This replaces the v1 implementation, which is now renamed to
legacy-xdg-foreign. Both implementations use the same data structures
internally, so that protocol version mismatches between
the importer client and exporter client don't fail.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2770>
Things like meta_compositor_destroy() and meta_compositor_add_window()
isn't intended to be used externally, and if they was, things would
probably fall apart rather quickly.
MetaCompositor also isn't introspected, meaning things that technically
belong to the compositing parts isn't easily available via some object,
but much take detours via other objects like MetaDisplay.
So move the API intended for internal usage to compositor-private.h, and
leave API that is meant to be expose in the public compositor.h.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
The "later" API is used to queue actions in relation to compositing,
thus is owned by the MetaCompositor instance. Make users of this
functionality get MetaLaters instance from the compositor, and stop
using the global meta_later() API.
display: Use non-singleton MetaLater API
tests: Use non-singleton MetaLater API
meta/common: Make docs refer to context aware MetaLater API
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
This means we can eliminate the use of scattered singletons that isn't
added by the tests or the test framework itself.
tests: Don't get backend from old singleton getter
Either use the ownership chain, or the explicit test context instance
pointer.
tests/wayland: Pass context to test client constructor
So that we can get the Wayland compositor directly from the context.
tests: Don't get display from singleton
tests/client: Make test client carry a context pointer
tests/runner: Have test cases carry a context pointer
tests/wayland/test-driver: Get backend from context
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
Instead of passing around state using GINT_TO_POINTER() pass around a
state struct that also carries a pointer to the context. This allows
avoiding using old singletons for getting a window list.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
On the path towards clear ownership chains and always using them to find
other components, do the same for X11 client support paths too.
x11-display: Don't get backend from signleton
x11/selection: Don't get display from singleton
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
The API has no concept of user data, and requires the user to some how
get an instance without context, i.e. via static globals. Limit this to
the file where this is needed.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
As with the backend commit, this means all objects can reach the
MetaContext by walking up the chain, thus can e.g. get the backend from
the context, instead of the global singleton.
This also is a squashed commit containing:
compositor: Get backend via the context
The MetaCompositor instance is owned by MetaDisplay, which is owned by
MetaContext. Get the backend via that chain of ownership.
dnd: Don't get backend from singleton
window-actor: Don't get backend from singleton
dnd: Don't get Wayland compositor via singleton
background: Don't get the monitor manager from the singleton
plugins: Don't get backend from singleton
This applies to MetaPlugin, it's manager class, and the default plugin.
feedback-actor: Pass a compositor pointer when constructing
This allows getting to the display.
later: Keep a pointer to the manager object
This allows using the non-singleton API in idle callbacks.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
As elsewhere, make sure objects that need to have a ownership up to the
context, and use this ownership chain to find relevant components, such
as the backend or the Wayland compositor object instance.
wayland/data-device: Hook up data devices to seats
They are tied to a seat - make that connection in struct fields too, so
that related objects can get to the context via it.
wayland: Don't get Wayland compositor via singleton getter
This means via the ownership chain or equivalent.
xwayland: Hook up manager to Wayland compositor
Same applies to the drag-n-drop struct.
xwayland: Make X11 event handling compositor instance aware
This avoids finding it via singletons in the callee.
xwayland: Don't get Wayland compositor from singleton
xwayland: Pass manager when handling dnd event
window/xwayland: Don't get Wayland compositor from singleton
xwayland/grab-keyboard: Don't get backend from singleton
xwayland: Don't get backend from singleton
wayland: Always get the backend from the context
This means traveling up the ownership chain or equivalent when
necessary.
wayland: Hook up data devices, offers and sources to the compositor
This allows tying them to a context without going through any
singletons.
wayland: Don't get display from singleton
xwayland: Don't get display from singleton
tablet: Don't get display from singleton
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
As with other parts, make objects have the ability to walk up the
ownership chain to the context, to get things like the Wayland
compositor or backend instances.
Contains these squashed commits:
display: Don't get backend from singleton
window: Don't get backend from singleton
keybindings: Don't get backend from singleton
workspace: Don't get backend from singleton
display: Don't get Wayland compositor from singleton
selection: Add display getter
context/main: Get backend directly from the context
clipboard-manager: Don't get display from singleton
stack-tracker: Don't use singleton MetaLater API
startup-notification: Hook up sequences and activations to display
This allows using context aware API directly.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
This means objects have an owner, where the chain eventually always
leads to a MetaContext. This also means that all objects can find their
way to other object instances via the chain, instead of scattered global
singletons.
This is a squashed commit originally containing the following:
cursor-tracker: Don't get backend from singleton
idle-manager: Don't get backend from singleton
input-device: Pass pointer to backend during construction
The backend is needed during construction to get the wacom database.
input-mapper: Pass backend when constructing
monitor: Don't get backend from singleton
monitor-manager: Get backend directly from monitor manager
remote: Get backend from manager class
For the remote desktop and screen cast implementations, replace getting
the backend from singletons with getting it via the manager classes.
launcher: Pass backend during construction
device-pool: Pass backend during construction
Instead of passing the (maybe null) launcher, pass the backend, and get
the launcher from there. That way we always have a way to some known
context from the device pool.
drm-buffer/gbm: Get backend via device pool
cursor-renderer: Get backend directly from renderer
input-device: Get backend getter
input-settings: Add backend construct property and getter
input-settings/x11: Don't get backend from singleton
renderer: Get backend from renderer itself
seat-impl: Add backend getter
seat/native: Get backend from instance struct
stage-impl: Get backend from stage impl itself
x11/xkb-a11y: Don't get backend from singleton
backend/x11/nested: Don't get Wayland compositor from singleton
crtc: Add backend property
Adding a link to the GPU isn't enough; the virtual CRTCs of virtual
monitors doesn't have one.
cursor-tracker: Don't get display from singleton
remote: Don't get display from singleton
seat: Don't get display from singleton
backend/x11: Don't get display from singleton
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
While already cleaning up API, if this should ever be more non-static
than a constant, it's better if its a function on the monitor manager
instance than something static.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2718>
Since the Wacom panel rewrite, the "output" setting is handled as
a kind of tri-state for display-integrated tablets:
- If the setting is unset, the device is automatically mapped
to an output
- If the setting is set and not empty, the device is mapped to
the output defined by the EDID data
- If the setting is ['', '', ''], the device is mapped to the
span of all displays, like opaque tablets do.
This distinction for the unset setting fell through the cracks,
so both "Automatic" and "All displays" options were handled as
the former.
Add this distinction, so that display-integrated tablets can
be used like opaque tablets of sorts with no limitations.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2767>
These are the ones attached to a display, thus they are the ones that may need
help from this heuristic. Non-integrated tablets (e.g. Intuos) will default to
the span of all monitors.
Fixes mapping of opaque tablets if a display-integrated tablet of the same
brand is also plugged in.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2767>
This might happen when the workspace is not switched and
focus_default_window is called or when 'workspace on primary display
only' is enabled, a secondary display exists and the workspace is
switched.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2747>
There's 2 users of this, meta_display_sync_wayland_input_focus() which
does already perform these checks on its own, and MetaCursorTracker's
update_displayed_cursor() to determine whether it should go with the
Wayland client's cursor.
This second check should also consider the existing ClutterGrabs, so
make meta_display_windows_are_interactable() handle them for both
callers.
Fixes the cursor shown over windows while e.g. there are menus opened.
Close: https://gitlab.gnome.org/GNOME/mutter/-/issues/2553
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2754>
This code path is important for "empty" commits to ensure we schedule
frame callbacks even if previous commits didn't cause stage redraws.
There is, however, no reason to schedule updates on all stage views
instead of only those the actor is on.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2671>
Avoid some allocations, save some CPU cycles and make the code easier
to read.
