We are using internal API that has the benefit of checking that the
focus surface still matches, but has the drawback that it does not
check the MetaWaylandKeyboard state.

In order to fix this, look for keyboard focus and serial matches
specifically when triggering activation.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2148>


(cherry picked from commit d2998a8e)

This calculation wrongly applied a `x2 - x1` pattern.

Closes https://gitlab.gnome.org/GNOME/mutter/-/issues/1971

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2169>


(cherry picked from commit 6204769f)

When dealing with a faulty hardware or bugs in the driver, it might be
interesting to force the use of software cursors for debugging purposes.

Add a debug environment variable MUTTER_DEBUG_DISABLE_HW_CURSORS to
disable hardware cursors and force using software cursors.

Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2046
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2150>


(cherry picked from commit 56939abd)

Robert Mader authored Dec 18, 2021 and Robert Mader committed Dec 22, 2021

Process surface damage in the right order, simplifying the
calculations.

No functional change intended.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2168>

(cherry picked from commit f4717d0a)

Robert Mader authored Dec 18, 2021 and Robert Mader committed Dec 22, 2021

If no viewport is set, the neutral viewport is the surface size
without viewport destination size applied - i.e. transform and
scale applied to the buffer size. Change it accordingly, giving
us the same values we'd return in `get_width` in this case.

As result, this only changes cases where a viewport destination
size but no viewport source rectangle is set.

The change fixes exactly such cases, e.g. the Gstreamer Wayland
sink. Can be tested with: `gst-play-1.0 --videosink=waylandsink`.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2168>

(cherry picked from commit c1668116)

Robert Mader authored Dec 18, 2021 and Robert Mader committed Dec 22, 2021

Most clients nowadays switched to buffer damage, most notably Mesa
and Xwayland. Thus lets avoid the extra cost of allocating three
`cairo_region_t`s and doing some calculations.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2168>

(cherry picked from commit c498ae33)

In various places we retrieved the default seat from the ClutterBackend.
All the clutter backends implement this by calling
meta_backend_get_default_seat() which will then return
MetaBackendPrivate::default_seat.

Lets avoid this by fetching the default seat directly.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2002>

This will later be used to invalidate GPU state when closing device
nodes.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2147>
(cherry picked from commit 25ed64b6)

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2147>
(cherry picked from commit c87779f6)

During tear down, if anything teared down after the seat tries to get
the cursor renderer, we'd crash trying to get it as the seat would
already be gone. Avoid this by returning NULL when there is no seat.

It's assumed that any path that will happen during tear down that relies
on getting the cursor renderer will gracefully handle it not being
present, e.g. by relying on the cursor rendering cleaning up itself.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2147>
(cherry picked from commit ada52426)

This ensures we don't have any left over cursor GPU buffers (via
gbm_bo's) after destroying the corresponding gbm_device (owned by
MetaRenderDevice).

Fixes crashes with backtraces such as

  1) meta_drm_buffer_gbm_finalize at ../src/backends/native/meta-drm-buffer-gbm.c:450
  4) invalidate_cursor_gpu_state at ../src/backends/native/meta-cursor-renderer-native.c:1167
  9) update_cursor_sprite_texture at ../src/wayland/meta-wayland-cursor-surface.c:70
 10) meta_wayland_surface_role_apply_state at ../src/wayland/meta-wayland-surface.c:1869
 11) meta_wayland_surface_apply_state at ../src/wayland/meta-wayland-surface.c:832
 12) meta_wayland_surface_commit at ../src/wayland/meta-wayland-surface.c:993
 13) wl_surface_commit at ../src/wayland/meta-wayland-surface.c:1158
 14) ffi_call_unix64 at ../src/x86/unix64.S:76
 15) ffi_call at ../src/x86/ffi64.c:525

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2147>
(cherry picked from commit 4159c76a)

An incorrect assumption that after mode set there would be no pending
page flips was made. This meant that if there was a mode set, followed
by a page flip, if that page flip was for a CRTC on a now unused GPU,
we'd crash due to the renderer GPU data having already been freed. This
commit avoids that by keeping it alive as long as the page flips are
still in the air. It fixes crashes with backtraces such as

