We already swapped the front buffer, and even if it didn't get
presented, we should still swap our representation of the state, to not
get into a confused buffer tracking state.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1822>
All pointer a11y is a fabrication of Clutter backend-independent
code, with the help of a ClutterVirtualInputDevice and with some
UI on top.
On the other hand, MetaInputSettings is a backend implementation
detail, this has 2 gotchas:
- In the native backend, the MetaInputSettings (and pointer a11y
with it) are initialized early, before the ClutterSeat core
pointer is set up.
- Doing this from the MetaInputSettings also means another dubious
access from the input thread into main thread territory.
Move the pointer a11y into ClutterSettings, making this effectively
backend-independent business, invariably done from the main thread
and ensured to happen after seat initialization.
Fixes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1765
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1849>
Compositor behaviour when calling `wl_data_device_manager_get_data_device`
for the same seat multiple times is not very clearly defined in
the spec and both Mutter and Weston currently don't handle
the DnD case properly.
While Weston handles DnD only for the last created data device,
Mutter, because of some internal reshuffling, ends up toggling
between two devices.
Properly handling this case requires some bigger changes. So
in order to behave predictable and in line with Weston,
only take the last created data device into account while
still keeping the previous created ones around.
The main affect client here is Firefox, which gets very
confused by the toggling behaviour and becomes more stable
with this patch.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1841>
Virtual Kernel Mode Setting (vkms) is a virtual /dev/dri/card* device
not backed by any actual hardware. It's intended for testing purposes,
e.g. to run tests suites with a reproducable setup, or in continuous
integration pipelines.
Currently mutter don't have any tests that can run on top of vkms, but
will eventually get that. To prepare for the ability to do that, and
having said kernel module loaded without causing wierd issues with any
active session, add an udev rule that tells mutter to ignore any vkms
device.
Otherwise, when vkms is loaded, mutter would detect it, assume it's a
regular monitor, configure it as such, thus add a region of the stage
that ends up nowhere, which isn't very helpful. It might also conflict
with running actual tests that need to interact with vkms if the active
session has taken control of it.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1740>
With atomic mode setting, commits don't work when CRTCs aren't enabled,
which they aren't when we're power saving. This means the gamma state
fails to being update. To fix night light and for whatever other reason
gamma ramps was changed during power saving by marking the CRTC gamma
state as invalid when leaving power saving, as well as when resuming.
This means that the next frame will append the CRTC gamma state to the
KMS commit.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1755
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1835>
This GSource is not being properly unref nor the variable holding it
cleared. This on one hand leaks the GSource memory, on the other hand
may trigger warnings in keyboard_repeat() as the source may be
(reentrantly) cleared, yet we don't exit early as
seat_impl->repeat_source is never NULL.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1839>
When a viewport source rect or destination size is set, `stex->dst_width`
gives us the the cropped and/or scaled size. At this step, we need the
uncropped/unscaled size however.
Note: this is only ever relevant if buffer transform and viewport are
used together - otherwise the values are the same.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1836>
The shadow size is factored into the paint volume MetaWindowActorX11
returns in its get_paint_volume() vfunc override, so we should
invalidate the paint volume every time that shadow might change.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1829>
When using buffer transforms and viewports together, we currently
apply the transformation (read: rotation) first, resulting in
wrong buffer coordinates for viewport source rects.
Flip the order in whitch we apply our matrix transformations.
This can be tested e.g. via:
`weston-simple-damage --use-viewport --transform=flipped-180`
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1832>
If a subsurface first gets reordered and afterwards detached from
the parent before the parent surface got commited, we currently
would end up reattaching the subsurface to its previous parent.
While clients should avoid this behaviour, it's legit according
to the spec.
We already prevent similar cases where the subsurface is destroyed -
extend that check to detaching, which includes the destroy case.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1831>
Currently when reordering subsurfaces, we un- and reparent all child
actors of the window actor. This is unnecessarily wasteful and
triggers bugs in clutter. While the underlying issue should be fixed
eventually, simply reorder the actors with the tools clutter provides
us with, avoiding those bugs and likely being faster as well.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1831>
The X server generates a property change notification whenever it processes a
property change request, even if the value of the property is not changing. This
triggers libgdk to probe all display outputs, which can be slow depending on
which display driver and hardware are in use.
#0 0x00007f8e4d5e91a0 in XRRUpdateConfiguration () at /usr/lib/libXrandr.so.2
#1 0x00007f8e505208da in _gdk_x11_screen_size_changed (screen=0x5566e4b7e080, event=0x7ffe0e44bd60) at ../gdk/x11/gdkscreen-x11.c:1199
#2 0x00007f8e505066d1 in gdk_x11_display_translate_event (translator=0x5566e4b5b110, display=0x5566e4b5b110, event=0x7f8dec001b20, xevent=0x7ffe0e44bd60) at ../gdk/x11/gdkdisplay-x11.c:1201
#3 0x00007f8e505135a0 in _gdk_x11_event_translator_translate (translator=0x5566e4b5b110, display=0x5566e4b5b110, xevent=0x7ffe0e44bd60) at ../gdk/x11/gdkeventtranslator.c:51
#4 0x00007f8e50512c97 in gdk_event_source_translate_event (event_source=0x5566e4b764a0, xevent=0x7ffe0e44bd60) at ../gdk/x11/gdkeventsource.c:243
#5 0x00007f8e50512f57 in _gdk_x11_display_queue_events (display=0x5566e4b5b110) at ../gdk/x11/gdkeventsource.c:341
#6 0x00007f8e50497644 in gdk_display_get_event (display=0x5566e4b5b110) at ../gdk/gdkdisplay.c:442
#7 0x00007f8e5051301f in gdk_event_source_dispatch (source=0x5566e4b764a0, callback=0x0, user_data=0x0) at ../gdk/x11/gdkeventsource.c:363
#8 0x00007f8e516ecf9c in g_main_context_dispatch () at /usr/lib/libglib-2.0.so.0
#9 0x00007f8e51740a49 in () at /usr/lib/libglib-2.0.so.0
#10 0x00007f8e516ec503 in g_main_loop_run () at /usr/lib/libglib-2.0.so.0
#11 0x00007f8e508ef5fd in meta_run_main_loop () at ../src/core/main.c:928
#12 0x00007f8e508ef60e in meta_run () at ../src/core/main.c:943
#13 0x00005566e450842a in ()
#14 0x00007f8e50649b25 in __libc_start_main () at /usr/lib/libc.so.6
When GNOME is animating a display fade when the night light feature is toggled
on or off, it sends a lot of change requests for the CTM property in the
process, which triggers a lot of display probes from gdk. In the case of the
night light feature, the CTM itself is not actually changing, so these requests
are redundant. Fix this by caching the CTM value in the MetaOutputXrandr and
skipping the server requests if it's not being changed.
Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/3978
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1816>
When we set the matrix, we checked the device mapping mode in the main
thread, then passed along the calculated matrix to the input thread for
application. This could however be racy, as the mapping mode is managed
in the input thread. Fix this by sending the unaltered matrix, having
the input thread checking the mapping mode.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1806>
The connector state wasn't properly predicted, as it earlied out if
the connector wasn't part of a mode set connector list.
Instead use the old CRTC to check whether it was used in any mode set,
and whether the connector was part of any new mode set, to predict
whether the connector is inactive or active.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1821>
When a device only had mode sets which turned off monitors, not enabling
anything, there would be no KMS update created and posted, and the
active monitors would remain on.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1821>
On hybrid graphics system, the primary path used to transfer the stage
framebuffer onto the dedicated GPU's video memory preparing for scanout,
is using the dedicated GPU to glBlitFramebuffer() the content from the
iGPU texture onto the scanout buffer.
After we have done this, we reset the current EGL context back to the
one managed by cogl. What we failed to do, however, was to reset the
current EGL context when we inhibited the actual page flip due to having
entered power save mode.
When we later started to paint again, Cogl thought the current EGL
context was still the correct one, but in fact it was the one used for
the iGPU -> dGPU blit, causing various EGL surface errors, and as a side
effect, eventually hitting an assert.
Fix this by making sure we reset to the Cogl managed EGL context also
for this case.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1803>
Destroying the EGLSurface frees the underlying container structs. When
we call gbm_surface_release_buffer() with a gbm_surface the EGLSurface
was created from, doing that after the EGLSurface was destroyed results
in attempts to access freed memory. Fix this by releasing any buffer
first, followed by destroying the EGLSurface, and lastly, the
gbm_surface.
This was not a problem prior to CoglOnscreen turning into a GObject, as
in that case, the dispose-chain was not setup correctly, and the
EGLSurface destruction was done in the native backend implementation.
This also changes a g_return_if_fail() to a g_warn_if_fail(), as if we
hit the unexpected case, we still need to call up to the parent dispose
vfunc to not cause critical issues.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1803>
It's handled by CoglOnscreenEgl's dispose() implementation. It was
failed to be invoked in the past because the old non-GObject web of
vtables were not setup correctly, meaning the old generic EGL layer of
the CoglOnscreen de-init was never invoked.
When the type inheritence was cleaned up, this mistake was not cleaned
up, so do that now.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1803>
Mutter would deny the application the right to resize itself during an
interactive resize, to avoid the user and the client to fight for the
size.
When the client is not allowed to resize, it would use the client rect
rather than the buffer rect.
As a result, the client window with client side decorations would
quickly shrink to its minimum size.
Use the buffer rect instead, so that the size really remains the same.
https://gitlab.gnome.org/GNOME/mutter/-/issues/1674
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1777>
Meson doesn't seem to handle depending on generated headers, at least
when those headers are pulled in indirectly via another header file.
Luckily, we don't actually need to include the generated D-Bus boiler
plate in meta-monitor-manager-private.h, since the MetaMonitorManager
type no longer is based on the D-Bus service skeleton.
So, by moving the inclusion of the generated D-Bus header file into
meta-monitor-manager.c, we should hopefully get rid of the sporadic
build issues.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1682
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1819>
If there was only a single mode, add the common modes to provide options
to select other resolutions than the built in default. This avoids
issues where the connector listed multiple supported modes, but where
the common modes added would exceed the possible bandwidth. We could
probably make an attempt to filter out more modes from the common mode
list to avoid these issues, but it's likely that the driver already
lists suitable modes, meaning there is no point in adding the common
modes.
The common modes were initially added[0] to add modes to connectors with
a single bundled mode, so we shouldn't regress the original bug fix.
[0] https://bugzilla.gnome.org/show_bug.cgi?id=744544
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1232
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1824>
The intel DRM driver is known for not being able to handle multi head
setups when KMS modifiers are enabled, due to the implicitly selected
modifiers, while being more suitable for single head setups, cause
bandwidth issues when a certain number of monitor times resolution and
refresh rate is configured.
We don't yet support automatically finding a combination of modifiers
that work, and have because of this disabled KMS modifiers unless the
driver actually needs it.
Lets flip this configuration the other way around, changing the current
udev rule to decide wen to *disable* KMS modifier support, as it so that
only the Intel driver has this problem, while on the other hand, there
several drivers that requires modifiers to function at all.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1792>
The input thread is in deep water doing the meta_is_*() check itself,
as that pokes the MetaMonitorManager managed by the main thread. Use
the getter from the MetaViewportInfo instead.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1793>
We need to pass this info from the main thread, as that pokes the
MetaMonitorManager underneath. Store it in the MetaViewportInfo
so that the input thread can use this information.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1793>
Use the new API to make sure the shaped texture has a valid size
during the next layout phase.