Behaviourwise the only expected change is that now, if there are mapped
clones, we unconditionally choose the view with the highest refresh
rate the actor (or one of its clones) is on and don't check the
obscurred region any more.
Thus in some cases a client may receive a higher rate of frame callbacks
when obscured on a faster view while a clone is present on a slower
one. The assumption is that cases like this are relatively rare and
that the reduction of code complexity, the reduction of allocations in
`meta_surface_actor_is_obscured_on_stage_view()` whenever the actor is
not fully obscured and has clones on other views, as well as generally
fewer lookups and less code in most common cases, compensate for that.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2671>
We only support feedback-actors, such as DnD-icons, in the compositing
path at the moment.
The approach is similar to how we handle certain shell elements.
Implementations need to ensure no references to the object keep
around longer that necessary.
Arguably this should be replaced by a more robust and implicit actor
hierachy detection in the direct scanout code at some point.
Closes https://gitlab.gnome.org/GNOME/mutter/-/issues/2470
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2677>
The fields of 'priv->video_format.max_framerate' are all of type
uint32_t. Multiplying by G_USEC_PER_SEC can overflow, and equally,
dividing a large numerical type by uint32_t can err too.
Since the variable holding the result is int64_t, cast all uint32_t
fields to int64_t before doing any maths on it.
Spotted while trying to investigating an issue with framerates on
HDMI screencasts.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2762>
Add a helper function that ensures any queued virtual input events have
been flushed from the input thread. This works by posting a task to the
input thread, which will itself queue another callback back to the main
thread. Once the main thread callback is invoked, the flush call is
unblocked and the function returns. Upon this, any previously emitted
virtual input event should have already passed through the input thread
back into the main thread, however not necessarily fully processed.
For making sure it has been processed, one also have to make sure the
stage has been updated, e.g. via `meta_wait_for_paint()`.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2727>
After the commit "wayland/subsurface: Implement
meta_wayland_surface_get_window()" subsurfaces are supported. Adjust
some comments and fix a warning that could occur when closing a window.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2717>
The intention here was to check if the subsurface belongs to a window.
Thus it didn't behave as expected for subsurfaces belonging to non-toplevel
windows.
After the previous commit we can use `get_window()` to check for what we
actually want here.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2717>
Subsurfaces are special regarding windows as they don't have a window,
but usually have an ancestor which does. All current users of
`get_window()` are either used for known surface roles, such as xdg-*
ones, or, as is the case for pointer constrains, would actually want to
get the ancestors window.
Thus implement `get_window()` to allow pointer constrains to work.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2223
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2717>
The meta_prop_get_motif_hints() function was only used in the
old MetaUI frames code. The remaining code in mutter accesses
directly the MetaPropValue when loading properties for a window,
and does not use this API call.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2741>
Since we use XCB in the Mutter side, but Xlib in the frames client,
we cannot share the same struct definition since both libraries
will expect different type lengths (respectively, 32-bit ints vs.
longs).
Revert the changes that made both executables share the same
struct, since not both of them can get it right (and retrieve
correctly the struct with the contained flags) in reading the
Motif WM hints.
This reverts commit 2fb3c5a4f5.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2741>
Adding the 'default-decoration' CSS class to MetaFrameHeader after
it is set as the headerbar makes it not account for the minimum size
correctly sometimes. This is a bit racy though - if the window opens
very quickly, it works as expected.
Adding the CSS class before the widget is used guarantees it'll
always report the correct size though, so do that.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2753>
Add this CSS class both to the header bar itself, since it is what
actually contains the window controls, and to MetaFrameHeader too,
since it's what's directly attached to the window.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2740>
Previous commit added support for setting the GTK4 theme setting
according to the color scheme setting. That's cool. What it didn't
add, though, was initializing the GTK4 theme setting to the proper
value. That means if the desktop starts at dark style, you'd still
get a light titlebar.
Fix that by updating the GTK4 theme setting on init as well.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2740>
These are now referenced on the frames client side (in order to
track deletable state from the client window) and the mutter side
(pretty much everything else, like figuring out if a window wants
WM decorations).
It makes sense to make this a separate header, so that we don't
need to doubly define these flags/structs.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2735>
We use this for tracking the deletable state of the client window,
but forgot to check that the MWM_HINT_FUNCTIONS hint is set in
hints.flags before checking hints.functions.
This resulted in windows that do not specify this flag (and thus
should go with the defaults) in being mistakenly removed the close
button, as the functions flags would be typically 0 in that case.
Fixes issues with Chromium and Electron applications missing the
close button, since Chromium does this on X11.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2735>
Recalculating window features is a busy thing on the Mutter side, the
different properties being (re)set will overwrite the current state
and cause some side work. Between that is the rewriting of the
_MUTTER_NEEDS_FRAME property on the window being recalculated, which
throws the frames client off, by thinking the window does actually
require a new frame.
It is not sufficient to trust that PropertyNewValue means the property
or the value are new, also double check that the window did not have
in fact a frame, and avoid the busy work if it did.
Besides the busywork that can be easily avoided, this also fixes the
window close button state being stuck if the window changed its
deletable state, since the frame being respawn managed to miss the
property change.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2735>
meta_screen_cast_window_stream_src_set_cursor_metadata() relies
entirely on meta_screen_cast_window_transform_cursor_position()
to return the correct relative cursor position.
However, this function actually does not return the expected
values, since it does not apply the resource scale to the
transformed position.
Actually apply the cursor scale when calculating the cursor
position.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2737>
meta_screen_cast_stream_src_set_cursor_sprite_metadata() receives
the cursor sprite, position, and scale, and with that it downloads
the cursor sprite by drawing it into a separate framebuffer, then
calls cogl_framebuffer_read_pixels() in it - this is the offscren
path that is very common when using screen capturing applications
such as OBS Studio.
There's a sneaky issue in this code path though: the 'scale' value
is a float. The cursor size is then determined by multiplying the
sprite width and height - two integer variables - by scale, and
this relies on standard float-to-int conversions. This is problematic
as sometimes the rounded values disagree with what is expected by
cogl_framebuffer_read_pixels(). If the packing of either the cursor
width or height is off by one, glReadPixels() will try to write into
off bounds, which crashes.
This can be reproduced by enabling fractional scaling, setting a 150%
zoom level, on a 4K screen, and opening any commit with an image diff
in gitlab.gnome.org, all while screencasting. When hovering the new
image, the cursor sprite will be such that it triggers this code path,
and reproduces this issue.
Fix this by always ceiling the cursor sprite sizes.
