On Wed, 20 Dec 2017 12:26:51 +0000 Daniel Stone <[email protected]> wrote:
> When trying to assign planes, keep track of the areas which are > already occluded, and ignore views which are completely occluded. This > allows us to build a state using planes only, when there are occluded > views which cannot go into a plane behind views which can. > > Signed-off-by: Daniel Stone <[email protected]> > --- > libweston/compositor-drm.c | 37 ++++++++++++++++++++++++++++++++----- > 1 file changed, 32 insertions(+), 5 deletions(-) > > diff --git a/libweston/compositor-drm.c b/libweston/compositor-drm.c > index 845f43e5b..71027a5ae 100644 > --- a/libweston/compositor-drm.c > +++ b/libweston/compositor-drm.c > @@ -3054,7 +3054,7 @@ drm_output_propose_state(struct weston_output > *output_base, > struct drm_output *output = to_drm_output(output_base); > struct drm_output_state *state; > struct weston_view *ev; > - pixman_region32_t surface_overlap, renderer_region; > + pixman_region32_t surface_overlap, renderer_region, occluded_region; > struct weston_plane *primary = &output_base->compositor->primary_plane; > > assert(!output->state_last); > @@ -3076,9 +3076,12 @@ drm_output_propose_state(struct weston_output > *output_base, > * as we do for flipping full screen surfaces. > */ > pixman_region32_init(&renderer_region); > + pixman_region32_init(&occluded_region); > > wl_list_for_each(ev, &output_base->compositor->view_list, link) { > struct weston_plane *next_plane = NULL; > + pixman_region32_t clipped_view; > + bool occluded = false; > > /* If this view doesn't touch our output at all, there's no > * reason to do anything with it. */ > @@ -3090,13 +3093,27 @@ drm_output_propose_state(struct weston_output > *output_base, > if (ev->output_mask != (1u << output->base.id)) > next_plane = primary; > > + /* Ignore views we know to be totally occluded. */ > + pixman_region32_init(&clipped_view); > + pixman_region32_intersect(&clipped_view, > + &ev->transform.boundingbox, > + &output->base.region); > + > + pixman_region32_init(&surface_overlap); > + pixman_region32_subtract(&surface_overlap, &clipped_view, > + &occluded_region); > + occluded = !pixman_region32_not_empty(&surface_overlap); > + if (occluded) { > + pixman_region32_fini(&surface_overlap); > + pixman_region32_fini(&clipped_view); > + continue; > + } > + > /* Since we process views from top to bottom, we know that if > * the view intersects the calculated renderer region, it must > * be part of, or occluded by, it, and cannot go on a plane. */ > - pixman_region32_init(&surface_overlap); > pixman_region32_intersect(&surface_overlap, &renderer_region, > - &ev->transform.boundingbox); > - > + &clipped_view); > if (pixman_region32_not_empty(&surface_overlap)) > next_plane = primary; > pixman_region32_fini(&surface_overlap); > @@ -3104,6 +3121,9 @@ drm_output_propose_state(struct weston_output > *output_base, > if (next_plane == NULL) > next_plane = drm_output_prepare_cursor_view(state, ev); > > + if (next_plane == NULL && !drm_view_is_opaque(ev)) > + next_plane = primary; > + > if (next_plane == NULL) > next_plane = drm_output_prepare_scanout_view(state, ev); > > @@ -3116,9 +3136,16 @@ drm_output_propose_state(struct weston_output > *output_base, > if (next_plane == primary) > pixman_region32_union(&renderer_region, > &renderer_region, > - &ev->transform.boundingbox); > + &clipped_view); > + else if (output->cursor_plane && > + next_plane != &output->cursor_plane->base) Should this not be: if (!output->cursor_plane || next_plane != &output->cursor_plane->base) so that lack of a cursor plane goes straight to occluded? > + pixman_region32_union(&occluded_region, > + &occluded_region, > + &clipped_view); > + pixman_region32_fini(&clipped_view); > } > pixman_region32_fini(&renderer_region); > + pixman_region32_fini(&occluded_region); > > return state; > } I cannot find any other fault here, and I stared at this for a long time. Thanks, pq
pgpG17crgeJCa.pgp
Description: OpenPGP digital signature
_______________________________________________ wayland-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/wayland-devel
