Pass the backend instead of the compositor to the windowed output API
create_head() method and increment the API version.
That way the backend will not have to find the backend pointer from the
compositor. This is trivial now, but in the multi-backend case would
entail iterating over all backends to find the correct one.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
The output move listener removal was incomplete. Remove the remaining
bits to fix a segfault on shutdown.
Fixes: 40f5eaf401 ("backend-vnc: use output power_state to disable repainting while disconnected")
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
This was not found in the test suite, but if you run wayland-backend
manually with ASan, you see the same leaks as in
backend-headless: fully release pango and fontconfig
Fix them the same way.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
With weston_output_power_on/off() we can use power_state to disable
repainting completely while no VNC client is connected. This allows
to remove the initial repaint and per-output damage tracking.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Whenever a color transformation is being created, this debug scope
prints its pipeline before and after being optimized. It should be used
with the color-lcms-transformations scope.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
It prints the existent color profiles for new subscribers. Also prints
any creation/destruction of color profiles.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
It prints the existent color transformations for new subscribers. Also
prints any creation/destruction of color transformations.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
We have a string describing the ICC profile. cmsGetProfileVersion()
returns a float value, and we are converting that to string with "%f"
and saving to this description. Instead, use "%.1f" to restrict it to a
single decimal value, which is enough. With this change we have e.g.
"version 4.4" instead of "version 4.4000000".
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
LCMS API cmsStageAllocToneCurves uses cmsDupToneCurve which internally
re-allocates a new table of points. As a result, we have to free the old
table returned from lcmsJoinToneCurve.
Signed-off-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
We were declaring that the binding handler took an enum in the
declaration (good!), but then using a uint in the definition (oops).
cf. wayland/weston!1205
Signed-off-by: Daniel Stone <daniels@collabora.com>
wl_array_for_each() returns a pointer to each storage location; as we're
storing a pointer to drm_fb, this means that we have a drm_fb **, not a
drm_fb *.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Rather than setting the initial power state when adding
it (using weston_compositor_add_output), do that at the initilization
stage.
Reason being that the compositor can set up the output from the start as
FORCED_OFF, before enabling the output, rather than enabling the output
and then turning off the power of the output.
Signed-off-by: marius vlad <marius.vlad@collabora.com>
Rather than damaging the output before the output has been added with
weston_compositor_add_output, do that afterwards as to avoid scheduling
a repaint for that output *before* actually adding the output.
This would avoid the awkward case where we attempt to set initial power
state to normal, but we can't apply it at that stage.
Signed-off-by: marius vlad <marius.vlad@collabora.com>
For some reason we'd managed to have a mismatching header prototype and
implementation. Fix this up to consistently use enums everywhere.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Output repaint uses a pair of fence syncs to profile GPU execution by
retrieving their timestamps once signalled. While the end timestamp
can be rather inaccurate in some cases (drivers reusing sync objects
from previous command buffers), the begin timestamp is never correct
because fence syncs are signalled on command buffer completion.
Get rid of the begin fence sync and use the EXT_disjoint_timer_query
extension to measure the actual repaint duration and extrapolate the
begin timestamp from the end one.
Fixes#342
Signed-off-by: Loïc Molinari <loic.molinari@gmail.com>
Based on patches from:
Peter Hutterer <peter.hutterer@who-t.net>
Lyude Paul <thatslyude@gmail.com>
Bastian Farkas <bfarkas@de.adit-jv.com>
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
When it comes to a window frame, a tablet tool and cursor act almost
identical; they click things, drag things, etc. The tool type and extra
axes don't serve any use in the context of a window frame, so tablet
pointers share the frame_pointer structures used for the mouse pointer.
Co-authored-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Lyude Paul <thatslyude@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
Signed-off-by: Bastian Farkas <bfarkas@de.adit-jv.com>
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
The tablet is given a separate cursor. Most tablet interaction is an absolute
interaction and shouldn't need a cursor at all, but usually the cursor is used
to indicate the type of virtual tool currently assigned.
Based on patches from
Peter Hutterer <peter.hutterer@who-t.net>
Lyude Paul <thatslyude@gmail.com>
Bastian Farkas <bfarkas@de.adit-jv.com>
Maniraj Devadoss <Maniraj.Devadoss@in.bosch.com>
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
Based on a patches from
Peter Hutterer <peter.hutterer@who-t.net>
Lyude Paul <thatslyude@gmail.com>
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
Based on a patch from
Peter Hutterer <peter.hutterer@who-t.net>
Lyude Paul <thatslyude@gmail.com>
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
Closely modelled after the pointer focus handling
Co-authored-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Lyude Paul <thatslyude@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Bastian Farkas <bfarkas@de.adit-jv.com>
Based on a patch from
Peter Hutterer <peter.hutterer@who-t.net>
Lyude Paul <thatslyude@gmail.com>
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
Introduces three new structs, weston_tablet and weston_tablet_tool for the
respective devices, with the respective information as it's used on the protocol.
And weston_tablet_tool_id to track the tools of a tablet.
Note that tools are independent of tablets, many tools can be used across
multiple tablets.
The nesting on the protocol level requires a global tablet manager, a tablet
seat nested into weston_seat. The list of tablets and tools are also part of
the weston_seat.
Most functions are stubs except for the actual tablet and tablet tool
creation and removal.
This is based on patches from Peter Hutterer <peter.hutterer@who-t.net> and
Bastian Farkas <bfarkas@de.adit-jv.com>.
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
We have an optimization to skip composition if there's no damage on the
primary plane and we already have a renderer buffer active. But we don't
allow this optimization if there's a pending capture task for the
output. For the renderer-based sources, that is really necessary, but
for the writeback source we should allow this optimization.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
A popup grab is specified to have the top most popup surface gain
keyboard focus. This means the keyboard focus should always follow the
most recent xdg_popup.grab() surface. Make sure this happens by keeping
track of the parent surface in the libweston-desktop popup grab,
updating the keyboard focus when surfaces are added and removed from the
popup chain, and restoring the keyboard focus to the toplevel when there
are no popups anymore.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Popups should have keyboard focus when active, but the toplevel window
should still appear "active". Make sure this is the case by changing the
"active" tracking to see whether any child surface has keyboard focus.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
When doing plane selection for an output CRTC check if the plane
already has a CRTC attached and if so prefer that plane only for
the corresponding CRTC.