  0) meta_render_device_get_egl_display (render_device=0x0)
     at ../src/backends/native/meta-render-device.c:320
  1) secondary_gpu_state_free (secondary_gpu_state=0x1c8cc30)
     at ../src/backends/native/meta-onscreen-native.c:560
  2) meta_onscreen_native_dispose (object=0x1cb65e0)
     at ../src/backends/native/meta-onscreen-native.c:2168
  3) g_object_unref (_object=<optimized out>)
     at ../gobject/gobject.c:3540
  4) g_object_unref (_object=0x1cb65e0)
     at ../gobject/gobject.c:3470
  5) clutter_stage_view_finalize (object=0x1cbb450)
     at ../clutter/clutter/clutter-stage-view.c:1412
  6) g_object_unref (_object=<optimized out>)
     at ../gobject/gobject.c:3578
  7) g_object_unref (_object=0x1cbb450)
     at ../gobject/gobject.c:3470
  8) meta_kms_page_flip_closure_free (closure=0x1d47e60)
     at ../src/backends/native/meta-kms-page-flip.c:76
  9) g_list_foreach (list=<optimized out>, func=0x7fb3ada67111 <meta_kms_page_flip_closure_free>, user_data=0x0)
     at ../glib/glist.c:1090
 10) g_list_free_full (list=0x1cb4d20 = {...}, free_func=<optimized out>)
     at ../glib/glist.c:244
 11) meta_kms_page_flip_data_unref (page_flip_data=0x1c65510)
     at ../src/backends/native/meta-kms-page-flip.c:109
 12) meta_kms_callback_data_free (callback_data=0x227ebf0)
     at ../src/backends/native/meta-kms.c:372
 13) flush_callbacks (kms=0x18e2630)
     at ../src/backends/native/meta-kms.c:391
 14) callback_idle (user_data=0x18e2630)
     at ../src/backends/native/meta-kms.c

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2147>
(cherry picked from commit 287908e0)

This is so that it can unregister from it on tear down. The tracker owns
references to cursors too, but this cycle is already broken as the
backend calls 'g_object_run_dispose()' when tearing the cursor tracker
down.

Fixes a crash on shutdown.

Robert Mader authored Dec 30, 2021 and Robert Mader committed Jan 02, 2022

Otherwise we often advertise wrongly calculated modes.
Original patch from Jonas Ådahl and the xrandr project.

Closes https://gitlab.gnome.org/GNOME/mutter/-/issues/275

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2190>

The list returned by clutter_actor_get_children() is caller owned.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2193>


(cherry picked from commit 39415974)

When Xwayland was not initalized, we'd still clean things up. What this
accidentally meant was that the uninitialized display number 0 was
cleanud up, which very likely was main display of the host session.

What this meant in practice was that /tmp/.X11-unix/X0 was often
removed, causing every Flatpak X11 application to fail to start until
Xwayland was restarted nad the X0 socket file was restored.

Fix this in two ways: firstly only shutdown Xwayland if we ever started
it, i.e. if the X11 display policy was not 'disabled'. This should fix
the issue most of the times. Secondly only clean up the socket if it was
ever initialized. This should fix things if the socket creation failed,
as if it did, the name would be set.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2162>
(cherry picked from commit 0df12ebe)

This was always true and therefore resulting in all configurations being
considered as having a size, even when they don't change the size, such
as a state change.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2103>
(cherry picked from commit 67033b0a)

Signed-off-by: Ievgen Popovych <jmennius@gmail.com>
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2187>
(cherry picked from commit b596de26)

When rendering to a buffer that is not the stage view buffer, we can not
know where the buffer will be displayed on the screen. As a result we
also can not know what translation would need to be applied to culling.

This was causing glitches when the gnome-shell magnifier was applying
offscreen effects. ClutterOffscreenEffect causes MetaWindowGroup to be
rendered to an offscreen buffer at an offset, because it draws to a
slightly larger texture with an accordingly translated origin. This
translation then later is canceled out again when the offscreen buffer
is drawn. To meta_actor_painting_untransformed() however which only sees
the translation used when drawing to the buffer this looked like the
window group was being rendered at the offset. This then lead to
redraw_clip getting translated accordingly, resulting in wrong
coordinates used for culling.

Similarly this was leading to issues when taking area screenshots while
at 1x zoom.

Fixes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1678
Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4876
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2080>


(cherry picked from commit 450ab0fd)

Swap which windows are switched with which in the workspace switching
animation of the default plugin since this was backwards (resulting in
failures), but animate only those which are on either the current
workspace or the one to switch to. Have the animation span the entire
screen as intended.

Related: https://gitlab.gnome.org/GNOME/mutter/-/issues/2038
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2076>
(cherry picked from commit 888e69d3)

Since the completion callback (on_switch_workspace_effect_complete) sets
priv->tml_switch_workspace1 to NULL, the unref was trying to unref NULL,
and the reffed ClutterTimeline was not getting unreffed.

This could be triggered by rapidly switching workspaces, switching again
before the animation of the initial switch was done.

