Commit Graph

7291 Commits

Author SHA1 Message Date
Erik Kurzinger
056c45fe0c wayland/buffer: Try realizing EGLStream before EGLImage buffer
Currently, it is assumed that if querying the EGL_TEXTURE_FORMAT of a
Wayland buffer succeeds it is an EGLImage. However, this assumption will no
longer hold on upcoming versions of the NVIDIA EGL Wayland driver which
will include support for querying this attribute for EGLStream buffers as
well. Hence, we need to check if buffers are EGLStreams first.

Fixes #488
https://gitlab.gnome.org/GNOME/mutter/merge_requests/477
2019-03-14 14:39:52 -07:00
Pablo Barciela
318164779c boxes: Actually check for rectangle containment
Fixes condition duplicated:

          /* If a contains b, just remove b */
          if (meta_rectangle_contains_rect (a, b))
            {
              delete_me = other;
            }
          /* If b contains a, just remove a */
          else if (meta_rectangle_contains_rect (a, b))
            {
              delete_me = compare;
            }

Closes https://gitlab.gnome.org/GNOME/mutter/issues/480
2019-03-12 05:03:53 +01:00
Florian Müllner
7bd33e7b00 frame: Remove flashing support
It's now unused.

https://gitlab.gnome.org/GNOME/mutter/issues/491
2019-03-12 00:27:07 +00:00
Florian Müllner
f8d62da2dc bell: Always flash window instead of frame
Traditionally visual alerts were implemented by flashing the focus
window's frame. As that only works for windows that we decorate,
flashing the whole window was added as a fallback for client-decorated
windows.

However that introduces some confusing inconsistency, better to just
always flash the entire window.

https://gitlab.gnome.org/GNOME/mutter/issues/491
2019-03-12 00:27:07 +00:00
Marco Trevisan (Treviño)
303e02bdac monitor: Fix indentation style on foreach functions
https://gitlab.gnome.org/GNOME/mutter/merge_requests/488
2019-03-12 00:15:43 +00:00
Marco Trevisan (Treviño)
e2525f286d monitor: Fix indentation style on calculate_crtc_pos and friends
https://gitlab.gnome.org/GNOME/mutter/merge_requests/488
2019-03-12 00:15:43 +00:00
Marco Trevisan (Treviño)
9b8510ac56 monitor: Fix indentation style on calculate_supported_scales and friends
https://gitlab.gnome.org/GNOME/mutter/merge_requests/488
2019-03-12 00:15:42 +00:00
Florian Müllner
8a87e87a05 shadow-factory: Update shadow of attached modals
When commit 91c6a144da synced shadows with Adwaita, it removed the
shadow completely from attached modal dialogs. However Adwaita uses
the same shadow for all dialogs (modal or not), so do the same here.

https://gitlab.gnome.org/GNOME/mutter/issues/490
2019-03-12 00:05:42 +00:00
Florian Müllner
32504ae917 meta-monitor: Remove useless variable
Since commit f76b3edf9c, the variable is never set to TRUE and
we can simply remove it.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/475
2019-03-11 23:44:46 +00:00
Benjamin Berg
80ceeb2848 keybindings: Fix theoretical memory leak while ungrabbing
In the unlikely event that one tries to ungrab an action which does not
exist, a small leak could occur. Fix this by using g_autofree.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/478
2019-03-11 23:39:25 +00:00
Florian Müllner
1574099449 prefs: Change default of show-fallback-app-menu
The app menu always was a GNOME-only thing, so after it was removed this
cycle, assuming that it is not displayed by the environment is a better
default.

https://gitlab.gnome.org/GNOME/mutter/issues/493
2019-03-11 23:30:10 +00:00
Andrea Azzarone
8856a396fd sound-player: Don't deadlock in finish_cb
The function finish_cb can be called as a result of a call to ca_context_cancel
in cancelled_cb. This will result in a deadlock because, as per documentation,
g_cancellable_disconnect cannot be called inside the cancellable handler.