This prevents changing a CRTC's primary plane when it is active
which is not allowed by the DRM framework.
Based-on-patch-by: Eric Ruei <e-ruei1@ti.com>
Signed-off-by: Andrew F. Davis <afd@ti.com>
For some reason we'd managed to have a mismatching header prototype and
implementation. Fix this up to consistently use enums everywhere.
Signed-off-by: Daniel Stone <daniels@collabora.com>
We map view alpha(0.0-1.0) to plane state's alpha
by using the max plane alpha value got from drm.
Signed-off-by: Hsuan-Yu Lin <hlin@jp.adit-jv.com>
Signed-off-by: Veeresh Kadasani <external.vkadasani@jp.adit-jv.com>
Signed-off-by: Vinh Nguyen Trong <Vinh.NguyenTrong@vn.bosch.com>
This checks whether plane alpha is supported.
We get range of alpha value supported for plane
which is required for mapping view's alpha(0.0-1.0)
with drm plane alpha. No functional change.
Signed-off-by: Hsuan-Yu Lin <hlin@jp.adit-jv.com>
Signed-off-by: Veeresh Kadasani <external.vkadasani@jp.adit-jv.com>
Signed-off-by: Vinh Nguyen Trong <Vinh.NguyenTrong@vn.bosch.com>
This prevents to trigger an assert within
weston_view_set_rel_position(), introduced with commit 'libweston: Split
weston_view_set_position() into rel and abs variants', which is hit when
a subsurface attempts to commit without having a parent surface set.
Fixes: #730
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
Reported-by: Colin Kinloch <collin.kinloch@collabora.com>
We need only check that the region is not empty. If either the input region or
the constraint region have degenerate extents, the intersection from the
previous instruction will set confine_region->data to pixman_region_empty_data.
Fixes: b6423e59
Signed-off-by: Sergio Gómez <sergio.g.delreal@gmail.com>
We log the reasons why the fb of a certain view was not placed in an
overlay plane and use that for debug purposes. With these reasons we
also decide if the scanout tranche should be included on the dma-buf
feedback or not. For instance:
1. If the reason is the incompatibility between the format/modifier
pair of the fb and those supported by the KMS device, the scanout
tranche is added and feedback is re-sent (so that the client can
re-allocate with parameters that makes it eligible for direct
scanout).
2. If the reason is because we have no overlay planes available, the
scanout tranche is useless. So the scanout tranche is removed and
the feedback re-sent (so that clients can re-allocate with
parameters optimal for the render device).
Also, when we detect that a view is eligible for direct scanout, we
don't even consider sending new feedback, as our interpretation of the
dma-buf feedback spec was that we should avoid bothering clients with
new feedback when they are already hitting direct scanout.
After some discussions and clarifications regarding the spec, we've
realized that Weston should start to also include the scanout tranche
even when the compositor is able to place client's content on overlay
planes. Basically, because this gives a chance for clients to
re-allocate with the proper parameters (not only format/modifier pair,
but also the target_device and the flags) from the scanout tranche. In
this patch we start doing this.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
It makes no sense to keep the scanout tranche on the dma-buf feedback if
there are no overlay planes available. So start to remove it.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
wayland_output_destroy_shm_buffers() is called immediately before
output_destroy() of the renderer is called. And for the pixman renderer all
renderbuffers must be destroyed before the output can be destroyed.
Also, weston_renderbuffer_unref() is not called when the buffer is released
because buffer->output is now NULL, so the renderbuffer would be leaked.
So just unref the renderbuffer immediately. Set it to NULL to avoid unreffing it
again should wayland_output_destroy_shm_buffers() be called again before the
buffer is released. This can happen during an xdg-shell resize.
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
In commit "drm-backend: add writeback connector screenshooter to
DRM-backend" we were failing the writeback screenshot when the DRM/KMS
driver would take longer than the atomic commit to finish. In this patch
we address such case.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
In this patch, we add the writeback connector screenshooter to the
DRM-backend.
This will be useful to create plane composition tests that will run in
our CI, as VKMS already supports writeback connectors.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
With this change, we expose the DRM-backend writeback source through the
output capture interface, making it available to clients.
For now we'll always fail writeback screenshots requests, because we
still don't have the writeback screenshooter implementation on the
DRM-backend. We add that in the following commits.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
In the following commits we add a writeback screenshooter. For that,
we'll need the formats supported by the writeback connectors. So include
the supported formats in struct drm_writeback.
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
Allow VNC clients that support the cursor pseudo encoding to render
the cursor themselves. This reduces observable latency of cursor
movement.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
The state does not own the fd. This is usually not a problem, because the
in_fence_fd of the state is assigned during drm_assign_planes() and then
immediately used in drm_repaint_flush(). It cannot be closed in-between.
However, in the fallback path in drm_output_start_repaint_loop(), the state is
duplicated. At this point in time, the in_fence_fd may be invalid because it was
replaced in a new commit of the corresponding surface.
The plane state was already committed to the kernel when it is copied, so the
fence is no longer needed. So just clear it.
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
With multiple DRM devices, the state for one device may be empty during
repaint_flush(). This can happen for example if an output of one device triggers
the repaint and there are no screens attached to the other device and therefore
no active outputs.
The atomic commit will actually fail because the commit contains the
DRM_MODE_PAGE_FLIP_EVENT flag but no CRTCs.
Avoid this by skipping the commit entirly. There is nothing to to anyways.
