[Intel-gfx] [PATCH 2/4] drm/i915/vrr: Fix guardband/vblank exit length calculation for adl+

[Navare, Manasi]

On Fri, Dec 02, 2022 at 03:44:10PM +0200, Ville Syrjala wrote:
> From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> 
> We are miscalculating both the guardband value, and the resulting
> vblank exit length on adl+. This means that our start of vblank
> (double buffered register latch point) is incorrect, and we also
> think that it's not where it actually is (hence vblank evasion/etc.
> may not work properly). Fix up the calculations to match the real
> hardware behaviour (as reverse engineered by intel_display_poller).
> 
> Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_vrr.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/display/intel_vrr.c b/drivers/gpu/drm/i915/display/intel_vrr.c
> index 6655dd2c1684..753e7b211708 100644
> --- a/drivers/gpu/drm/i915/display/intel_vrr.c
> +++ b/drivers/gpu/drm/i915/display/intel_vrr.c
> @@ -78,10 +78,10 @@ static int intel_vrr_vblank_exit_length(const struct intel_crtc_state *crtc_stat
>  	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
>  	struct drm_i915_private *i915 = to_i915(crtc->base.dev);
>  
> -	/* The hw imposes the extra scanline before frame start */
>  	if (DISPLAY_VER(i915) >= 13)
> -		return crtc_state->vrr.guardband + crtc_state->framestart_delay + 1;
> +		return crtc_state->vrr.guardband;

This makes sense since with guardband, there is no framestart delay

>  	else
> +		/* The hw imposes the extra scanline before frame start */
>  		return crtc_state->vrr.pipeline_full + crtc_state->framestart_delay + 1;
>  }
>  
> @@ -151,7 +151,7 @@ intel_vrr_compute_config(struct intel_crtc_state *crtc_state,
>  		 * number of scan lines. Assuming 0 for no DSB.
>  		 */
>  		crtc_state->vrr.guardband =
> -			crtc_state->vrr.vmin - adjusted_mode->crtc_vdisplay;
> +			crtc_state->vrr.vmin + 1 - adjusted_mode->crtc_vdisplay;

Why are we adding + 1 here? The bspec says guardband should be :
Guardband = Vmin - Vactive - Window2 where in our case Window2 = 0
If we need that + 1 to get this working, then perhaps we need to update
Bspec?

I kind of want to see if this is still breaking if we dont have that +
1?

Manasi

>  	} else {
>  		crtc_state->vrr.pipeline_full =
>  			min(255, crtc_state->vrr.vmin - adjusted_mode->crtc_vdisplay -
> -- 
> 2.37.4
>