Commit Graph

29195 Commits

Author SHA1 Message Date
Jonas Ådahl
4095c2431c barrier: Get rid of private instance struct
The struct is private to begin with, so no need for it.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2442>
2022-07-26 20:27:24 +00:00
Jonas Ådahl
3c6c1af0a0 barrier: Add GInitable based constructor method
This one does proper error reporting. Via Javascript, barriers are
constructed directly via GObject construction, which currently can't
handle error reporting, but when calling from C we can. However, if we
initialize using GInitable, and use that in our constructor method, once
gjs gains support for construction using GInitable, including the error
reporting, we'll automatically get proper error reporting to Javascript.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2442>
2022-07-26 20:27:24 +00:00
Jonas Ådahl
27b75d3c96 barrier: Modernize type implementation
This means using G_DECLARE*() to declare the type, using
`_get_private_instance()` everywhere, and other smaller cleanups.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2442>
2022-07-26 20:27:24 +00:00
Jonas Ådahl
1992aa1567 x11-display: Stop special casing barriers when iterating window table
In the past, barries were added to the window management X11 display
instance window table, and then special cased when iterating over the
list.

Since then, barriers, which are really part of the backend, has stopped
being added to the window hash table, instead being managed by the
backend. Lets clean up the left-over special casing that is no longer
needed.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2442>
2022-07-26 20:27:24 +00:00
Jonas Ådahl
0debb24e12 barriers: Make barriers fully part of the backend
Prior to this commit, barriers were created with a MetaDisplay pointer,
despite being entities related and owned by the backend. In the X11
case, it was also not hooked up to the backend X11 connection, but the
clutter one, meaning for example that the logic was active (but dormant)
also for the Xwayland connection.

Fix this by moving X11 barrier management and event processing fully to
the backend. Also replace passing a display pointer with passing a
backend pointer. Keep the display pointer around for a release, but mark
it as deprecated.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2442>
2022-07-26 20:27:24 +00:00
Neil Moore
877dc33545 shaped-texture: Move texture mipmap implementation to 'meta-texture-mipmap'
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2506>
2022-07-26 19:16:50 +00:00
Daniel van Vugt
16fa2100dc shaped-texture: Stop using MetaTextureTower and use GL mipmapping instead
But only in cases where it's needed; when we are displaying a texture at
less than half its native resolution. We also keep the existing
optimization that avoids mipmapping of textures while they're animating.

Fixes: https://gitlab.gnome.org/GNOME/mutter/-/issues/849
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2506>
2022-07-26 19:16:50 +00:00
Daniel van Vugt
b29a6c4d0c compositor: Return more details from meta_actor_painting_untransformed
This adds x_scale and y_scale information, as well as simplifying the
parameters.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2506>
2022-07-26 19:16:50 +00:00
Robert Mader
0c12903352 renderer-view: Drop bogus meta_monitor_transform_invert()
It papered over wrong `meta_rectangle_transform()` behaviour for
non-flipped output transforms. Also there is no obvious reason
why we would need inverted values here.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2459>
2022-07-26 18:41:43 +00:00
Robert Mader
0c34908132 cursor-renderer/native: Various fixes for rotated cursors
- Drop bogus `meta_monitor_transform_invert()`. It papered over
  wrong `meta_rectangle_transform()` behaviour for non-flipped
  output transforms.
- Update `scale_and_transform_cursor_sprite_cpu` to match the GL
  pipeline matrix in `MetaShapedTexture`, fixing several of the
  flipped cases. Note: the rotation applied is the one a client would
  need to apply to the buffer for a given monitor transform.
- While on it, drop a redundant `return`.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2459>
2022-07-26 18:41:43 +00:00
Robert Mader
d644a40cf5 shaped-texture: Swap 90 and 270 degree transforms in pipeline matrix
With `META_MONITOR_TRANSFORM` values matching their `WL_OUTPUT_TRANSFORM`
counterparts, the definition from the Wayland spec applies: the
`META_MONITOR_TRANSFORM` value tells us how the output was rotated
and that the buffer was drawn by the client to compensate for that.

