Summary:
With nouveau driver it can happen that KWin gets frozen when first trying
to render with OpenGL. This results in a freeze of the complete desktop
as the compositor is non functional.
Our OpenGL breakage detection is only able to detect crashes, but not
freezes. This change improves it by also added a freeze protection.
In the PreInit stage a thread is started with a QTimer of 15 sec. If the
timer fires, qFatal is triggered to terminate KWin. This can only happen
if the creation of the OpenGL compositor takes longer than said 15 sec.
In the PostInit stage the timer gets deleted and the thread stopeed
again.
Thus if a freeze is detected the OpenGL unsafe protection is written into
the config. KWin aborts and gets restarted by DrKonqui. The new KWin
instance will no longer try to activate the freezing OpenGL as the
protection is set.
If KWin doesn't freeze the protection is removed from the config as
we are used to.
Check for freezes for the first n frames, not just the first
This patch changes the freeze detection code to detect freezes in the
first 30 frames (by default, users can change that with the
KWIN_MAX_FRAMES_TESTED environment variable). This detects
successfully the freezes associated to nouveau drivers
in https://bugzilla.suse.com/show_bug.cgi?id=1005323
Reviewers: davidedmundson, #plasma, #kwin, graesslin
Reviewed By: #plasma, #kwin, graesslin
Subscribers: luebking, graesslin, kwin, plasma-devel, davidedmundson
Tags: #plasma
Differential Revision: https://phabricator.kde.org/D3132
This fixes a regression found in TestSceneQPainter for restarting the
Compositor. The internal clients were never added to the Scene.
We cannot just call setupCompositing because there is a phase during
startup where setupCompositing already passes but the Compositor is not
yet fully initialized. Thus it could happen that it's called twice which
makes the Scene assert.
To solve that setupCompositing and finishCompositing are now overriden
in ShellClient and track whether they successfully called
setupCompositing.
We should be able to restart the Compositor also on Wayland.
As the Compositor might be terminated after Workspace we need to
ensure to not call into Workspace while doing finishCompositing.
Reviewed-By: bshah
Summary:
Rational: unredirect fullscreen windows is a weird beast. It's intended
to make fullscreen windows "faster" by not compositing that screen. But
that doesn't really work as KWin jumps out of that condition pretty
quickly. E.g. whenever a tooltip window is shown. KWin itself has a
better functionality by supporting to block compositing completely.
The complete code was full of hacks around it to try to ensure that
things don't break.
Overall unredirect fullscreen has always been the odd one. We had it
because a compositor needs to have it, but it never got truly integrated.
E.g. effects don't interact with it properly so that some things randomly
work, others don't. Will it trigger the screenedge, probably yes, but
will it show the highlight: properly no.
By removing the functionality we finally acknowledge that this mode is
not maintained and has not been maintained for years and that we do not
intend to support it better in future. Over the years we tried to make
it more and more hidden: it's disabled for Intel GPUs, because it used
to crash KWin. It's marked as an "expert" option, etc.
It's clearly something we tried to hide from the user that it exists.
For Wayland the whole unredirect infrastructure doesn't make sense
either. There is no such thing as "unredirecting". We might make use
of passing buffers directly to the underlying stack, but that will be
done automatically when we know it can be done, not by some magic is
this a window of specific size.
Test Plan:
Compiles, cannot really test as I am an Intel user who never
had that working.
Reviewers: #kwin, #plasma, #vdg
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D2180
Summary:
This change ensures that KWin doesn't crash in the QPainter scene if
the compositor gets restarted and if there are Wayland clients.
BUG: 365471
Test Plan:
Test case added to scene qpainter which triggers a restart
of the Compositor with a window being shown. Verifies that rendering
is correct afterwards.
Reviewers: #kwin, #plasma_on_wayland
Subscribers: plasma-devel, kwin
Tags: #plasma_on_wayland, #kwin
Differential Revision: https://phabricator.kde.org/D2185
Summary:
KWin needs to support restarting the OpenGL compositor in case of a
graphics reset event.
On Wayland the tricky part is that the applications should not notice
this. Most importantly KWin cannot just destroy the EGLDisplay and create
a new one. But this is how a restart works: the complete compositor gets
torn down and recreated - including the EGLDisplay.
This change moves ownership of the EGLDisplay to the Platform.
The AbstractEglBackend subclasses query the Platform whether there is
already an EGLDisplay. Only if there is no EGLDisplay the EGLDisplay is
created and only if no EGLDisplay is registered with Wayland the bind
is performed.
