On Tue, Jan 11, 2022 at 06:45:31PM +0200, Jani Nikula wrote:
> On Tue, 11 Jan 2022, Stanislav Lisovskiy <[email protected]> 
> wrote:
> > Currently we only recalculate CDCLK if active plane mask changes
> > or if we do a full modeset, however according to BSpec
> > required Dbuf bandwidth calculations also depend on pipe/plane
> > scaling ratio, which means that CDCLK must be recalculated
> > everytime plane scaling ratio changes, because it affects
> > display buffer bandwidth requirements.
> >
> > Signed-off-by: Stanislav Lisovskiy <[email protected]>
> > ---
> >  drivers/gpu/drm/i915/display/intel_display.c | 63 +++++++++++++++++++-
> >  1 file changed, 60 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c 
> > b/drivers/gpu/drm/i915/display/intel_display.c
> > index bf7ce684dd8e..2c616348e993 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -7499,13 +7499,65 @@ static int 
> > intel_bigjoiner_add_affected_planes(struct intel_atomic_state *state)
> >     return 0;
> >  }
> >  
> > +static bool scaling_affects_cdclk(struct intel_plane_state 
> > *old_plane_state,
> > +                             struct intel_plane_state *new_plane_state)
> > +{
> > +   int old_src_w = drm_rect_width(&old_plane_state->uapi.src) >> 16;
> > +   int old_src_h = drm_rect_height(&old_plane_state->uapi.src) >> 16;
> > +   int old_dst_w = drm_rect_width(&old_plane_state->uapi.dst);
> > +   int old_dst_h = drm_rect_height(&old_plane_state->uapi.dst);
> > +   int new_src_w = drm_rect_width(&new_plane_state->uapi.src) >> 16;
> > +   int new_src_h = drm_rect_height(&new_plane_state->uapi.src) >> 16;
> > +   int new_dst_w = drm_rect_width(&new_plane_state->uapi.dst);
> > +   int new_dst_h = drm_rect_height(&new_plane_state->uapi.dst);
> > +   int old_hscale_ratio, new_hscale_ratio;
> > +   int old_vscale_ratio, new_vscale_ratio;
> > +
> > +   if (needs_scaling(old_plane_state) != needs_scaling(new_plane_state))
> > +           return true;
> > +
> > +   if (!old_dst_w || !old_dst_h)
> > +           return true;
> > +
> > +   DRM_DEBUG_KMS("old_dst_w %d old_dst_h %d\n", old_dst_w, old_dst_h);
> > +
> > +   old_hscale_ratio = DIV_ROUND_UP(old_src_w, old_dst_w);
> > +   old_vscale_ratio = DIV_ROUND_UP(old_src_h, old_dst_h);
> > +
> > +   if (!new_dst_w || !new_dst_h)
> > +           return true;
> > +
> > +   DRM_DEBUG_KMS("new_dst_w %d new_dst_h %d\n", new_dst_w, new_dst_h);
> > +
> > +   new_hscale_ratio = DIV_ROUND_UP(new_src_w, new_dst_w);
> > +   new_vscale_ratio = DIV_ROUND_UP(new_src_h, new_dst_h);
> > +
> > +   DRM_DEBUG_KMS("new_hscale_ratio %d new_vscale_ratio %d "
> > +                 "old_hscale_ratio %d old_vscale_ratio %d\n",
> > +                 new_hscale_ratio, new_vscale_ratio,
> > +                 old_hscale_ratio, old_vscale_ratio);
> 
> All of the debug logging seem excessive? Also, please use drm_dbg_atomic
> or drm_dbg_kms instead of DRM_DEBUG_KMS for the ones that need to stay.

Yes, I should probably leave only message repoting that cdclk has to be 
recalculated. Otherwise indeed seems excessive - was just using it for
debugging.

Stan

> 
> BR,
> Jani.
> 
> > +
> > +   if ((old_hscale_ratio != new_hscale_ratio) ||
> > +       (old_vscale_ratio != new_vscale_ratio)) {
> > +           DRM_DEBUG_KMS("Scaling ratios changed from %dx%d"
> > +                         " to %dx%d - need cdclk recalc\n",
> > +                         old_hscale_ratio, old_vscale_ratio,
> > +                         new_hscale_ratio, new_vscale_ratio);
> > +           return true;
> > +   }
> > +
> > +   return false;
> > +}
> > +
> >  static int intel_atomic_check_planes(struct intel_atomic_state *state)
> >  {
> >     struct drm_i915_private *dev_priv = to_i915(state->base.dev);
> >     struct intel_crtc_state *old_crtc_state, *new_crtc_state;
> >     struct intel_plane_state *plane_state;
> > +   struct intel_plane_state *old_plane_state;
> >     struct intel_plane *plane;
> >     struct intel_crtc *crtc;
> > +   bool need_cdclk_calc = false;
> >     int i, ret;
> >  
> >     ret = icl_add_linked_planes(state);
> > @@ -7516,7 +7568,7 @@ static int intel_atomic_check_planes(struct 
> > intel_atomic_state *state)
> >     if (ret)
> >             return ret;
> >  
> > -   for_each_new_intel_plane_in_state(state, plane, plane_state, i) {
> > +   for_each_oldnew_intel_plane_in_state(state, plane, old_plane_state, 
> > plane_state, i) {
> >             ret = intel_plane_atomic_check(state, plane);
> >             if (ret) {
> >                     drm_dbg_atomic(&dev_priv->drm,
> > @@ -7524,6 +7576,9 @@ static int intel_atomic_check_planes(struct 
> > intel_atomic_state *state)
> >                                    plane->base.base.id, plane->base.name);
> >                     return ret;
> >             }
> > +
> > +           if (scaling_affects_cdclk(old_plane_state, plane_state))
> > +                   need_cdclk_calc = true;
> >     }
> >  
> >     for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
> > @@ -7539,18 +7594,20 @@ static int intel_atomic_check_planes(struct 
> > intel_atomic_state *state)
> >              * the planes' minimum cdclk calculation. Add such planes
> >              * to the state before we compute the minimum cdclk.
> >              */
> > -           if (!active_planes_affects_min_cdclk(dev_priv))
> > +           if (!active_planes_affects_min_cdclk(dev_priv) && 
> > !need_cdclk_calc)
> >                     continue;
> >  
> >             old_active_planes = old_crtc_state->active_planes & 
> > ~BIT(PLANE_CURSOR);
> >             new_active_planes = new_crtc_state->active_planes & 
> > ~BIT(PLANE_CURSOR);
> >  
> > -           if (hweight8(old_active_planes) == hweight8(new_active_planes))
> > +           if ((hweight8(old_active_planes) == 
> > hweight8(new_active_planes)) && !need_cdclk_calc)
> >                     continue;
> >  
> >             ret = intel_crtc_add_planes_to_state(state, crtc, 
> > new_active_planes);
> >             if (ret)
> >                     return ret;
> > +
> > +
> >     }
> >  
> >     return 0;
> 
> -- 
> Jani Nikula, Intel Open Source Graphics Center

Reply via email to