The matrix describes the transformation from surface- to buffer-
coordinates, so the operation we need here is the same one that
the client applied (not from buffer- to surface-coordinates, i.e.
the inverse).

While on it fix `FLIPPED_90` and `FLIPPED_270` to use the correct
axes: flip on the x-axis, rotation on the z-axis.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2459>
2022-07-26 18:41:43 +00:00
Robert Mader
2ade26ebf8 boxes: Swap 90 and 270 degree transforms in meta_rectangle_transform()
`meta_rectangle_transform()` is used in the stack to *compensate* for a
`MetaMonitorTransform` applied to a output, not to apply it again.
Change the function accordingly.

Context:
Experimenting with direct scanout on offscreen-rotated outputs revealed
that the 90/270 degree cases were actually interchanged.
Further digging revealed that we use `meta_rectangle_transform()` with
those values swapped in every single case, papering over the issue.

Either a unintuitive and unexplained `meta_monitor_transform_invert()`
was added, in which case "flipped" values would be wrong, or, in case
of Wayland buffer transforms, the values were swapped by interpreting
the Wayland enums accordingly, see commit 8d9bbe10.

Swapping the 90/270 degree values in `meta_rectangle_transform()`:
1. fixes hardware cursor positioning with flipped output transforms
2. fixes rendering issues with offscreen-rotated flipped output transforms
3. allows us to drop unexplained `meta_monitor_transform_invert()`s in
follow-up commits
4. allows us to make `META_MONITOR_TRANSFORM` and `WL_OUTPUT_TRANSFORM`
enums match again (reverting 8d9bbe10, as already done)

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2459>
2022-07-26 18:41:43 +00:00
Robert Mader
29cafe6f6c monitor-transform: Fix transform() and drop relative_transform()
The following implicit definition for `transform()` did not
correctly apply:
```
a * b = c
c * invert(b) = a
```

Crucially the following did not apply for `FLIPPED-90`
and `FLIPPED-270`:
```
a * invert(a) = identity
```

Fix this by applying the operations, first the flip, then the
rotation, in this order and add tests to ensure correct results
for the requirement above.

Also drop `relative_transform()` as it only had a single user and
can be replaced by `transform()`:
```
invert(a) * b = c
a * c = b
```

As this is not very intuitive, ensure in tests as well.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2459>
2022-07-26 18:41:43 +00:00
Robert Mader
c6cf4bd9c3 backends/native: Fix checks for 'reflect-x' and 'reflect-y' bits
Checking for both bits at once means only one matching bit is
sufficient - very likely in case of `rotate-0'.

This fixes crashes on hardware that does not support 'reflect-'
bits when setting a flipped output transform.

While on it, also update the check for `reflect-y` instead of
`reflect-x` + `rotate-180`. They are logically equivalent,
however some hardware may support `reflect-y` but not both
other bits.

Fixes commit 4e3f3842a1

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2459>
2022-07-26 18:41:43 +00:00
Robert Mader
9e71278488 Revert "wayland/surface: Swap 90 and 270 degree transforms"
As testing of direct scanout revealed, `META_MONITOR_TRANSFORM`
does actually match `WL_OUTPUT_TRANSFORM` enums. The fact that
things rendered correctly with 90/270 degree values swapped
was because other parts of the stack got the interpretation
wrong, most notably `meta_rectangle_transform()`.
Thus lets revert this change and fix the stack accordingly.