Found while working on #2038.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2120>
(cherry picked from commit 7bb0055a)

We save the window rect before going fullscreen to a dedicated variable,
so we can go back to the correct dimension. We also have a dedicated
variable for returning from other window states, e.g. maximized, and
this one we initialized when creating the MetaWindow. This meant that we
could always rely on this being up to date on X11 windows that were
mapped maximized or fullscreen.

What the commit that introduced the saved rect dedicated for going
unfullscreen missed was to initialize the new saved rectangle too when
creating the MetaWindow. This resulted in windows mapped as fullscreen
often ending up misbehaving when unfullscreening, as mutter would tell
them to unfullscreen to 0x0.

Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1786
Fixes: a51ad8f9
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2210>


(cherry picked from commit 13f35ab0)

Georges Basile Stavracas Neto authored Dec 27, 2021 and Robert Mader committed Jan 07, 2022

Next commits will reintroduce a certain behavior of stage
capturing that can only happen with DMA-BUF buffers. To
control this, add a new flag tp MetaScreenCastRecordFlags
for this behavior.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2186>

(cherry picked from commit 8fe45d77)

Georges Basile Stavracas Neto authored Dec 27, 2021 and Robert Mader committed Jan 07, 2022

When the before-paint function is executed, it's only purpose
is to check if there's any scanout queue, and immediately
record it if any.

However, since [1], we regressed in this specific case with the
introduction of an idle callback in the before-paint function.
The regression only happens when the PipeWire stream is using
DMA-BUF buffers, and it would operate as follows:

 1. In before-paint, when there's a scanout available, we queue
    an idle callback to capture the monitor. The idle callback
    (almost always) executes after the scanout is pulled from
    the stage view

 2. meta_screen_cast_stream_src_maybe_record_frame() is called
    by the idle callback. In the DMA-BUF case, it then runs
    meta_screen_cast_monitor_stream_src_record_to_framebuffer()

 3. In meta_screen_cast_monitor_stream_src_record_to_framebuffer(),
    because the stage view doesn't have a scanout anymore, it
    ends up calling cogl_blit_framebuffer() with the stage view
    framebuffer. This is the regression bug.

This regression presents itself in the form of the screencast
stream showing the desktop when there's an unredirected fullscreen
application window running.

Revert before-paint - and only that - back to immediately capturing
any available scanout. Only record these frames when the target
buffer is a DMA-BUF handle. Nothing is captured on before-paint if
the stream is not using DMA-BUF, since the regular paint routine
will handle these frames regularly post-paint.

[1] https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1914

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2186>

(cherry picked from commit 77d5c8c9)

If there are any pending updates, for example if we painted one of
multiple monitors but without having posted the update due to waiting
for another monitor to be painted, but before we paint all of them and
post the update, another hotplug event happens, we'd have stale pending
KMS update. When that update eventually would be processed, we'd try to
apply out-of-date updates which may contain freed memory.

Fix this by discarding any update when we're rebuilding the views. We
can be sure not to need any of the old updates since we're rebuilding
the whole content anyway.

Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1928
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2216>


(cherry picked from commit 8e91c629)

There was a sanity check that complained if there was still a "next
framebuffer" when disposing an onscreen. This is correct to complain
about under normal operation, as we always wait until receiving the page
flip callback before cleaning up the onscreen and their state.

However, when there are many hotplugs occurring, we might end up with
race conditions when the above sanity check is not valid: when we have
more than one monitor active, paint 1 one of them, but receive a hotplug
event before we paint the other(s), we will discard the already painted
onscreen before really issuing a page flip.

In this situation, we will have the "next framebuffer", but having that
is not a bug, it's a race condition, thus to not leak in this situation,
make sure to clean up the next framebuffer here too.

Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2081
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2225>


(cherry picked from commit bb6ae40a)

Commit 2289f561 ("monitor-manager: Don't apply unneeded orientation
changes") added an early return to handle_orientation_change () in case
the transform is unchanged.

But this did not take the correction of the transform for devices
with 90° mounted panels into account causing a desired orientation
change to get skipped if the new orientation matches the corrected
logical orientation from the previous transform setting.

Fix this by calling meta_monitor_crtc_to_logical_transform () on the
transform before comparing it, matching the
meta_monitor_crtc_to_logical_transform () call in
create_for_builtin_display_rotation ().

Related: https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1233
(cherry picked from commit cc9bb7c5)

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2223>

When checking panel orientation on logical monitors we should take
panel orientation transform to check it's properly applied, so ensure
that we're checking the right one.

(cherry picked from commit 54b36d77)

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2223>

Update NEWS.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2223>