This is needed here because, quoting the previous commit:
When the texture size is invalidated using `invalidate_size()`, the new
size will only get calculated the next time `update_size()` is
called. This happens e.g. in `meta_shaped_texture_get_preferred_size()`
via `ensure_size_valid()`.
`update_size()` can chain up to `clutter_content_invalidate_size()`
as well as emitting a `size-changed` signal. If this happens during
layout, the result is a 'change the layout conditions during layout'
issue, causing heavy breakage in e.g. the Shell overview.
To fix this, expose `ensure_size_valid()` as API so callers can make
sure the texture has a valid size without creating redundant size
invalidations calls.
Note that if a buffer with a new size is attached we already trigger
`update_size()` explicitely, avoiding such situations.
Closes https://gitlab.gnome.org/GNOME/mutter/-/issues/1718
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1799>
When the texture size is invalidated using `invalidate_size()`, the new
size will only get calculated the next time `update_size()` is
called. This happens e.g. in `meta_shaped_texture_get_preferred_size()`
via `ensure_size_valid()`.
`update_size()` can chain up to `clutter_content_invalidate_size()`
as well as emitting a `size-changed` signal. If this happens during
layout, the result is a 'change the layout conditions during layout'
issue, causing heavy breakage in e.g. the Shell overview.
To fix this, expose `ensure_size_valid()` as API so callers can make
sure the texture has a valid size without creating redundant size
invalidations calls.
Note that if a buffer with a new size is attached we already trigger
`update_size()` explicitely, avoiding such situations.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1799>
If only a viewport destination size is set, the noop viewport has
to take the buffer scale into account.
If a viewport source but no viewport destination size is set, the
destination size is that of the viewport source, not of the whole
texture.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1786>
If a client is naive enough to assume that it can set the selection while
it has got no surfaces, mutter will simply ignore the request and leave
the selection unchanged.
This is good and the expected behavior, however the poor client that did
this will enter in an inconsistent state where it "claimed" the selection,
but nobody told it that the wl_data_source is not current.
So, when the client is focused the next time, it will receive wl_data_offers
as usual, but it will still think all the time that it is owning the
selection. In the case of GTK, that takes client-side shortcuts, so any
attempted paste will still bring back the client-side aborted selection.
To fix this, cancel the selection right away if it happened while unfocused,
the client will be able to undo its own failed selection, and not assume
that future offers are its own.
Fixes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1469
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1772>
It seems to be the preferred format of the Mesa V3D driver on
Raspberry Pi 4. If the compositor doesn't advertise it then Mesa will
fallback from `zwp_linux_dmabuf_v1` to `wl_drm`, incorrectly. Meaning
it will keep using a buffer with modifiers on an interface that does
not have modifiers.
Add support for `DRM_FORMAT_ABGR2101010`. It works, and prevents Mesa
from taking its broken fallback path.
Fixes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1520
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1798>
As documented in g_once_init_enter(): "While @location has a volatile qualifier,
this is a historical artifact and the pointer passed to it should not be
volatile.". And effectively this now warns with modern glibc.
Drop the "volatile" qualifier from these static variables as it's expected.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1785>
For Xwayland, mutter creates the sockets in the standard /tmp/.X11-unix
directory.
Yet, if that directory already exists, it may have been created by
another user with full control over the created socket.
To avoid that issue, if the directory /tmp/.X11-unix already exists,
check that the permissions are as we expect, i.e. the directory belongs
to either root or the user herself, is writable and has the sticky bit.
Thanks to fabian@ritter-vogt.de for reporting that issue.
https://gitlab.gnome.org/GNOME/mutter/-/issues/1708
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1787>
Unlike other subsurface state, placement operations need to get
applied in order. As per spec:
```
Requests are handled in order and applied immediately to a pending
state. The final pending state is copied to the active state the
next time the state of the parent surface is applied.
```
Having placement operations being part of the subsurface state
makes it difficult to support arbitrary orderings. Make them
part of the parents surface pending state instead.
Closes https://gitlab.gnome.org/GNOME/mutter/-/issues/1691
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1768>
MetaInputSettings unref:ed the seat on destruction, but it never ref:ed
it on construction, meaning it "stole" the reference from the rightful
owner. Make MetaInputSettings less of a thief.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1775>
XIQueryPointer allocates the button state mask that we were leaking in
some places. We need to manually free this, because there is no XI
function that would do this for us.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1728>
In case the shell ignores or can't accept the restart request we should
hide the message that has been just requested to be shown.
As per ::show-restart-message signal documentation, this has to be done by
emitting the signal with a NULL message.
So follow the API properly in such case
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1780>
In non-systemd managed session we are unable to start services on
demand. Instead, gnome-session will start everything at login time,
including any X11 related service (i.e. gsd-xsettings).
However, in order to start gsd-xsettings, Xwayland needs to be started
already. Otherwise it will connect to GNOME_SETUP_DISPLAY and login will
hang at that point.
Fix this by detecting whether mutter is running in a systemd unit. If it
is, we assume that we are systemd managed and the machinery to start the
services works fine. If not, we assume that the session management may
unconditionally try to start X11 related services and Xwayland must be
started in order to not block this.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1771>
On Wayland MetaInputSettings is part of the input thread. Connecting
a GSettings binding to the default ClutterSettings could result in the
change notification being emitted on the input thread. This then could
end up triggering the same handler from two different threads at the
same time. In the case of the ClutterText layout cache it was attempting
to unref the same layout twice, leading to a crash.
This can be avoided by simply removing the GSettings bind. This does not
cause changes to this setting to be missed by ClutterSettings because it
itself already sets up a bind.
Fixes https://gitlab.gnome.org/GNOME/mutter/-/issues/1696
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1776>
When deciding if `configure` event should be sent,
`meta_window_wayland_move_resize_internal` compares requested window size
with `window->rect` size. However, `window->rect` is only updated when `commit`
is received. So the following sequence produces incorrect result:
1. a window initially has size `size1`
2. `move_resize_internal` is called with `size2`. `configure` is sent
3. `move_resize_internal` is called with `size1` to restore original size,
but `commit` for `size2` haven't arrived yet. So `window->rect` still has size
`size1`, and thus new `configure` is not sent
4. `commit` for `size2` arrives, window changes size to `size2`
Expected window size in the end: `size1`
Actual: `size2`
To fix the issue, take size from pending `configure` events into account.
Fixes https://gitlab.gnome.org/GNOME/mutter/-/issues/1627
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1755>
This eliminates the need for any render node or device nodes, thus can
be used without any graphics hardware available at all, or with a
graphics driver without any render node available.
The surfaceless mode currently requires EGL_KHR_no_config_context to
configure the initial EGL display.
This also means we can enable the native backend tests in CI, as it
should work without any additional privileges.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
Tests that creating and starting a virtual screen cast monitor works,
and that at least one one buffer is processed.
Currently the content of the buffer isn't checked more than it can be
mmap():ed. Only MemFd buffers are tested for for now, as DMA buffers
would need a surfaceless EGL context to check properly.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
The new RecordVirtual API creates a virtual monitor, i.e. a region of
the stage that isn't backed by real monitor hardware. It's intended to
be used by e.g. network screens on active sessions, virtual remote
desktop screens when running headless, and scenarios like that.
A major difference between the current Record* API's is that
RecordVirtual relies on PipeWire itself to negotiate the refresh rate
and size, as it can't rely on any existing monitor, for those details.
This also means that the virtual monitor is not created until the stream
negotiation has finished and a virtual monitor resolution has been
determined.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
The area source, window source, and monitor source, currently set up the
stream size up front, given the area, maximum allowed window size or
monitor resolution, but for to be introduced sources, the size will be
negotiated using PipeWire, instead of specified via the D-Bus API. This
commit changes the internal source API to allow for this. There are
currently no users of this new behaviour.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
There may be a race between the ability to turn stream relative input
coordinates and turning them into screen coordinates, due to the future
scenario where the entity backing a stream is created and managed ad-hoc
depending on PipeWire stream negotiations.
When an input event is sent during this time, drop it.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
Make it possible to create persintent virtual monitors using command
line argument. This will not be the only way to create virtual monitors,
the primary way will be using the screen cast API, but using command
line argumenst is convenient for debugging purposes.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
The testing currently done is:
* Creating a virtual monitor succeeds and gets the right configuration
* Painting a few times results in the expected output
* Changing the content of the stage also changes the painted content
accordingly
* Destroying the virtual monitor works as expected
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
This adds a test framework that makes it possible to compare the result
of painting a view against a reference image. Test reference as PNG
images are stored in src/tests/ref-tests/.
Reference images needs to be created for testing to be able to succeed.
Adding a test reference image is done using the
`MUTTER_REF_TEST_UPDATE` environment variable. See meta-ref-test.c for
details.
The image comparison code is largely based on the reference image test
framework in weston; see meta-ref-test.c for details.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
Virtual monitors are monitors that isn't backed by any monitor like
hardware. It would typically be backed by e.g. a remote desktop service,
or a network display.
It is currently only supported by the native backend, and whether the
X11 backend will ever see virtual monitors is an open question. This
rest of this commit message describes how it works under the native
backend.
Each virutal monitor consists of virtualized mode setting components:
* A virtual CRTC mode (MetaCrtcModeVirtual)
* A virtual CRTC (MetaCrtcVirtual)
* A virtual connector (MetaOutputVirtual)
In difference to the corresponding mode setting objects that represents
KMS objects, the virtual ones isn't directly tied to a MetaGpu, other
than the CoglFramebuffer being part of the GPU context of the primary
GPU, which is the case for all monitors no matter what GPU they are
connected to. Part of the reason for this is that a MetaGpu in practice
represents a mode setting device, and its CRTCs and outputs, are all
backed by real mode setting objects, while a virtual monitor is only
backed by a framebuffer that is tied to the primary GPU. Maybe this will
be reevaluated in the future, but since a virtual monitor is not tied to
any GPU currently, so is the case for the virtual mode setting objects.
The native rendering backend, including the cursor renderer, is adapted
to handle the situation where a CRTC does not have a GPU associated with
it; this in practice means that it e.g. will not try to upload HW cursor
buffers when the cursor is only on a virtual monitor. The same applies
to the native renderer, which is made to avoid creating
MetaOnscreenNative for views that are backed by virtual CRTCs, as well
as to avoid trying to mode set on such views.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
The order of which function argument expressions are executed is
undefined, so don't rely on this for setting the background colors, as
it results in different colors on different architectures.
For example, it has been observed that the order of execution is
reversed comparing x86_64 and aarch64, making these two architectures
having different background color.
Fix this confusion, and also reproduceability in future reference tests,
by making the order of execution predictable.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
It's useful to be able to have very very tiny monitors (e.g. 60x60
pixels) when doing reference testing, as tests have reference images
that the output is compared to. Smaller reference images the less
storage they use.
To avoid annoying pointless warnings when this is done, change the
pedantic workspace work area code to be more forgiving if the work area
happens to match the display size.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
When rebuilding the monitors (e.g. during hotplug), make sure to detach
the disposed monitors from any outputs before creating the new monitors.
While this isn't currently needed, as outputs are too being recreated,
with the to be introduced virtual outputs that are created for virtual
monitors, this is not always the case anymore, as these virtual outputs
are not regenerated each time anything changes.