This reverts commit 8d9bbe109b.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2459>
2022-07-26 18:41:43 +00:00
Carlos Garnacho
550b66d4e6 clutter/actor: Handle repicks when actors become reactive
The notification list in the GNOME Shell calendar popup triggers some
interesting interactions when closing a notification:

- Close button is clicked
- The notification animates to be hidden
- The next notification ends up hovered as a result of the animation
- The notification being hovered sets its close button as non-transparent
  and reactive
- The pointer is now again over a close button

At this point the reactiveness change should trigger a repick, so that
the new notification's close button is picked, and future button presses
are directed to it, but we do not handle this situation.

To fix this, handle actors becoming reactive so that if the closest
reactive parent has a pointer, it will be repicked again just in case
the pointer is over the newly reactive actor.

Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2364
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2532>
2022-07-26 14:18:13 +00:00
Carlos Garnacho
6dabdec10e clutter/stage: Add function to maybe invalidate pointers on an actor
The function that currently invalidates pointers over an specific actor
also asserts for the situations where this invalidation makes sense to
happen (i.e. the actor became unmapped, or non-reactive).

We want to have a function that is more forgiving, and that doesn't
enforce any guarantees about the pointer focus actually changing.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2532>
2022-07-26 14:18:13 +00:00
Jonas Ådahl
d985069542 kms/crtc: Don't report predicted gamma changes
We started to report resource changes using prediction when an update
had been successfully committed. While at it, gamma changes were
reported too, but this was problematic, as gsd-color will listen for the
MonitorsChanged D-Bus signal and naively set the gamma again, even if it
didn't change. There aren't currently any actual use cases for being
told when gamma changes from a prediction, so just ignore it and just
report privacy screen changes.

This avoids a feedback loop between mutter and gsd-color.

Fixes: 81b28a1d97
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2531>
2022-07-26 11:04:52 +02:00
Michel Dänzer
d928b6baa6 wayland: Clean up tablet manager state
Fixes memory leak:

==995170== 936 (40 direct, 896 indirect) bytes in 1 blocks are definitely lost in loss record 15,090 of 15,641
==995170==    at 0x48445EF: calloc (vg_replace_malloc.c:1328)
==995170==    by 0x4B211D0: g_malloc0 (gmem.c:155)
==995170==    by 0x4A56693: meta_wayland_tablet_manager_new (meta-wayland-tablet-manager.c:109)
==995170==    by 0x4A56693: meta_wayland_tablet_manager_init (meta-wayland-tablet-manager.c:126)
==995170==    by 0x4A3FA95: meta_wayland_compositor_new (meta-wayland.c:626)
==995170==    by 0x49C7FA7: meta_context_start (meta-context.c:412)
==995170==    by 0x10F065: main (mutter.c:148)

Fixes: 745cb67988 ("wayland: Initialize the MetaWaylandTabletManager")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2527>
2022-07-25 14:55:59 +00:00
Michel Dänzer
f3cd2b8ce7 wayland: Clean up output state
Fixes memory leak:

==995170== 383 (96 direct, 287 indirect) bytes in 1 blocks are definitely lost in loss record 14,733 of 15,641
==995170==    at 0x483F7B5: malloc (vg_replace_malloc.c:381)
==995170==    by 0x4B21178: g_malloc (gmem.c:125)
==995170==    by 0x4B395C0: g_slice_alloc (gslice.c:1072)
==995170==    by 0x4B0766D: g_hash_table_new_full (ghash.c:1071)
==995170==    by 0x4A4A8B4: meta_wayland_compositor_update_outputs (meta-wayland-outputs.c:483)
==995170==    by 0x4A4ABAB: meta_wayland_outputs_init (meta-wayland-outputs.c:716)
==995170==    by 0x4A3FA65: meta_wayland_compositor_new (meta-wayland.c:620)
==995170==    by 0x49C7FA7: meta_context_start (meta-context.c:412)
==995170==    by 0x10F065: main (mutter.c:148)

v2:
* Use meta_backend_get_monitor_manager. (Jonas Ådahl)

Fixes: 9a4783e364 ("Integrate the monitor manager with wayland")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2527>
2022-07-25 14:55:59 +00:00
Michel Dänzer
806e64dcb0 wayland: Clean up presentation-time state
Fixes memory leak:

==995170== 288 (96 direct, 192 indirect) bytes in 1 blocks are definitely lost in loss record 14,607 of 15,641
==995170==    at 0x483F7B5: malloc (vg_replace_malloc.c:381)
==995170==    by 0x4B21178: g_malloc (gmem.c:125)
==995170==    by 0x4B395C0: g_slice_alloc (gslice.c:1072)
==995170==    by 0x4B0766D: g_hash_table_new_full (ghash.c:1071)
==995170==    by 0x4A4F973: meta_wayland_init_presentation_time (meta-wayland-presentation-time.c:222)
==995170==    by 0x4A3FB04: meta_wayland_compositor_new (meta-wayland.c:635)
==995170==    by 0x49C7FA7: meta_context_start (meta-context.c:412)
==995170==    by 0x10F065: main (mutter.c:148)

Fixes: dccc60ec3e ("wayland: Implement stub presentation-time")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2527>
2022-07-25 14:55:59 +00:00
Michel Dänzer
b329dce070 wayland: Clean up xdg-activation state
Fixes memory leak:

==995170== 240 (48 direct, 192 indirect) bytes in 1 blocks are definitely lost in loss record 14,428 of 15,641
==995170==    at 0x48445EF: calloc (vg_replace_malloc.c:1328)
==995170==    by 0x4B211D0: g_malloc0 (gmem.c:155)
==995170==    by 0x4A3CDB3: meta_wayland_activation_init (meta-wayland-activation.c:383)
==995170==    by 0x4A3FB0C: meta_wayland_compositor_new (meta-wayland.c:636)
==995170==    by 0x49C7FA7: meta_context_start (meta-context.c:412)
==995170==    by 0x10F065: main (mutter.c:148)

Fixes: ec390b68c5 ("wayland: Implement the xdg-activation protocol")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2527>
2022-07-25 14:55:59 +00:00
Michel Dänzer
4ddefb9a69 wayland: Clean up scheduled_surface_associations hash table
Fixes memory leak:

==995170== 192 (96 direct, 96 indirect) bytes in 1 blocks are definitely lost in loss record 14,260 of 15,641
==995170==    at 0x483F7B5: malloc (vg_replace_malloc.c:381)
==995170==    by 0x4B21178: g_malloc (gmem.c:125)
==995170==    by 0x4B395C0: g_slice_alloc (gslice.c:1072)
==995170==    by 0x4B0766D: g_hash_table_new_full (ghash.c:1071)
==995170==    by 0x4A3F3A4: meta_wayland_compositor_init (meta-wayland.c:477)
==995170==    by 0x4E1F509: g_type_create_instance (gtype.c:1929)
==995170==    by 0x4E03DFC: g_object_new_internal (gobject.c:2011)
==995170==    by 0x4E0538C: g_object_new_with_properties (gobject.c:2181)
==995170==    by 0x4E05D40: g_object_new (gobject.c:1821)
==995170==    by 0x4A3F8C4: meta_wayland_compositor_new (meta-wayland.c:590)
==995170==    by 0x49C7FA7: meta_context_start (meta-context.c:412)
==995170==    by 0x10F065: main (mutter.c:148)

Fixes: 8df2a1452c ("wayland: Notify actively of xwayland window/surface associations")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2527>
2022-07-25 14:55:59 +00:00
Michel Dänzer
39fd32c362 wayland: Disconnect signals when Wayland compositor object is finalized
Fixes potential use-after-free during mutter shutdown, e.g.:

==993876== Invalid read of size 8
==993876==    at 0x4A4FCA3: meta_wayland_presentation_time_ensure_feedbacks (meta-wayland-presentation-time.c:373)
==993876==    by 0x4A3F07F: on_presented (meta-wayland.c:282)
==993876==    by 0x661B7E9: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.0)
==993876==    by 0x661A922: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.0)
==993876==    by 0x4DFF4BC: g_cclosure_marshal_generic_va (gclosure.c:1648)
==993876==    by 0x4DFE948: _g_closure_invoke_va (gclosure.c:893)
==993876==    by 0x4E17498: g_signal_emit_valist (gsignal.c:3406)
==993876==    by 0x4E176BE: g_signal_emit (gsignal.c:3553)
==993876==    by 0x51D9DB5: clutter_stage_view_notify_presented (clutter-stage-view.c:1226)
==993876==    by 0x499ACD2: frame_cb (meta-stage-view.c:83)
==993876==    by 0x499ACD2: frame_cb (meta-stage-view.c:43)
==993876==    by 0x50CAA41: notify_event (cogl-onscreen.c:175)
==993876==    by 0x50CAA41: _cogl_onscreen_notify_complete (cogl-onscreen.c:545)
==993876==    by 0x4A877F5: meta_onscreen_native_notify_frame_complete (meta-onscreen-native.c:211)
==993876==  Address 0x24b7be58 is 296 bytes inside a block of size 344 free'd
==993876==    at 0x484217B: free (vg_replace_malloc.c:872)
==993876==    by 0x4E1F88B: g_type_free_instance (gtype.c:2001)
==993876==    by 0x49C793C: meta_context_dispose (meta-context.c:675)
==993876==    by 0x4E037E0: g_object_unref (gobject.c:3636)
==993876==    by 0x4E037E0: g_object_unref (gobject.c:3553)
==993876==    by 0x10F145: glib_autoptr_clear_GObject (gobject-autocleanups.h:27)
==993876==    by 0x10F145: glib_autoptr_clear_MetaContext (meta-context.h:32)
==993876==    by 0x10F145: glib_autoptr_cleanup_MetaContext (meta-context.h:32)
==993876==    by 0x10F145: main (mutter.c:126)
==993876==  Block was alloc'd at
==993876==    at 0x483F7B5: malloc (vg_replace_malloc.c:381)
==993876==    by 0x4B21178: g_malloc (gmem.c:125)
==993876==    by 0x4B395C0: g_slice_alloc (gslice.c:1072)
==993876==    by 0x4B39C29: g_slice_alloc0 (gslice.c:1098)
==993876==    by 0x4E1F544: g_type_create_instance (gtype.c:1901)
==993876==    by 0x4E03DFC: g_object_new_internal (gobject.c:2011)
==993876==    by 0x4E0538C: g_object_new_with_properties (gobject.c:2181)
==993876==    by 0x4E05D40: g_object_new (gobject.c:1821)
==993876==    by 0x4A3F864: meta_wayland_compositor_new (meta-wayland.c:585)
==993876==    by 0x49C7FA7: meta_context_start (meta-context.c:412)
==993876==    by 0x10F065: main (mutter.c:148)

Fixes: 2ce3a050f0 ("wayland: Wire up presentation-time machinery")
Fixes: 8cff3b84f7 ("wayland/compositor: Process frame callbacks on 'after-update'")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2527>
2022-07-25 14:55:59 +00:00
Jonas Ådahl
81b28a1d97 kms: Notify about privacy screen changes via predictions
When we change the privacy screen, we added a result listener to the KMS
update object to notify the upper layer about the privacy screen state
change. This was slightly awkward as one might have changed the state
multiple times for a single update, thus it was necessary to remove any
old result listeners to an update before adding a new one.

Doing this will not be possible when updates are fully async and managed
by the KMS impl device.

To handle this, instead make the post-commit prediction notify about
changes that happens in response to a successfully committed update. We
already predicted the new privacy screen state, so the necessary change
was to plumb the actual change into a callback which emits the signal if
there actually was a privacy screen change.

