Commit Graph

29799 Commits

Author SHA1 Message Date
Leônidas Araújo
1637272b8b Update Brazilian Portuguese translation 2022-09-06 00:27:01 +00:00
Matej Urbančič
76516319a9 Update Slovenian translation 2022-09-05 19:59:19 +00:00
Aurimas Černius
8d8fbdee45 Updated Lithuanian translation 2022-09-05 22:46:05 +03:00
Daniel Mustieles
311a79e150 Updated Spanish translation 2022-09-05 14:30:33 +02:00
Fran Dieguez
3f4736e044 Update Galician translation 2022-09-04 22:24:39 +00:00
Aleksandr Melman
7eff0c711e Update Russian translation 2022-09-04 20:46:13 +00:00
Florian Müllner
383c32e756 Bump version to 43.rc
Update NEWS.
2022-09-04 15:42:39 +02:00
Asier Sarasua Garmendia
830006d2ac Update Basque translation 2022-09-04 07:13:41 +00:00
Boyuan Yang
474a111406 Update Chinese (China) translation 2022-09-03 21:29:38 +00:00
Anders Jonsson
4ebc7e457b Update Swedish translation 2022-09-03 20:10:49 +00:00
Jonas Ådahl
be4e6484ae render-device: Unmake the EGLContext after checking whether hw accelerated
When creating a render device, we create a temporary EGLContext where we
then query the `GL_RENDERER` string to check whether the renderer is any
of the known software renderers. After we're done, we destroy the
context and move on.

This should be fine as according to specification eglDestroyContext(),
with the context being actually destroyed a bit later when it's no
longer current, but mesa, when running RK3399 (Pinebook Pro), this
results in a crash in a future eglMakeCurrent():

  #0 in dri_unbind_context () at ../src/gallium/frontends/dri/dri_context.c:266
  #1 in driUnbindContext () at ../src/gallium/frontends/dri/dri_util.c:763
  #2 in dri2_make_current () at ../src/egl/drivers/dri2/egl_dri2.c:1814
  #3 in eglMakeCurrent () at ../src/egl/main/eglapi.c:907
  ...

We can avoid this, however, by calling eglMakeCurrent() with
EGL_NO_CONTEXT on the EGLDisplay, prior to destroying, effectively
avoiding the crash, so lets do that.

Related: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7194
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2414
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2612>
2022-09-03 18:34:49 +02:00
Jonas Ådahl
b1a687ae57 tests/color: Test that we handle devices with no GAMMA_LUT
We should create the color device etc, but we shouldn't attempt to set
the GAMMA_LUT.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2611>
2022-09-03 14:41:36 +00:00
Jonas Ådahl
c724cb3720 tests/monitor-test-utils: Add way to make the GAMMA_LUT zero sized
This will be used to test devices that has no gamma capabilities.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2611>
2022-09-03 14:41:36 +00:00
Jonas Ådahl
b29604c3f8 tests/color-management: Move some utility functions higher up
That means they'll be re-usable without having to declare them on top.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2611>
2022-09-03 14:41:36 +00:00
Jonas Ådahl
255cd9a9ff color-device: Don't attempt to set GAMMA_LUT if LUT size is zero
This might happen on e.g. virtual machines. If we don't skip trying,
we'll end up crashing because we'll fail to create a look up table since
the size must be greater than zero to make any sense.

Fixes: baef39e603
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2415
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2611>
2022-09-03 14:41:36 +00:00
Hugo Carvalho
1bd06e645e Update Portuguese translation 2022-09-03 14:22:34 +00:00
Rūdolfs Mazurs
c89f262d39 Update Latvian translation 2022-09-03 13:14:12 +00:00
Piotr Drąg
79479868ae Update Polish translation 2022-09-03 12:56:51 +00:00
Danial Behzadi
bcb105eb68 Update Persian translation 2022-09-03 11:42:58 +00:00
Emin Tufan Çetin
2bde3402b3 Update Turkish translation 2022-09-03 06:46:22 +00:00
Balázs Úr
5e75d84181 Update Hungarian translation 2022-09-03 00:39:08 +00:00
Jonas Ådahl
1d4cdd1eb5 wayland/dma-buf: Always send modifiers when using surfaceless
We'll never scan out, which is why ADDFB2 is required otherwise, and we
won't enable the DMA buffer extension if
'EGL_EXT_image_dma_buf_import_modifiers' is missing, so send modifiers
in this case.

