Re: [PATCH 1/2] drm/display: hdmi: Only allow BPC values of 8, 10, 12 and 16

[Nicolas Frattaroli <nicolas.frattaroli@collabora.com>]

On Monday, 8 June 2026 13:46:27 Central European Summer Time Nicolas Frattaroli wrote:
> On Monday, 8 June 2026 13:30:22 Central European Summer Time sashiko-bot@kernel.org wrote:
> > Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> > 
> > Pre-existing issues:
> > - [High] The function `sink_supports_format_bpc` explicitly permits 16 BPC for RGB444 and YCBCR444 formats without verifying if the sink's EDID actually supports 48-bit deep color.
> > --
> > 
> > commit 767fa3cc832e8b93f697ad91bcc66e5b3b1ed85d
> > Author: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
> > 
> > drm/display: hdmi: Only allow BPC values of 8, 10, 12 and 16
> > 
> > As per the comment in sink_supports_format_bpc(), CTA-861-F defines that
> > only bits-per-channel values of 8, 10, 12 and 16 are allowed for HDMI.
> > Allowing more than this has surprising consequences for the atomic check
> > phase.
> > 
> > > diff --git a/drivers/gpu/drm/display/drm_hdmi_state_helper.c b/drivers/gpu/drm/display/drm_hdmi_state_helper.c
> > > index 4867edbf2622b..b06499dbfe8b3 100644
> > > --- a/drivers/gpu/drm/display/drm_hdmi_state_helper.c
> > > +++ b/drivers/gpu/drm/display/drm_hdmi_state_helper.c
> > > @@ -419,6 +419,16 @@ sink_supports_format_bpc(const struct drm_connector *connector,
> > >  		return false;
> > >  	}
> > >  
> > > +	switch (bpc) {
> > > +	case 8:
> > > +	case 10:
> > > +	case 12:
> > > +	case 16:
> > 
> > [Severity: High]
> > This is a pre-existing issue, but does explicitly allowing 16 BPC here
> > expose a missing validation check for 48-bit deep color further down in
> > sink_supports_format_bpc()?
> > 
> > When checking DRM_OUTPUT_COLOR_FORMAT_RGB444 and
> > DRM_OUTPUT_COLOR_FORMAT_YCBCR444, the code validates 10 BPC and 12 BPC
> > against the EDID flags, but completely omits checking 16 BPC against
> > DRM_EDID_HDMI_DC_48:
> > 
> > drivers/gpu/drm/display/drm_hdmi_state_helper.c:sink_supports_format_bpc() {
> > 	...
> > 	case DRM_OUTPUT_COLOR_FORMAT_RGB444:
> > 		...
> > 		if (bpc == 10 && !(info->edid_hdmi_rgb444_dc_modes & DRM_EDID_HDMI_DC_30)) {
> > 			drm_dbg_kms(dev, "10 BPC but sink doesn't support Deep Color 30.\n");
> > 			return false;
> > 		}
> > 
> > 		if (bpc == 12 && !(info->edid_hdmi_rgb444_dc_modes & DRM_EDID_HDMI_DC_36)) {
> > 			drm_dbg_kms(dev, "12 BPC but sink doesn't support Deep Color 36.\n");
> > 			return false;
> > 		}
> > 
> > 		drm_dbg_kms(dev, "RGB format supported in that configuration.\n");
> > 		return true;
> > 	...
> > }
> > 
> > If the TMDS clock limit permits it, the driver might incorrectly conclude
> > that the sink supports 16 BPC and program the hardware accordingly, which
> > could result in a black screen if the monitor doesn't actually support
> > 48-bit deep color.
> > 
> > Should a check for (info->edid_hdmi_rgb444_dc_modes & DRM_EDID_HDMI_DC_48)
> > be added for the RGB444 and YCBCR444 cases?
> 
> Finally, we have automated the ultimate Linux kernel review response:
> "while you're at it, can you fix everything else wrong in this code?"
> 
> (No, this does not expose any pre-existing issue; the validation is
> more restrictive than it was before. That it's still not restrictive
> enough in all cases does not introduce a new problem.)

Correction: there isn't even any problem here. DC48 is the highest
bpc, so the check is redundant anyway, which is likely also why it
was omitted.

Bad clanker! Bad bad bad!

> 
> > 
> > > +		break;
> > > +	default:
> > > +		return false;
> > > +	}
> > > +
> > >  	if (!info->is_hdmi &&
> > >  	    (format != DRM_OUTPUT_COLOR_FORMAT_RGB444 || bpc != 8)) {
> > >  		drm_dbg_kms(dev, "DVI Monitors require an RGB output at 8 bpc\n");
> > 
> > 
> 
>