at least initially. So this should just are about
source_can_output().
> + return true;
> +
> + return false;
> +}
> +
> static enum drm_mode_status
> intel_dp_sink_format_valid(struct intel_connector *connector,
> const struct drm_display_mode *mode,
> @@ -1364,6 +1378,13 @@ intel_dp_sink_format_valid(struct intel_connector
> *connector,
>
> return MODE_OK;
Again, would prefer a cleanr 420->444->rgb order.
> case INTEL_OUTPUT_FORMAT_RGB:
> + return MODE_OK;
> + case INTEL_OUTPUT_FORMAT_YCBCR444:
> + if (!(info->color_formats &
> BIT(DRM_OUTPUT_COLOR_FORMAT_YCBCR444)))
> + return MODE_BAD;
> + if (!intel_dp_can_ycbcr444(intel_dp))
> + return MODE_BAD;
These two ifs are swapped when compared to the HDMI version for no
good reason that I can see.
And missing the has_hdmi_sink check here as well.
> +
> return MODE_OK;
> default:
> MISSING_CASE(sink_format);
>
> --
> 2.54.0
--
Ville Syrjälä
Intel