This also happens to avoid crashing when the GPU is null, since we'd
otherwise attempt to dereference it.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2578>
2022-09-02 20:43:05 +00:00
Jonas Ådahl
44dae3e160 kms: Only force legacy mode setting when using EGLStream
EGLStream is incompatible with atomic mode setting, but nvidia-drm when
using libgbm is not, so lets only deny using atomic mode setting when
the render device is an EGLStream based one.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2578>
2022-09-02 20:43:05 +00:00
Jonas Ådahl
58c3734d78 backend/native: Prepare render devices earlier than KMS
The type of render device used for a specific GPU affects the mode
setting backend that can be used, more specifically, when the render
device is an EGLStream based one, atomic mode setting isn't possible, as
page flipping is done via EGL, not via atomic mode setting commits.

Preparing the render devices before KMS devices means can make a more
informed decision whether to deny-list atomic mode setting for when
a certain GPU uses a EGLStream based render device instance.

This also means we need to translate mode setting devices to render node
devices when creating the render device itself, as doing it later when
creating the mode setting device is already too late.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2578>
2022-09-02 20:43:05 +00:00
Jonas Ådahl
1a0bd44694 backend: Hook up to the parent constructed()
Doing an early out in a constructed() is a bit awkward, and unexpected,
and makes it tricky to call the parents constructed() method (which we
didn't), so clean that up.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2578>
2022-09-02 20:43:05 +00:00
Marek Černocký
6ac106d3d3 Updated Czech translation 2022-09-02 22:25:24 +02:00
Jordi Mas
298c6a4d89 Update Catalan translation 2022-09-02 21:48:36 +02:00
Carlos Garnacho
7a1beb817d backends: Expose InputMapping D-Bus path to determine tablet mapping
Currently, the peripheral "output" setting will be unset if Mutter is
deciding automatically the mapped output of a tablet device. In that
case, gnome-control-center will have a hard time figuring out itself
the better output to show the tablet calibration UI, unless it's hand
held by Mutter.

Add this private D-Bus interface so that gnome-control-center can look
up the output as determined by Mutter to bring the missing harmony
between both. This interface consists of a simple method to get the
mapped output for a input device node.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2605>
2022-09-02 18:34:36 +00:00
Jonas Ådahl
60b382c4e3 x11/session: Set the context as client data on connection watch
It was already assumed to be the context, but we never set it as such,
so things just crashed instead. Now when we set it up correctly,
hopefully that won't happen anymore.

Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2406
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2267
Closes: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/5078
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2609>
2022-09-02 17:52:41 +00:00
Jonas Ådahl
b1cc591ef5 tests: Extract D-Bus runner as reusable python module
The D-Bus runner used by tests, including installed tests, is made to be
reusable from GNOME Shell. To do this, install it and the templates in
the pkgdatadir (e.g. /usr/share/mutter-APIVERSION/tests/), generate a
custom runner for the installed tests that uses the installed script and
templates, and change the non-installed original runner to use the
non-installed templates.

The end goal is to reuse the D-Bus session runner and templates used for
mutter when test running GNOME Shell.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1354>
2022-09-02 17:18:43 +00:00
Simon McVittie
4bc5833144 Consistently pass timestamp as uint64 when creating MetaStartupSequence
Passing the wrong type to a varargs function can cause a mismatch
between marshalling and unmarshalling on the stack, particularly on
32-bit architectures. This results in a build-time test failure.

Resolves: https://gitlab.gnome.org/GNOME/mutter/-/issues/2065
Signed-off-by: Simon McVittie <smcv@debian.org>
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2182>
2022-09-02 15:21:27 +00:00
Sebastian Wick
449327093e backends: Implement gamma LUT virtual function for MetaCrtcDummy
Fixes a crash in the nested backend.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2606>
2022-09-02 14:53:56 +00:00
Dor Askayo
01846ffd91 clutter/frame-clock: Simplify early presented event handling
A few calculations and assignments are done unnecessarily when the
last next presentation time is invalid. This increases the cognitive
complexity of the function for no reason.

No change in behavior.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2486>
2022-09-02 13:54:30 +00:00
Alessandro Bono
33dc8ca187 window: Notify about size-change when changing monitor
When moving to another monitor the window size may change in some
cases. While unconditionally notifying a size change is not always
correct, it animates the window when moved to another monitor in
GNOME Shell.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2558>
2022-09-02 13:12:09 +00:00
Yuri Chornoivan
98adff03da Update Ukrainian translation 2022-09-02 12:05:59 +00:00
Erico Nunes
932351c2bb backends: Set up swap_region without damage history
Following the EGL_KHR_swap_buffers_with_damage specification, the
surface damage used by eglSwapBuffersWithDamage does not need to
contain the damage history.
Rework that to initialize swap_region earlier, before appending the
damage history.
This may help optimizing the composition process in some cases (at least
on X11 when EGL_KHR_swap_buffers_with_damage is available) by not
accumulating additional regions as damaged unnecessarily.

