For Xwayland we need to have the Scene (and EglDisplay) created prior
to starting Xwayland and having X11. This requires creating the
Compositor before creating Workspace and starting Xwayland.
To support this the startup of Compositor is split into two parts:
prior and after Workspace creation.
The change might also be interesting for the kwin_x11 case as it could
result in the compositor being up in a quicker way.
Don't start the composite timer at the end of performPaint() when a
buffer swap is pending and vsync is enabled. Instead set
m_composeAtSwapCompletion to true so performPaint() gets called again
as soon as the swap completes.
This makes the repaint cycle look like this:
scene->paint()
SwapBuffers()
Process events
·
·
Swap completes
Fetch and reset damage (if applicable)
scene->paint()
SwapBuffers()
Process events
·
·
Swap completes
...
This results in a noticeable improvement in animation smoothness with
drivers that support GLX_INTEL_swap_event, since we're now able to
consistently render at the monitor refresh rate.
The Compositor is destroyed before the Client and Decorations are
destroyed on shutdown. This meant the Decorations reacted needlessly
on the alpha channel supported. E.g. Aurorae recreated the Decoration
and most likely crashed in Qt.
With this change the signal gets disconnected and the Decorations
just don't do anything.
Instead of getting size from displayWidth() and displayHeight() use
the information we have from Screens. This means there is only one
place to have the information and by that we can ensure that all
components use the same data to rely on. displayWidth/displayHeight
seem to provide the wrong information when unplugging an output
without disabling the output. This results in rendering artefacts.
But KWin::Screens has the correct information available.
Move the buffer-swap-pending state from the compositing backends to
the Compositor class. The Compositor is the only class that needs to
access the state, and this way it to do it without calling through
a chain of virtual functions. This commit adds two new functions to
Compositor; aboutToSwapBuffers() and bufferSwapComplete(). The
backends call these functions to set and reset the buffer-swap-pending
state.
This commit also renames a number of functions and variables to make
their meaning clear.
The act of promoting the contents of the back buffer to become the
contents of the front buffer is referred to as posting the buffer,
presenting the buffer, or swapping the buffers; rendering the buffer
is what paintScreen() does.
Create and import X sync fences into GL and use them to synchronize
the kwin command stream with the X command stream.
This prevents damaged windows from being composited by kwin before
the rendering that triggered the damage events have finished on
the GPU.
Requires GL_EXT_x11_sync_object.
Tested-by: Marco Martin <notmart@gmail.com>
The Wayland event queue is moved into a dedicated thread and a
new class is created for just creating the connection and listening
for events. The WaylandBackend creates the thread and uses an event
queue for the main thread.
REVIEW: 119761
So far the DecorationRenderer got destroyed and recreated after
the signal compositing toggled was emitted. But that's too late for
e.g. the OpenGL Textures to be destroyed. So lets trigger the destroy
directly in setup/finish compositing. The new renderer will still be
created after the compositing toggled is emitted - we don't want to
have it recreated for the still active compositer type.
NOTE: this is not working completely yet, lots of code is still ifdefed
other parts are still broken.
The main difference for the new decoration API is that it is neither
QWidget nor QWindow based. It's just a QObject which processes input
events and has a paint method to render the decoration. This means all
the workarounds for the QWidget interception are removed. Also the paint
redirector is removed. Instead each compositor has now its own renderer
which can be optimized for the specific case. E.g. the OpenGL compositor
renders to a scratch image which gets copied into the combined texture,
the XRender compositor copies into the XPixmaps.
Input events are also changed. The events are composed into QMouseEvents
and passed through the decoration, which might accept them. If they are
not accpted we assume that it's a press on the decoration area allowing
us to resize/move the window. Input events are not completely working
yet, e.g. wheel events are not yet processed and double click on deco
is not yet working.
Overall KDecoration2 is way more stateful and KWin core needs more
adjustments for it. E.g. borders are allowed to be disabled at any time.
Similar to the already existing DBusInterface wrapper for the
org.kde.KWin interface a new CompositorDBusInterface is introduced for
org.kde.kwin.Compositing.
That way the DBus interface is split from the implementation and DBus
specific methods are no longer required in the Compositor class.
The deprecated DBus methods
* toggleCompositing(bool)
* setCompositing(bool)
are removed.
REVIEW: 118463
The Scene connects to the deletedRemoved signal and tries to remove the
Deleted from it's list of Windows, which will fail because a SceneWindow
doesn't get created for an existing Deleted in the Scene setup.
It doesn't make sense to create a SceneWindow for a Deleted as there is
no pixmap around anyway. Thus the best approach is to ensure that there
are no Deleted by just discarding all prior to creating the Scene.
BUG: 333275
REVIEW: 117557
So far the Unmanaged got released after an XCB_UNMAP_NOTIFY. This event
gets created after xcb_unmap_window or after xcb_destroy_window. In the
latter case the window is already distroyed and any of KWin's cleanup
calls will cause a BadWindow (or similar) error.