Prepare for this by making sure that cleaning up disposed monitors
detach themself properly from the outputs, so new ones can attach
themself to outputs without running into conflicts.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
This makes it possible to pass custom properties to backends when
constructing tests. This will be used to create "headless" native
backend instances for testing the headless native backend.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
With this commit, it's possible to run mutter without being DRM master.
It's not yet possible to add virtual monitors, but one can for example
already add virtual input devices.
This currently doesn't try to hook up to any logind session, thus will
not have a real seat assigned. Currently it's hard coded to "seat0".
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
Currently our only entry point for DRM devices is MetaKms*, but in order
to run without being DRM master, we cannot use /dev/dri/card*, nor can
we be either of the existing MetaKmsImplDevice implementation (legacy
KMS, and atomic KMS), as they both depend on being DRM master.
Thus to handle running without being DRM master (i.e. headless), add a
"dummy" MetaKmsImplDevice implementation, that doesn't do any mode
setting at all, and that switches to operate on the render node, instead
of the card node itself.
This means we still use the same GBM code paths as the regular native
backend paths, except we never make use of any CRTC backed onscreen
framebuffers.
Eventually, this "dummy" MetaKmsImplDevice will be replaced separating
"KMS" device objects from "render" device objects, but that will require
more significant changes. It will, however, be necessary for e.g. going
from being headless, only having access to a render node, to turning
into a real session, with a seat, being DRM master, and having access to
a card node.
This is currently not hooked up, but will be in a later commit.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
Add a flag to MetaSeatNative and MetaSeatImpl that tells it not to
attempt to create a libinput context. This is intended to be used when
mutter is to run headless, as in without any input devices other than
virtual ones.
Currently not hooked up.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
This leaves only the atomic mode setting cap check before creating the
impl device, aiming to make it possible to create a non-mode-setting
MetaKmsImplDevice implementation.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
Make it possible to pass --headless as a command line argument in order
to turn the native backend "headless". This currently doesn't do
anything, but the intention is that it should not use logind nor KMS,
and work completely headless with only virtual outputs.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1698>
Input settings requires a valid seat in order to initialize the a11y
settings (since commit 1609d145), however in X11 we never set it and
even if we create the input settings early (as per commit 7547891a) we
never initialize the seat for it.
This leads to startup critical errors on X11:
clutter_seat_get_pointer_a11y_settings: assertion
'CLUTTER_IS_SEAT (seat)' failed
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1763>
This commit adds the events created in the function
`meta_seat_x11_notify_devices` to the clutter events queue, which
are currently only added to the stage queue making the events not
being picked up by the `clutter_seat_handle_event_post` function.
This results in devices not getting added to the device-list of
`MetaInputSettings`.
Fixes the bug in which mouse and touchpad settings are not working in
the settings app during x11 session.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1767>
Since commit 2ceac4a device-related X11 events aren't processed anymore,
causing the input settings not to handle the devices.
This is due to the fact that we may never call clutter_seat_handle_event_post()
for such events.
While this is always happening for the native backend, it doesn't happen in
X11 because the events are removed from the queue as part of
meta_x11_handle_event(), and thus no event was queued to the stage by the
backend events source.
This also makes sure that the event post handler is called after the
event is actually processed, and not before an event is queued.
Fixes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1564
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1769>
The nested backend may need to have an input setting implementation,
while we don't want to change the host settings (re-using an X11 input
settings) we can add a dummy implementation, until something more
complex is needed.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1769>
The sync ring has an API about "frames", where it is notified about
the end of frames. However, its "insert wait" call is done before
updates, meaning that some "insert waits" will never see the "after
frame" if there was no frame drawn. This will cause mismatching in the
frame counting, causing freezes in the synchronization until something
else triggers an actual frame, effectively "unfreezing" the sync ring.
Fix this by not only notifying the sync ring about frames when there
were actual frames drawn, but also on plain updates which didn't result
in a drawn frame.
Related: https://gitlab.gnome.org/GNOME/mutter/-/issues/1516
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1754>
When a gtk theme uses larger shadows for the unfocused state than for
the focused one, this can cause a crash in meta_frame_left_click_event.
Since whether to call meta_frame_left_click_event is decided based on
the decoration size before focusing and the control that was clicked on
after focusing, this can result in an event handled in
meta_frame_left_click_event being on the client area.
Fixes https://gitlab.gnome.org/GNOME/mutter/-/issues/1668
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1748>
With commit 7d78768809 we switched to
storing pointer coordinates in MetaInputDeviceNative instead of
ClutterInputDevice, and while we had set the coordinates of the
ClutterInputDevice in ClutterStage when queueing an event, we now set
the MetaInputDeviceNative coordinates in new_absolute_motion_event().
Here a small mistake snuck in: new_absolute_motion_event() only
translates the coordinates of the event, but we call
meta_input_device_native_set_coords() using the x and y variables
(which remain untranslated), so now the input device coordinates are no
longer translated.
Fix that by translating the coordinates of the x and y variables in case
we're we handling a tablet/stylus event instead of only translating the
event coordinates.
Fixes https://gitlab.gnome.org/GNOME/mutter/-/issues/1685
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1760>
This fixes the interpolate test to not use the wall clock, but the
monotonic clock. It also cleans up the timestamp granularity naming, so
that the different granularity is clearer, as in the test, different
timestamps have different granularity.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1751>
This more or less rewrites this test so that it explicitly tests the
"interpolation" when a timeline loops, i.e. that if something occupies
the thread when a timeline was supposed to have looped, we end up in the
right place "in the middle" of the next timeline cycle.
The test more or less does this:
* Start a 3 second looping timeline
* Sleep so that we're in the middle of the first cycle
* Sleep again so that we end up in the middle of the next cycle
The semantics checked are that we see the following frames:
* The first frame with timestamp 0
* The second frame in the middle of the first cycle (timestamp ~= 1.5
sceonds)
* The third frame in the end of the first cycle (timestamp == 3.0
seconds)
* The fourth frame, first in the second cycle, with timestamp ~= 1.5
seconds)
This means we can increase the "grace period" to the double (from 0.5 s
to 1 s), while at the same time decrease the time spent running the test
(from 10 s to 4.5 s). This should hopefully make the test less flaky,
especially in slower runners, e.g. aarch64.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1751>
With commit c985753442 the support for
multiple hardware cursors broke, but those were never properly supported
anyway as we usually assume there's only one hardware cursor around.
With the introduction of the KMS thread in the future, we'll only have
one KMS cursor that gets updated directly from the input thread. So
apart from the fact that it never really makes sense to have two cursors
visible, in this new model having multiple cursors won't work anyway.
So make the cursor we show for stylii a software cursor again.
Eventually the plan is to make the input device that's driving the KMS
cursor interchangeable, so that we can always use hardware cursors.
This reverts commit 165b7369c8.
Fixes https://gitlab.gnome.org/GNOME/mutter/-/issues/1645
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1758>
This commit assumes that cursor surfaces work in a "mailbox" fashion. If
they are painted multiple times before a successful flip, all commits
but the last get discarded, and the last commit gets presented after the
flip succeeds. This is more or less how it works in the atomic backend,
and also more or less how it works in other backends, with the exception
that the cursor painting might fail without any way of knowing. This
assumption is still better than unconditionally discarding all cursor
surface feedbacks as if the cursor painting always fails.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1484>
Regarding the sequence = 0 fallback: in some cases (moving a cursor
plane on atomic amdgpu) we get sequence = 0 in the page flip callback.
This seems like an amdgpu bug, so work around it by assuming a sequence
delta of 1 (it is equal to 1 because of the sequence != 0 check above).
Sequence can also legitimately be 0 if we're lucky during the 32-bit
overflow, in which case assuming a delta of 1 will give more or less
reasonable values on this and next presentation, after which it'll be
back to normal.
Sequence is also 0 on mode set fallback and when running nested, in
which case assuming a delta of 1 every frame is the best we can do.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1484>
The presentation-time protocol allows surfaces to get accurate
timestamps of when their contents were shown on screen.
This commit implements a stub version of the protocol which correctly
discards all presentation feedback objects (as if the surface contents
are never shown on screen). Subsequent commits will implement sending
the presented events to surfaces shown on screen.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1484>
This concerns only the cases when the presentation timestamp is received
directly from the device (from KMS or from GLX). In the majority of
cases this timestamp is already MONOTONIC. When it isn't, after this
commit, the current value of the MONOTONIC clock is sampled instead.
The alternative is to store the clock id alongside the timestamp, with
possible values of MONOTONIC, REALTIME (from KMS) and GETTIMEOFDAY (from
GLX; this might be the same as REALTIME, I'm not sure), and then
"convert" the timestamp to MONOTONIC when needed. An example of such a
conversion was done in compositor.c (removed in this commit). It would
also be needed for the presentation-time Wayland protocol. However, it
seems that the vast majority of up-to-date systems are using MONOTONIC
anyway, making this effort not justified.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1484>
KMS and GLX device timestamps have microsecond precision, and whenever
we sample the time ourselves it's not the real presentation time anyway,
so nanosecond precision for that case is unnecessary.
The presentation timestamp in ClutterFrameInfo is in microseconds, too,
so this commit makes them have the same precision.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1484>
A flag indicating whether the presentation timestamp was provided by
the display hardware (rather than sampled in user space).
It will be used for the presentation-time Wayland protocol.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1484>
A flag indicating whether the presentation timestamp was provided by the
display hardware (rather than sampled in user space).
It will be used for the presentation-time Wayland protocol.
This is definitely the case for page_flip_handler(), and I'm assuming
this is also the case for the two instances in the GLX code.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1484>
The old calculation was introduced to improve the precision
with commit c16a5ec1cf.
Here, I call the calculation as "revision 2", and the
calculation even older as "revision 1", and the new
calculation introduced with this commit as "reivion 3".
Revision 2 has two problems:
1. The calculation is mixed with fixed-point numbers and
floating-point numbers.
To overcome the precision loss of fixed-point numbers division,
it first "calculates refresh rate in milliHz first for extra
precision", but this requires converting the value back to Hz.
An extra calculation has performance and precision costs.
It is also hard to understand for programmers.
2. The calculation has a bias.
In the process, it does:
refresh += (drm_mode->vtotal / 2);
It prevents the value from being rounded to a smaller value in
a fixed-point integer arithmetics, but it only adds a small
bias (0.0005) and consumes some fraction bits for
floating point arithmetic.
Revision 3, introduced with this commit always uses
double-precision floating-point values for true precision and
to ease understanding of this code. It also removes the bias.
Another change is that it now has two internal values, numerator
and denominator. Revision 1 also calculated those two values
first, and later performed a division with them, which minimizes
the precision loss caused by divisions. This method has risks of
overflowing the two values and revision 1 caused problems due to
that, but revision 3 won't thanks to double-precision. Therefore,
revision 3 will theoretically have the result identical with
the calculation with infinite-precision.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1737>
Handle the case of a TOUCH_BEGIN event during window dragging separately
instead of treating it like a TOUCH_UPDATE event: Simply return TRUE to
make Clutter stop event propagation if it's the pointer emulating
sequence and let Clutter propagate the event if it isn't.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/427>
Window dragging should be cancelled when the touch sequences we're using
are no longer available. Also listen to TOUCH_CANCEL events if the
window is grabbed and cancel the grab op when a TOUCH_CANCEL event
happens.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/427>
This removes the responsibility of tracking these from the backend to
the base object. The backends are instead responsible for calling the
function to update the values.
