[Intel-gfx] [PATCH v4 2/4] drm: Add aspect ratio parsing in DRM layer

[Ville Syrjälä]

On Mon, Oct 17, 2016 at 07:10:42PM +0530, Sharma, Shashank wrote:
> Regards
> 
> Shashank
> 
> 
> On 10/17/2016 6:01 PM, Ville Syrjälä wrote:
> > On Mon, Oct 17, 2016 at 05:34:38PM +0530, Shashank Sharma wrote:
> >> Current DRM layer functions don't parse aspect ratio information
> >> while converting a user mode->kernel mode or vice versa. This
> >> causes modeset to pick mode with wrong aspect ratio, eventually
> >> causing failures in HDMI compliance test cases, due to wrong VIC.
> >>
> >> This patch adds aspect ratio information in DRM's mode conversion
> >> and mode comparision functions, to make sure kernel picks mode
> >> with right aspect ratio (as per the VIC).
> >>
> >> Signed-off-by: Shashank Sharma <shashank.sharma at intel.com>
> >> Signed-off-by: Lin, Jia <lin.a.jia at intel.com>
> >> Signed-off-by: Akashdeep Sharma <akashdeep.sharma at intel.com>
> >> Reviewed-by: Jim Bride <jim.bride at linux.intel.com>
> >> Reviewed-by: Jose Abreu <Jose.Abreu at synopsys.com>
> >> Cc: Daniel Vetter <daniel.vetter at ffwll.ch>
> >> Cc: Emil Velikov <emil.l.velikov at gmail.com>
> >>
> >> V2: Addressed review comments from Sean:
> >> - Fix spellings/typo
> >> - No need to handle aspect ratio none
> >> - Add a break, for default case too
> >> V3: Rebase
> >> V4: Added r-b from Jose
> >> ---
> >>   drivers/gpu/drm/drm_modes.c | 31 +++++++++++++++++++++++++++++++
> >>   1 file changed, 31 insertions(+)
> >>
> >> diff --git a/drivers/gpu/drm/drm_modes.c b/drivers/gpu/drm/drm_modes.c
> >> index 53f07ac..fde927a 100644
> >> --- a/drivers/gpu/drm/drm_modes.c
> >> +++ b/drivers/gpu/drm/drm_modes.c
> >> @@ -997,6 +997,7 @@ bool drm_mode_equal_no_clocks_no_stereo(const struct drm_display_mode *mode1,
> >>   	    mode1->vsync_end == mode2->vsync_end &&
> >>   	    mode1->vtotal == mode2->vtotal &&
> >>   	    mode1->vscan == mode2->vscan &&
> >> +	    mode1->picture_aspect_ratio == mode2->picture_aspect_ratio &&
> >>   	    (mode1->flags & ~DRM_MODE_FLAG_3D_MASK) ==
> >>   	     (mode2->flags & ~DRM_MODE_FLAG_3D_MASK))
> >>   		return true;
> >> @@ -1499,6 +1500,21 @@ void drm_mode_convert_to_umode(struct drm_mode_modeinfo *out,
> >>   	out->vrefresh = in->vrefresh;
> >>   	out->flags = in->flags;
> >>   	out->type = in->type;
> >> +	out->flags &= ~DRM_MODE_FLAG_PIC_AR_MASK;
> >> +
> >> +	switch (in->picture_aspect_ratio) {
> >> +	case HDMI_PICTURE_ASPECT_4_3:
> >> +		out->flags |= DRM_MODE_FLAG_PIC_AR_4_3;
> >> +		break;
> >> +	case HDMI_PICTURE_ASPECT_16_9:
> >> +		out->flags |= DRM_MODE_FLAG_PIC_AR_16_9;
> >> +		break;
> >> +	case HDMI_PICTURE_ASPECT_RESERVED:
> >> +	default:
> >> +		out->flags |= DRM_MODE_FLAG_PIC_AR_NONE;
> >> +		break;
> >> +	}
> >> +
> >>   	strncpy(out->name, in->name, DRM_DISPLAY_MODE_LEN);
> >>   	out->name[DRM_DISPLAY_MODE_LEN-1] = 0;
> >>   }
> >> @@ -1544,6 +1560,21 @@ int drm_mode_convert_umode(struct drm_display_mode *out,
> >>   	strncpy(out->name, in->name, DRM_DISPLAY_MODE_LEN);
> >>   	out->name[DRM_DISPLAY_MODE_LEN-1] = 0;
> >>   
> >> +	/* Clearing picture aspect ratio bits from out flags */
> >> +	out->flags &= ~DRM_MODE_FLAG_PIC_AR_MASK;
> >> +
> >> +	switch (in->flags & DRM_MODE_FLAG_PIC_AR_MASK) {
> >> +	case DRM_MODE_FLAG_PIC_AR_4_3:
> >> +		out->picture_aspect_ratio |= HDMI_PICTURE_ASPECT_4_3;
> >> +		break;
> >> +	case DRM_MODE_FLAG_PIC_AR_16_9:
> >> +		out->picture_aspect_ratio |= HDMI_PICTURE_ASPECT_16_9;
> >> +		break;
> >> +	default:
> >> +		out->picture_aspect_ratio = HDMI_PICTURE_ASPECT_NONE;
> >> +		break;
> >> +	}
> > Hmm. So now we have the mode aspect ratio infromation duplicated
> > in two different places. Not sure that won't lead to some confusion.
> In drm layer, this is the only place. Actually till now, DRM layer dint 
> have the support for aspect ratio at all. This was causing
> HDMI compliance tests like 7-27 fail, which expect a particular unique 
> VIC in AVI infoframe on modeset.
> 
> I have given some details about this in cover-letter.
> > Although we do want the user to be able to override via the property I
> > suppose, so we'd have to change that (+ the inforframe code) to
> > look at the mode flags instead if we would eliminate
> > 'picture_aspect_ratio'.
> I had a small discussion on this with Daniel, and we discussed that it 
> doesnt make sense to override just the aspect ratio if the monitor 
> doesn't even support that aspect ratio.
> And currently the was how this property is implemented is, we just 
> override the aspect ratio without any validity check.
> 
> Now as we have all the supported aspect ratio added properly in the mode 
> info itself, we need not to have this property at all, So Daniel 
> suggested me to remove that property too.
> >
> > And that brings me to the other point. At least i915 will simply
> > override the mode's aspect ration with the property. So this looks like
> > a big no-op for now on i915.
> Yes, This is a bug in I915. When I published the first version of this 
> series, I had one patch, which was overriding the value only when the 
> property is set.
> This should be the right case. And then Daniel suggested to remove the 
> property all together (and it makes sense as we have proper aspect 
> ratios in mode information
> itself) So I kept that patch separate, to be merged separately.

Yeah, removing the property entirely seems like an OKish solution since
userspace can just stuff that information into the mode flags instead.
The only slight concern is that someone's setup might depend on the
property, and now we're removing it.

> >   It's looking like we might need a new
> > propetty value to differentiate between "auto" and "none" for real.
> Now we have the exact aspect ratio given by monitor EDID, so IMHO it 
> would be better if we can just have real aspect.

Well, "real" isn't quite the right term. It's still a user specified
thing, which means the user can ask for somehting the display didn't
even advertise. Which is fine as such, however since the AVI infoframe
lacks the capability to transmit these new aspect ratios we have a bit
of problem on our hands if the user asks for something we can't even
tell the display to do. I guess we would just need to return an error
to the user in that case.

-- 
Ville Syrjälä
Intel OTC