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
With some displays connect, disconnect, connect events can happen is a very
short amount of time. When this happens, the output global may already be
destroyed when a client tries to bind it. As a result, the client is
disconnected with a protocol error. See [1] for more details on the general
problem.
To mitigate this problem call wl_global_remove() first and call
wl_global_destroy() several seconds later. This is inspired by the
implementation for the same problem in wlroots.
[1] https://gitlab.freedesktop.org/wayland/wayland/-/issues/10
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
Since the logic of pointer constraints assumes a valid view throughout, add a
signal to disable constraints when its current view is unmapped by Weston.
The assumption that a previously unmapped view is valid already leads to the
constraints code crashing. This can happen when attaching a NULL buffer to the
surface and commiting, which effectively unmaps the view with the side effect of
clearing the surface's input region, which is then assumed valid inside
maybe_warp_confined_pointer().
Fixes: #721
Signed-off-by: Sergio Gómez <sergio.g.delreal@gmail.com>
Currently, the surface destroy listener in pointer constraints is redundant,
since surface destruction already handles pointer constraints destruction (see
libweston/compositor.c:weston_surface_unref()).
Signed-off-by: Sergio Gómez <sergio.g.delreal@gmail.com>
This makes flight recorder creation faster by using wider store
instructions (depending on the memset() implementation).
Signed-off-by: Loïc Molinari <loic.molinari@gmail.com>
This is a flag used to track whether the position has changed, not
whether the position is set.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Add the --additional-devices parameter to Weston to add secondary drm devices
that will only be used as outputs, but not for rendering.
We can only fail the repaint for the entire backend, but not for single
devices. Thus, if one of the devices fail, we have to fail the repaint for the
entire backend.
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Additional devices don't have a gbm device. Therefore, we cannot create gbm bos
for the cursor.
If the output device differs from the gbm device, fall back to the allocation of
a dumb buffer for the cursor on the output device. Update the cursor sprite with
a memcpy to the already mapped dumb buffer that belongs to the current cursor.
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
If the GBM bo was allocated on a different device than the device that is used
for the fb, we have to import the fd first and update the handle.
Use drmPrimeFDToHandle directly instead of using a gbm device for the scanout
device, since a gbm device would require a gbm implementation, which is often
not available for devices that only support scanout.
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
If we are using multiple GPUs and are not able to use modifiers to ensure that
the formats are compatible, we have to use linear buffers for the transfer.
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Weston uses a cached drm_fb when a view is shown multiple times. If the view is
shown on multiple outputs backed by different DRM devices, Weston returns the
cached drm_fb for the first device that was used for the import. This causes a
failure when adding the fb to the other device.
Use a list of all drm_fbs to cache the buf_fb per device, and check for the
device before reusing a drm_fb.
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
The faked z position must be created for each device. Therefore, the device
itself must be passed to the function. If only the backend is passed, the faked
z position would be only created for the primary device.
Signed-off-by: Michael Tretter <m.tretter@pengutronix.de>
Now that struct weston_renderbuffer is refcounted, hold a reference for
renderbuffers on the pixman_output_state::renderbuffer_list. This allows
backends to destroy the renderer output state and release renderbuffer
references in any order without running into an assert().
To avoid breaking resizing, We also have to drop the renderbuffer list
during pixman_renderer_resize_output(). The backends have to create new
renderbuffers afterwards.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Align the function name and arguments of vnc_convert_damage() with
weston_region_global_to_output(). It does not support rotation and
stores the result in a pixman_region16_t, but otherwise it serves the
same purpose.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
I also snuck in a trivial change to drag_surface_configure at the same
time to avoid yet another micro patch.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
weston_renderer::repaint_output must be called from the weston_output::repaint
callback. When called from the weston_output::enable callback, a black frame
is produced. Instead of painting an invalid buffer and then working around it
by setting output damage, just don't skip the first real repaint even though
no VNC client is connected yet.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Add a debug scope "vnc-backend" and use it to log per-renderbuffer
accumulated damage and new repaint damage before repainting.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Since neatvnc frame buffers are in system memory, using a shadow
buffer just causes an unnecessary copy in the pixman renderer.
Stop using the shadow buffer.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
It is enough to report new repaint damage instead of accumulated
per-renderbuffer damage to nvnc_display_feed_buffer().
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
While not repainting, all buffers are damaged exactly the same.
Avoid unnecessary work by tracking this damage separately on struct
vnc_output.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
In IVI, there are several displays connected to a SoC. These displays
are just driven by differential pairs (LVDS, FPD-Link, GMSL) and powered
centrally. To reduce power comsumption when user inactivity timeout
happended on the display, there is a need to cut down pixel clock from
SoC. Then, if any input events happend on the display, it should become
active again.
Currently, controlling the compositor outputs doesn't happen independently
but rather globally, and outputs repaints are based on the compositor state
This is necessary to have an API that can force the power state of an
output to off via DPMS mode while all other compositor outputs remain
unaffected.
Signed-off-by: Rajendraprasad K J <KarammelJayakumar.Rajendraprasad@in.bosch.com>
Signed-off-by: Vinh Nguyen Trong <Vinh.NguyenTrong@vn.bosch.com>
This is needed by drm_output_fini_egl() to be able to retrieve the
backend out of the drm_output on the shutdown path of the compositor.
Both the remoting plug-in and the pipewire plug-in are users of the
drm-virtual API and as such they would trigger a crash when shutting
down the compositor, as we're not setting up any backend whatsoever.
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
Convert the bare x,y coordinates into struct weston_coord and update all
users.
We keep the surface position in wl_fixed_t for now so it still exactly
matches the position most recently sent to clients.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
When a view is destroyed then the views of subsurfaces remain until the view
list is rebuilt for the next repaint.
During that time view->parent_view contains an invalid pointer and weston will
crash when it tries to access the view.