For the native backend, it's important that this happens on the correct
thread, so each time either of these states may change, post a idle
callback on the main thread that sets the, at the time of queuing said
callback, up to date state. This means that things on the main thread
will always be able to get a "new enough but not too new" state when
listening on the 'notify::' signals and getting the property value
after.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1739>
Make sure to reset all the state that was set for an interactive grab op
back to the defaults after a grab op has ended.
Especially important here is setting grab_frame_action back to FALSE,
since this will constrain window-titlebars to the panel. We set this to
TRUE on some grabs, for example when resizing, but not when moving
windows. Since this remained being set to TRUE, it would also constrain
non-grab window movements, like calling MetaWindow.move_frame(), which
is used by gnome-shells OSK. By resetting it back to FALSE after a grab,
the OSK can now always move non-maximized windows to the position it
wants.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1736>
The cache had the size 9, which was "big enough" in the past, but when
more ways pipelines could be constructed, the size was not enough. The
need to increase the cache size was hard to spot though, since adding
pipeline flag didn't give any hints about the cache being directly tied
to these flag values.
So, when enough flag bits were set when attempting to retrieve and put a
pipeline in the cache, it'd instead overwrite some arbitrary stack
memory, which would sooner or later result in a memory corruption
induced crash. Valgrind could not detect this particular memory
corruption, as it messed up stack memory, not e.g. freed heap memory, so
it instead got confused and thought plain stack values were unreadable.
Fix these two issues by making the cache size the combination of all
pipeline flags + 1, so that we can safely put any flag combination in
the cache.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1747>
If `meta_xwayland_start_xserver()` returned via an error path, some of
the socket FDs were leaked.
Fix that, and add `steal_fd()` calls to make it clearer that FDs passed
to `g_subprocess_launcher_take_fd()` are transferred to it. There were
no bugs with how `GSubprocessLauncher` was being used.
Spotted while working on
https://gitlab.gnome.org/GNOME/glib/-/issues/2332.
Signed-off-by: Philip Withnall <pwithnall@endlessos.org>
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1738>
Each next and current scanout buffer has a reference on them making sure
they stay alive. When dumb buffers were used on the secondary GPU state,
this didn't happen, leading to crashes due to unref:ing one time too
many, with backtraces such as
0) g_type_check_instance_is_fundamentally_a ()
1) g_object_unref ()
2) secondary_gpu_release_dumb ()
3) import_shared_framebuffer ()
4) update_secondary_gpu_state_post_swap_buffers ()
5) meta_onscreen_native_swap_buffers_with_damage ()
6) cogl_onscreen_swap_buffers_with_damage ()
7) swap_framebuffer ()
8) clutter_stage_cogl_redraw_view_primary ()
9) clutter_stage_cogl_redraw_view ()
10) _clutter_stage_window_redraw_view ()
11) handle_frame_clock_frame ()
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1746>
gnome-shell has this hack where it sets the environment variable
"NO_AT_BRIDGE" to "1" before calling meta_init() and then unsets it
after meta_init() returns.
This variable being set to "1" will then cause the ATK bridge in
at-spi2-gtk to fail to load, which GTK then ignores. This is on purpose,
since accessibility is supposed to be done done by GNOME Shell via
Clutter, not via GTK.
The problem is that, now, by default, setting "NO_AT_BRIDGE" to
"1" during meta_init() only has the desired effect on an X11 session,
where we always connect to the X11 server on startup (i.e. during
meta_init()). With Xwayland on-demand, we do not attempt to create the
GDK display during meta_init(), thus this hack falls apart.
Since there are no real altenatives to this hack, just move it to
mutter, which have a better idea when GDK displays are created or not.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1744>
Commit afa43154 tried to make sure the focus was properly changed when
calling focus_default_window() by checking the focused window just after
trying to set the focus.
However, the X11 “Inter-Client Communication Conventions Manual” version
2.0 (ICCCM 2 for short) states that some X11 client may want to use a so
called “globally active input” model in which case the client expects
keyboard input and set input focus even when it's not one of its own
window.
To comply with this, when dealing with such clients, mutter will not
change the focus and send a WM_TAKE_FOCUS message instead.
That mechanism will defeat the logic introduced by commit afa43154
because the focused window is not changed in this case. As a result, the
input focus will fallback to the no-focus window.
To avoid this, only check that the focus change occurred for windows
using a synchronous focus model.
v2: Split specific test for "globally active input" model (Florian).
v3: Remove the check for window->unmanaging which is useless (Jonas).
Fixes: afa43154 - "core: Make sure focus_default_window() worked"
Close: https://gitlab.gnome.org/GNOME/mutter/-/issues/1620
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1716>
X11 clients can use different models of input handling, of which some
may not result focus being set synchronously.
For such clients, meta_focus_window() will not change the focus itself
but rely on the client itself to set the input focus on the desired
window.
Add a new MetaWindow API to check when dealing with such a window.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1716>
Without these devices, things that depend on the existance of input
device classes won't know about the existance of e.g. pointer devices,
if the only pointer device is from a virtual one.
This requires handling situations where e.g. a device doesn't have a
device node thus can't be matched against a udev device.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1688>
Libinput will queue a few initial events when a seat is assigned to the
udev backend; a result of it probing udev adding detected devices. For
us to see these events, we need to dispatch libinput before going idle,
as nothing will show up on the libinput file descriptor until something
else (e.g. keyboard event or mouse movement) wakes us up.
Do this by adding a prepare() function to the libinput GSource, that
checks whether there are any events in the queue already, and return
TRUE if so is the case, causing us to dispatch before going fully idle.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1688>
The API for NotifyKeyboardKeycode() does not mention what type of
keycode is expected to be submitted.
So, clarify in the API that the keycode to submit is expected to be an
evdev keycode.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1732>
Currently, when a remote desktop user submits a keycode, it will be
interpreted differently, when using the x11 session, instead of a
wayland session.
In a wayland session, submitting a keycode will have the expected
result (as if the key was pressed locally).
In a x11 session, this is not the case. Instead of getting the expected
key, some other key will be pressed (or sometimes even none).
The reason for this is that the native backend interprets the keycode
as evdev keycode and the x11 backend interprets the keycode as xkb
keycode.
To ensure that both backends produce the same behaviour when submitting
a keycode, fix the x11 backend to always interpret the keycode as evdev
keycode, instead of a xkb keycode.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1732>
GObject signals pass the emitting GObject as the first argument to
signal handler callbacks. When refactoring the grab-op-begin/end signals
to remove MetaScreen with commit 1d5e37050d,
the "screen" argument was replaced with a "display" argument instead of
being removed completely. This made us call the signal handlers with two
identical MetaDisplay arguments, which is very confusing and actually
wasn't handled in a grab-op-begin handler in gnome-shell.
So fix this by not adding the MetaDisplay as an argument to those
signals, GObject will take care of that for us.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1734>
We're going to round the workspace backgrounds in the new overview for
gnome-shell 40.
So far corner-rounding was only possible for StWidgets because the
rounded clipping was done using cairo drawing. We now need rounded
clipping for ClutterActors too because backgrounds are drawn using
ClutterActors (or more specifically a ClutterContent). To implement
that, first a ClutterOffscreenEffect subclass together with a fragment
shader from GSK (see gskSetOutputColor() [1] in the GSK GL renderer
code) was investigated, and while that was generic and worked quite
well, it was extremely slow for the case of drawing wallpapers because
of all the FBOs that had to be allocated.
This is the new, more performant approach: Use the same fragment shader,
but perform the rounded clipping right in MetaBackgroundContent while
we're painting the wallpaper. This has almost no performance impact,
with the downside of not being a generic solution.
To allow for rounded clipping not only at the edges of the wallpaper,
but using any given bounding rectangle, the API exposes not only the
radius, but also a bounding rect.
[1] https://gitlab.gnome.org/GNOME/gtk/-/blob/master/gsk/resources/glsl/preamble.fs.glsl
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1717>
Currently, the documentation for NotifyPointerAxis() just mentions that
a smooth scroll event is emitted.
However, this is not entirely correct. For each NotifyPointerAxis(),
mutter emits an emulated discrete scrolling event based on the
submitted accumulated smooth scrolling deltas.
Additionally, it doesn't mention how the motion deltas need to be
interpreted.
So, document the NotifyPointerAxis() notification better by mentioning
the emulation of discrete scroll events, how these discrete scroll
events are calculated and how the motion deltas need to be interpreted.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1727>
When a remote desktop user emits a virtual smooth scrolling event, a
smooth scroll event, that is not emulated, is emitted and on occasion
a discrete scroll event, that is emulated, is emitted.
As base for the discrete scrolling event, the smooth scrolling steps
are accumulated.
When the accumulated smooth scrolling steps surpass the
DISCRETE_SCROLL_STEP, the discrete scrolling event is emitted.
Currently, mutter uses for DISCRETE_SCROLL_STEP the value 10, which is
a terrible value to work with, especially for high resolution mouse
wheels.
When a triple resolution mouse wheel is used, each scrolling step will
have the value 3 1/3.
Three of such events won't however surpass the DISCRETE_SCROLL_STEP.
To fix this situation, add DBL_EPSILON to the calculation step, when
checking for the discrete scroll event to ensure that 3 smooth scroll
events, with each having the value 3 1/3, emit a discrete scrolling
event.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1727>
MetaVirtualInputDeviceX11 currently doesn't handle smooth scroll events
at all.
So, if a user of the remote desktop API uses smooth scroll events, then
only the wayland backend handles these events.
The user of the remote desktop API however, might not know which
backend is being used and actually the user should not even have to
care about it.
Actual smooth events cannot be emulated in the X11 events.
What can be done however is accumulating smooth events and then when
the accumulated steps surpass the DISCRETE_SCROLL_STEP value, emit a
discrete scroll event.
So, do exactly that, to make smooth scroll events work when the remote
desktop API is used with the x11 backend.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1727>
Given X11 nature, the pointer "leaves" the stage anytime it wanders into
a client window, or any other areas that are not deemed part of the
stage input region.
Yet we want to stay correct in those situations, e.g. have the clutter
side reasonably in sync, picking and highlighting to work properly, etc.
In order to achieve that, emulate motion events on XI_RawMotion. These
are as much throttled as our pointer tracking for a11y, in order to avoid
too many XIQueryPointer sync calls. This emulation only kicks in anytime
that X11 notifies us that we are not "on" the stage.
This replaces some sync_pointer() calls in GNOME Shell code that are
there just to compensate for this trait of X11, e.g. in the message tray
code.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1659>
Ensure we issue a motion event for the current pointer position,
as there might be situations where compositor modals get X grabs
from other clients stacked on top, or missed events in between
otherwise.
Ensure the Clutter state is still up-to-date afterwards here. This
replaces some sync_pointer() calls done in GNOME Shell code, always
done after modality changes.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1659>
Do these Wayland operations (that apply on both native and nested backends)
in the MetaCompositorServer subclass. We want to add more backend specific
behavior here in the future.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1659>
In the case that DnD is performed and succeeds, we want to release
the grab early, and let the transfer IPC happen in the back. For
that to happen without a hitch, drag source and offer must be left
related to each other after undoing the grab, even though the default
ungrabbing code does that automatically (indirectly, by unsetting the
drag focus).
In these cases, we used to manually unset the current source, so
this decoupling was skipped. Notably, one missed case is X11 DnD,
so we might end up with the situation there that DnD did succeed,
transfer is ongoing, but the source and offer are already decoupled,
this confused the machinery and missed the finishing XdndFinished
to be emitted to the X11 drag source.