It is possible to detect if the call to finish_cb is caused by ca_context_cancel
checking if error_code == CA_ERROR_CANCELED. To avoid the deadlock we should
call g_signal_handler_disconnect instead g_cancellable_disconnect if this is the
case.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/474
2019-03-05 12:07:41 +00:00
Florian Müllner
7abceb434d x11-display: Split out restoring of active workspace
Splitting out the X11 display initialization from display_open() broke
restoring the previously active workspace in two ways:

 - when dynamic workspaces are used, the old workspaces haven't
   been restored yet, so we stay on the first workspace

 - when static workspaces are used, the code tries to access
   the compositor that hasn't been initialized yet, resulting
   in a segfault

Fix both those issues by splitting out restoring of the active workspace.

https://gitlab.gnome.org/GNOME/mutter/issues/479
2019-03-04 19:42:00 +00:00
Carlos Garnacho
3e472faf5c wayland: Minor refactor
We use the input_method on both branches, but only check for its existence
when enabling the text_input. The case of focusing out shouldn't happen in
practice as we couldn't have focused in ever before, but still make the
check one level above so it's clearer that the text_input's IM focus cannot
be enabled without an IM implementation.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/432
2019-03-04 18:17:08 +00:00
Carlos Garnacho
eabb789381 wayland: Only enable OSK if receiving .enable when already active
Enable the OSK if receiving .enable consecutively (i.e. the
ClutterInputFocus was already focused). We specifically want to avoid
enabling the panel just because of focus changes within a surface (where
the .disable request across focus change would previously unfocus the
ClutterInputFocus). Prior state should be preserved if possible in that
situation.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/432
2019-03-04 18:17:08 +00:00
Carlos Garnacho
3fd0e23ed9 clutter: Make ClutterInputFocus API to set panel state explicit
Before we just had API to toggle the OSK panel state. Make this API
generic so the upper layers may set the state as they see fit.
All callers have been updated.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/432
2019-03-04 18:17:08 +00:00
Marco Trevisan (Treviño)
8b09542fd9 boxes: Crop rectangle converting to integer with grow strategy
Reuse meta_rectangle_from_clutter_rect with growing strategy to properly cast
the clutter floating rectangle to integer MetaRectangle.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/3
2019-03-01 19:42:08 +01:00
Marco Trevisan (Treviño)
a8c972cd6b boxes: Add function to create a rectangle from floating clutter rect
Meta rectangles are integer based while clutter works in floating coordinates,
so when converting to integers we need a strategy.

Implement the shrink strategy by ceiling the coordinates and flooring the width,
and the grow strategy reusing clutter facility for this.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/3
2019-03-01 19:42:07 +01:00
Marco Trevisan (Treviño)
ceb4fe2151 wayland-tablet-tool: Downscale the sprite texture in FB mode
When using scaled framebuffer we need to downscale the texture size in order
to get the cursor properly drawn at its real size and in good quality.

https://bugzilla.gnome.org/show_bug.cgi?id=765011
https://gitlab.gnome.org/GNOME/mutter/merge_requests/3
2019-03-01 17:46:49 +00:00
Jonas Ådahl
ad5555bf42 clutter: Add API to get the resource scale of an actor
A clutter actor might be painted on a stage view with a view scale
other than 1. In this case, to show the content in full resolution, the
actor must use a higher resolution resource (e.g. texture), which will
be down scaled to the stage coordinate space, then scaled up again to
the stage view framebuffer scale.

Use a 'resource-scale' property to save information and notify when it
changes.

The resource scale is the ceiled value of the highest stage view scale a
actor is visible on. The value is ceiled because using a higher
resolution resource consistently results in better output quality. One
reason for this is that rendering is often not perfectly pixel aligned,
meaning even if we load a resource with a suitable size, due to us still
scaling ever so slightly, the quality is affected. Using a higher
resolution resource avoids this problem.

For situations inside clutter where the actual maximum view scale is
needed, a function _clutter_actor_get_real_resource_scale() is provided,
which returns the non-ceiled value.

Make sure we ignore resource scale computation requests during size
requests or allocation while ensure we've proper resource-scale on
pre-paint.

https://bugzilla.gnome.org/show_bug.cgi?id=765011
https://gitlab.gnome.org/GNOME/mutter/merge_requests/3
2019-03-01 17:46:49 +00:00
Marco Trevisan (Treviño)
3512647419 background: Use monitor scale to generate the texture at real monitor size
We need to use pixel size of the monitor in order to generate a valid
texture with full quality for current monitor

In spanned case the background should cover all the differently scaled monitors
thus we scale the texture up to the maximum scaling level and then we resample
it drawing only each side in the monitor it should occupy using the proper
scaling level.