This happens for a surface with subsurfaces with views on two different outputs
with the ivi-shell:
When the surface is destroyed then the destroy handler of the ivi-shell
(shell_handle_surface_destroy()) may be called first. It will (indirectly)
destroy the view of the main surface with weston_view_destroy().
Next the surface destroy handler of the subsurfaces
(subsurface_handle_parent_destroy() is called. It will unmap the first view of
the subsurface. Here weston_surface_assign_output() is called which tries to
find the output of the second view and accesses the now invalid
view->parent_view in the process.
There are probably other ways to trigger similar crashes.
To avoid this, clear view->parent_view when the parent view is destroyed.
Fixes 0669d4de4f ("libweston: Skip views without a layer assignment in
output_mask calculations")
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
The pixel format stored in backends->format[0] is effectively looked up
via pixel_format_get_info(DRM_FORMAT_ARGB8888). Reuse that instead of
looking up the same via pixel_format_get_info_by_pixman(PIXMAN_a8r8g8b8)
in multiple places.
There are still two instances of hard-coded CAIRO_FORMAT_ARGB32 in
wayland_output_get_shm_buffer() that currently can not be obtained from
pixel_format_info.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
If the wl_surface resource has version 5 or newer, we should always
ignore the offset parameters in wl_surface.attach.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Push weston_coord into the notification functions instead of passing
two doubles.
The touch handlers are passed a pointer to a weston_coord, which is
unusual. This is done so we can pass a NULL pointer instead of a
fabricated invalid coordinate when the touch type is TOUCH_UP.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
This creates a global coordinate from a device coordinate.
Replace it with weston_coord_global_from_output_point() which
does the same thing and returns a weston_coord_global.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Remove the independent x, y floats from the clipping code and replace them
with struct weston_cord. This includes the polygon8 structure as well.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
If a window is clicked with a mouse while it's being interacted with via
touch input, the assert from 2dc8680d will fire.
This is a leftover from d611ab24
Update the transform before converting the coordinate to fix this.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Starting with commit 4cde507be6 "backend-drm: fix plane sorting" the
plane list will have a descending order of the planes rather than ascending.
This reversed order had the side-effect of exposing the fact that we
don't set-up a plane index when creating the drm_plane using the DRM
virtual API. Without settting a plane index for that drm_plane we
effectively overwrite the plane index which has the 0 (zero) entry.
This wasn't an issue before commit 4cde507be6 "backend-drm: fix
plane sorting" as it seems we never picked up that plane index as
being a suitable one due to the fact that those were assigned to primary
planes, but after that commit, the cursor plane will be one getting
the 0 (zero) plane index.
Finally, this would trip over because we attempt to place a (cursor)
view on a primary plane (where it would've normally be a cursor
plane) and we end up with no framebuffer ref.
This is fixed trivially by assigning a plane index, different than the
ones already created by create_spirtes().
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
Split x11_output_get_pixel_format() out of x11_output_init_shm() so the
pixel format can already be known when calling pixman->output_create().
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Reorder pixman renderer output state and SHM renderbuffer creation and
removal to make sure the pixman renderbuffer list is empty when the
output state is destroyed.
Fixes: 4d96635a3f ("pixman-renderer: track damage in weston_renderbuffer")
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Pixman output state creation requires a valid pixel format to be set for
weston_output_update_capture_info().
Fixes: c67773bc5c ("pixman-renderer: use pixel_format_info instead of pixman_format_code_t")
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Reorder pixman renderer output state and SHM renderbuffer removal to
make sure the pixman renderbuffer list is empty when the output state
is destroyed.
Fixes: 4d96635a3f ("pixman-renderer: track damage in weston_renderbuffer")
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Pixman output state creation requires a valid pixel format to be set for
weston_output_update_capture_info().
Fixes: c67773bc5c ("pixman-renderer: use pixel_format_info instead of pixman_format_code_t")
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Whilst GLSL requires highp for the vertex shader stage, highp is
optional for the fragment shader.
Make sure that we work in highp by default during the vertex shader
stage, using either highp or mediump for the texcoord varying according
to what the fragment shader supports.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Search for planes that support the rotation required to properly display
a paint node, and properly set coordinates and rotation properties.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
For now we just always set the plane up to have a "normal" rotation, so
no new features are added with this commit.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
From our perspective, a bitmask is pretty much the same as an enum, so
allow it to use the same path.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
We're pushing more and more mutable state into paint nodes, but this state
has a non-zero cost to rebuild every render.
Let's take care to track when we need to rebuild the state.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Every shells have an implementation of weston_desktop_api structures.
It includes some callbacks to listen to xdg_surface signals. Committed callback
is one of them.
Currently, the xdg_shell don't invoke the weston_desktop_api->committed
callback when the xdg_surface is initial stage or no surface content.
In the case: the client attached a null buffer after the valid buffer,
the shell isn't going to invoke to xdg_surface committed, and don't know
the surface is disappeared. If the surface is fullscreen, we will get
a black background on the screen until a new valid frame come. That
should not happen.
Signed-off-by: Tran Ba Khang(MS/EMC31-XC) <Khang.TranBa@vn.bosch.com>
Add the weston_surface_is_unmapping() api, this will help the shell
to detect the commit of a surface is unmapping or not.
Suggested-by: Morgane Glidic <sardemff7+git@sardemff7.net>
Signed-off-by: Tran Ba Khang(MS/EMC31-XC) <Khang.TranBa@vn.bosch.com>
Reorder pixman renderer output state and renderbuffer removal to make
sure the pixman renderbuffer list is empty when the output state is
destroyed.
Fixes: 4d96635a3f ("pixman-renderer: track damage in weston_renderbuffer")
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
We can clear this via drm_plane_state_put_back() at the end of
drm_output_propose_state(). We need to set it back to the minimum zpos
when rendering.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Update users of the old coordinate space conversion functions that take
x, y pairs to the new weston_coord versions.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Possibly the least useful place to use this, as the input comes directly
from pixman rects, and the output is more complicated than usual, but
I guess consistency counts for something.