Closes https://gitlab.gnome.org/GNOME/mutter/-/issues/2542
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2736>
The uninitialized fields in this event causes use of uninitialised
data as seen in valgrind:
==71864== Syscall param writev(vector[0]) points to uninitialised byte(s)
==71864== at 0x5026EBD: __writev (writev.c:26)
==71864== by 0x5026EBD: writev (writev.c:24)
==71864== by 0x6482A3B: UnknownInlinedFun (xcb_conn.c:296)
==71864== by 0x6482A3B: _xcb_conn_wait.part.0 (xcb_conn.c:551)
==71864== by 0x6482BAF: UnknownInlinedFun (xcb_out.c:469)
==71864== by 0x6482BAF: _xcb_out_send (xcb_out.c:470)
==71864== by 0x6483DD7: UnknownInlinedFun (xcb_out.c:416)
==71864== by 0x6483DD7: xcb_writev (xcb_out.c:409)
==71864== by 0x53B79B4: _XSend (xcb_io.c:587)
==71864== by 0x53BBF38: _XReply (xcb_io.c:679)
==71864== by 0x53AFFC9: XQueryTree (QuTree.c:47)
==71864== by 0x4982A5F: query_xserver_stack (stack-tracker.c:508)
==71864== by 0x4EA1F5F: g_closure_invoke (gclosure.c:832)
==71864== by 0x4ECFD45: signal_emit_unlocked_R.isra.0 (gsignal.c:3796)
==71864== by 0x4EC0129: g_signal_emit_valist (gsignal.c:3549)
==71864== by 0x4EC03B2: g_signal_emit (gsignal.c:3606)
==71864== Address 0x287d5900 is 32 bytes inside a block of size 16,384 alloc'd
==71864== at 0x4849444: calloc (vg_replace_malloc.c:1340)
==71864== by 0x53A5FE8: XOpenDisplay (OpenDis.c:240)
==71864== by 0x6100E3C: _gdk_x11_display_open (gdkdisplay-x11.c:1565)
==71864== by 0x60CF675: gdk_display_manager_open_display (gdkdisplaymanager.c:462)
==71864== by 0x49D59F1: open_gdk_display (meta-x11-display.c:1041)
==71864== by 0x49D5D64: meta_x11_display_new (meta-x11-display.c:1156)
==71864== by 0x49564AD: meta_display_init_x11_finish (display.c:743)
==71864== by 0x495679D: on_x11_initialized (display.c:818)
==71864== by 0x4D67558: g_task_return_now (gtask.c:1232)
==71864== by 0x4D67782: UnknownInlinedFun (gtask.c:1301)
==71864== by 0x4D67782: g_task_return (gtask.c:1258)
==71864== by 0x495663C: on_xserver_started (display.c:788)
==71864== by 0x4D67558: g_task_return_now (gtask.c:1232)
==71864== Uninitialised value was created by a stack allocation
==71864== at 0x49D4A59: take_manager_selection (meta-x11-display.c:640)
==71864==
To fix this, fully initialize the event struct before sending it.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2535
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2724>
Since the frames are now rendered by a separate process, we no longer
can guarantee at this point that all updates were handled. Engaging
in a new synchronous resize operation will again freeze the actor,
so sometimes we are left with a not-quite-current buffer for the
frame+window surface.
In order to ensure that the right changes made it onscreen, delay
this next synchronous resize step until the moment the surface was
repainted. This avoids those glitches, while still ensuing the
resize operation ends up in sync with the pointer.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
Let the frames client render its own shadow. In order to do that,
avoid double painting a shadow on the compositor side, and extend
the mask area of the frame, so it does unveil the (so far)
hidden frames-client-side shadows.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
There's two meanings of "frame" there! Since SSD frames are now
rendered by an external client, and there are no actual mechanism
that ensures the frame did already get painted when the client did
respond to its NET_WM_FRAME_SYNC_REQUEST request, there may be
artifacts when resizing windows.
In order to get always the best visual result, we should actually
synchronize rendering with both the client window and the window
frame window.
This commit adds these mechanisms, so a sync alarm update is
expected on both windows until further resizes are allowed, this
ensures window and frame stay in sync, even after moving rendering
elsewhere.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
It will become necessary to track properties and changes from frame windows,
and it will be more convenient to have this managed by the common property
tracking mechanisms.
Add this source_xwindow parameter so property handler functions can check
whether the property belonged to the client Window or the frame Window.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
Store the alarms in a different hashtable, and look up the MetaSyncCounter
right away. It so far avoids the MetaWindow middle man, but will also be
simpler when each window can possibly have more than one active alarms.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
Replace the in-process implementation of frames with the external
frames client.
When a client window is created and managed by Mutter, Mutter will
determine whether it is a window that requires decorations and
hint the creation of a frame for it by setting the _MUTTER_NEEDS_FRAME
property on the client window.
After the frames client created a window that has the _MUTTER_FRAME_FOR
property, Mutter will proceed to reparent the client window on the
frame window, and show them as a single unit.
Rendering and event handling on the frame window will be performed by
the external client, Mutter is still responsible for everything else,
namely resizing client and frame window in synchronization, and
managing updates on the MetaWindowActor.
In order to let the frame be managed by the external client, Mutter
needs to change the way some properties are forwarded to the client
and/or frame windows. Some properties are necessary to keep propagating
to the client window only, some others need to happen on the frame
window now, and some others needs to be propagated on both so they
are synchronized about the behavior.
Also, some events that were previously totally unexpected in frame
windows are now susceptible to happen, so must be allowed now.
MetaFrame in src/core/frame.c now acts as the wrapper of foreign
windows created by the frames client, from the Mutter side. Location,
size, and lifetime are still largely in control of Mutter, some
details like visible/invisible borders are obtained from the client
instead (through the _MUTTER_FRAME_EXTENTS and _GTK_FRAME_EXTENTS
properties, respectively).
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
This small X11 client takes care of creating frames for client
windows, Mutter will use this client to delegate window frame
rendering and event handling.
The MetaWindowTracker object will keep track of windows created
from other clients, and will await for _MUTTER_NEEDS_FRAME property
updates on those (coming from Mutter), indicating the need for a
frame window.
This process is resilient to restarts of the frames client, existing
windows will be queried during start, and the existence of relevant
properties checked. Mutter will be able to just hide/show
SSD-decorated windows while the frames client restarts.
The frames are created through GTK4 widgets, the MetaWindowContent
widget acts as a replacement prop for the actual client window,
and the MetaFrameHeader wraps GtkHeaderBar so that windows can be
overshrunk, but otherwise a MetaFrame is a 100% true GTK4 GtkWindow.
After a frame window is created for a client window, the
_MUTTER_FRAME_FOR property will be set on the frame window,
indicating to mutter the correspondence between both Windows.
Additionally, the pixel sizes of the visible left/right/top/bottom
borders of the frame will be set through the _MUTTER_FRAME_EXTENTS
property, set on the frame window.
In order to make the frame window behave as the frame for the
client window, a number of properties will be tracked from the
client window to update the relevant frame behavior (window title,
resizability, availability of actions...), and also some forwarding
of events happening in the frame will be forwarded to the client
window (mainly, WM_DELETE_WINDOW when the close button is clicked).
Other than that, the frames are pretty much CSD GTK4 windows, so
window drags and resizes, and window context menus are forwarded for
the WM to handle.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
This check dates all the way back to commit ac2aa5337d. At the time, the
window switcher was an actual X window, that could generate crossing events
if popped up under the pointer. Checking for this kind of crossing events
made sense back at the time in order not to break focus-follows-mouse as
it's been behaving for long.
But now, this UI is all Clutter widgetry, which in the worst case (X11
sessions, of course) it will update the stage window shape to make these
parts clickable. This happens in other places of code that do already
check for ignoring crossing events.
Underneath, this looked up for a Mutter-local GdkWindow of type
GDK_WINDOW_TEMP, only the main MetaFrames window matches those characteristics
nowadays, notably no window switcher popups. Since the remaining window is
never unmapped (until perhaps shutdown), the paths were functionally dead.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
If the window is not managed, it's weird that it asks for _NET_FRAME_EXTENTS,
it's even weirder that mutter replies with a frame border that would only
apply if the window were managed. Stop doing the latter, and drop the
MetaUI call that calculates borders from the theme settings.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
Put the helper to use, in order to lift MetaWindow itself from this
accounting. As a bonus, the data itself now moved to the MetaWindowX11
private struct, since this may only happen with X11 windows (or its
Xwayland subclass).
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
This helper struct takes care of the handling of requests and alarms
in order to satisfy NET_WM_SYNC_REQUEST. It will be necessary to
decouple rendering of windows and frames in future commits, so each
window may need its own synchronization and accounting.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
This may result in a view of the stack in MetaStackManager that does not correspond
to reality, since the window is already being unmanaged, there is no point either in
notifying the stack manager about it.