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2241>
2022-09-02 11:06:04 +00:00
Goran Vidović
f305e4fd3f Update Croatian translation 2022-09-02 10:51:16 +00:00
Alessandro Bono
db0f12135e output-xrand: Plug a memory leak
Fixes: 99d84ae1cc
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2608>
2022-09-02 09:51:47 +00:00
Carlos Garnacho
a9ae43ff43 wayland: Ignore text_input requests from unfocused clients
This is what the protocol says we should do, and even though normally
an out of focus client should not have any reason to create IM requests,
there is a bit of a grey area around focus changes, as both the client
losing focus and the client gaining focus may respectively try to
disable/enable in an undetermined order.

Anyways, since in that situation the client losing focus is not aware
of the requests being ignored, the serial should always be incremented
in order not to break accounting of .done/.commit for that specific
client.

Fixes the IM focus being possibly "lost" after changing focus between
clients, if the race condition turned the odds in that direction.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2585>
2022-09-02 09:19:42 +00:00
Daniel van Vugt
cc729def90 monitor-config-store: Format <rate> using the C locale (. decimal point)
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2602>
2022-09-02 08:43:00 +00:00
Bilal Elmoussaoui
02110976dc workspace: Mark a string as translatable
This also replaces the wording as a desktop is no longer used to refer
to a workspace

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2375>
2022-09-02 08:04:31 +00:00
Bilal Elmoussaoui
7902fa3f9f core: Make sound player feature optional
Mutter can play sounds in some contexts and also provides an API
for libmutter users to do so using libcanberra internally.

In some specific use cases of Mutter, we would like to not depend
on libcanberra and not have any sound playing feature by default.

The changes keeps the sound player API but make it no-op if the
sound_player feature is disabled to not make it possible to break
a gnome-shell build.

See https://gitlab.gnome.org/GNOME/gnome-shell/-/merge_requests/2270
for relevant discussion

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2375>
2022-09-02 08:04:31 +00:00
Daniel van Vugt
f94189d4eb clutter/utils: Remove magic numbers from fixed point calculations
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2587>
2022-09-02 07:33:26 +00:00
Alberts Muktupāvels
3c50a9f8aa output-xrandr: Fix memory leak
Introduced in 99d84ae1cc.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2589>
2022-09-02 07:02:30 +00:00
Jonas Ådahl
eb764e0f0d barrier/x11: Let the seat bind the input events for us
Without doing this, the events bound by the X11 barrier manager will get
overridden by the ones later bound by the seat.

Fixes: 0debb24e12
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2394
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2604>
2022-09-02 06:35:00 +00:00
Jonas Ådahl
73bb333000 backend: Remove unused barrier vfunc
It was added in error, and has been unused since its introduction. Lets
correct the error.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2604>
2022-09-02 06:35:00 +00:00
Sebastian Keller
3e22eb35ed color-store: Fix two small leaks in init_profile_directory
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2603>
2022-09-01 21:39:38 +00:00
Sebastian Keller
48ff211762 color-store: Fix small file path leak in ensure_device_profile
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2603>
2022-09-01 21:39:38 +00:00
Sebastian Keller
57c7d75c43 color-device: Don't wrongly use Unknown vendor in ICC profile from EDID
If the vendor_name was previously successfully determined, we would end
up in the else case, overwriting it with "Unknown vendor" and leaking
the previous vendor_name.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2603>
2022-09-01 21:39:37 +00:00
Jonas Ådahl
9aa9cacb3d color-device: Pass calibration state separately when creating profile
This allows using two separate ICC profiles for one "color profile",
which is necessary to properly support color transform
calibration profiles from an EFI variable.

These types of profiles are intended to be applied using the color
transformation matrix (CTM) property on the output, which makes the
presented output match sRGB. In order to avoid color profile aware
clients making the wrong assumption, we must set the profile exposed
externally to be what is the expected perceived result, i.e. sRGB, while
still applying CTM from the real ICC profile.

The separation is done by introducing a MetaColorCalibration struct,
that is filled with relevant data. For profiles coming from EFI, a
created profile is practically an sRGB one, but the calibration data
comes from EFI, while for other profiles, the calibration data and the
ICC profile itself come from the same source.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2568>
2022-09-01 21:09:28 +00:00