There is some small benefit in switching to weston_matrix_transform_coord
to hide the perspective normalization step and the homogenous coords.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
All through weston we have code that passes int x, y or
float x, y or wl_fixed_t x, y pairs. These pairs are frequently
converted to/from wl_fixed_t and other types.
We also have struct vec2d and struct weston_geometry which also
contain coordinate pairs.
Let's create a family of coordinate vector structures for coordinate
pairs and use it anywhere we sensibly can.
This has a few benefits - it helps remove intermediate conversion
between fixed/float/int types. It lets us roll the homogenous
coordinate normalization bits into helper functions instead of
needing them open coded throughout the source.
Possibly most importantly, it also allows us to do some compile time
validation of what coordinate space we're working in.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
One variant is used when a view is being positioned relative to a parent,
the other is when the view is being given an absolute position in the
global space.
This will help later when surface and global coordinates are different
data types, but for now the two functions do essentially the same thing.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Clean up the code a little by dropping the now unnecessary switch case
in drm_output_init_pixman(). As a side effect, this enables all formats
that have a pixman_format entry in the pixel format table:
- rgb565
- xrgb8888
- argb8888
- xbgr8888
- abgr8888
- rgbx8888
- rgba8888
- bgrx8888
- bgra8888
- xrgb2101010
- argb2101010
- xbgr2101010
- abgr2101010
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Use const struct pixel_format_info *format instead of uint32_t
gbm_format for backend and output pixel format.
Since create_gbm_surface() is never called without output->format being
set, drop the unnecessary error message.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Make use of meson subprojects support. Allow to optionally build the
libraries required for the VNC backend as subproject of Weston.
Signed-off-by: Stefan Agner <stefan@agner.ch>
[philipp.zabel@gmail.com: update neatvnc and aml versions]
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Use struct pixel_format_info pointers instead of uint32_t drm fourcc
values at the API surface for output and image creation.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Add a helper function to turn an array of DRM fourccs into an array of
corresponding struct pixel_format_info pointers.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Use struct pixel_format_info pointers instead of pixman_format_code_t
values at the API surface for output and image creation.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Xwayland pop-up menus aren't displayed until the mouse moves, or some other
action causes the compositor to repaint.
This is because the shell knows nothing about xwayland override redirect
windows, so it won't assign them an output. Without an output, the
surface commit won't cause a repaint.
Fix this with brute force by updating their geometry on commit. If the
geometry is dirty (and it will be for new surfaces), this will cause an
output to be assigned before the upcoming weston_surface_schedule_repaint()
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Found-by: Hideyuki Nagase <hideyukn@microsoft.com>
Add weston_renderbuffer_ref/unref() functions and use them to
eventually destroy the weston_renderbuffer. Drop the explicit
renderbuffer_destroy vfunc from the pixman renderer interface.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
This pixman image is not actually used anymore, drop it.
Fixes: 89e1831cd7 ("pixman-renderer: add weston_renderbuffer and create/destroy interface")
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Given that pixman_image_create_bits_no_clear() is asked to allocate the
buffer on its own, the _no_clear part is not important. Drop it.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Move the call to weston_output_update_capture_info() from the headless
backend into pixman_renderer_output_create(). For this, add an
uint32_t drm_format parameter to struct pixman_renderer_output_options.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Let weston_output_update_capture_info() take a uint32_t drm_format
parameter directly instead of const struct pixel_format_info *format.
No other fields apart from the format were used from this structure.
Without this, callers may have to unnecessarily look up the pixel
format info in cases where the DRM fourcc is already available.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Stop calling pixman_renderer_init() from backends directly.
Call it via weston_compositor_init_renderer() instead.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Add a private struct pixman_renderbuffer that derives from struct
weston_renderbuffer and move the pixman renderer specific image and link
fields into it.
Add a pixman_renderbuffer_get_image() helper for the backends that need
to access the contained pixman image, RDP and X11.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Add a damage region to struct weston_renderbuffer and use it to replace
the previous_damage tracking in the drm backend.
Keep renderbuffers on a list in struct pixman_output_state and use it
to accumulate damage on all renderbuffers during repaint_output.
Now renderbuffers have to be created when pixman output state already
exists.
Reorder renderer output state and renderbuffer creation accordingly.
With this, pixman_renderer_output_set_hw_extra_damage() can be removed.
This can not yet replace the external damage tracking in the VNC
backend, which needsto know the accumulated damage that is not returned
from repaint_output.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Add a struct weston_renderbuffer parameter to repaint_output and make
backends set the pixman image renderbuffer through this parameter
instead of using pixman_renderer_output_set_buffer()
Turn pixman_renderer_output_set_buffer() static.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Add a create_image_from_ptr vfunc to struct pixman_renderer_interface,
which wraps weston_renderbuffer creation for the pixman renderer via
pixman_image_create_bits(), as well as a renderbuffer_destroy vfunc
to dispose of the pixman image renderbuffer.
Also add create_image_no_clear using pixman_image_create_bits_no_clear()
instead.
Make the backends create and destroy their pixman image renderbuffers
through this interface.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Add a struct pixman_renderer_interface with output_create and
output_destroy vfuncs and store a pointer to it in struct
weston_renderer.
Make all backends access the pixman_renderer_output_create/destroy
functions through this interface and turn them static.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
The blit_shadow_to_output() function leaves the generic vertex attrib
arrays 0 and 1 enabled. This commit disables them for consistency with
the other drawing calls of the renderer.
Signed-off-by: Loïc Molinari <loic.molinari@gmail.com>
Generic vertex attrib arrays 0 and 1 are constantly enabled across the
entire renderer. This commit enables them:
- once per output repaint instead of once per view repaint,
- once for the 4 borders instead of once for each border,
- once for all the shadow regions instead of once for each region.
Signed-off-by: Loïc Molinari <loic.molinari@gmail.com>
The blending func is constant across the entire renderer. This commit
sets it once per output repaint instead of once per view repaint.