This will then be communicated via the same signal listener that already
listens to the 'resources-changed' signal.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2340>
2022-07-25 11:02:35 +00:00
Jonas Ådahl
1b0aa0b3ad kms: Rename MetaKmsUpdateChanges to MetaKmsResourceChanges
The name had a bit conceptual conflicts with MetaKmsUpdate, as it shared
its namespace but had no relation to it. Fix this by renaming it
MetaKmsResourceChanges (and the corresponding META_KMS_UPDATE_CHANGE_*
to META_KMS_RESOURCE_CHANGE_*). The term "resource" is used since that's
already used in the signal, and the fact that the changes partly comes
from changes in the DRM resource as retrieved by drmModeGetResources.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2340>
2022-07-25 11:02:35 +00:00
Carlos Garnacho
633635d608 backend: Fix cursor visibility on X11 with only tablets as input
Since tablets on X11 drive the Virtual Core Pointer as every other
pointing device, we should consider it as a pointer device when
determining whether the pointer cursor should be visible.

Fixes the cursor being hidden if there are only tablets for input.

Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2283
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2518>
2022-07-25 10:12:50 +00:00
columbarius
b17a510309 screen-cast-stream: Timestamp buffers
With spa_meta_header we can timestamp our buffers so clients can relate
and synchronize media from different sources.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2516>
2022-07-21 22:29:45 +00:00
columbarius
30b5229e0e screen-cast-stream: Use SPA_META_Header
With this header we can set a flag to signal that the whole buffer is
corrupt and should be ignored. With this we can cater to all cases:
* Window buffer fine, but cursor broken:
  Use the spa_meta_cursor properties like id or offset accordingly
* Window buffer broken, but cursor fine:
  Use the chunk flags
* Both are broken / the dequeued buffer is not usable
  Use the spa_meta_header flag

Additionally clients can now check if a buffer contains spa_meta_header
data and can thus only check for the new or the old behaviour.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2516>
2022-07-21 22:29:45 +00:00
Michel Dänzer
10c7ab2dec cursor-renderer/native: Fix type in unset_crtc_cursor_renderer_privates
Since the commit below, meta_crtc_kms_get_cursor_renderer_private has
returned a CrtcCursorData pointer, but this code was still treating it
as a MetaDrmBuffer pointer.

Fixes: fea8ebcca9 ("cursor-renderer/native: Store struct in CRTC private")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2524>
2022-07-21 21:03:18 +00:00
Michel Dänzer
709d06e89e wayland/surface: The same sub-surface is not a valid sibling
The Wayland protocol documentation at
https://wayland.app/protocols/wayland#wl_subsurface:request:place_above
says:

 The reference surface must be one of the sibling surfaces, or the
 parent surface. Using any other surface, including this sub-surface,
 will cause a protocol error.

v2:
* Leave it as a single line. (Georges Basile Stavracas Neto)

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2525>
2022-07-21 17:03:12 +02:00
Michel Dänzer
ddb7f84291 wayland/surface: Simplify is_child & is_sibling
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2525>
2022-07-21 17:03:12 +02:00
Carlos Garnacho
f685567afe clutter: Let ClutterInputMethods pass the anchor position for preedit text
This is an API break, since users (i.e. GNOME Shell) provide the input method
implementation.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2517>
2022-07-19 11:04:56 +00:00
Carlos Garnacho
ffac294520 clutter: Pass anchor position to ClutterInputFocus
Some implementations may need this (namely, the one that drives
the Wayland protocol), so pass this along from the IM events.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2517>
2022-07-19 11:04:56 +00:00
Carlos Garnacho
190301f66b clutter: Make ClutterIMEvent able to store preedit anchor position
This is something that is exported through the protocol, but we don't
pass forward. Make the events able to store this anchor so that we
can do that.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2517>
2022-07-19 11:04:56 +00:00
Daniel van Vugt
3d94c7cc2d clutter/actor: Do clutter_actor_destroy_all_children without an iterator
`clutter_actor_iter_destroy` will try to match up the iterator's `age`
with that of the parent ("root") actor:

```
g_return_if_fail (ri->age == ri->root->priv->age);
```

In a simple actor graph that's completely reasonable but somewhere in the
more complex graph of gnome-shell the parent's `age` was skipping ahead
faster than that of the iterator. This could happen in theory if the
destroy indirectly leads to more children being destroyed than the
iterator has visited.