The prior commits prepared for this source/offer decoupling being
a manual operation, this commit avoids doing this automatically
when ungrabbing.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1720>
This object is just being detached, with no code unref'ing it. Do
this whenever the XDnD selection goes unowned, usually a good
indication that the drag source no longer is one.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1720>
g_set_error_literal() asserts that the provided message is not NULL.
If it is NULL, the function is entirely no-op.
This resulted in a NULL dereference of the GError, which remained
NULL in this case, when trying to print a warning in
clutter_stage_cogl_redraw_view().
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1715>
PipeWire recently introduced busy buffers, which actually fixes the last remaining
issue that blocked us from downgrading these cogl_framebuffer_finish() calls into
cogl_framebuffer_flush() ones.
Switch to cogl_framebuffer_flush() in all three stream sources.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1701>
When a transfer request is done to the MetaSelectionSourceRemote source,
it's translated to a SelectionTransfer signal, which the remote desktop
server is supposed to respond to with SelectionWrite.
A timeout (set to 15 seconds) is added to handle too long timeouts,
which cancels the transfer request.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1552>
Nothing is hooked up, it only does basic sanity checking i.e. whether
the clipboard was enabled when interacting with it. No actual clipboard
integration is hooked up yet.
This also syncs org.gnome.Mutter.RemoteDesktop.xml from
gnome-remote-desktop.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1552>
The original implementation of ::touch-mode tested for keyboard
presence to know whether the OSK and other touch-only features were
enabled.
However that didn't pan out, every webcam, card reader and kitchen
sink like to live a second life as EV_KEY devices. This made the
detection of actual external keyboards a much harder task than it
sounds, and was thus removed in commit f8e2234ce5.
Try a different approach here, and test for pointer devices, it
doesn't matter if internal or external devices, the rationales:
- It is significantly easier to get this right, there's virtually
no devices with abs/rel axes that don't try to be a real input
device of some sorts.
- It's not as good as testing for keyboard presence, but it's the
next best thing. These usually come in pairs, except in weird
setups.
- It is better than not having anything for a number of situations:
- Non-convertible laptops with a touchscreen will get touch-mode
disabled due to touchpad presence (plus keyboard). There's
been complains about OSK triggering with those.
- Same for desktop machines with USB touchscreens, the mouse
(and presumably keyboard) attached would make touch-mode
get in the middle.
- Convertible laptops with a broken tablet-mode switch get a
chance to work on tablet modes that do disable input devices
(e.g. detachable keyboards, or via firmware)
- Kiosk machines, tablets, and other devices that have a
touchscreen but will not regularly have a mouse/keyboard
will get the touch-mode enabled.
All in all, this seems to cover more situations the way we expect it,
there's only one situation that the OSK would show where it might
not be desirable, and one that might not show when it better should:
- Tablets and kiosk machines that get one keyboard plugged, but not a
mouse, will still show the OSK, despite being able to type right
away.
- Convertible laptops with broken/unreliable tablet-mode switch (e.g.
ignored by the kernel) rely entirely on the device/firmware
characteristics to work. If after folding into tablet mode the
touchpad remains active, touch-mode will not turn on.
Fixing the tablet-mode switch on these devices should be preferred,
as that'll also make libinput magically disable the touchpad.
The latter can be worked around with the a11y toggle. The former is
merely inconvenient, and nothing prevents the user from plugging a mouse
in addition.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1710>
Technically this is still wrong if the source actor or dnd actor are
transformed in other ways. However geometry scale is the by far most
common case and we currently lack convenience API in Clutter to
easily compute the right values.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1683>
When recording the screen and real time encoding it using a gstreamer
pipeline, that pipeline can stall when the encoder is too slow. Log a
debug message using the new SCREEN_CAST debug topic in that case so we
know when framedrops are happening.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1709>
Since commit c255031b6d, we allow some modifier+scroll events to
pass through to Clutter to enable gnome-shell to handle them. That
action shouldn't trigger a modifier-only action at the same time, so
reset the corresponding tracking just like we do for modifier+click.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1695>
Since commit c255031b6d we pass scroll-events through to
the compositor if the window_grab_modifiers are pressed;
in order to allow gnome-shell to check for those events,
expose the struct member as a MetaDisplay property.
Also take the opportunity to pick a more generic name, now
that the modifier is no longer used exclusively for mouse
clicks (unless we maintain the notion of scroll events as
button 4 and 5 "clicks").
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1695>
The other end of the PipeWire stream can set the buffer data type to a
bitmask of supported buffer types. We should respect this, and not
attempt to allocate a DMA buffer if it isn't asked for.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1697>
Instead of getters, pass the width, height and stride around when
relevant. This also removes the redudant "stream_size" and
"stream_height" variables from the src struct, as they are already part
of the video format.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1697>
Mutter freezes Xwayland commits when resizing windows, and thaw them in
the window actors' after_paint() for X11.
Yet, after_paint() could be never called, as when a new window is mapped
while the overview is active in gnome-shell.
As a result, the content of the X11 window will remain invisible to the
overview.
Add a new window actor API to tell whether commits can be frozen. For
Wayland window actors, this always return FALSE, whereas for X11 window
actors, it checks whether the Clutter actor is mapped.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1678>
When (un)maximizing, (un)fullscreening, the move/resize action is
flagged with 'ACTION_MOVE' and 'ACTION_RESIZE' , while e.g.
'appears-focus' does not.
When a client misbehaved and didn't immediately reply to a configure
request with a commit with the corresponding ack_configure, the
following commit would trigger a oddly timed move, making the window
appear to move back to a previous position.
Avoid this issue by only carrying over the target window position if the
configuration actually contained a new position.
We cannot only rely on the flags however, as e.g. a new position should
be respected during interactive resize, even though only 'ACTION_RESIZE'
is passed in such scenarios.
Do the same for the size, except if the window state dictates that the
size is fixed to a certain size, e.g. being fullscreen or maximized.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1445>
Mutter needs to fetch the X11 Window ID from the onscreen and did that
by using an X11 specific API on the CoglOnscreen, where the X11 type was
"expanded" (Window -> uint32_t). Change this by introducing an interface
called CoglX11Onscreen, implemented by both the Xlib and GLX onscreen
implementations, that keeps the right type (Window), while avoiding X11
specific API for CoglOnscreen.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1514>
Instead of calling "init_onscreen()" on two different separate vtables
from the allocate() funtion, just have the CoglOnscreen sub types
themself implement allocate() and initialize in there.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1514>
Thins means that e.g. MetaOnscreenNative now inherits CoglOnscreenEgl,
which inherits CoglOnscreen which inherits CoglFramebuffer, all being
the same GObject instance.
This makes it necessary to the one creating the onscreen to know what it
wants to create. For the X11 backend, the type of renderer (Xlib EGL or
GLX) determines the type, and for the native backend, it's currently
always MetaOnscreenNative.
The "winsys" vfunc entries related to onscreens hasn't been moved yet,
that will come later.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1514>
To get meta-renderer-native.c down to a bit more managable size, and to
isolate "onscreen" functionality from other (at least partly), move out
the things related to CoglOnscreen to meta-onscreen-native.[ch].
A couple of structs are moved to a new shared header file, as
abstracting those types (e.g. (primary, secondary) render devices) will
be dealt with later.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1514>
The mutter naming convention for types and their instance variables is:
Type name:
[Namespace][BaseName][SubType]
Instance name:
[base_name]_[sub_type]
This means that e.g. CoglOnscreenGLX is renamed CoglOnscreenGlx, and
glx_onscreen is renamed onscreen_glx. This is in preparation for
GObjectification.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1514>
Makes sure that monitor specs which may be read from EDID data do not
contain characters that are invalid in XML. Makes it possible to restore
monitor configs of monitor models with characters such as '&' in them.
To make this change not break any tests, the sample monitor configs need
to be adjusted as well. Apostrophes don't strictly have to be escaped in
XML text elements. However, we now do escape the elements in
`<monitorspec>` specifically.
Closes: <https://gitlab.gnome.org/GNOME/mutter/-/issues/1011>
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1685>
Previously the wl_resource and MetaWaylandGtkSurface corresponding to
any client gtk_surface have been kept around until the exit of the
client due to the client side destroy method not signaling the
destruction to the server. Ideally the protocol would have specified a
destroy request marked as destructor to handle this automatically,
however this is no longer possible due to the destroy method being
implicitly generated in the absence of an explicit request in the
protocol. Adding a destroy request marked as destructor now would
generate a new destroy method that unconditionally would send the
request to the server, which would break clients running on servers not
supporting that request.
So instead of modifying the destroy request add a new "release"
destructor, that indicates to the server that it can release the
resource. This can be optionally be used by clients depending on the
server protocol version.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1307>
The MetaWaylandSurface corresponding to a MetaWaylandGtkSurface can be
destroyed before the MetaWaylandGtkSurface is destroyed. In its destroy
function MetaWaylandSurface however was unsetting the destructor of the
correspnding resource along with the gtk_surface1 interface
implementation. This was done to prevent further gtk_surface1 requests
on a NULLed MetaWaylandSurface, if it has been destroyed before the
MetaWaylandGtkSurface.
It would be enough to just unset the resource implementation, while
keeping the destructor to fix this leak. However the following commit
will rely on the implementation being available after the
MetaWaylandSurface has been destroyed. So instead introduce NULL checks
for all functions that can be called on the gtk_surface1 interface and
do not unset the implementation.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1307>
If the monitor configuration changed, even though the streamed monitor
didn't change, we'd still fail to continue streaming, as we failed to
update the stage watchers, meaning we wouldn't be notified about when
the stage views were painted.
Fix this by reattaching the stage watches, i.e. update the painted
signalling listeners to listen to the right views, when monitor changes
happens.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1691>
Make the API used more shared and better named.
meta_monitor_manager_on_hotplug() was renamed
meta_monitor_manager_reconfigure(), and meta_monitor_manager_reload()
was introduced to combine reading the current state and reconfiguring.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1689>
It was named "backend_native" and "backend" which is easily confused with
MetaBackendNative and MetaBackend which tends to have those names.
Prepare for introducing the usage of a MetaBackendNative and MetaBackend
pointers here by cleaning up the naming.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1689>
Fullscreen X11 windows that attempt to change the resolution on Wayland
use a surface viewport to achieve this without affecting the resolution
of the display. This however also means that pointer events will be
delivered in the display coordinates while the code handling the window
frame is not aware of any such viewport scaling. So a right click
outside of the area corresponding to the new resolution will not be
considered to be on the client area. And since the only area that is
ignored when determining whether to perform the right click action, such
as opening the context menu, is the client area, this will result in the
action being performed, despite happening on the (scaled) client area.
While it would be possible to scale the event coordinates so that
get_control() correctly determines the frame element the cursor is on,
viewport scaling only affects fullscreen windows. Since fullscreen
windows have no frame, we can always assume that if the window gets
delivered an event for a fullscreen window, it is on the client area
without doing any additional calculations.
Fixes https://gitlab.gnome.org/GNOME/mutter/-/issues/1592
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1661>
We remove pending pings when unmanaging a window, but currently
don't prevent new pings to be scheduled after that.
The previous commit fixed a code path where this did indeed happen,
but as the result of gnome-shell trying to attach a Clutter actor
to a non-existent window actor is pretty bad, also guard can_ping()
against being called for an unmanaging window.
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/2467
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1676>
This seems to have been the default in the past, but was (accidentally?) modified
by 8adab0275.