This slight divergence with reality in the MetaStackManager may produce a non-accurate
view if querying its state has to go through the predicted branches. Later synchronization
with the X11 stack may even this out, but the result really depends on when it is asked.
Fixes some intermittent failures in the stacking/closed-transient-only-take-focus-parents
unit test.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
The meta_window_show() method internally relies on window->mapped being
up-to-date, or attempting to focus it may fail since the window is not
mapped yet, resulting on the window being mapped, but not focused as
it would be expected.
This is moot so far, since windows with frames are created sort-of
synchronously and showing them will result in the focus attempt happening
when the window is already mapped, but things will break when this
becomes an asynchronous step.
Ensure to synchronize client state before showing, so any attempts to
focus the window are able to succeed despite the initial state when
calling meta_window_update_visibility().
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
The test does simply "wait" which apparently is not enough to ensure the
client window did resize to the expected dimensions. Use "wait_reconfigure"
and assert that the size after resize is the expected, before going further
at testing its behavior after maximize/unmaximize; it might end up with the
unexpected size after the whole operation.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175>
Move the use count from a separate MetaWaylandBufferRef struct to the
MetaWaylandBuffer class, and remove the former.
The buffer use count is now incremented already in
meta_wayland_surface_commit, since the Wayland protocol defines the
buffer to be in use by the compositor at that point. If the buffer
attachment ends up being dropped again before it is applied to the
surface state (e.g. because another buffer is committed to a
synchronized sub-surface before the parent surface is committed),
the use count is now decremented, and a buffer release event is sent if
the use count drops to 0.
Buffer release events were previously incorrectly not sent under these
circumstances. Test case: Run the weston-subsurfaces demo with the -r1
and/or -t1 command line parameter. Resize the window. Before this
change, weston-subsurfaces would freeze or abort after a few resize
operations, because mutter failed to send release events and the
client ran out of usable buffers.
v2:
* Handle NULL priv->buffer_ref in
meta_wayland_cursor_surface_apply_state.
v3:
* Remove MetaWaylandBufferRef altogether, move the use count tracking
to MetaWaylandBuffer itself. Much simpler, and doesn't run into
lifetime issues when mutter shuts down.
v4:
* Warn if use count isn't 0 in meta_wayland_buffer_finalize.
* Keep pending_buffer_resource_destroyed for attached but not yet
committed buffers. If the client attaches a buffer and then destroys
it before commit, we ignore the buffer attachement, same as before
this MR.
v5:
* Rebase on top of new commit which splits up surface->texture.
* MetaWaylandSurfaceState::buffer can only be non-NULL if
::newly_attached is TRUE, simplify accordingly.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
Until all dma-buf file descriptors for all buffers in the transaction
are readable, which corresponds to when the client drawing to the
buffers has finished.
This fixes https://gitlab.gnome.org/GNOME/mutter/-/issues/1162 if the
GPU & drivers support high priority contexts which can preempt lower
priority contexts.
v2:
* Also remove dma-buf fds from transaction and try applying it from
pending_buffer_resource_destroyed. Avoids freeze due to leaving a
GSource based on a closed fd attached if a client destroys a wl_buffer
which is part of a transaction which was committed but not applied
yet. (Robert Mader)
* Tweak transaction cleanup logic in wl_surface_destructor.
v3:
* Adapt to meta_wayland_dma_buf_get_source.
v4:
* Adapt to new commits using transactions for (sub-)surface destruction,
drop code to remove destroyed surfaces from pending transactions.
v5:
* Use g_clear_pointer in meta_wayland_transaction_destroy.
(Georges Basile Stavracas Neto)
* Add spaces between type casts and values. (Carlos Garnacho)
* Use (gpointer *) instead of (void**). (Carlos Garnacho)
* Use gpointer instead of void * in
meta_wayland_transaction_dma_buf_dispatch.
v6:
* Use g_hash_table_remove in meta_wayland_transaction_dma_buf_dispatch.
(Carlos Garnacho)
v7: (Jonas Ådahl)
* Move include of glib-unix.h below that of meta-wayland-transaction.h.
* Split up g_hash_table_iter_next call to multiple lines in
meta_wayland_transaction_commit.
* Call g_source_destroy as well as g_source_unref when freeing a
committed but not yet applied transaction (during mutter shutdown).
v8:
* Drop dma_buf_source_destroy, can use g_source_destroy directly.
(Jonas Ådahl)
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
From xdg_surface_constructor_get_popup / xdg_popup_reposition (called
during Wayland protocol processing) to finish_popup_setup /
meta_wayland_xdg_popup_apply_state (called when the popup state is
applied).
This makes sure that the parent window frame rectangle is up to date in
meta_wayland_xdg_positioner_to_placement.
v2:
* Use meta_wayland_surface_state_new () in
meta_wayland_transaction_add_xdg_popup_reposition.
v3:
* Move xdg_popup_repositioned handling to
meta_wayland_xdg_popup_apply_state.
v4:
* Do not steal pending->xdg_positioner in
meta_wayland_xdg_popup_apply_state, fixes leaking the corresponding
memory.
* Drop MetaWaylandSurfaceState::xdg_popup_repositioned, just use
::xdg_positioner.
v5:
* Reformat meta_wayland_xdg_positioner_to_placement calls to stay within
80 columns. (Jonas Ådahl)
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
This makes sure that finish_popup_setup is called after any previous
transactions for the parent surface have been applied, so the parent
window geometry is up to date.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
Preparation for potentially calling meta_wayland_transaction_apply some
time after surface commit, in which case doing it in the former would be
too late: The client may legally destroy the attached wl_buffer
immediately after commit, in which case meta_wayland_buffer_attach would
spuriously fail and disconnect the client (or possibly even crash mutter
due to NULL error).
Requires splitting up the surface texture between protocol and output
state, and propagating from the former to the latter via
MetaWaylandSurfaceState.
v2: (Jonas Ådahl)
* Move meta_wayland_surface_get_texture call to separate line.
* Use g_autoptr for GError.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
And keep track of the hierarchy separately for the Wayland protocol and
for output. Protocol state is updated immediately as protocol requests
are processed, output state only when the corresponding transaction is
applied (which may be deferred until the next commit of the parent
surface).
v2:
* Directly add placement ops to a transaction, instead of going via
pending_state.
* Use transaction entry for the sub-surface instead of that for its
parent surface.
v3:
* Use transaction entry for the parent surface again, to ensure proper
ordering of placement ops, and call
meta_wayland_surface_notify_subsurface_state_changed only once per
parent surface.
* Drop all use of wl_resource_add_destroy_listener, transactions are
keeping surfaces alive as long as needed.
v4:
* Rebase on https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2501
* Drop ClutterActor code from meta_wayland_surface_apply_placement_ops.
(Robert Mader)
v5:
* Rename MetaWaylandSubSurfaceState to MetaWaylandSurfaceSubState, since
the next commit adds not sub-surface specific state to it.
v6:
* Move include of meta-wayland-subsurface.h from
meta-wayland-transaction.c to .h, since the latter references
MetaWaylandSubsurfacePlacementOp.
v7:
* Drop superfluous !entry check from meta_wayland_transaction_apply.
v8:
* Rename output/protocol fields to output/protocol_state. (Jonas Ådahl)
v9:
* Use meta_wayland_surface_state_new in
meta_wayland_transaction_add_placement_op.
v10:
* Fix a few style issues per check-style.py.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
Destroy Wayland protocol related state immediately when the Wayland
resource is destroyed, but keep the rest alive by any transaction which
references the surface.