Another change is regarding the destruction: the AbstractEglDisplay does
no longer unbind the Wayland display and does no longer destroy the
EGLDisplay. The EGLDisplay is destroyed by the Platform - so very late
on application exit. The Wayland display is unbound when the Compositor
terminates.
Test Plan:
Limited testing with the added auto-test. This one needs to
be extended to fully verify that OpenGL applications continue to work.
But this requires build.kde.org to support OpenGL on Wayland.
Reviewers: #kwin, #plasma_on_wayland
Subscribers: plasma-devel, kwin
Tags: #plasma_on_wayland, #kwin
Differential Revision: https://phabricator.kde.org/D2202
This is responsible for plasmashell freezing when screen is locked.
The change itself was not wrong, but we first need to enable the
infrastructure to "fake" rendering when screen is locked/disabled, etc.
Summary:
So far when a ShellClient got unmapped ready_for_painting was set to
false. That is the ShellClient was treated in the same way as a not
yet shown window. It was completely excluded from painting, a close
animation impossible.
This change makes use of the functionality available in
Client::hiddenInternal(). The window is considered as hidden, thus
still excluded from e.g. getting input events, but could be rendered
any time as we still have a previous window pixmap (if referenced).
This allows to have it considered in the rendering pass, but effects
still cannot make use of it as that state is not yet exposed to the
effects.
Reviewers: #kwin, #plasma_on_wayland
Subscribers: plasma-devel, kwin
Tags: #plasma_on_wayland, #kwin
Differential Revision: https://phabricator.kde.org/D2083
On Wayland tear down the Workspace gets destroyed before the Compositor
gets destroyed. If the timer fires in that area, the Compositor would
crash. It's not possible to start the timer if the Workspace is
destroyed, thus it should also be stoped on Workspace destroy.
Reviewed-By: Bhushan Shah
Summary:
A new virtual method createOpenGLSafePoint is added to Platform.
This is invoked through the Compositor with a PreInit and a PostInit
argument pre and post creating the SceneOpenGL.
The Platform plugin can implement this and use it for detecting whether
creating the OpenGL compositor on this platform crashed in the past.
Thus it's the base for the openGLIsBroken platform check.
The x11 standalone plugin is the first to implement this functionality
using the previous code which was designed for X11.
This also means that a crash of the OpenGL compositor during init on
Wayland won't result in OpenGL being disabled.
Reviewers: #plasma
Subscribers: plasma-devel
Projects: #plasma
Differential Revision: https://phabricator.kde.org/D1582
Summary:
So far the OpenGL is unsafe check functionality in Compositor disabled
OpenGL compositing if glx is not available and we are in standalone X11
mode.
This is technically no longer correct for quite some time. Just because
GLX is not available doesn't mean that OpenGL doesn't work. We have an
EGL backend. So let's try to use that if glx is not available.
This change removes the check completely from Compositor. Instead the
standalone x11 plugin checks whether glx is available prior to createing
the glx backend. If not available it falls through to the egl backend.
Reviewers: #plasma
Subscribers: plasma-devel
Projects: #plasma
Differential Revision: https://phabricator.kde.org/D1581
Summary:
CompositingPrefs is only relevant for X11 standalone. It had some
"hacks" to make it not block Compositing on Wayland. Thus it was in
its current form not really useful.
Now all the functionality is provided through Platform with a default
implementation which is sensible for Wayland platforms.
The X11 standalone platform implements the new methods with the
Wayland checks removed.
In addition all calls to CompositingPrefs now go through the platform
directly and CompositingPrefs is completely dropped.
Reviewers: #plasma
Subscribers: plasma-devel
Projects: #plasma
Differential Revision: https://phabricator.kde.org/D1576
Summary:
This change reduces the windowing system specific code pathes. Instead
of checking whether we are on X11 or Wayland to decide whether
compositing is required, we just ask the Platform.
The default is true, only x11 standalone allows to not require
compositing.
Reviewers: #plasma
Subscribers: plasma-devel
Projects: #plasma
Differential Revision: https://phabricator.kde.org/D1575
Summary:
This ensures that frameRendered is also sent if we start a Wayland Server
on X11.
Reviewers: #plasma
Subscribers: plasma-devel
Projects: #plasma
Differential Revision: https://phabricator.kde.org/D1597
Summary:
Instead of checking the operation mode for X11, we check whether the
damage_handle is created. This fixes a bad damage warning for Wayland
windows in kwin_x11.