For GNOME 40, we'll be returning to our root with horizontal workspaces, so instead
of overriding it in GNOME Shell side, change the default back to what it once was.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1684>
We'll have two persistent client connections alive for the whole test,
one X11 client, and one Wayland client. So in order to be able to set up
the async waiter, do so after setting up the X11 client, as after that
we know we'll have a MetaX11Display ready to use.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1681>
One for the public channel, and one for the private maintainance
channel. Use the public one for test clients, otherwise tests become
flaky, and the private one for MetaX11Display.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1681>
This adds a MetaKmsImplDevice backend using atomic drmMode* API in constrast to
non-atomic legacy drmMode* API used in MetaKmsImplDeviceSimple.
This has various behavioral differences worth noting, compared to the
simple backend:
* We can only commit once per CRTC per page flip.
This means that we can only update the cursor plane once. If a primary
plane composition missed a dead line, we cannot commit only a cursor
update that would be presented earlier.
* Partial success is not possible with the atomic backend.
Cursor planes may fail with the simple backend. This is not the case
with the atomic backend. This will instead later be handled using API
specific to the atomic backend, that will effectively translate into
TEST_ONLY commits.
For testing and debugging purposes, the environment variable
MUTTER_DEBUG_ENABLE_ATOMIC_KMS can be set to either 1 or 0 to
force-enable or force-disable atomic mode setting. Setting it to some
other value will cause mutter to abort().
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/548
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
In order to reliably manage the reference count of the user data passed
to page flip listeners - being the stage view - make the ownership of
this data travel through the different objects that take responsibility
of the next step.
Initially this is the MetaKmsPageFlipListener that belongs to a
MetaKmsUpdate.
When a page flip is successfully queued, the ownership is transferred to
a MetaKmsPageFlipClosure that is part of a MetaKmsPageFlipData. In the
simple impl device, the MetaKmsPageFlipData is passed to
drmModePageFlip(), then returned back via the DRM event. In the future
atomic impl device, the MetaKmsPageFlipData is stored in a table, then
retrieved when DRM event are handled.
When the DRM events are handled, the page flip listener's interface
callbacks are invoked, and after that, the user data is freed using the
passed GDestroyNotify function, in the main context, the same as where
the interface callbacks were called.
When a page flip fails, the ownership is also transferred to a
MetaKmsPageFlipClosure that is part of a MetaKmsPageFlipData. This page
flip data will be passed to the main context via a callback, where it
will discard the page flip, and free the user data using the provided
GDestroyNotify.
Note that this adds back a page flip listener type flag for telling the
KMS implementation whether to actively discard a page flip via the
interface, or just free the user data. Avoiding discarding via the
interface is needed for the direct scanout case, where we immediately
need to know the result in order to fall back to the composite pipeline
if the direct scanout failed. We do in fact also need active discard via
the interface paths, e.g. in the simple impl device when we're
asynchronously retrying a page flip, so replace the ad-hoc discard paths
in meta-renderer-native.c and replace them by not asking for no-discard
page flip error handling.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Suspending might have changed the CRTC configuration, turning some off,
some on, etc. We need to update our internal representation of this
state, so that we know how to reconfigure upon resuming, e.g. what CRTCs
to turn off again.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Destroying an onscreen destroyes the gbm_surface, the gbm_bo's, and the
fb_id's. Doing this (drmModeRmFB() of the fb_id specifically), may on
some hw implicitly disable the CRTC of the plane that framebuffer was
assigned to. This would cause following atomic commit that attempts to
disable the CRTC to fail as disabling an already disabled CRTC is not
allowed.
It'd also mean we'd always disable the plane before having finished next
mode set, leaving it monitor content potentially empty when not really
necessary.
Solve this by keeping the CoglOnscreens (thus the gbm_surface, gbm_bo
and fb_id) alive until the following global mode set has completed, i.e.
the new state has been fully committed and applied.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
This makes "power save" (i.e. when you make a monitor go into power save
mode, or make it come out of power save mode), a per device action when
turning on power saving (power save being set to 'off'), and implicitly
handled when turning off power saving (power save being set to 'on')
when doing a mode set.
This is needed as with atomic mode setting, the configuration of DPMS
(Display Power Management Signaling), is replaced by directly turning on
or off CRTCs, and via the CRTC drm properties. Thus in order to handle
both with a common API, make that API high level enough for both cases
being covered.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Before we received new gamma updates via D-Bus and posted the update to
KMS directly. This won't be possible with atomic KMS, since one can only
update the state of a CRTC once per cycle.
Thus, to handle this, when configured by D-Bus, only cache the value,
and mark it as invalid. The next frame, the native renderer will pick
up the newly cached gamma value and configure the CRTCs accordingly.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
We cannot switch DPMS state to 'on' first, then mode set later, when
using atomic KMS. So when we're turning it on, just let the eventual
mode set handle DPMS too.
When switching DPMS to 'off', do it directly, synchronously, both by
setting the DPMS state and switching off CRTCs.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Before each frame is maybe redrawn, push any new cursor KMS state to the
pending update. It'll then either be posted during the next page flip,
or when the same frame finishes, in case nothing was redrawn.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
This makes it possible to post a symbolic page flip and frame callback,
meant to be used by immediate symbolic page flip reply when emulating
cursor plane changes using legacy drmMode* functions.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
In constrast to notify_presented(), notify_ready() also returns the
state machine to the idle state, but without providing new frame
information, as no frame was actually presented.
This will happen for example with the simple KMS impl backend will do a
cursor movement, which will trigger a symbolic "page flip" reply in
order to emulate atomic KMS behavior. When this happen, we should just
try to reschedule again.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Don't mode set each CRTC in separate KMS updates, as reconfiguring one
CRTC might cause other CRTCs to be implicitly reset thus as well,
causing KMS return EBUSY if using atomic modesetting.
Prepare for this by compositing each CRTC first including adding steps
to the KMS update, but wait until all views has rendered at least once
before posting the initial update. After this each CRTC is posted
separately.
Using EGLStreams instead of normal page flipping seems to fail when
doing this though, so handle that the old way for the EGLStream case,
i.e. eglSwapBuffers() -> mode set with dumb buffer -> eglStream
"acquire" (resulting in page flip under the hood).
For this we also introduce a new error code so that we don't use client
buffers when doing mode sets, which could accidentally configure the
CRTC in a way that is incompatible with the primary plane buffers.
Do the same also when we're in power save mode, to only have one special
case path for this scenario in the regular swap-buffer path.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Instead of setting the frame result in the most generic layer, have the
backends do it themselves. This is necessary to communicate that a
swap-buffer call didn't really succeed completely to present the swapped
buffer, e.g. errors from KMS.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
This argument is intended to be used by clutter to be able to
communicate with the onscreen backend, that happens to be the native
backend. It will be used to pass a ClutterFrame pointer, where the
result of page flips, mode sets etc can be communicated whenever it is
available.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
ClutterFrame aims to carry information valid during dispatching a frame.
A frame may or may not include redrawing, but will always end with a
result.
A asynchronous page flip, for example, will result in a
CLUTTER_FRAME_RESULT_PENDING_PRESENTED, while a frame that only
dispatched events etc will result in CLUTTER_FRAME_RESULT_IDLE. Instead
of this being implicit, make the ClutterStageWindow implementation
handle this itself.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
The way drm events are handled depends on whether we're using atomic or
not. Lets move the handling to the implementation, so that later the
atomic backend can handle the event they it need to.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
If we reassign e.g. a cursor plane twice before it's updated, we need to
make sure the 'fb-unchanged' flag is correctly handled, so that if we
changed the fb first, then updated the assignment again only changing
the position, the new assignment should not be flagged with
fb-unchanged.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
When we e.g. try to post an direct client buffer scanout update, it
might arbitrarily fail; when this happen we still will want to post the
rest of the update when we try again after having composited the primary
plane. To do this, add a way to preserve the metadata of an update if it
failed, only dropping the failed plane assignments. This involves
unlocking a previously locked MetaKmsUpdate, so that e.g. a new primary
plane can be assigned.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Sealing is a one way operation, but in the next commit, the "seal" will
be broken, so to avoid missusing the "seal" terminology, rename related
methods and variables to use the term "lock" instead. E.g.
meta_update_is_sealed() is now meta_update_is_locked().
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
If a modeset is pending, it's likely that the cursor update will not
work; thus, wait with updating the cursor so that it's applied together
with the mode set update.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Something might want to affect the next update that is going to be
posted, but without posting it immediately. For example, changing the
cursor might need to wait for mode setting. Make it possible to get
feedback from posting the update, in order to gracefully handle any
errors.
Note, the API for notifiying about results take out the result listener
from the update, and notifies them in an open coded for loop. The reason
for this is that in the next commit we'll sometimes reuse updates, and
we only want notify about the results once.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Page flipping shouldn't necessarily be an actively requested action, but
happen implicitly depending on the given state. Thus, change the "page
flip" update into adding listeners for page flip feedback instead.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
This will later make it possible to pass cursor plane assignments,
together with a complete update including the primary plane, but not
failing the whole update if just processing the cursor plane failed.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
If posting an update resulted in an immediate error, don't communicate
this failure using the page flip feedback callbacks, but directly as a
return value.
This makes it possible for the direct client buffer scanout path not to
pass around flags triggering this behavior, meaning we can handle such
direct scanouts better.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Instead of a "post all pending updates", pass an update specific to a
single device. This gets rid of the awkward "combine feedback" function,
and makes it possible to queue updates to a multiple devices without
always posting them together.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Custom page flips are meant to allow using e.g. EGLStream API to
indirectly trigger page flip queueing, when the KMS API cannot be used
directly. This is really something that is specific to a device, so
instead of making part of the page flip API, make it a configuration of
the update itself.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Eventually the type of impl device will depend on the driver details, so
get that information before constructing the impl device. This commit
doesn't introduce any new usage of the details, it just prepares for
the future.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
This commit consolidates DRM buffer management to the MetaDrmBuffer
types, where the base type handles the common functionality (such as
managing the framebuffer id using drmModeAdd*/RMFb()), and the sub types
their corresponding type specific behavior.
This means that drmModeAdd*/RmFB() handling is moved from meta-gpu-kms.c
to meta-drm-buffer.c; dumb buffer allocation/management from
meta-renderer-native.c.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
For now feedbacks from an update are combined, meaning we might lose
error information. The feedback API may have to be reconsidered and
redesigned when planes gets a more front seat position.
This means we need to avoid trying to post updates if we're in power
save mode, as it may be empty.
Note that this is an intermediate state during refactoring that aims to
introduce atomic mode setting support, and we'll stop combining
feedbacks completely in the future.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Instead of a constructor method, use the type directly and handle error
reporting using GInitable.
The DRM capability setting is done before construction, as later it'll
determine what type of impl device should be constructed.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Instead of telling MetaKmsConnector fill a MetaKmsUpdate with connector
property changes, make the update itself aware of the changes, making
the impl side translate that to property changes.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Instead of having MetaKmsPlaneAssignment carry a low level property
list, set the actual state change, and then have the implementation
translate that into the necessary property changes.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
With the aim at always using the property table to fetch and parse
property metadata, move IN_FORMATS handling to the property table, using
the newly introduced parse vfunc.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Values may need to be processed and parsed in custom ways; make this
possible via the property table infrastructure using a callback.