This makes it easier and cleaner to deal with a surface getting
destroyed while it's still referenced by transactions.
v2:
* No more need to keep references for surfaces in the entries hash
table.
v3:
* Do not use surface->sub.transaction in wl_surface_destructor, just
destroy it.
v4:
* No need for wl_surface_destructor to use its own transaction.
v5:
* Use g_steal_pointer & (more) g_clear_pointer in wl_surface_destructor.
v6:
* Leave SURFACE_DESTROY signal emission in wl_surface_destructor.
v7:
* Use finalize instead of dispose callback.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
This keeps all surfaces referenced by a transaction alive until the
transaction is destroyed, and makes sure transactions are applied in
the same order as they were committed with respect to all surfaces
they reference.
v2:
* Guard against NULL entry in meta_wayland_transaction_apply.
v3:
* Keep single entries hash table.
v4:
* Unref the surface in the meta_wayland_transaction_merge_into while
loop only if the "to" transaction didn't already have an entry for it,
to prevent premature finalization of the surface (likely followed by a
crash).
v5:
* Unref the surface (implicitly via g_hash_table_iter_remove) in the
meta_wayland_transaction_merge_into while loop even if the "to"
transaction already had an entry for it, or we leak a reference.
* Use g_clear_object & g_steal_pointer to not leave behind a dangling
from->state pointer in meta_wayland_transaction_entry_merge_into.
v6:
* Add curly braces around
meta_wayland_transaction_add_placement_surfaces calls. (Jonas Ådahl)
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
Instead of cached_state.
surface_commit for a synchronized sub-surface either commits the
transaction or merges it into the parent surface's transaction (if
the parent is a synchronized sub-surface itself).
This should fix or at least improve the behaviour of nested synchronized
sub-surfaces.
Also change wl_subsurface_set_desync:
* Commit sub-surface transactions separately. This may allow some of
them to be applied earlier in some cases.
* Commit transaction only for descendant sub-surfaces which become
newly de-synchronized themselves.
v2:
* Drop unused function prototypes
v3:
* Use g_clear_pointer for surface->sub.transaction.
v4:
* Use g_steal_pointer instead of g_clear_pointer. (Sebastian Wick, Jonas
Ådahl)
v5: (Carlos Garnacho)
* Add spaces between type casts and values.
* Use (gpointer *) instead of (void**).
v6: (Jonas Ådahl)
* Use g_clear_object in meta_wayland_transaction_entry_merge_into.
* Use meta_wayland_transaction_entry_free in
meta_wayland_transaction_merge_into.
* Fix alignment of meta_wayland_transaction_merge_pending_state
parameters.
* Remove unused meta_wayland_transaction_add_state declaration.
v7:
* Use meta_wayland_surface_state_new in
meta_wayland_transaction_merge_pending_state.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
A transaction contains the committed state for a surface, plus any
cached state for synchronized subsurfaces.
v2:
* Handle sub-surface positions separately from surface states.
v3:
* Sync child states only for surfaces with state in the transaction.
v4: (Jonas Ådahl)
* Drop unnecessary g_object_new call from wl_subsurface_set_desync. (me)
* Fix indentation & formatting in meta_wayland_surface_commit.
* Add meta_wayland_surface_state_new helper function.
* Fix alignment of meta_wayland_transaction_apply_subsurface_position
parameters.
* Add curly braces around meta_wayland_transaction_sync_child_states
call in meta_wayland_transaction_apply.
v5:
* Make meta_wayland_surface_state_new an inline function.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
If multiple transactions have entries for the same surface, they are
applied in the same order as they were committed. Otherwise, they can
be applied in any order.
This is preparation for following changes, transactions are still
applied as soon as they're committed.
v2:
* Move GQueue for transactions to MetaWaylandCompositor (Jonas Ådahl)
v3
* Say "entry for" instead of "state for", since there can be transaction
entries with no state (for surfaces which are getting destroyed).
v4:
* Use a hash table to keep track of all candidate transactions which
might be newly ready to be applied.
* Use clearer function / variable names.
v5:
* Use custom single-linked list instead of hash table for candidate
transactions, ordered by the transaction commit sequence number, so
that they're attempted to be applied in the same order as they were
committed.
* Rename transaction->queue to transaction->committed_queue, and
simplify its handling.
v6: (Carlos Garnacho)
* Add spaces between type casts and values.
* Use (gpointer *) instead of (void**).
v7: (Jonas Ådahl)
* Use G_MAXSIZE instead of ULONG_MAX.
* Fix indentation of meta_wayland_transaction_apply &
meta_wayland_transaction_maybe_apply_one parameters.
* Refactor find_next_transaction_for_surface & ensure_next_candidate
helper functions out of meta_wayland_transaction_apply.
* Refactor has_unapplied_dependencies helper function out of
meta_wayland_transaction_maybe_apply_one.
* Make while (TRUE) loop in meta_wayland_transaction_maybe_apply
consistent with general usage.
* Drop unused value local from meta_wayland_transaction_commit.
* Store pointer to compositor object in transactions, instead of
pointer to the queue of committed transactions.
* Drop tautological g_assert from meta_wayland_transaction_apply. (me)
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
v2:
* Use single hash table with struct which will contain all kinds of
state handled by a transaction.
v3:
* Add meta_wayland_transaction_destroy.
v4 (Georges Basile Stavracas Neto)
* Fix struct _MetaWaylandTransaction(Entry) formatting.
* Explicitly test against NULL.
* Use gpointer insteadof void * for
meta_wayland_transaction_entry_destroy.
v5: (Robert Mader)
* Use for loop in is_ancestor.
* Include meta-wayland-transaction.h first in
meta-wayland-transaction.c.
v6:
* Use g_autofree & g_clear_object.
v7: (Jonas Ådahl)
* Rename meta_wayland_transaction_entry_destroy to
meta_wayland_transaction_entry_free.
* Drop g_autofree use from meta_wayland_transaction_entry_free again.
* Make meta_wayland_transaction_entry_free take a
MetaWaylandTransactionEntry pointer.
* Rename meta_wayland_transaction_destroy to
meta_wayland_transaction_free.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
Need to deal with surface->resource == NULL and
surface->pending_state == NULL in some places.
v2:
* Avoid expanding conditions to multiple lines.
(Georges Basile Stavracas Neto)
v3:
* Use a single bailout condition in meta_wayland_client_owns_window as
well.
v4:
* Remove spare empty line in meta_wayland_surface_apply_state.
(Robert Mader)
* Add wl_resource_post_error calls in xdg-shell request handlers.
(Robert Mader)
* Drop checks in functions which can only be called if there's a valid
resource.
* Drop more checks which are unnecessary due to leaving the
SURFACE_DESTROY signal emission in wl_surface_destructor later.
v5:
* Move resource = surface->resource assignments to if (!resource) tests.
(Jonas Ådahl)
v6:
* Fix style issue per check-style.py.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
Creates a GSource which will call the specified dispatch callback when
all dma-buf file descriptors for the buffer have become readable.
v2:
* Hold a reference to the buffer in the source, to prevent the buffer
from getting destroyed before the source.
v3:
* Do not use check callback, handle everything in dispatch callback.
(Dor Askayo)
v4: (Georges Basile Stavracas Neto)
* Define and use MetaWaylandDmaBufSource & MetaWaylandDmaBufSourceDispatch
types.
* Fix meta_wayland_dma_buf_source_dispatch &
meta_wayland_dma_buf_source_funcs formatting.
* Use gpointer instead of void*.
* Rename meta_wayland_dma_buf_get_source to
meta_wayland_dma_buf_create_source. (Carlos Garnacho)
v5:
* Explicitly handle NULL return value. (Jonas Ådahl)
v6:
* Fix style issue per check-style.py.
v7:
* Fix code style harder. (Jonas Ådahl)
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
And call it from meta_wayland_buffer_realize. This makes dma-buf fds
available for EGL image type buffers as well.
v2:
* Move buffer->dma_buf.dma_buf assignment value to next line.