In wallpaper mode (or color mode) for example we don't need to scale the area,
also the texture size we return should be unscaled, not to confuse
MetaBackgroundActor making it use more space than needed.

https://bugzilla.gnome.org/show_bug.cgi?id=765011
2019-03-01 17:46:49 +00:00
Marco Trevisan (Treviño)
42953a50bb stage: Queue redraw previous Rect only if it has changed
No need to queue redraw the same area multiple times, so we can avoid the
computation involved with it.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/3
2019-03-01 17:46:49 +00:00
Marco Trevisan (Treviño)
f4c2b69934 stage: Clamp to pixel the redraw clip taking care of the floored coords
When we floor the quad coordinates then we've also to enlarge the quad by the
difference between the floored value and the actual coordinate, otherwise
we'd end up in a smaller quad.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/3
2019-03-01 17:46:49 +00:00
Marco Trevisan (Treviño)
d7ec5d3022 stage: Add utility function to queue draw a ClutterRect
Remove duplicated code to clamp to pixel

https://gitlab.gnome.org/GNOME/mutter/merge_requests/3
2019-03-01 17:46:49 +00:00
Marco Trevisan (Treviño)
7442de81bb display: Add meta_display_get_monitor_scale
This will return the monitor scaling for the requested logical screen

https://bugzilla.gnome.org/show_bug.cgi?id=765011
2019-03-01 17:46:49 +00:00
Marco Trevisan (Treviño)
f76b3edf9c monitor: Define scale_steps globally
No need to compute the scale steps multiple times, since
it's just a defined value, so let's use a define for this
avoiding to pass around.

https://bugzilla.gnome.org/show_bug.cgi?id=782742
2019-03-01 17:46:49 +00:00
Jonas Dreßler
8d9a5e3c7e wayland/touch: Remove undelivered touches earlier
When the touch_down event was not delivered to Wayland clients, there's
no point in keeping the touchpoint in our list, so remove it early
inside update() instead of removing it after the touch ended.

This fixes a crash inside touch_handle_surface_destroy() where the
assertion to make sure the surface is removed fails because the
touch_count of the surface never reached 0. This in turn happened
because a new sequence was added, while a (already ended one) wasn't
removed from the touch->touches list before. This caused the touch
counter to get incremented by 1 while no new sequence was added to the
list (because Clutter reuses sequence IDs, the old sequence is equal to
the new one, i.e. the new sequence already is present in the list).

Fixes https://gitlab.gnome.org/GNOME/mutter/issues/200

https://gitlab.gnome.org/GNOME/mutter/merge_requests/426
2019-02-28 09:52:23 +00:00
Niels De Graef
1c6ea5d1db Use a consistent style for enum braces
https://gitlab.gnome.org/GNOME/mutter/merge_requests/361
2019-02-28 09:31:01 +01:00
Niels De Graef
b67394dcd1 clutter: StageWindow: Use G_DECLARE_INTERFACE()
It cuts away a bit of the GObject boilerplate, gives us support for
`g_autoptr`, and removes the typedef hack inside clutter-stage-window.c.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/380
2019-02-27 16:44:24 +00:00
Jonas Ådahl
9e75ce2ad7 monitor-config-store: Read system wide config files
Before introducing the new config store, the old monitor configuration
system read system wide monitor configuration files. Add back that
feature.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/253
2019-02-26 21:04:27 +00:00
Jonas Ådahl
7c807c6e2a org.gnome.Mutter.ScreenCast: Document what was added in API version 2
RecordWindow and the cursor-mode property are only available if the
advertised version is at least 2.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
9a20271f90 screen-cast-window-stream: Add support for cursor modes
Make the RecordWindow method also understand the 'cursor-mode' property.
For 'embedded' the cursor is drawn onto the pixel buffer using cairo,
otherwise it works similarly to how RecordMonitor deals with it.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
1760a6e560 screen-cast: Bump API version to 2
https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
8c2b805d88 screen-cast-src: Add helper to draw cursor sprite into pixel buffer
It makes sure the cursor sprite is correctly scaled and positioned in
stream coordinate space.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
2f01418d45 screen-cast-window: Add API to check if content has damage
https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
a76107a19f screen-cast-window: Add API to transform cursor state
To be used to translate absolute cursor positions to relative positions,
as well as to determine whether a cursor sprite is inside the stream or
not. It also helps calculating the scale the cursor sprite needs to be
scaled with to be in stream coordinate space.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
c80ba819e7 screen-cast-window-src: Keep a screen cast window instead of an actor
Practically it's the same object, but ideally, we should not deal
with "actors" anywhere here.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
14c6a347c1 screen-cast-window-src: Rename painted callback to after_paint
We'll add a before_paint soon, so better have the names somewhat match.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
11bd847896 screen-cast-window-stream: Don't set the position property
It was not meant to be set for window streams, and was set to (0, 0)
anyway. This removes the corresponding MetaScreenCastWindow API, as it's
not needed anymore.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
870f38542d screen-cast-window-stream: Set the size property to the logical size
As the stream size is the logical monitor size multiplied with the ceil
of the logical monitor scale, the corresponding logical size, which is
what should be passed via the size property on the D-Bus object, should
be the logical monitor size.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
c22edeff1f screen-cast-window-stream: Use initable to initialize
Move the initialization from _new() to an initable implementation. This
will allow us to initialize fields before MetaScreenCastStream
initializes.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
183de60d91 screen-cast-src: Move back MetaSpaType to C file
It's not needed by the monitor source anymore.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
176b508029 screen-cast-src: Change offscreen cursor helper to take a uint8_t *
Instead of a `uint32_t *`. Eventually we shouldn't assume pixels are
always 32 bit, and this gets rid of some casts while at it.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Jonas Ådahl
9be189daa7 screen-cast: Move cursor metadata setting to generic source
Make the monitor implementation do things strictly related to its own
source type, leaving the Spa related logic and cursor read back in the
generic layer, later to be reused by the window source type
implementation.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/413
2019-02-26 13:39:11 +00:00
Carlos Garnacho
8a9a78ec8a wayland: Check whether Xwayland window had a prior surface in generic code
Since commit 8df2a1452c (As pointed out by Robert Mader) we just happened
do this check when doing the first lookup for a Wayland surface for a
XWayland window, when we are later notifying upon surface creation we just
set the relation with no further checks.