Will be used for e.g. parsing rotation and formats.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Instead of relatively verbosely going through the DRM properties finding
the properties we care about and saving their ID's, add a more
declarative way to fetch property metadata. This'll allow for fetching
more property IDs with relatively less code, which will be useful for
the atomic backend.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
This contains a copy of a drmModeModeInfo, describing a mode. It also
has an unused pointer to the impl device it is associated with. It'll
later be used to get a blob ID for the mode.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1488>
Current Xwayland has marked the command line option "-listen" as
deprecated in favor of "-listenfd".
Use the pkg-config variable "have_listenfd" (if available) from Xwayland
to determine if we should use that option, to avoid a deprecation
warning when spawning Xwayland.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1682>
Mutter listens to two display connections, one for regular X11 clients
and another one for the so called "managed services".
Once an available display number is found for the regular X11 clients,
mutter would then redo the work to find another available display number
for the managed services.
Yet, it does so starting from the same initial display, which is a waste
of time since it just tried all displays to find the first available
one, so all these, including the regular display it just took, are now
in use.
So instead of starting over from the beginning when looking for a
display available for the managed services, continue from the next
display immediately after the one we found precedently.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1680>
Some X11 servers may not always create a lock file, yet mutter uses the
lock file to find a possible display number and then tries to bind to
the socket corresponding to that display number.
If it fails to bind, it will simply bail out. As a result, if an X11
server is already listening on that display but hadn't created a lock
file, mutter won't be able to start Xwayland.
To avoid that possible issue, make mutter retry with another display
for a given number of tries when binding fails even though the display
was supposed to be available based on the lock file presence.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1604
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1669>
The function choose_xdisplay() calls open_display_sockets() which calls
ensure_x11_unix_dir().
We don't need to do that from within the loop though, as the directory
/tmp/.X11-unix is the same regardless of the display number.
Move the call to ensure_x11_unix_dir() from open_display_sockets() to
choose_xdisplay() prior to enter the display loop.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1669>
In case of failure to open the display sockets, we would not propagatre
the error which can cause a crash when trying to show the error message.
Properly propagate the error to avoid the crash.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1669>
In the shutdown paths we check with the X11 display whether there's
remaining clients. However this happens in paths that happen after
the MetaX11Display vanished in the case of Xwayland crash.
Since in that situation the clients are forcibly vanishing too,
skip the client check.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1677>
This means backend implementations can have more control of the order of
how things are destroyed. To be precise, this will, in the next commit,
allow us to destroy the logind integration after the clutter backend
thus the libinput owning seat, that uses the logind integration to
release input devices.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1670>
We can't post tasks to the input thread when cleaning up the
MetaSeatImpl, as that will make the GTask complain about adding
references to a to be purged object. Avoid this by adding an explicit
meta_seat_impl_destroy() function that handles the destruction of the
MetaSeatImpl properly.
This also does more of the cleanup in the input thread, as that is where
it was managed. Will likely not make a difference as before this
happened after tearing down the thread, but lets tear down things in the
thread they were managed for good measure.
This fixes the last log spew I see right now when terminating mutter.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1670>
Closing automatically Xwayland once all relevant X11 clients are gone is
inherently racy, if a new client comes along right at the time we're
killing Xwayland.
Fixing the possible race conditions between mutter, Xwayland and the X11
clients may take some time.
Meanwhile, make that an experimental feature "autoclose-xwayland".
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1673>
<super> is considered a system modifier, and applications cannot use it
for keyboard shortcuts or as button modifier. It doesn't seem too much
of a loss taking <super>+scroll-event away as well, so that it becomes
available to gnome-shell/extensions.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1674>
Currently, mutter checks for the presence of X11 windows to decide
whether or not Xwayland can be terminated, when Xwayland is started on
demand.
Unfortunately, not all X11 clients will map a window all the time, an
X11 client may keep the X11 connection opened after closing all its
windows. In that case, we may terminate Xwayland while there are some
X11 client connected still, and terminating Xwayland will also kill
those X11 clients.
To avoid that issue, check the X11 clients actually connected using the
XRes extension. The XRes extension provides the PID of the (local) X11
clients connected to the Xserver, so we need to match that against the
actual executable names, and compare with a list of known executables
that we can safely ignore, such as ibus-x11 or gsd-xsettings.
We also check against our own executable name, considering that the X11
window manager is also an X11 client connected to the Xserver.
Also, XRes returning the PID of local clients only is not a problem
considering that Xwayland does not listen to remote connections.
However, if the user spawns a client remotely on another system using
ssh tunneling (ssh -X), only clients which actually map a window will
be accounted for.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1537
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1671>
Changes in games between fullscreen and windowed modes may trigger
chaotic situations where the buffer and the frame size temporarily
disagree, producing rectangles with negative width/height. This is
usually followed by other updates that bring the pointer constraint
up to date.
This makes cairo panic and return an "error" empty region, which breaks
deeper down when using the region rectangles to apply the pointer
constraint.
If we hit this situation, ignore the frame rectangle, and just go with
the buffer rectangle.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1655>
We didn't tear down the libinput objects in the right thread when
exiting, but did so after the input thread exited.
We also tried to destroy the libinput devices after the libinput context
was destroyed, which isn't allowed.
Fix these two issues by tearing down the libinput objects in a input
thread task that when done exits the input thread. This effectively
"flushes" the input thread tasks while destroying the libinput objects
just before the thread exits.
While it might fine to tear down libinput objects in an arbitrary (main
in this case) thread while making sure nothing pokes at it in parallel
(e.g. the input thread is gone), libinput is by definition single
threaded, and could theoretically make assumptions about this, and we
shouldn't cause any possible surprises here, so make sure to destroy it
all in the right thread.
This fixes an abort() on exit caused by an assert about invalid object
destruction in libinput.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1663>
The "seat" usually refers to the ClutterSeat (MetaSeatNative) object,
and "seat_impl" to the MetaSeatImpl object, but there were still a few
places where this wasn't adhered to so fix those.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1663>
When supported, this property allows the window system to apply a 3x3 color
correction matrix in order to transform colors from the window system's native
color space to the measured color space of a display device.
Query for this property and set the 'supports-color-transform' property in the
GetResource reply. Add support for the SetOutputCTM DBus method and plumb that
through to the server's CTM property.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1048>
Add a test case to check whether the stage views and frame clocks are
all correctly freed after working with timelines a bit and then
hotplugging and removing all monitors.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1632>
Currently, the NotifyPointerAxis method always assumes that the scroll
source is CLUTTER_SCROLL_SOURCE_FINGER.
This is however not always true and in some cases a remote desktop
client needs to submit a PointerAxis event with a custom axis step.
This is for example the case with high resolution mouse wheels, where
the NotifyPointerAxisDiscrete method is unsuitable.
In such cases NotifyPointerAxis needs to be called, but with the
intention that the scroll source is still a mouse wheel.
To solve this situation, don't assume the scroll source always to be
CLUTTER_SCROLL_SOURCE_FINGER.
Instead, add further flag options to NotifyPointerAxis, which allow a
remote desktop client to choose the scroll source.
This way a remote desktop client can choose what scroll source is the
most suitable one for the current scroll event.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1636>
The monitor texture is the final background image. It doesn't need to
have any alpha channel. Cross-fades (which is the process of rendering
*into* the monitor texture) still work just fine.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1665>
Unfortunately there's situations where we can't fully rely on only
crossing events here. One such situation is pointer visibility changes
due to touch interaction, or e.g. after closing the lid.
In these situations the pointer position stays the same, picks the
same actor, yet we want to see the right surface as the pointer focus
again in the wayland side.
This used to happen on the first motion event after the pointer
visibility change before commit ad3f2b0b86, use motion events again
for picking so we don't break these usecases.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1657>
These events may be emitted for touchpoints (in which case they contain
an event sequence). Ignore those as they are not relevant for pointer
picking, and shouldn't influence its focus.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1657>
This signal may be left dangling when disconnecting a device, and be executed
later on if the device is connected again, and mapped to other output. Make it
sure the signal handler is disconnected when unplugging the device.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1657>
Just because of implementation details, this is only relevant to Wayland,
and is done via ::effects-completed handlers there. Ideally, Clutter should
notice by itself about effects starting, finishing, and affecting picking.
Doing this in generic code seems slightly cleaner in the interim.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1654>
In the wayland code we are in synchronization with the events that
the ClutterStage is managing at the moment. Asking the ClutterSeat for
the pointer position gets ahead of the current events, and may result
in imprecise coordinates sent in wl_pointer.enter.
To be in consistence with the motion events that might be already
queued, we should ask the stage for the last known coordinates.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1654>
To clear a pointer constraint, the Wayland backend passes a NULL
constraint to the native input backend.
The new async API however tries to reference/un-reference the given
object to use it while running in a separate task, which leads to a
warning from GLib trying to g_object_ref()/g_object_unref() a non
GObject pointer.
To avoid that issue, simply set the data only if the given constraints
pointer is not NULL.
Suggested-by: Carlos Garnacho <carlosg@gnome.org>
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1587
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1652>
find_focusable_ancestor() may pick an ancestor window which is not
mapped or hidden, and setting focus on that window will fail.
Be a tad more selective when looking for a focusable ancestor, to reduce
the chance of meta_window_focus() not focusing the happy chosen one.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1643>
The function focus_default_window() optionally takes a MetaWindow
argument denoting a window that should not be focused.
That function calls focus_ancestor_or_top_window() which in turn
calls meta_window_focus() to pass focus to another window.
However meta_window_focus() gives no guarantee that the given window
will end up being the one focused, and can fail in various and creative
ways.
If that fails, we could possibly end up with the focus window being the
one to avoid, while the caller assumes focus was changed, going as far
as asserting that fact like meta_window_unmanage() does.
As a result, mutter may abort simply because meta_window_focus() failed
to set focus on the expected window.
To avoid that issue, check that the focus did not end up on the window
that we explicitly did not want, and if that's the case, simply fallback
to the default focus window.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/862
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1643>
As planned and prepared with the last commits, let ClutterStage take
care of tracking input devices and their respective actors. This means
we now can remove the old infrastructure for this from
ClutterInputDevice.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1633>
Power saving changes in laptop panels enable/disable the attached
touchscreen input device, this is an asynchronous operation that
may be happening while the device is disappearing.
In fact, closing the lid is such perfect storm where both things
happen around the same time.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1644>
Currently, the MetaInputDeviceNative owns the libinput_device, with the
small catch that it is eventually finished in the main thread (as the
CLUTTER_DEVICE_REMOVED event keeps the last reference to it).
Make it sure that the libinput_device is destroyed in the input thread,
before giving away the last extra input device references.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1639>
There was an attempt to remove an unnecessary inclusion of a header
file, but only got so far as compile testing after having commented out,
but didn't remove the comment before creating a commit. This commit
fixes that mistake.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1640>
Add a new ClutterPaintNode parameter to the paint_target() vfunc.
For now, create a temporary ClutterEffectNode that is passed to
paint_target() and immediately painted; next commits will move
this to upper layers.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1355>
To find XWayland output that should be the primary one, iterate through all
XWayland outputs, and compare their geometry to the geometry of the primary
logical monitor.
To avoid possible race conditions (Mutter's monitor configuration already
updated, but Xrandr not yet), set the output both after Randr notifications and
after 'monitors-changed' signal.
https://gitlab.gnome.org/GNOME/mutter/-/issues/1407
Signed-off-by: Aleksandr Mezin <mezin.alexander@gmail.com>
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1558>
When we're running under a polyinstantiated SELinux environment, we'll
likely start with an isolated and empty /tmp, meannig no /tmp/.X11-unix
directory to add things to. To make it possible to still function in
this kind of setup, make sure said directory exists.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1626>
Instead of using native backend platform data specifically, store
this info in ClutterMotionEvent. This includes time in usec since
it's just used for motion events, in the future it could make sense
to make these general to all events again, but it could make sense
to make ClutterEvent structs private before.
