[Intel-gfx] [v2] drm/i915/display: Add an extra vblank wait before fbc activation

[Shankar, Uma]

> -----Original Message-----
> From: Anshuman Gupta <anshuman.gupta at intel.com>
> Sent: Monday, August 17, 2020 5:49 PM
> To: Shankar, Uma <uma.shankar at intel.com>
> Cc: intel-gfx at lists.freedesktop.org
> Subject: Re: [Intel-gfx] [v2] drm/i915/display: Add an extra vblank wait before fbc
> activation
> 
> On 2020-08-17 at 16:55:29 +0530, Uma Shankar wrote:
> > Add an extra vblank before fbc is activated.
> > WA: 1409689360
> > Corruption with FBC around plane 1A enabling. In the Frame Buffer
> > Compression programming sequence "Display Plane Enabling with FBC"
> > add a wait for vblank between plane enabling step 1 and FBC enabling
> > step 2.
> >
> > v2: Add wait only for active crtc
> >
> > Signed-off-by: Uma Shankar <uma.shankar at intel.com>
> > Signed-off-by: Stanislav Lisovskiy <stanislav.lisovskiy at intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_fbc.c | 7 +++++--
> >  1 file changed, 5 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_fbc.c
> > b/drivers/gpu/drm/i915/display/intel_fbc.c
> > index 2ab32e6532ff..2a9d4796c4a8 100644
> > --- a/drivers/gpu/drm/i915/display/intel_fbc.c
> > +++ b/drivers/gpu/drm/i915/display/intel_fbc.c
> > @@ -1085,10 +1085,13 @@ static void __intel_fbc_post_update(struct
> intel_crtc *crtc)
> >  	if (!intel_fbc_can_activate(crtc))
> >  		return;
> >
> > -	if (!fbc->busy_bits)
> > +	if (!fbc->busy_bits) {
> > +		if (IS_TIGERLAKE(dev_priv))
> > +			intel_wait_for_vblank_if_active(dev_priv, crtc->pipe);
> IMHO with this we are also adding a VBLANK wait when fbc is getting enabled
> from frontbuffer flush flow intel_fbc_flush(), i am being afraid if there could be
> unnecessary VBLANK wait here.

This is just ensuring that we let the update happen after vblank go, so that double buffered
registers get latched to the new values of watermarks etc, later the fbc gets activated.

This is a bit delicate I agree, and we need to check for corner cases if this causes any issue.
Best solution for this will be to activate fbc very close to vblank or within the vblank if we want to be
100% sure.

Regards,
Uma Shankar
 
> Thanks,
> Anshuman Gupta.
> >  		intel_fbc_hw_activate(dev_priv);
> > -	else
> > +	} else {
> >  		intel_fbc_deactivate(dev_priv, "frontbuffer write");
> > +	}
> >  }
> >
> >  void intel_fbc_post_update(struct intel_atomic_state *state,
> > --
> > 2.22.0
> >
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx at lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/intel-gfx