Reviewers: #plasma
Subscribers: plasma-devel
Projects: #plasma
Differential Revision: https://phabricator.kde.org/D1490
Summary:
If the Toplevel has a Wayland Surface we don't want an X11 damage
handle. Ideally we would just check for the Surface and get rid of
the operation mode check. But setupCompositing is called for Xwayland
clients before the surface is set.
Reviewers: #plasma
Subscribers: plasma-devel
Projects: #plasma
Differential Revision: https://phabricator.kde.org/D1486
Summary:
Instead of binding to the OperationMode whether the damage_handle
should be destroyed, we bind it to whether damage_handle got created.
Reviewers: #plasma
Subscribers: plasma-devel
Projects: #plasma
Differential Revision: https://phabricator.kde.org/D1483
So far if the Scene creation failed kwin_wayland went into a shutdown,
but didn't succeed because the thread to start Xwayland was already
running: it froze.
This change introduces a new signal in Compositor: sceneCreated. The
startup of Xwayland is bound to this signal. If it gets fired KWin can
startup Xwayland. If it does not get fired, KWin terminates correctly.
When screen is locked,
- No window other then screenlocker or inputmethods gets rendered
- Only screenlocker gets keyboard events
- Only screenlocker and inputmethods get mouse events
Things that are not secured/tested are :
- Touch events
- Global shortcuts for screenlocker
- Fallback/emergency screen not yet working
REVIEW: 126015
During Compositor tear down Xwayland is already destroyed. Thus it
doesn't make sense to try to delete the support properties: either
it freezes in xcb or it crashes because the connection is null.
At the same time we also ensure that the connection internally is
reset to null. Note: the one in kwinglobals.h caches and could cause
use-after-free errors. Any tear-down code must be migrated to
kwinApp()->x11Connection().
-use qstringliteral only when necessary (i.e. not in concat or comparison)
-use qbytearray instead of qstring when dealing with latin1 input and output (glplatform)
-use qstringref to extract numbers from strings (glplatform)
-define qt_use_qstringbuilder to optimize all string concatenations
-anidata: use ctor init lists, add windowType member initialization
REVIEW: 125933
The check whether Workspace is created is not sufficient. There's a
time when Workspace is already created but Compositor not ready yet
and where we can hit this code path.
This ensures that the Compositor doesn't perform compositing and doesn't
run the compositing timer if e.g. all outputs are dpms disabled. Thus
we don't render any more, don't trigger wakeups and block applications
from rendering if they properly implement the frame rendered callback.
The creation of PlasmaWindowInterface is moved from WaylandServer into
AbstractClient. This allows the sub classes to better control when to
create/destroy the Client.
For creation it's bound to becoming visible - that is Windows which are
only created but never shown are not announced at all.
For Client it's destroyed with the normal tear-down of a Client, for
ShellClient it's destroyed on unmapped (which also means a new one
will be created again in case of another mapping of the surface).
As a side effect, this works around the problem that ShellClients do not
yet get destroyed for QtWayland's menus (needs further investigation).
It's possible that the Workspace doesn't get created at all (e.g.
Xwayland failed to start). In that case we must ensure to not call into
Workspace calls during tear down.
only return early if we cannot create a
selectionwatcher, otherwise the claiming
code turns unreachable after the creation
of cm_selection
BUG: 347813
REVIEW: 123826
The ShellClient is a Toplevel subclass for a
KWayland::Server::ShellSurfaceInterface. It gets created when a new
ShellSurfaceInterface is created and destoryed when it gets unmapped.
So far the usage is still rather limited. The ShellClient is opened
at position (0/0). While it's possible to pass pointer events to it,
it's not yet possible to activate it, so no keyboard focus.
Replaces the functionality of the WaylandBackend and makes it available
to all backends by providing the functionality directly in
AbstractBackend. By default a backend is not ready and the implementation
must call setReady(true) to indicate that setup has finished
successfully. The compositor won't start till the backend indicates that
it is ready.
The aim is to be able to create a plugin for each of the backends.
The following directories are created:
* backends/drm
* backends/fbdev
* backends/wayland
* backends/x11
This new backend allows to start a kwin_wayland server nested on an
X-Server by using a normal X11 window as output. This allows testing
kwin_wayland without needing to start another Wayland server first.
The behavior is triggered by using new command line arguments:
--windowed
--x11-display=<:0>
With optional --width and --height arguments.
In this mode the WaylandBackend is not created at all.
So far the backend is not fully integrated yet and only the QPainter
backend supports this mode.
On Wayland we get the damage from the SurfaceInterface instead of
using a damage handle. This change ensures that the damage handle
interaction is only used on platform X11, while on Wayland we get
the damage from the SurfaceInterface directly.