Introduce properties for the main renderer and allocator the compositor
wants wlroots to use when it needs to blit to a staging buffer during
multigpu. We need to perform an additional blit if the modifier is not
compatible with the target GPU.
We need this really slow path if the user is using GPUs that don't have
common compatible modifiers. One example of a vendor that doesn't support
rendering to a LINEAR modifier (which otherwise should always exist)
is NVIDIA.
We also need to introduce allocator variants to wlr_raster_upload_texture
and wlr_raster_attach so that the given allocator can be used to
allocate a stanging buffer to aid in blitting
If the compositor is running without a renderer, that means that
the compositor must be driven by something external that may or may not
be there. So we have two scenarios:
1. This compositor is currently being watched and driven by some
external source that is consuming buffers. This is okay, because
during the commit handler `surface->current.buffer` and
`surface->buffer_damage` will be usable and things will be handled
like normal.
2. Things break however if the compositor is not currently driven. This
however is commonly temporary. Something may not be interested right now,
but later it can be. In this case we have to accumulate state until
this external consumer is ready. Here, we have to accumulate the
`buffer_damage` and keep the buffer locked until the consumer is ready.
`wlr_surface_consume` needs to be called when the state of this surface
was consumed so that it is safe to release these resources.
wlr_compositor will now wait for the current buffer to be released before
clearing relevant state. For now, this will always happen at the end of
the commit so there should be no functional change here.
Since wlr_raster supports wlr_compositor usage with and without a renderer,
use it for wlr_curosr so cursors support running on a surface without
a renderer.
The prerelease signal lets users do things things at the last moment
that would be inappropriate to do on the release signal. Inside the
prerelease signal, it is allowed to lock the buffer and also
upload/import the contents of the buffer to a texture.
If a surface with an existing buffer has a syncobj surface state created
without committing a new buffer with associated timelines, callers will
see the surface as having a syncobj state and may try to use it, but
calling the signal_release_with_buffer helper at this time will assert
on the lacking release timeline.
As this is a valid situation, remove the assert and replace it with an
early return so that callers do not need to explicitly check for the
presence of valid timelines.
Fixes: https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3895
A struct wlr_xdg_toplevel_configure is passed in with the whole
state. This makes it a lot clearer that the size and WM state are
always sent to the client.
It was completely wrong: according to the protocol, the effective
geometry is only updated on commit time if there pending state has
new state from xdg_surface.set_window_geometry or
xdg_surface.set_window_geometry has never been sent at all.
This commit adds wlr_xdg_surface.geometry which correctly matches the
effective window geometry and removes now-useless
wlr_xdg_surface_get_geometry().
It seems that some scene compositors want to avoid wlr_scene_output_commit
and use the lower lever wlr_scene_output_build_state. However, build
state does not early return if a frame is not needed so compositors will
implement the check themselves. Let's add a helper function that compositors
can use to implement the check.
Technically pending_commit_damage is a private interface, so this lets
compositors not interface with private interfaces to implement the check.
This fixes direct scanout VRR. As direct scanout buffers are not part
of the swapchain, we would mistakenly union instead of subtract the damage
meaning it will just accumulate indefinitely.
The reason for this existing in the first place is for compositors that
might want to sidestep scene and commit their own buffers to the output.
In this case, scene could theoretically acknowledge that and update the
damage. Except, this really didn't work because WLR_OUTPUT_STATE_DAMAGE
would need to be defined which is optional. This patch also properly
acknowledges commits without damage.
In the use case of a weird compositor that might want to sidestep scene,
they can just trash the damage ring themselves.
Fixes: #3871
