We don't want to iterate the subtree for trivial update scenarios. We only
care when a scene is reparented or disabled in which both cases provides
us with explicit damage
This is useful in these cases:
- The same surface is added to two different scene-graphs. wlroots
can't figure out on its own which scene-graph should drive the
Xwayland stacking.
- A compositor uses multiple Xwayland servers (e.g. one per app).
wlroots will try to restack surfaces from different Xwayland
instances and this will not go well.
Storing the frame pacing output in a per-scene and per-surface
struct doesn't play well with multiple scenes. outputs_update is
only triggered for outputs the scene knows about, but operates on
all outputs the surface has entered regardless of the scene. Thus
leaving an output on one scene will not refresh the frame pacing
output on other scenes, and these other scenes will operate with
a stale frame pacing output. The surface will not receive any more
wl_surface.frame done events.
This also avoids keeping a dangling pointer around when the frame
pacing output is destroyed but the output isn't added in the scene.
References: https://github.com/swaywm/sway/issues/8885
Before this patch, when a surface became occluded on all outputs,
we'd reconfigure it with a base configuration (scale set to 1,
transform set to NORMAL, image description set to gamma 2.2/sRGB).
As a result, when quickly hiding a toplevel and showing it again,
the client would render to switch to the base configuration, then
render again to switch to the output configuration.
Avoi this needless back-and-forth by retaining the last sent
preferred configuration when a surface leaves all outputs.
Fixes incorrectly rejecting scanout for gamma2.2 buffers when the output
has no image description set. This happens on `hdr off` mode on sway.
Also refactor the scanout check into its own function while at it to
make it easier to follow.
We were incorrectly doing comparison with `!= 0` to detect non-sRGB
tf/primaries. Since these enums are bit flags, the default sRGB values
are 1, not 0, so sRGB buffers were incorrectly rejected.
Fixes: bf40f396bf ("scene: grab image description from output state")
We cache whether buffers are single-pixel buffers (and if so what color
they are) to allow rendering optimizations. But this breaks if the
client changes out the single-pixel buffer for one with a different
color, because this updates the texture in-place instead of actually
changing the buffer.
We can fix this by blocking in-place texture updates for single pixel
buffers.
Original bug: https://codeberg.org/ifreund/waylock/issues/121
See also: !5092
If a surface is mirrored on two outputs, we don't want to pick the
first output if the second has a higher refresh rate.
Also fixes duplicate frame/feedback events when a surface is added
to multiple scenes.
This lets the surface handler decide which output to send frame
callbacks from. The output_sample event already works this way.
Introduce wlr_scene_surface_send_frame_done() as a replacement for
wlr_scene_buffer_send_frame_done() when a compositor doesn't have
an output at hand.
If a surface appears on two outputs with the same intersection
area, or even if a surface appears on an output with a small
intersection area, we want to use the highest scale.
Fixes flip-flop when a surface is added to multiple scenes.
References: https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3901
scene_entry_try_direct_scanout returns a tristate value, but the log
message was not updated to account for this.
Compare whether or not the state is specifically SCANOUT_SUCCESS for
logging purposes.
Fixes: c450991c4b
Move single-pixel buffer status cache from wlr_scene_surface to
wlr_scene_buffer, it makes more sense there and means the optimisations
will still work if wlr_scene_buffer is used without wlr_scene_surface.
Direct scanout can be enabled and disabled on a frame-by-frame basis,
and so we could end up sending different feedback to a surface on every
other frame. Reacting to new feedback is expensive, as the client may
need to reallocate their swapchain.
Debounce the state change a number of frames, for now set to 30, to
avoid immediate reaction to scanout (or composition) that only lasts a
few frames.
A timer could be used instead, but it did not seem worth the complexity.
What just want to know that the state has been stable across a
reasonable number of samples, and a counter seems sufficient for that.
The single-pixel buffer protocol is used to allow wayland clients to
easily draw solid-color rectangles by presenting a 1x1-pixel buffer and
scaling it to the desired size. This patch improves how these buffers
are then handled in the scene-tree renderer.
We already ignore opaque black rectangles at the very bottom (and
anything under them) because we assume we'll be rendering on a black
background. This patch detects black opaque single-pixel buffers and
handles them in the same way as black opaque rectangles. It also
renders single-pixel buffers as rectangles rather than buffers because
this is probably more efficient in the underlying renderer.
In wlr_scene_surface we cache whether the attached buffer is a
single-pixel buffer. This is done because the
wlr_single_pixel_buffer_v1 will be destroyed after texture upload, after
which it becomes much more annoying to check if the buffer is a
single-pixel buffer.
We were signaling the release timeline point when the
wlr_client_buffer was released. However, the wlr_client_buffer isn't
necessarily released at the same time as the underlying source
wlr_buffer. For instance, with wl_shm the source buffer is released
before the wlr_client_buffer, and with linux-dmabuf-v1 the source
buffer is released after the wlr_client_buffer. However, we want
to signal the release timeline point exactly at the same time we
send the wl_buffer.release event to the client.
Use surface->buffer->source instead of &surface->buffer->base to
fix this.
linux-drm-syncobj-v1 can only be used with DMA-BUFs, and
wlr_client_buffer.texture will keep the source locked, so
surface->buffer->source is guaranteed to be non-NULL and unreleased.
Closes: https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3940
Fixes: 9e71c88467 ("scene: unwrap wlr_client_buffer for direct scan-out")
Remove unneeded includes of wlr_output.h from wlr_compositor.h and
wlr_cursor.h (unneeded now that we forward-declare struct wlr_surface)
and put the actually-required includes in the right places.