(Jonas Ådahl)
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1880>
Add internal state (starting, running, stopping), and use this instead
of MetaDisplay struct fields to determine whether to start animations.
This fixes issues when we try to animate things when shutting down.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2716>
On hotplug, the events we receive from the kernel are async, and
connectors in the kernel come and go as they please. In practice, this
means that calling drmModeGetConnector() twice more or less directly
after each other, there is no guarantee that the latter call will return
anything if the former did.
When updating the connector in response to hotplugs, we'd first update
the list of existing connectors, and following that, query each and
every one again for their current state, to update our internal
representation; only the former handled drmModeGetConnector() returning
NULL, meaning if unlucky, we'd end up doing a null pointer dereference
when trying to update the state.
Handle this by querying the kernel for the current connector state only
once per connector, updating the list of connectors and their
corresponding state at the same time.
Fixes the following crash:
#0 meta_kms_connector_read_state at ../src/backends/native/meta-kms-connector.c:684
#1 meta_kms_connector_update_state at ../src/backends/native/meta-kms-connector.c:767
#2 meta_kms_impl_device_update_states at ../src/backends/native/meta-kms-impl-device.c:916
#3 meta_kms_device_update_states_in_impl at ../src/backends/native/meta-kms-device.c:267
#4 meta_kms_update_states_in_impl at ../src/backends/native/meta-kms.c:604
#5 update_states_in_impl at ../src/backends/native/meta-kms.c:620
#6 meta_kms_run_impl_task_sync at ../src/backends/native/meta-kms.c:435
#7 meta_kms_update_states_sync at ../src/backends/native/meta-kms.c:641
#8 handle_hotplug_event at ../src/backends/native/meta-kms.c:651
#9 on_udev_hotplug at ../src/backends/native/meta-kms.c:668
Related: https://bugzilla.redhat.com/show_bug.cgi?id=2131269
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2709>
There is no need to use the 'bypass-*' method of event processing in the
changed function since in all cases the 'bypass-*' variable was set, any
following event processing functions would ignore the event anyway.
Simplify things a bit by just returning TRUE if the event is consumed.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2714>
We can land inside meta_window_focus() in the middle of changing the
window workspace, because some signal handler of MetaWorkspace's
"window-removed" signal triggers a focus. This can cause a crash in
`g_assert (link)` when updating the MRU list because we still think
we're on the old workspace when actually we are already removed from
this workspaces MRU list.
To avoid crashes like this, bail out of meta_window_focus() when we're
in the middle of a workspace change.
Fixes https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/5368
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2691>
It's a bad idea to have data like this in the middle of a struct, as it
will easily cause everything behind it to be badly aligned and thus
increase memory access times.
So move all those bitfield booleans to the end of the struct.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2691>
It is generally assumed here and there that the pointer at all point in
time is within some logical monitor, if there is any logical monitor to
be within.
With the input thread, this was for a short amount of time not reliable,
resulting in crashes in combination with hotplugging or suspend/resume,
where monitors come and go quickly.
What happens is that the pointer at first is within a logical monitor,
but when that logical monitor is removed, while the new monitor
viewports are handed to the input thread, the constraining happens
asynchronously, meaning there is a time between between the new
viewports are sent, and before clutter_seat_query_state() starts
reporting the constrained position.
If a new client mapped a maximized window during this short time frame,
we'd crash with
#0 meta_window_place at ../src/core/place.c:883
#1 place_window_if_needed at ../src/core/constraints.c:562
#2 meta_window_constrain at ../src/core/constraints.c:310
#3 meta_window_move_resize_internal at ../src/core/window.c:3869
#4 meta_window_force_placement at ../src/core/window.c:2120
#5 xdg_toplevel_set_maximized at ../src/wayland/meta-wayland-xdg-shell.c:429
#6 ffi_call_unix64 at ../src/x86/unix64.S:105
#7 ffi_call_int at ../src/x86/ffi64.c:672
#8 wl_closure_invoke at ../src/connection.c:1025
#9 wl_client_connection_data at ../src/wayland-server.c:437
The fix for this is to make sure that the viewports are updated and
pointers constrained synchronously, i.e. the main thread will wait until
after the input thread is done constraining before continuing.
Related: https://bugzilla.redhat.com/show_bug.cgi?id=2147502
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2711>
We put a DEVICE_ADDED or DEVICE_REMOVED event into Clutters event queue
here, so we should also wait for Clutter to process events once.
Just putting an event into the queue doesn't mean it gets processed
immediately (especially when the commit after this one is applied), so
wait for a stage update here.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2697>
Currently, we will notify the display about a new window being created
during the constructed phase of the GObject. During this time,
property-change notifications are frozen by GObject, so we'll emit a few
::notify signals only after the window-created signal, although
the actual property change happened before that.
This caused confusion in gnome-shell code where a notify::skip-taskbar =
true emission was seen when the property already was true inside a
window-created handler before.
In order to fix that that, we notify the window creation
post-construction
of the GObject on GInitable.init vfunc
Details
https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/6119#note_1598983
Fixes https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/6119
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2703>
If two X11 windows were the last two, we'd remove them from the stack
while unmanaging them. That'd hit an assert in
meta_stack_tracker_restack_managed(), resulting in the following crash
when Xwayland exited unexpectedly with two or more X11 windows being the
only windows on the stack:
#1 g_assertion_message() at ../glib/gtestutils.c:3256
#2 g_assertion_message_expr() at ../glib/gtestutils.c:3282
#3 meta_stack_tracker_restack_managed() at ../src/core/stack-tracker.c:1210
#4 on_stack_changed() at ../src/core/stack.c:142
#5 _g_closure_invoke_va() at ../gobject/gclosure.c:895
#6 g_signal_emit_valist() at ../gobject/gsignal.c:3456
#7 g_signal_emit() at ../gobject/gsignal.c:3606
#8 meta_stack_changed() at ../src/core/stack.c:265
#9 meta_stack_remove() at ../src/core/stack.c:324
#10 meta_window_unmanage() at ../src/core/window.c:1542
#11 meta_x11_display_unmanage_windows() at ../src/x11/meta-x11-display.c:111
#12 meta_x11_display_dispose() at ../src/x11/meta-x11-display.c:141
#13 g_object_run_dispose() at ../gobject/gobject.c:1448
#14 meta_display_shutdown_x11() at ../src/core/display.c:831
The added test specifically checks that this scenario is handled
gracefully.
Related: https://bugzilla.redhat.com/show_bug.cgi?id=2143637
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2704>
Instead of having users of the test client manually deal with alarm
filters, let the test client automatically add itself as filters. This
changes the MetaX11Display a bit, to handle an array of filters instead
of a single filter.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2704>
The CRTC cursor sprite scale was incorrectly assumed to be always 1.0
when using the default not-scale-monitor-framebuffer mode. This is
harmless in most cases, as most clients provide HiDPI capable cursors,
but for the ones that didn't, we'd end up drawing their cursors
unscaled, when using the cursor planes.
Fix this by using the "texture scale" which is what is intended for
this.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2477
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2698>
Cursor planes tend to be ARGB8888 and support no other format (ideally
we should not hard code this, but un-hard-coding that is for another
day), and if we put e.g. a XRGB8888 buffer in there, it'll either result
in the gbm_bo allocation failing (it doesn't allow USE_CURSOR with any
other format) or mode setting failing if using dumb buffers directly.
In the former case, we'll fall back to OpenGL indefinitely, and in the
latter, we'll have failed mode sets as long as we try to set the invalid
cursor buffer as the cursor plane.
Change things to process all buffers that are not ARGB8888 using the
scale/rotate machinery we already have, turning XRGB8888 into ARGB8888.
Related: https://gitlab.gnome.org/GNOME/mutter/-/issues/2477
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2698>
Attaching a new buffer with a different size than the old one means
that the viewport needs to be recalculated.