So there's no evidence of actual corruption, only the age check might
fail in a `clutter_actor_iter_destroy` loop because the age check itself
can't handle all possible valid scenarios.

Since our only mandate is to destroy all children, we can do that reliably
without an iterator and thus without assuming anything about the parent's
`age` counter.

Fixes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/4747
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2074>
2022-07-18 17:13:01 +08:00
Daniel van Vugt
4a70a73e42 clutter/actor: Deduplicate child destruction loop
It's identical in clutter_actor_real_destroy and
clutter_actor_destroy_all_children, other than the latter having
some extra error checking.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2074>
2022-07-18 17:13:01 +08:00
columbarius
82fd5cbdbd screen-cast-stream: Use flag to signal failed recording
PipeWire supports flags to signal a corrupted buffer. We should use the
flag SPA_CHUNK_FLAG_CORRUPTED for `chunk->flags` instead of setting
`chunk->size = 0` since the size isn't well defined for arbitrary dmabufs
and should be set to 0.
Sadly clients like obs are using a chunk size of 0 to decide if a buffer
should be imported. Thus we should offer both until clients are using
the flag.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2323>
2022-07-15 02:03:44 +02:00
José Expósito
9dd6268d13 wayland/pointer: Send high-resolution scroll data
Upgrade the seat protocol to version 8 and change the scroll handler to
support new clients.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1962>
2022-07-14 22:29:56 +00:00
José Expósito
5171e5b795 backends/native: Migrate discrete scroll
Change meta_seat_impl_notify_discrete_scroll_in_impl to receive 120
based values and report high-resolution scroll values as smooth scroll.

Notify discrete scroll only when the accumulated value reach 120.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1962>
2022-07-14 22:29:56 +00:00
José Expósito
a1c50194a9 backends/native: Migrate continuous scroll
In order to get the delta X/Y value of the
LIBINPUT_EVENT_POINTER_SCROLL_FINGER
or LIBINPUT_EVENT_POINTER_SCROLL_CONTINUOUS events the new function
libinput_event_pointer_get_scroll_value should be used instead of
libinput_event_pointer_get_axis_value.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1962>
2022-07-14 22:29:56 +00:00
José Expósito
5d884d49b5 backends/native: Deprecate LIBINPUT_EVENT_POINTER_AXIS
Ignore deprecated LIBINPUT_EVENT_POINTER_AXIS events and handle
LIBINPUT_EVENT_POINTER_SCROLL_WHEEL,
LIBINPUT_EVENT_POINTER_SCROLL_FINGER and
LIBINPUT_EVENT_POINTER_SCROLL_CONTINUOUS instead.

The scroll source is now encoded in the event type making
libinput_event_pointer_get_axis_source and translate_scroll_source
redundant.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1962>
2022-07-14 22:29:56 +00:00
Carlos Garnacho
80db591534 build: Update Wayland dependency to 1.21
This is needed for wl_seat v8.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1962>
2022-07-14 22:29:56 +00:00
José Expósito
cddf30c1df ci: Update Wayland to v1.21.0
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1962>
2022-07-14 22:29:56 +00:00
Pascal Nowack
ab3931fec8 backends/screen-cast-stream-src: Set correct stride when using dmabufs
Currently, mutter hardcodes the stride of all recorded images in a
screencast to width * 4.
While this stride is valid for general memory buffers, it is not
necessarily valid for dmabuf buffers, as the driver may need a specific
alignment of the stride.
This can currently result into import failures, e.g. when using the AMD
driver.