Signed-off-by: Loïc Molinari <loic.molinari@gmail.com>
Move the struct gl_renderer_interface pointer from the backends into
the weston_renderer structure. The interface struct only contains
function pointers that never change, so make it const.
Load and initialize the GL renderer in libweston instead of in the
backends, using the new weston_compositor_init_renderer() function.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
As seen in some instances, subsurfaces do not have an entry in their
layer_link, as we bring them into existence rather directly in the
view_list and not using the layer list approach.
This adds two messages to the debug scene graph to point out if the
views aren't really in any layer or if they're indirectly in the view
list using an ancestor (the main parent view actually).
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
We can save a bit of work by using the output_to_buffer_matrix we've
already calculated for the paint node to transform the destination
rectangle, instead of starting over from global regions and doing
individual steps.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
We should never hit this because drm_view_transform_supported() filters
out any transforms where it would occur.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
It should harmlessly fail or do nothing, but seeing attempts to set
uniform -1 can be confusing when looking for real shader issues.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
To avoid retrieving the backend from the compositor all the time, store
a pointer to the x11 backend on its x11_output structures.
This will be useful once the compositor contains more than one backend.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
To avoid retrieving the backend from the compositor all the time, store
a pointer to the wayland backend on its wayland_output structures.
This will be useful once the compositor contains more than one backend.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
To avoid retrieving the backend from the compositor all the time, store
a pointer to the vnc backend on its vnc_output structures.
This will be useful once the compositor contains more than one backend.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
To avoid retrieving the backend from the compositor all the time, store
a pointer to the rdp backend on its rdp_output structures.
This will be useful once the compositor contains more than one backend.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
To avoid retrieving the backend from the compositor all the time, store
a pointer to the headless backend on its headless_output structures.
This will be useful once the compositor contains more than one backend.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
To avoid retrieving the backend from the compositor all the time, store
a pointer to the drm backend on its drm_output structures.
This will be useful once the compositor contains more than one backend.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Passing the backend as a parameter to the weston_backend function
pointers seems more natural and will be very useful once there can be
more than one backend.
Since all backends already store a pointer to the compositor instance,
replace the compositor parameter with the backend in all functions.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
We already store the buffer_to_output, and this is just the inverse.
The pixman renderer will use the inverted version.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Fix the incorrect fb_pool stride set in vnc_switch_mode.
Also replace nvnc_fb_pool_unref/new with nvnc_fb_pool_resize.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
The import_simple_dmabuf() function does not modify the contents of its
struct dmabuf_attributes parameter, so make it const.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Fix the RDP backend to handle to_rdp_head() returning NULL.
In practice, this should only happen once multi-backend support
is enabled.
Fixes: 70b03b2928 ("rdp: Add preliminary rdp multihead support")
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
This is an odd corner case where the surface doesn't yet have an output
assigned -- noticed when starting up with the RDP backend and
fullscreen-shell, and we attempt to emit a timeline point for a surface
without an output assigned, causing weston to crash when that
happens.
Rather than trying to catch this in the timeline code, still emit the
flush damage timeline but instead of using the output the surface is on,
use the output that accumulates damage.
Suggested-by: Derek Foreman <derek.foreman@collabora.com>
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
This sets up monitor layout callbacks, and enables input event translation
between the RDP space and the weston desktop. The RDP backend now uses
a heads changed callback instead of the simple head configurator.
We only allow a single monitor for now, but in the future RAIL will make
use of multi-head.
As a side effect, scaling is now supported in RDP sessions.
It should be noted that due to differences between RDP and wayland
representation of their global coordinate spaces, mixing DPI leads to
RDP monitor layouts that can't properly be represented in weston.
Co-authored-by: Steve Pronovost <spronovo@microsoft.com>
Co-authored-by: Brenton DeGeer <brdegeer@microsoft.com>
Signed-off-by: Hideyuki Nagase <hideyukn@microsoft.com>
Signed-off-by: Steve Pronovost <spronovo@microsoft.com>
Signed-off-by: Brenton DeGeer <brdegeer@microsoft.com>
Instead of passing a name, pass a FreeRDP rdpMonitor struct. For now the
only caller will pass a NULL to use default values.
Also, drop static from the function declaration as new code will call it
from another file shortly.
Co-authored-by: Steve Pronovost <spronovo@microsoft.com>
Co-authored-by: Brenton DeGeer <brdegeer@microsoft.com>
Signed-off-by: Hideyuki Nagase <hideyukn@microsoft.com>
Signed-off-by: Steve Pronovost <spronovo@microsoft.com>
Signed-off-by: Brenton DeGeer <brdegeer@microsoft.com>
In some upcoming changes rdp_output_set_size will be removed. There's
no need to have it do this setup, so move that into output_create.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
In the absence of scale factors, this has been accurate. Once we allow
scaling we need to apply a full transform.
Co-authored-by: Steve Pronovost <spronovo@microsoft.com>
Co-authored-by: Brenton DeGeer <brdegeer@microsoft.com>
Signed-off-by: Hideyuki Nagase <hideyukn@microsoft.com>
Signed-off-by: Steve Pronovost <spronovo@microsoft.com>
Signed-off-by: Brenton DeGeer <brdegeer@microsoft.com>
Output dimensions are only correct here with unity scaling, but we're
going to allow scale factors shortly. The mode matches what RDP
expects.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
If a client connects to an RDP session it should replace the existing
mode, not add a new one.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
ensure_matching_mode() pays no attention to flags, we must set them on the
returned result afterwards.