Not doing this caused the viewport to be incorrectly applied when
viewport_src_rect remained the same after attaching such buffer.
Pipeline reset usually happens when applying a new viewport,
but it doesn't happen when the viewport values remain the same.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2689>
A client may provide a positioner that places the window outside of its
parent. This isn't allowed, according to spec, so we hide the window and
log a warning. This, however, leads these affected clients with an
incorrect view of what is mapped or not, meaning it becomes harder to
recover.
Fix this by sending xdg_popup.done when we hide the popup due to an
invalid position. Don't error out the client, let the bug slide, as
that's a less jarring experience for existing applications that
reproduce this than being disconnected, which practically feels like a
crash.
Related: https://gitlab.gnome.org/GNOME/mutter/-/issues/2408
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2645>
In constrast to x11, Wayland has sane handling for touch events and
allows the compositor to handle a touch event while the clients are
already seeing it. This means we don't need the REJECTED state on
Wayland, since we can also grab sequences after the client has seen
them.
So disallow moving sequences to the REJECTED state on Wayland.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2508>
It's not really a backend thing, and we'll want to profile e.g. loading
the backend too, so create it very early and destroy it very late and
let MetaContextMain own it.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2678>
This change fixes the issue where the cursor is always
embedded in the frames even when the client has requested
the cursor information be sent as metadata in the stream.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2629>
This reverts commit eac227a203.
Currently, Flatpak applications can bypass the X11 permission setting
and access the X server through abstract sockets because X11 authentication
is not enforced for the current user ID.
Fix this by always requiring X11 authentication for Xwayland. This also
means applications without XAUTHORITY set to the file with Mutter's
Xwayland credentials cannot connect to X, including apps launched from
VT or SSH.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2633>
When deciding if a window should be unredirected because it was causing
fullscreen damage in the past, it was not considered whether the window
is still fullscreen. This could result in a floating window being
unredirected if it was chosen for unredirection because of
_NET_WM_BYPASS_COMPOSITOR = 1 and was previously fullscreened for >= 100
frames, long enough to change does_full_damage, before getting
unfullscreened.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2434
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2651>
Back in 2014 sending pressed keys to Wayland clients caused issues,
because at least Xwayland didn't handle that gracefully, causing issues
like ghost-pressed keys. A way it was reproduced was quickly alt-tab:ing
to and from a Firefox window, which would cause the File menu bar
incorrectly appearing.
While this was reported to the Xwayland component back then, it was,
probably by mistake, assumed to be an issue in mutter, and mutter
stopped sending pressed key events on enter.
The following year, Xwayland was eventually fixed, but the work around
in mutter has been kept around until it was again noticed as an
inconsistency between compositor implementations.
Lets remove the work around, and follow the spec, again.
This reverts commit c39f18c2d4.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2457
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2657>
We'd set the capabilities to 'none', meaning all previously enabled
device classes would be disabled. That means we shouldn't re-disable
them directly after.
This ensures '..disable()' is only called once for every '..enable()'.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2657>
With high frequency mouse devices, we would send very many configure
events per each update cycle, which had the end result that some clients
constantly re-allocating and redrawing their buffers far too often, if
they did this in direct response to xdg_toplevel configure events.
Lets throttle the interactive resize updates to stage updates, to avoid
having these clients doing the excessive buffer reallocation.
This also removes some old legacy X11 client resize throttling, that
throttled a bit arbitrarily on 25 resizes a second; it is probably
enough to throttle on stage updates for these clients.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2652>
Some mice send a value slightly lower than 120 for some detents. The
current approach waits until a value of 120 is reached before sending a
low-resolution scroll event.
For example, the MX Master 3 sends a value of 112 in some detents:
detent detent
| | |
^ ^ ^
112 REL_WHEEL 224
As illustrated, only one event was sent but two were expected. However,
sending the low-resolution scroll event in the middle plus the existing
heuristics to reset the accumulator solve this issue:
detent detent
| | |
^ ^ ^ ^
REL_WHEEL 112 REL_WHEEL 224
Send low-resolution scroll events in the middle of the detent to solve
this problem.
Fix https://gitlab.gnome.org/GNOME/mutter/-/issues/2469
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2668>
Previously, when scroll was received in a remote session, it was handled
as continuous scroll.
This generated issues with clients without high-resolution scroll
support as the code path in charge of accumulating scroll until 120 is
reached was not used and therefore discrete scroll events were not being
generated.
Handle scroll generated in a remote session as discrete scroll when the
source is CLUTTER_SCROLL_SOURCE_WHEEL to fix this issue.
Fix https://gitlab.gnome.org/GNOME/mutter/-/issues/2473
Fixes: 9dd6268d13 ("wayland/pointer: Send high-resolution scroll data")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2664>
In fcfe90aa, multiple for loops were replaced with
META_WAYLAND_SURFACE_FOREACH_SUBSURFACE.
However, this substitution was not side-effect free, and introduced a
null-pointer dereference risk as shown in the example below:
Old:
for (n = g_node_first_child (surface->subsurface_branch_node);
n;
n = g_node_next_sibling (n))
{
if (G_NODE_IS_LEAF (n))
continue;
meta_wayland_surface_update_outputs_recursively (n->data);
}
n is checked for NULL during each loop in the condition expression.
Therefore, when `G_NODE_IS_LEAF (n)` is called, `n` is guaranteed not to
be NULL. Note also that g_node_first_child is also NULL-safe since it
performs a NULL check internally.
New:
META_WAYLAND_SURFACE_FOREACH_SUBSURFACE (surface, subsurface_surface)
meta_wayland_surface_update_outputs_recursively (subsurface_surface);
=
for (GNode *G_PASTE(__n, __LINE__) = meta_get_first_subsurface_node ((surface)); \
(subsurface = (G_PASTE (__n, __LINE__) ? G_PASTE (__n, __LINE__)->data : NULL)); \
G_PASTE (__n, __LINE__) = meta_get_next_subsurface_sibling (G_PASTE (__n, __LINE__)))
In the new logic `subsurface` is still checked for NULL in the loop
condition. However, in the new loop init:
...
meta_get_first_subsurface_node (MetaWaylandSurface *surface)
...
n = g_node_first_child (surface->subsurface_branch_node);
if (!G_NODE_IS_LEAF (n))
...
The above implementation performs a `G_NODE_IS_LEAF` call, which
performs a dereference on `n`, without first checking for NULLs.
This NULL dereference triggers the following gnome-shell crash:
Core was generated by `/usr/bin/gnome-shell'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0 meta_get_first_subsurface_node (surface=0x55d589623450) at ../src/wayland/meta-wayland-surface.h:399
#1 pointer_can_grab_surface (pointer=0x7f6dc4012700, surface=0x55d589623450) at ../src/wayland/meta-wayland-pointer.c:1306
#2 0x00007f6ddb94d509 in meta_wayland_pointer_can_grab_surface (pointer=<optimized out>, surface=surface@entry=0x55d589623450, serial=serial@entry=996) at ../src/wayland/meta-wayland-pointer.c:1321
#3 0x00007f6ddb950d05 in meta_wayland_seat_get_grab_info (seat=seat@entry=0x55d586c24f20, surface=0x55d589623450, serial=996, require_pressed=require_pressed@entry=0, x=x@entry=0x0, y=y@entry=0x0)
at ../src/wayland/meta-wayland-seat.c:467
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2655>
Unlike the multi-view path, the optimized/single-view one doesn't check
if the surface-actor is really present on the view. That is the case
whenever it's hidden - e.g. when the window is minimized.