In order to express that a motion event has relative motion info,
the CLUTTER_EVENT_FLAG_RELATIVE_MOTION event flag has been added
for it.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1623>
We had code in both backends that sort of independently associated
sequences to slots. Make both transform slots to sequences the same
way, so they may share the implementation convert those back to slots.
This helper now lives in Clutter API.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1623>
We have this as platform-dependent data in the native backend, and
a bunch of fallback code done in place in the evcode users. Stop
making this platform-dependent data, and move it to the relevant
ClutterEvents.
The fallback code for the X11 backend case is about the same, but
now it is done directly by the X11 backend.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1623>
Making this an event is overly convoluted, accounting that we
emit the event, then convert it to a ClutterStage signal, then
its only consumer (a11y) sets the active ATK state.
Take the event out of the equation, unify activation/deactivation
of the stage in MetaStage, and use it from the X11 backend too.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1623>
Ensure that color_ptr gets set, and avoid color_char usage too in
that case. Fixes:
../../../../Source/gnome/mutter/src/backends/native/meta-monitor-manager-kms.c: In function ‘meta_monitor_manager_kms_set_crtc_gamma’:
../../../../Source/gnome/mutter/src/backends/native/meta-monitor-manager-kms.c:370:7: warning: ‘color_char’ may be used uninitialized in this function [-Wmaybe-uninitialized]
370 | g_string_append_printf (string, " %c: ", color_char);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../../../../Source/gnome/mutter/src/backends/native/meta-monitor-manager-kms.c:351:12: note: ‘color_char’ was declared here
351 | char color_char;
| ^~~~~~~~~~
../../../../Source/gnome/mutter/src/backends/native/meta-monitor-manager-kms.c:391:36: warning: ‘color_ptr’ may be used uninitialized in this function [-Wmaybe-uninitialized]
391 | (*color_ptr)[i]);
| ~^~~~~~~~~~~
../../../../Source/gnome/mutter/src/backends/native/meta-monitor-manager-kms.c:350:24: note: ‘color_ptr’ was declared here
350 | unsigned short **color_ptr;
| ^~~~~~~~~
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1624>
Assert that the region is created, thus we passed a valid enum value
to the get_scaled_region() function. Fixes:
../../../../Source/gnome/mutter/src/compositor/meta-surface-actor.c: In function ‘get_scaled_region’:
../../../../Source/gnome/mutter/src/compositor/meta-surface-actor.c:113:10: warning: ‘scaled_region’ may be used uninitialized in this function [-Wmaybe-uninitialized]
113 | return scaled_region;
| ^~~~~~~~~~~~~
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1624>
It was a test case in the Wayland test client directory, but it wasn't a
Wayland test client but a standalone test linking to libmutter. Since it
uses rlimit to implement certain aspects of the test, it can't be made
part of the regular unit tests, as that means any test running after
being stuck with the rlimit set, thus keep it standalone, but at least
run it as part of the test suite.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1557>
The XIOErrorExitHandler expects (Display *, void *) whereas mutter uses
(Display *, MetaX11Display *).
That causes a warning at build time:
warning: passing argument 2 of ‘XSetIOErrorExitHandler’ from
incompatible pointer type [-Wincompatible-pointer-types]
813 | XSetIOErrorExitHandler (xdisplay, x_io_error_exit, display);
Actually, the MetaX11Display is not even used, so we can just use the
expected API and ignore the value.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1621>
The stack and stack tracker tend to cause missed frames from time to
time, especially when there are many open windows. Add some
instrumentation to make it this easily verifiable when profiling.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1616>
While browsing sysprof profiling reports, I saw surface-commit taking
significant times sometimes; trace attach too, to see whether such
things are due to e.g. texture uploads.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1616>
Constantly manipulating the stack caused severe stalls (several seconds)
with many open windows when switching workspaces. The cause for this was
that each show/hide call dealt with the stack in isolation, meaning if
you hid N windows, we'd manipulate and synchronize the stack N times,
potentially doing synchronous calls to the X server while doing so.
Avoid the most severe stalls by freezing the stack while calculating
showing; this made the worst case go from several seconds to around
10-20 ms, which is still bad, but by far not as bad.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1616>
Figuring out the MetaSeatImpl this much indirectly is fairly awkward when
the keymap is only updated from the MetaSeatImpl, pass instead the seat
impl's xkb_state, as we have it handy in all the places this is called.
This will not break on NULL seats during initialization, should the numlock
state be restored from previous sessions.
Fixes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1556
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1612>
The API allows for invalid barriers to be created; in an X11 session,
this could result in involutary early exit, so guard against those with
soft asserts. These will be logged in the journal as warnings, but will
avoid the crash unless compiled out.
Note that this doesn't fix the bug, it just makes it more detectable.
Related: https://bugzilla.redhat.com/show_bug.cgi?id=1901610
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1611>
Make it impossible to add individual includes of input thread objects.
This must go through meta-input-thread.h now, which should be enough
to make anyone think it twice.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
They're a dime a dozen. If it gets called exclusively from the
input thread, it got one. Hopefully these breadcrumbs will be
enough so people don't lose their way here.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
This (now) doesn't change anything in regards to the API that the UI
thread should access from the MetaSeatImpl. The MetaInputDeviceNative,
MetaInputSettings and MetaKeymap objects are now considered owned by
the input thread, as well as all of libinput objects.
The MetaEventSource now dispatches events in a GMainContext that is
the thread default to this thread, and all UI-thread-accessible API
(seat and virtual input device API) will be handled in a serialized
manner by that same input thread.
The MetaSeatImpl itself is still considered to be owned by the caller
thread, and all the signals that this object emits will be emitted in
the GMainContext that is default at the time of calling
meta_seat_impl_new().
The MetaInputSettings configuration changes will likewise be handled
in the input thread, close to libinput devices.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Instead of going through the event queue, stage handling code, and
back to the input device via a vmethod call, do this directly in the
MetaSeatImpl. This is not too different from X11, where everything
happens inside the backend.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
All that is left in the "public" struct is all state that ClutterStage
delegates on ClutterInputDevice. That should move somewhere else, but
not here, not now.
All private fields belong to construct-only properties, with only getter
API, and idempotent vmethods (except keyboard a11y, atm). This should
be enough to make ClutterInputDevice obviously thread safe, outside the
backend.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
This API is the one accessed from different bits of the UI thread,
make it "async" (it's basically one-way setters, so API stays the same
in the surface) and able to run in the MetaSeatImpl main context.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Simplify the handling of numlock state, so it can be entirely handled
within the input thread. Since the saving/restoring is triggered inside
each backend code, there's no need anymore for meta_backend_set_numlock().
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Push it a little bit down to the MetaSeatNative. As both the UI thread
and the input thread are interested in dealing with the xkb_keymap and
it is not meant to be used in different threads, keep 2 separate copies
around.
The keyboard map will always be set from the UI thread, so the xkb_keymap
owned by the MetaSeatNative (owned by the UI thread) can be considered
canonical.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Don't let the vfuncs (meant to be called from the UI thread) deal with
xkb state itself. Instead store the current state in struct fields, which
is then fetched in vfuncs.
This makes the keymap able to be used from the UI thread, while being
maintained by the input thread. Same caveats apply than
clutter_seat_query_state(), you are asking for the most up-to-date state,
but it still may be changing under your feet.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Wrap all keyboard state updates, and all pointer/stylus/touch cursor
position with a write lock, and ::query_state() (The only entrypoint
to this state from other threads) with a read lock.
The principle is that query_state may be called from different threads
(UI so far, but maybe KMS too in the future), while the input thread
may (or may not) be updating it. This state is fetched "atomically"
(eg. x/y will be consistently old or new, if the input thread were
updating it at the same time).
There's other places deep in backends/native that read this state,
they all will run in the input thread, so they count as "other readers"
to the other thread. Those changes are already mutually exclusive with
updates, so they don't explicitly need the RW lock.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
While barriers will be added from the main thread, the native barrier
manager will sit close to the MetaSeatImpl in its own thread. Add the
necessary locking so that we can pass MetaBarrierImplNative from the
UI thread to the input thread, and ensure the MetaBarrier signals are
still emitted in the UI thread.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Depending on the backend, we want to integrate this object at different
levels. It will sit close to the MetaBackendX11/MetaSeatX11 in X11, but
it will be put deep down with MetaSeatImpl in the native backend, in a
separate thread.
Since we can't depend on a single object type, nor are able to track
ClutterSeat signals neatly, make this API something to be called
explicitly by backends.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
These changes will happen in the input event management code, so let them
be emitted via the MetaSeatImpl, as that's what we'll have neat access to.
The ClutterSeat signals are now emitted from there.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Move most of the functional bits (those meant to run on a standalone
thread) to a MetaSeatImpl object. This object is managed by the MetaSeatImpl
and not exposed outside the friend MetaSeatNative/MetaInputDeviceNative/
MetaInputSettings classes.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Banish MetaInputSettings from MetaBackend "public" API, it's now meant to
spend the rest of its days in the backend dungeons, maybe hanging
off a thread.
MetaInputMapper replaces all external uses.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Delegate on the MetaInputMapper all matching of inputs and outputs,
including configuration. Other secondary aspects, like output
aspect ratio for tablet letterbox mode, or toggling attached devices
with power saving changes, are also moved here.
This makes MetaInputSettings independent of MetaMonitorManager,
all interaction with it happens indirectly via public API entrypoints.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Rename the set_tablet_keep_aspect() vfunc into a set_tablet_aspect_ratio()
one that takes an aspect ratio double, instead of leaking monitor info
into subclasses to let them all figure out this number themselves.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
We have 2 sources (this one in MetaSeatNative, and the one in
MetaBackend) dispatching ClutterEvents to the stage. Make the
MetaSeatNative one exclusively about dispatching the libinput
queue, and leave ClutterEvents to the other.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
This will resort to SW rendering if this cursor renderer does not
own the MetaKmsCursorRenderer, so it's pretty much equivalent thus
far, except we may now implement logic to flip the kms cursor renderer
around.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
We actually have a set_send_events() vfunc that can enable or disable
devices at the libinput and X11 input driver level, so use that. A
positive side effect is that those layers will leave the device at
a consistent idle state (as opposed to going mute maybe amid user
input).
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
We are aiming for a split of HW and SW cursor rendering management.
Given the HW plane is a limited resource and the amount of cursor
renderers may be >1 (due to tablets, even though we currently use an
always-software cursor renderer there), it would ideally be able to
switch between renderers.
Being MetaCursorRenderer not really a singleton, having cursor
inhibitor accounting here doesn't pan out. Make it MetaBackend API
so all cursor renderers get the same picture.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
These use now more of a "pull" model, where they receive update
notifications and the relevant input position is queried, instead
of the coordinates being passed along.
This allows to treat cursor renderers all the same independently
of the device they track. This notifying of position changes should
ideally be more backend-y than core-y, a better location will be
figured out in future commits.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
Instead of letting the wayland bits maintain an always-software
cursor renderer, let the cursor renderer be managed by the backend,
and only hook to it (as we do for pointer cursor) in the wayland
bits.
ATM, make the cursor renderer still always-software, although
ideally we should allow moving the HW cursor management between
renderers.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1403>
