If a very large number of clip rects are accumulated in rect_union_add,
rect_union_evaluate can end up being disproportionately expensive, and
as an extreme numbers of clip rects are not beneficial for drawing, this
is without any benefit.
Limit the number of rects to 1024 in rect_union_add, switching over to
bounding box mode instead when the limit is exceeded. This leads to a
small 70% reduction in CPU time in the Vulkan renderer on the
stacked/clip200/1024 benchmarks.
Signed-off-by: Kenny Levinsen <kl@kl.wtf>
rect_union_add takes a pixman_box32_t by value, and passes it along by
value to internal helpers. render_pass_mark_box_updated which is the
only caller receives the pixman_box32_t by reference, so just plumb it
through that way.
Results in a 13% improvement in CPU time when using the Vulkan renderer
on the stacked/clip200/1024 benchmarks on my machine.
Signed-off-by: Kenny Levinsen <kl@kl.wtf>
Implement a ring-buffer that uses timeline points to track and release
allocated spans. We add larger ring-buffers when it fills, and remove
them when they have been unused for many collection cycles.
Signed-off-by: Kenny Levinsen <kl@kl.wtf>
Implement fifo-v1 protocol: the capability for clients to explicitly queue wl_surface
commits.
The design of the wl_surface interface defines a double buffer, which is intended to support only
(from the point of view of the client) what is sometimes called the 'mailbox' presentation mode:
each request on a wl_surface operates on an accumulated pending state until a .commit request is
issued, immediately and atomically replacing the wl_surface's current state. This means that only
the state related to the latest .commit request will be latched onto.
fifo-v1 is a protocol that aims to support an important use case, namely the FIFO display
presentation mode, which allows to queue the accumulated pending state until the next latching
deadline occurs.
Signed-off-by: Sergio Gómez <sergio.g.delreal@gmail.com>
The returned fence is not required to be signalled at the earliest
possible time. It is not intended to replace the drm flip event, and is
expected to be signalled only much later
This changes the behavior of wlr_linux_drm_syncobj_surface_v1 to
automatically signal release of previous commits as they are replaced.
Users must call wlr_linux_drm_syncobj_v1_state_add_release_point or
wlr_linux_drm_syncobj_v1_state_signal_release_with_buffer to delay the
signal as appropriate.
Add color_representation to wlr_output_state, holding color
representation metadata about the primary buffer. This can be set
using wlr_output_state_set_primary_color_representation() and a
new enum value WLR_OUTPUT_STATE_COLOR_REPRESENTATION in
wlr_output_state.committed indicates when this data is present.
Also add color-representation to wlr_output, and discard
color-representation in wlr_output_state if it matches what's already
been committed to the output.
Currently we send wl_surface.enter/leave when a surface is hidden
and shown again on the same output. In practice, this happens very
often since compositors like river and sway enable and disable
the scene nodes of surfaces as part of their atomic transaction
strategy involving rendering saved buffers while waiting for
clients to submit new buffers of the desired size.
The new strategy documented in the new comments avoids sending
redundant events in this case.
wlr_box_intersection generates a new box based on the intersection of
two boxes. Often we simply want to know *if* two boxes intersected,
which we can answer much cheaper.
Add wlr_box_intersects, in similar vein as wlr_box_contains_box but
returning true for any overlap.
This allows using the vulkan renderer on platforms that provide all
the necessary Vulkan extensions.
Tested on a Mali G52 platform with Mesa 26.0.0 and 25.3.5, which only
support Vulkan API 1.0.
We'll need to grab textures from there in the next commit.
Also rename it to better reflect what it does: synchronize release
fences after a render pass has been submitted.
When libinput introduces new enum entries, we'd abort or send bogus
events to the compositor. Instead, log a message and ignore the
event.
Keep all enums without a default case so that the compiler warns
when we're missing a case.
Exposing libinput_tablet_tool from libinput allows compositors
to set tool specific configuration options like pressure range
and eraser button behavior.
See 'libinput_tablet_tool_*()' functions which require a
'libinput_tablet_tool' handle:
https://wayland.freedesktop.org/libinput/doc/1.30.0/api/group__config.html
(libinput 1.30.0, latest at the time of writing).
schedule_frame is only called when the client has submitted damage
and a new frame should be rendered immediately. schedule_frame is
not called when the capture client is waiting for the next frame
but hasn't submitted damage.
Sources implementing a rendering loop based on the capture rate
need to know when a capture client is ready to accept a new frame.
Such sources want to trigger a redraw if and only if (1) they are
dirty (their contents have changed) and (2) the capture client is
ready to accept a new frame.
Replace schedule_frame with request_frame, triggered each time a
client sends a capture request. A flag indicates whether the capture
client has submitted damage.
This makes it so callers no longer need to juggle with raw pixel
pointers anymore, and only need a single wlr_buffer-based codepath.
Additionally, cursors can be unloaded without risking use-after-free.