To fix this issue, retrieve the actual dmabuf stride, when using dmabuf
screencasts, and set it as the stride value in the PipeWire buffer.

Fixes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1913
Fixes: https://gitlab.gnome.org/GNOME/gnome-remote-desktop/-/issues/101
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2513>
2022-07-14 20:09:07 +02:00
Dor Askayo
f037c9df66 wayland/dma-buf: Don't require modifiers support for scanout tranche
When building the list of formats to be sent as part of the scanout
tranche, avoid requiring modifier support by the DRM driver for
formats relying on implicit modifiers (DRM_FORMAT_MOD_INVALID).

Specifically, the previous check required the DRM driver to have
advertised some modifier support for the given format in its
IN_FORMATS KMS plane property, regardless of modifier it was. If it
hadn't, the format was left out of the list of formats to be sent
in the scanout tranche.

When no formats remained to be sent in the scanout tranche, the
tranche simply wasn't sent.

This resulted in the scanout tranche never being sent for GPUs where
modifiers aren't supported. In those cases, no formats are advertised
using the IN_FORMATS property, and thus the list of formats to be sent
in the scanout tranche remained empty.

Since Mesa doesn't use scanout-compatible buffers for native Wayland
clients unless specifically requested to do so using the "scanout"
tranche flag, it effectively means that direct scanout of native
Wayland clients wasn't supported for GPUs without modifiers support.

Sending a tranche with formats paired with the implicit modifier
(DRM_FORMAT_MOD_INVALID) is both allowed by the protocol and is
already done by default for GPUs with modifiers support, unless the
experimental support for explicit modifiers is enabled in Mutter.

So instead of requiring modifiers to be supported for each format
being evaluated for the scanout tranche, when processing formats
which rely on implicit modifiers, only check if the format in
question is supported by the DRM driver for scanout on the primary
plane.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2510>
2022-07-12 18:53:30 +03:00
Robert Mader
1a56a53909 wayland/actor-surface: Consider clones in is_on_logical_monitor()
While the check for `clutter_actor_has_mapped_clones` clearly indicates
an intention to take clones into account, the following code
does not do so, likely because it predates the introduction of
`clutter_actor_is_effectively_on_stage_view()`.

Switch to that newer API in order to take clones into account. This
avoids unnecessary `wl_surface_send_enter()` and `wl_surface_send_leave()`
events when entering the overview, reducing client work.

This also avoids unnecessarily allocating a `cairo_region_t`.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2502>
2022-07-11 18:18:56 +00:00
Robert Mader
0f8f30c93f window-actor/wayland: Skip unnecessary set_child_at_index() calls
`clutter_actor_set_child_at_index()` is far from a no-op, even if
the current index is equal to the new one - presumably for good
reasons. For the use-case here we want it to be a no-op though, so
skip calling it if the index already matches.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2501>
2022-07-11 17:23:10 +00:00
Robert Mader
57881ba612 wayland/subsurface: Move actor unparenting back to rebuild_surface_tree()
Unparenting the surface actor when the subsurface object is destroyed
has several issues:
 - subsurface actors can be unparented while a close animation is
   still ongoing, breaking the animation for e.g. Firefox.
 - adding and removing the actor to/from the parent is not handled in
   one place, making the code harder to follow.
 - if the destroyed subsurface had children of its own, they potentially
   stick around until a surface-tree rebuild. This makes the Firefox
   hamburger menu not close with the "compositor" backend.

Move the unparenting back to
`meta_window_actor_wayland_rebuild_surface_tree()` and instead just
notify the parent of a state change, if it still exist. This will ensure
a correct mapping between the subsurface node tree and the flat surface
actor list. In case of the closing animation the parent will already be
removed and the call is skipped.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2501>
2022-07-11 17:23:10 +00:00
Florian Müllner
60e13fde09 Bump version to 43.alpha
Update NEWS.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2505>
2022-07-10 13:26:00 +02:00