The idea to circumvent these errors is to try to wait for the
DESTROY_NOTIFY event. To do so the processing of the release is slightly
delayed. If KWin gets the destroy notify before the delay times out the
Unamanged gets released immediately but with a Destroy flag. For this a
new enum ReleaseToplevel is introduced and Unmanage::release takes this
as an argument instead of the bool which indicated OnShutdown. Also this
enum is added to Toplevel::finishCompositing so that it can ignore the
destroyed case and not generate an error.
REVIEW: 117422
The Wayland Backend watches the socket it uses for communicating with the
Wayland compositor. If the socket is removed we have to perform a kind of
emergency stop. The backend tears down all data structures created from
the Wayland display and emits a signal that the system compositor died.
In addition the Wayland Backend starts to monitor the XDG_RUNTIME_DIR for
the socket to be added again. If the socket is created again the backend
reinitializes the Wayland connection.
This also requires the Compositor to restart. Therefore it connects to
the new signals emitted by the Wayland Backend to stop and start
compositing.
KWin already has a de facto OpenGL 2 dependency through QML. Combined
with the fact that the OpenGL 1 backend is basically unmaintained and
also unused, it's better to remove it for the new major release.
This change includes:
* Removal of cmake option KWIN_BUILD_OPENGL_1_COMPOSITING
* Removal of KWIN_HAVE_OPENGL_1 compile option and all code
ifdef'ed with it (partially removal of if-else constructs)
* Removal of CompositingType::OpenGL1Compositing (flags are kept
as a core flag should get introduced)
* Driver recommendation for OpenGL1Compositing changed to XRender
(should be evaluated whether the drivers can provide GL2)
* Removal of configuration option "GLLegacy"
* Removal of fooMatrix function in kwinglutils
* Removal of ARBBlurShader
* Removal of legacy code path in GLVertexBuffer
* Removal of GLShaderManager::disable
* if-blocks with ShaderManager::instance()->isValid() removed
REVIEW: 116042
In the Wayland world we need to have a compositor. This means we have to
enforce that the compositor is running. If the setup fails we have to
quit, because it doesn't make any sense any more to be running.
A new method requiresCompositing() is added to the Application. If it
returns true the useCompositing option will always return true and the
unredirect fullscreen option will always return false. That way
compositing is enforced at startup and cannot end by unredirecting.
In addition this method is checked if actions are performed which would
suspend compositing. E.g. the shortcut to toggle compositing. Restarting
the compositor is still possible in order to change the selected
compositing backend without a restart. But if it fails KWin will quit.
This compositor uses only the QPainter API to perform rendering. The
window's X Pixmap is mapped to a QImage using XShm. As rendering backend
a QImage is used.
The new compositing type "QPainterCompositing" is introduced. Effects
need to be adjusted to explicitly check the compositing type and no
longer assume the compositing type is XRender if it's not OpenGL.
This compositor can be selected with using "Q" as the value for
KWIN_COMPOSE env variable or setting the config value to "QPainter".
The GUI is not yet adjusted to select this compositor.
The QPainter scene provides currently the following features:
* 2D transformations (translation and scalation)
* opacity modifications
* rendering of decorations (new PaintRedirector sub class)
* rendering of shadows
* rendering of effect frames
* rendering to a Wayland surface
The following features are currently not provided:
* saturation changes
* brightness changes
* 3D transformations
* rendering to X Overlay window
* offscreen rendering (e.g. needed for screen shot effect)
* custom rendering in the effects to the current back buffer
Only the X based Scenes need an overlay window, so the Compositor doesn't
need to check for it in the Wayland case.
OverlayWindow is moved from OpenGLBackend to the sub classes which need
to provide it.
The egl wayland backend registers for the callback for a rendered frame.
This allows to throttle KWin's compositor so that we don't render frames
which wouldn't end up on the screen.
For this the Scene provides a method to query whether the last frame got
rendered. By default this returns true in all backends. The Egl Wayland
backend returns true or false depending on whether the callback for the
last frame was recieved.
In case the last frame has not been renderd when performCompositing is
tried to be called, the method returns just like in the case when the
overlay window is not visible. Once the frame callback has been recieved
performCompositing is invoked again.
An abstract backend is split out of SceneXRender which takes care of
managing the render pictures and swapping them after a frame is rendered.
Having this abstract allows to implement further backends for XRender
which do not use the Overlay Window for compositing.
To have it consistant the SceneXRender is now also created by a factory
method.
The ActionCollection was only used for two features:
* setting the object name
* finding the action for retrieving it's shortcut
This can also be achieved by just setting the object name and searching
for the children of the Workspace singleton.
The main purpose of the opengl testapp was to set the environment
variable LIBGL_ALWAYS_INDIRECT if direct rendering is not supported
before glx gets initialized.
With Qt5 we may no longer set this environment variable. QtQuick
requires direct rendering. On IvyBridge QtQuick is crashing if the
variable is set. Thus we are no longer allowed to set it and thus the
complete test becomes pointless.
The test app basically whitelisted most drivers anyway, the only
drivers which were problematic are the proprietary Catalyst drivers.
It that's still a problem we can also disable OpenGL compositing on
those drivers through the recommendation in the GLPlatform.