This could be reproduced by making the initial connection to an rdp session
using the same mode it picked at startup.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
We don't need this anymore. Since we only support one output,
we can use the first output for any place we used to use it.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
If an output exists, it should be cleaned up automatically when we release
all our heads later anyway.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
By the time we get to this failure we may (or may not) have implanted a
listener.
freerdp_listener_free() safely handles passed NULL pointers, so we don't
have to add any additional checks on our end.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Currently, this is what backend->output would be anyway. This is another
step towards removing the singleton output stored in the backend.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
The RDP backend wants to be able to change scale for existing outputs,
so try to hook this up mostly in the same way a mode switch works.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
When changing to/from the native mode, or when changing the native
mode we need to damage the changed output to ensure it's redrawn.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Some backends have special head specific state that doesn't fit into the
existing generic head setter functions, and is too specific to make more
functions for.
RDP's primary output flag is an example.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
This triggers my new favourite assert sometimes when making RDP
connections, so just update the transform here.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
We are assuming that CRTC_CTM is pass-through, so better ensure it
really is pass-through rather than whatever the previous KMS client left
there.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
We are assuming that CRTC_DEGAMMA_LUT is pass-through, so better ensure
it really is pass-through rather than whatever the previous KMS client
left there.
This too falls under deprecated_gamma_is_set check, because the legacy
gamma could use either GAMMA or DEGAMMA.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
We are assuming that gamma LUT is pass-through, so better ensure it
really is pass-through rather than whatever the previous KMS client left
there.
Unfortunately the legacy ioctl does not offer any way to reset the LUT
without actually crafting an identity LUT.
If the legacy gamma libdrm function indicates the feature is not
supported, do not try to use it again. This avoids hammering the legacy
gamma every frame when deprecated drm_output_set_gamma() is not used.
drm_output_set_gamma() is not updated to check/set this flag in order to
maintain its old behavior.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
We are assuming that CRTC_GAMMA_LUT is pass-through, so better ensure it
really is pass-through rather than whatever the previous KMS client left
there.
Unfortunately, we have old drm_output_set_gamma() API that cms-static
and cms-colord plugins are using. To avoid trampling over them, do not
touch gamma after they did. Those plugins are deprecated, so there is no
reason to make set_gamma work through atomic.
drm_output_set_gamma() is called from weston_compositor_add_output()
through the output_created_signal. This is after drm_output_enable() and
before any KMS modeset or atomic commit on the CRTC. Therefore it is not
possible that there would be any KMS action in flight when
drm_output_set_gamma() is called, and so setting deprecated_gamma_is_set
should be non-racy - like setting CRTC_GAMMA_LUT itself.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Move the debug printing before the bail-out if the property does not
exist. This means that trying to set a missing property will be logged,
and it can be identified by the property id being 0.
We are starting to program more KMS properties, and if KMS state
building fails, this gives better chances to figure out what happened.
For example, if we accidentally assume that some property always exists
when it doesn't.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Add an explicit request to the backend config to choose the renderer.
Currently, only Pixman remains supported, with auto defaulting to that.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Add an explicit request to the backend config to choose the renderer.
Currently, only Pixman remains supported, with auto defaulting to that.
Signed-off-by: Daniel Stone <daniels@collabora.com>
When we're selecting our renderer, use the enum rather than two
mutually-exclusive booleans to not use the no-op renderer.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Add an 'auto' or unspecified renderer type, so we can use enum
weston_renderer_type during the configuration stage, where the target
renderer may be unspecified or unknown.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Similar to the backend type, also expose the renderer type enum as ABI.
This makes it possible to implement a more consistent config API, as
opposed to every backend hand-rolling its own use-the-other-one bool.
The enums are explicitly numbered to avoid 0, so 0 can be used as a
'not-specified' sentinel value to allow backwards compatibility with the
old config interfaces.
Signed-off-by: Daniel Stone <daniels@collabora.com>
libweston contains weston_config and weston_shell_utils utilities
functions so include these in the sphinx documentation as well.
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
This doesn't really belong into shell-utils, so better move it out to
shared/config-parser. Renamed to weston_config_get_binding_modifier
to maintain the same namespace.
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
These shell utils functions are potentially useful to other shells as
well, so make them widely available.
Renamed all functions to weston_shell_utils namespace.
No functional change, copied ad litteram.
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
We've passed the paint node deeply enough that we can now use the cached
matrix in it to save some calculations.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Rename it as well.
By passing the paint node we'll have access to the cached buffer to output
matrix later.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
When the pointer is confined and we are handling motion, we need
up-to-date view transforms in order to respect the confinement. So
update view transforms in such case.
This fixes an issue in which we'd try to transform views with dirty
transforms in weston_pointer_clamp_event_to_region(), hitting an assert.
This is a leftover of d611ab24 "libweston: Update view transforms more
often".
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
When the output is moved, move its per-framebuffer accumulated damage
with it. The alternative would be to track accumulated damage in the
local output coordinate system, but that would require translating back
into global coordinates for repaint_output.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Currently linux-dmabuf.h implicitly depends on libweston.h being
included before it. Instead of adding missing includes for bool,
dev_t, struct timespec, struct weston_compositor,
struct weston_drm_format_array, struct wl_array, and struct wl_list,
just include libweston.h.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
The main surface commit includes the xdg surface commit. Here, the
accumulated surface size is validated. All subsurfaces must be comitted
first to ensure that the corrent current values are used.
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
Use the gl-renderer border code shared with headless-backend. We can
drop all this open-coded stuff.
In the output disable path, make sure to call this only when gl-renderer
is used. It will also reset the border state in gl-renderer, which is
harmless here, and it's necessary in the resize path.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Move this code from headless-backend to a helper library, so it can be
shared with wayland-backend.
gl-renderer.h was missing #pragma once, which made the build fail.
Unfortunately gl-borders needs gl-renderer.h which will attempt to
include EGL headers if gl-renderer is enabled in the build, so we must
get the EGL build flags too, just for the headers.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Refactor this code into two backend-agnostic functions, that in the next
step can be moved into code shared between backends.
Pure refactoring, no changes.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
GL-renderer keeps the pointer for the border image data for later use.