Fixes 3b7137cb35
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2662>
Fullscreen Wayland toplevel surfaces don't need to respect the
configured size in which case the window content get centered on a black
background which covers the whole monitor.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2338>
Fullscreen Wayland toplevel surfaces don't need to respect the
configured size in which case it should be shown centered on the monitor
with a black background. The black background becomes part of the window
geometry.
The surface container is responsible for correctly culling the surfaces
and making sure the surface actors are removed from the actor tree to
avoid destroying them.
The window actor culling implementation assumes all surfaces to be direct
children of said actor. The introduction of the surface_container actor
broke that assumption. This implements the culling interface in
MetaWindowActorWayland which is aware of the actor surface_container and
fullscreen state.
v2: Fix forwarding culling to surface even if there is a background.
v2: Don't alter passed geometry.
v2: Update window geometry code documentation to reflect these changes.
v2: Only use constrained rect if we're acked fullscreen.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2338>
Prepare for adding Wayland specific culling logic to the
MetaWindowActorWayland class by moving all the logic to the non-abstract
classes, since there will be no reason to keep the logic in
MetaWindowActor around.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2338>
This is helpful to know what current state a window actually have, in
contrast to the state in MetaWindow (e.g. MetaWindow::fullscreen) which
is the intended state, be it current or not yet so.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2338>
First make sure we call 'move_resize()' in all cases where the size or
position can change, then move the updating of the buffer rect to the
same place as we update the frame rect. This means keeping track of
surface size changes, in addition to geometry changes, and calling
finish_move_resize() whenever any of those changes, in addition to
acknowledged configurations.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2338>
A "window rect" in most places refers to the rectangle the window
corresponds to when it comes to window management. MetaWindow::rect also
refers to this window management related rectangle. However in the
geometry sync functions, it instead called what was to be the rectangle
the actor should have as "window rect", which is arguably a bit
confusing. Fix this by renaming it "actor_rect" so that it becomes clear
that it's the rectangle the actor should get on the stage.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2338>
MetaWindowActor previously peeked at the number of child Actors to
determine the number of surfaces. The following commit rearranged the
tree such that MetaWindowActorWayland always has two Actors. This change
lets the subclass determine if the main surface describes the whole
window.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2338>
When a window configuration is constructed for a Wayland surface it
contains a position, size and a scale. The scale is the geometry scale
for the configuration, i.e. before the size is sent the passed dimension
is divided with the passed scale.
When moving between monitors with different scales, if we use the
existing geometry scale, this means we will send a configure event with
incorrect dimensions. Fix this by calculating the scale used in the
configuration given the rect we're configuring with as this will mean
the correct size will be sent to the client.
v2: Removed the fullscreen condition. Don't know why it was added to
begin with.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2338>
There were some magic conditions that decided when
meta_window_constrain() was to be called or not. Reasoning about and
changing these conditions were complicated, and in practice the caller
knows when constraining should be done. Lets change things by adding a
'constrain' flag to the move-resize flags that makes this clearer. This
way we can, if needed, have better control of when a window is
constrained or not without leaking that logic into the generic
to-constrain-or-not expression.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2338>
We have no way to sanely add safe modes if there are no modes we can
compare with, thus don't try.
Fixes the following crash:
#0 are_all_modes_equally_sized at ../src/backends/native/meta-output-kms.c:284
#1 maybe_add_fallback_modes at ../src/backends/native/meta-output-kms.c:310
#2 init_output_modes at ../src/backends/native/meta-output-kms.c:347
#3 meta_output_kms_new at ../src/backends/native/meta-output-kms.c:414
#4 init_outputs at ../src/backends/native/meta-gpu-kms.c:332
#5 meta_gpu_kms_read_current at ../src/backends/native/meta-gpu-kms.c:368
#6 meta_gpu_kms_new at ../src/backends/native/meta-gpu-kms.c:403
#7 create_gpu_from_udev_device at ../src/backends/native/meta-backend-native.c:461
#8 init_gpus at ../src/backends/native/meta-backend-native.c:551
#9 meta_backend_native_initable_init at ../src/backends/native/meta-backend-native.c:632
Fixes: 877cc3eb7d44e2886395151f763ec09bea350444
Related: https://bugzilla.redhat.com/show_bug.cgi?id=2127801
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2646>
This just checks for any chromaticity being zero and gamma being in
range but we could do a better job at detecting bad data in the future.
Also check the return value of cmsCreateRGBProfileTHR which can be NULL.
Fixes gnome-shell#5875
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2627>
Make sure that if we wiggle a scan-out capable surface a bit, it won't
scan out if it's not exactly in the right position. Do this by first
making the window not fullscreen, then moving it back and forth,
verifying the correct scanout state for each presented frame.
This test addition reproduces the issue described in
https://gitlab.gnome.org/GNOME/mutter/-/issues/2387.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2624>
If we have a window that match the size (i.e. will pass the "fits
framebuffer" low level check), that doesn't mean it matches the
position. For example, if we have two monitors 2K monitors, with two 2K
sized windows, one on monitor A, and one on monitor both monitor A and
B, overlapping both, if the latter window is above the former, it'll end
up bing scanned out on both if it ends up fitting all the other
requirements.
Fix this by checking that the paint box matches the stage view layout,
as that makes sure the actor we're painting isn't just partially on the
right view.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2387
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2624>
Painting the swap region with CLUTTER_DEBUG_PAINT_DAMAGE_REGION happens
on the view framebuffer, so don't transform the region we paint to the
onscreen.
Fixes the swap region painting on rotated monitors.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2590>
Gnome-shell uses meta_display_focus_default_window() when shell elements
loose focus which is the case with Alt+Tab window switching. Globally
active input clients don't immediately gain focus though so if
meta_display_focus_default_window focuses a wrong window stacking and
focus don't behave as expected.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2489>
New commands to set the number of workspaces, activate a workspace, with
and without focus, move windows to specific workspaces, and check the
stacking on a specific workspace.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2489>
When switching workspaces we previously focused on whatever window is on
top of the stack. If a window is marked as "always on top" then it would
always receive focus when switching workspaces.
Fixes#2240
Fixes gnome-shell#5162
Fixes#178Fixes#678
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2489>
We want to use the workspace MRU list to decide the default focus but
Globally Active Input clients don't call
meta_window_set_focused_internal and therefore don't update the MRU
list. Move the update to meta_window_focus instead.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2489>
The completed signal is only emitted if the timeline actually completed
but when an actor is destroyed or removed from its parent the timeline
is stopped and not completed.
The workspace switch effect removes window actors from the window group
which destroys the timeline so on_$EFFECT_effect_stopped is never
called and the pointer to the timeline is dangling.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2489>
The workspace switch animation moves the WindowActors out of the
WindowGroup so if we shut down while the animation is playing the
WindowActors will have queued a destroy but will be disposed only after
the compositor is destroyed, leaving the WindowActor with a dangling
pointer.
Fix the issue by killing the workspace switch animation on shutdown.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2489>
This is an old relic from when ClutterStageView was being added, and
tests were somewhat prepared to be able to test the "X11 style" of
things, with the nested backend some how managing to emulate that.
Lets drop that stuff, it isn't used by the test suite, and isn't useful
anyway; if we want to test X11 configurations, we should use the actual
X11 backend, which didn't make use of this anyway.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2619>
It exposed unnecessary public and private API, and used a global static
variable instead of a return value, none which was necessary. Remove
both API and use a return value for communicating to the caller.
This doesn't remove a public symbol, lets do that for GNOME 44.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2619>
This was used gala to implement hot corners, and the way the barrier API
works, there isn't really any practical reasons to not make it
derivable, since the backend is a separate type and object.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2626>
This adds a copy of the calibration test profile and sets up a test to
first add it as a system profile, then setting up the XDG_DATA_HOME
directory so that the duplicate profile is detected, added, and later
discarded.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2622>