The cases pointed out in the comment (eg. window changing decoration) might
presumably happen in a quick enough sequence that we have two scheduled
associations on the fly, so move this check to the more generic
meta_xwayland_associate_window_with_surface() which is called on both
immediate and delayed paths.

https://gitlab.gnome.org/GNOME/mutter/issues/361
2019-02-25 17:53:52 +00:00
Robert Mader
803ffc19a4 shaped-texture: Clean-up deprecated NPOT check
There are most likely no GNOME users left still using hardware that
does not support NPOT textures. Further more, they would crash much
earlier and never hit this code-path. So remove the unnecessary check
here.

https://gitlab.gnome.org/GNOME/mutter/merge_requests/447
2019-02-25 17:51:48 +01:00
Robert Mader
ddd2ce3a88 wayland/buffer: Fall back to CoglTexture2DSliced
XWayland creates buffers of the combined size of all connected displays.
This can, especially on older but still in use hardware, exceed the limits
of the GPU.

If that is the case, use `CoglTexture2DSliced` instead of `CoglTexture2D`

https://gitlab.gnome.org/GNOME/mutter/merge_requests/447
2019-02-25 17:51:44 +01:00
Jonas Ådahl
979e689278 renderer/native: Retry page flip after one vsync if EBUSY
We might fail to page flip a new buffer, often after resuming, due to
the FIFO being full. Prior to this commit, we handled this by switching
over to plain mode setting instead of page flipping. This is bad because
we won't be synchronized to the refresh rate anymore, but just the
clock.

Instead, deal with this by trying again until the FIFO is no longer
full. Do this on a v-sync based interval, until it works.

This also changes the error handling code for drivers not supporting
page flipping to rely on them returning -EINVAL. The handling is moved
from pretending a page flip working to explicit mode setting in
meta-renderer-native.c.

Fixes: https://gitlab.gnome.org/GNOME/mutter/issues/460
2019-02-25 15:48:29 +00:00
Jonas Ådahl
2c1ab8b3cc renderer/native: Remove legacy non-stage-view code
A renderer view will, under the native backend, since long ago always
have a logical monitor associated with it, so remove the code handling
the legacy non-stage view case.

https://gitlab.gnome.org/GNOME/mutter/issues/460
2019-02-25 15:48:29 +00:00
Jonas Ådahl
252e64a0ea wayland: Move surface texture ownership to MetaWaylandSurface
Prior to this commit, MetaWaylandSurface held a reference to
MetaWaylandBuffer, who owned the texture drawn by the surface. When
switching buffer, the texture change with it.

This is problematic when dealing with SHM buffer damage management, as
when having one texture per buffer, damaged regions uploaded to one,
will not follow along to the next one attached. It also wasted GPU
memory as there would be one texture per buffer, instead of one one
texture per surface.

Instead, move the texture ownership to MetaWaylandSurface, and have the
SHM buffer damage management update the surface texture. This ensures
damage is processed properly, and that we won't end up with stale
texture content when doing partial texture uploads. If the same SHM
buffer is attached to multiple surfaces, each surface will get their own
copy, and damage is tracked and uploaded separately.

Non-SHM types of buffers still has their own texture reference, as the
texture is just a representation of the GPU memory associated with the
buffer. When such a buffer is attached to a surface, instead the surface
just gets a reference to that texture, instead of a separately allocated
one.

Fixes: https://gitlab.gnome.org/GNOME/mutter/issues/199
2019-02-25 15:35:38 +00:00