[Intel-gfx] [PATCH v4 2/2] drm: Add HDMI 2.0 VIC support for AVI info-frames

[Ville Syrjälä]

On Thu, Mar 23, 2017 at 06:31:33PM +0200, Sharma, Shashank wrote:
> Regards
> 
> Shashank
> 
> 
> On 3/23/2017 5:52 PM, Jose Abreu wrote:
> > Hi Ville,
> >
> >
> > On 23-03-2017 15:36, Ville Syrjälä wrote:
> >> On Thu, Mar 23, 2017 at 05:14:19PM +0200, Shashank Sharma wrote:
> >>> HDMI 1.4b support the CEA video modes as per range of CEA-861-D (VIC 1-64).
> >>> For any other mode, the VIC filed in AVI infoframes should be 0.
> >>> HDMI 2.0 sinks, support video modes range as per CEA-861-F spec, which is
> >>> extended to (VIC 1-107).
> >>>
> >>> This patch adds a bool input variable, which indicates if the connected
> >>> sink is a HDMI 2.0 sink or not. This will make sure that we don't pass a
> >>> HDMI 2.0 VIC to a HDMI 1.4 sink.
> >> The spec is unfortunately vague when it comes to the CEA-861-F VIC
> >> transmission when there is a corresponding HDMI VIC for the same mode.
> >> I'm not sure if it's telling us to set both or just one depending on
> >> whether we're transmitting 3D video or not. Or at least I can't parse
> >> that information from the spec. Anyone have a better crystal ball
> >> in their possession?
> >>
> > I've been working in HDMI receivers and this is what I've got in
> > a comment:
> >
> > 1282
> > /*
> >
> > 1283          * Update current VIC: When transmitting any
> > extended video format
> > 1284          * indicated through use of the HDMI_VIC field in
> > the HDMI Vendor
> > 1285          * Specific InfoFrame or any other format which is
> > not described in
> > 1286          * the above cases, an HDMI Source shall set the AVI
> > InfoFrame VIC
> > 1287          * field to
> > zero.
> >
> > 1288          */
> >
> > This was directly taken from the spec, can't remember exactly
> > were though.
> >
> > So, the VIC in AVIIF must be set to 0 and the HDMI_VIC field in
> > VSIF shall be set to the HDMI 4k VIC.
> >
> > Best regards,
> > Jose Miguel Abreu
> Even my understanding of the two specs seems similar
> If its a HDMI 1.4b monitor, 2D mode
>      - VIC field in AVI_IF should be set to appropriate VIC [if VIC is 
> not listed in CEA-861-D (VIC 1-64), make VIC=0]
> If its a HDMI 1.4b monitor, 3D mode
>      - VIC field in AVI_IF should be set to appropriate VIC, in 
> conjunction with 3D_structure field in HDMI VSIF
> If its a HDMI 2.0 monitor, 2D mode
>      - VIC filed in the AVI_IF should be set to appropriate VIC (1-107)
> If its a HDMI 2.0 monitor, 3D mode
>      - VIC field in AVI_IF should be set to appropriate VIC, in 
> conjunction with 3D_structure field in HDMI VSIF

You're overlooking the HDMI VIC entirely in that list.

OK, so our code even refuses to send both 3D stuff and HDMI VIC
at the same time. I presume this was an illegal combination in HDMI 1.4.
And since there was no overlap between the CEA VICs and HDMI VICs this
was all that we needed.

But now that there is overlap, I think we'll need to set CEA VIC=0
whenever HDMI VIC!=0. Or at least that's my strictest interpretation of
the spec. I wish they would have stated this stuff more explicitly.

So either we need to pass the AVI IF to
drm_hdmi_vendor_infoframe_from_display_mode() and have it clear the AVI
IF VIC if HDMI VIC is specified, or we need to pass the HDMI IF to
drm_hdmi_avi_infoframe_from_display_mode() and have it not set the AVI
IF VIC if HDMI VIC is specified. Or we need to make the caller handle
this, which would probably lead to more bugs.

-- 
Ville Syrjälä
Intel OTC