When we destroy our cairo_surface, that data gets freed. Therefore,
reset the pointer stored by GL-renderer too, to ensure use-after-free
cannot happen.
This is not really necessary in this particular case, because the
renderer output is destroyed immediately after, so there is no chance of
UAF.
However, this is needed for code sharing purposes. Wayland-backend will
do exactly this when an output is resized. To share this code with
wayland-backend, we also need to reset the pointers. It's harmless and
more correct in the output disable path.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Replace a bunch of copied code with a loop over a simple array. This
makes the code easier to read, and allows further refactoring.
This is pure refactoring, no changes to results or behavior.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Make sure dep_egl is always a valid dependency object, even if not
found. Ensure it is not found when not wanted, to avoid linking when
found but not wanted.
Using a not-found dependency in Meson is defined to be a safe no-op, so
use that to simplify the backend dependencies.
libweston/meson.build already errors out if renderer-gl is enabled and
EGL is not found, so the same checks can be removed from the backends.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Now that struct weston_head contains a pointer to the backend owning
the head, choose to call the same backend's create_output callback.
This will be necessary to support loading multiple backends
simultaneously.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Compositor code can use opaque pointer comparison to determine whether
a head belongs to a given backend. Store a backend pointer in struct
weston_head to enable the compositor to select the correct backend
specific output configuration code.
This also allows to use the backend pointer instead of the opaque
backend_id pointer to check whether a head belongs to a backend, so
replace the checks in all to_xyz_head() functions and drop backend_id.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
No output is created at this point. Even if an output existed,
it should be relased when the head ist destroyed.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
vnc_ensure_matching_mode() does not transfer flags from init_mode.
Set flags on the returned mode instead.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Let xzalloc warn and abort when running out of memory.
With this, vnc_insert_new_mode() and vnc_ensure_matching_mode() can not
return NULL anymore. Remove the now unnecessary error handling.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Let xzalloc warn and abort when running out of memory.
With this, vnc_head_create() doesn't need an error return value anymore.
Remove the now unnecessary error handling.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Let xzalloc warn and abort when running out of memory and remove the
now unnecessary error handling.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Now that the renderer type is stored in struct weston_renderer,
use that instead of use_pixman to determine the renderer type.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Now that the renderer type is stored in struct weston_renderer,
use that instead of use_pixman to determine the renderer type.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Now that the renderer type is stored in struct weston_renderer,
use that instead of use_pixman to determine the renderer type.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Move the renderer type from struct headless_backend into struct
weston_renderer to store the chosen renderer type in a unified manner.
This will later allow secondary backends to determine the renderer type
chosen by the primary backend.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
When the pointer is constrained and the surface in which it is
constrained gets committed, we may warp the pointer in some
circumstances. In order to do that, we need the associated view
transforms up-to-date. So update view transforms when this surface gets
committed.
This fixes an issue in which we'd hit an assert when trying to warp the
pointer, as we were trying to transform views with dirty transforms.
This is a leftover of d611ab24fd
"libweston: Update view transforms more often".
Signed-off-by: Leandro Ribeiro <leandro.ribeiro@collabora.com>
This should prevent extra focus signal emission the similarly to
how default_grab_pointer_focus() does, though we don't have the
surface jumping logic here.
This stops xdg pings from being sent every output repaint during
a grab when the pointer isn't in any windows belonging to the
grab parent.
An example would be running weston-terminal, bringing up the
right click pop-up, and moving the mouse onto the desktop while
another client causes repaints.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
This prevents a spurious pointer focus clear at the start of a grab.
This would, for example, cause an extra pointer leave when bringing
up a right-click pop-up in a ttk app like weston-terminal.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
When the new option is enabled, headless backend will draw decorations
around its outputs. This makes the actual "framebuffer" larger by the
thickness of the decorations to keep the video mode area free for
clients.
This will be needed for a future test, that will ensure that GL-renderer
will paint the output decorations correctly.
The output title is deliberately NULL, because text rendering is
unpredictable and depends on e.g. what fonts are installed in the
system. Therefore screenshot testing of any text would be really
painful, so let's avoid that.
The decorations setup code is mostly copied from wayland-backend.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Do not call weston_seat_release() if Weston aborts before
weston_seat_init() could be called. This fixes a possible
segfault due to uninitialized list traversal in the error
path.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
There are no internal users left for this protocol, they have been
migrated to the new weston-output-capture protocol. There are no
external users, because this protocol was private and never installed.
Remove this dead code.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This services output capture tasks for the 'framebuffer' and 'full
framebuffer' pixel sources.
Both pixel sources come from the same source: the EGLSurface. The only
difference is the area. The EGLSurface contains the borders used for
output decorations, hence 'full framebuffer' is possible to capture.
We use GL_ANGLE_pack_reverse_row_order extension to make glReadPixels
return the image data in the layout we need for wl_shm buffers directly.
Without the extension we have to flip manually.
Another extension to the same effect is MESA_pack_invert, but this is
not specified for GL ES. It also uses a different token value, so it
cannot be directly substituted even if supported.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This services output capture tasks for the 'framebuffer' and 'blending'
pixel sources.
Just like the old screenshooting path, the 'framebuffer' pixel source is
the hardware buffer, whether a shadow is used or not. This may not be
the best for performance, but you do get the real framebuffer contents.
Maybe it's rgb565, or even less.
When the shadow buffer is used, I realized it is effectively the same as
the intermediate blending buffer in GL-renderer when color management is
used. Pixman-renderer does non-linear blending only, so the shadow
buffer is in the blending space. The shadow buffer is also always 8 bpc
regardless of the hardware framebuffer, so the read-back may be
different from the hardware framebuffer. Read-back from the shadow is
optimal for performance, but not what the hardware gets.
'full-framebuffer' source cannot yet be implemented, because backends do
not tell Pixman-renderer about the margins where the wayland-backend
blits the output decorations. The target "hardware" buffer handed to
pixman-renderer does not allow accessing the decorations area.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