This also means that the KWIN_DIRECT_GL variable is no longer useful.
repaints caused by effects so far polluted the timing calculations
since they started the timer on the old vsync offset
This (together with undercut timing) lead to multiple frames in
the buffer queue, and ultimately to a blocking swap
For unsynced painting, it simply caused wrong timings - leading to
"well, kinda around 60Hz - could be 75 as just well".
REVIEW: 112368
CCBUG: 322060
that part is fixed in 4.11.2
1. when adding a full damange, that must not replace existing (larger) repaints
2. emit geometryChanged before invoking and to update shadowGeometry through addRepaintFull
BUG: 324560
FIXED-IN: 4.11.2
when the option changes, the compositor needs to update
the state as it will later not act (in the disabled case)
and should to reflect the option change.
REVIEW: 111867
BUG: 322633
FIXED-IN: 4.11
* "" needs to be wrapped in QStringLiteral
* QString::fromUtf8 needed for const char* and QByteArray
* QByteArray::constData() needed to get to the const char*
Qt 5 only supports raster which means our pixmaps are always non native,
so we don't need the Extension information any more and can drop all
special code handling for mapping a native QPixmap to an X11 pixmap.
The behavior for creating a pixmap for a window is moved from Toplevel
into a dedicated class WindowPixmap. Scene::Window holds a reference to
this class and creates a new WindowPixmap whenever the pixmap needs to be
discarded. In addition it also keeps the old WindowPixmap around for the
case that creating the new pixmap fails. The compositor can in that case
use the previous pixmap which reduces possible flickering. Also this
referencing can be used to improve transition effects like the maximize
windows effect which would benefit from starting with the old pixmap.
For XRender and OpenGL a dedicated sub-class of the WindowPixmap is
created which provides the additional mapping to an XRender picture and
OpenGL texture respectively.
BUG: 319563
FIXED-IN: 4.11
REVIEW: 110577
This was currently basically broken:
* Screen number got always attached
* openGLIsBroken did not check for screen number
-> KCM reported "everything is fine" while it wasn't
Now changed to:
* only attach screen number if it is a multi-head setup
* use same logic in both Composite and CompositingPrefs
Still problematic:
* kcm is not multi-head aware so it will report everything is fine in
case of a broken multi-head setup
REVIEW: 110631
We always reset with the complete window geometry, so the subtracting
doesn't make any sense. We can just always set the damage to an empty
region.
REVIEW: 110438
The CM selecton withdraw delay was introduced to mostly unburden
plasma from recreating the theme on temporary changes, but since
plasma also watches some properties hinting supported features
and acts in consequence, this isn't sufficient and actually causes
two theme changes instead of one in the case of a regular suspend
REVIEW: 110232
CCBUG: 179042
stealing it from ourself is reported to be able
to confuse KWindowSystem about the state.
Also it causes minor but unnecessary overhead.
REVIEW: 110231
CCBUG: 179042
Overall all notifications except compositing suspended by DBus were
configured by default to not have any action. This means all the time we
emit a notification we keep DBus and KDED busy for nothing.
All the cases when a notification is triggered ire also exported to
KWin scripting, so if one really needs to handle something in case a
window is moved, it could be done through a KWin script with much more
context about the event.
REVIEW: 110113
BUG: 258097
FIXED-IN: 4.11
The define KWIN_SINGLETON adds to a class definition:
public:
static Foo *create(QObject *parent = 0);
static Foo *self() { return s_self; }
protected:
explicit Foo(QObject *parent = 0);
private:
static Foo *s_self;
There is an additional define KWIN_SINGLETON_VARIABLE to set a different
name than s_self.
The define KWIN_SINGLETON_FACTORY can be used to generate the create
method. It expands to:
Foo *Foo::s_self = 0;
Foo *Foo::create(QObject *parent)
{
Q_ASSERT(!s_self);
s_self = new Foo(parent);
return s_self;
}
In addition there are defines to again set a different variable name and
to create an object of another inheriting class.
All the classes currently using this pattern are adjusted to use these
new defines. In a few places the name was adjusted. E.g. in Compositor
the factory method was called createCompositor instead of create.
REVIEW: 109865
Many headers included KLocale to use i18n and co. But those methods are
defined in KLocalizedString and not in KLocale.
With KF5 klocale.h does no longer include KLocalizedString causing lots
of compile errors.
either by
- forcing fullrepaints unconditionally
- turning a repaint to a full one beyond a threshhold
- completing the the backbuffer from the frontbuffer after the paint
BUG: 307965
FIXED-IN: 4.10
REVIEW: 107198
With Qt5 QCursor does no longer provide ::handle() which was used to
set a cursor on a native XWindow for which we do not have a QWidget.
Also KWin has had for quite some time an optimized version to get the
cursor position without doing XQueryPointer each time ::pos() is called.
These two features are merged into a new class Cursor providing more or
less the same API as QCursor.
In addition the new class provides a facility to perform mouse polling
replacing the implementations in Compositor and ScreenEdges.
For more information about the new class see the documentation for the
new class in cursor.h.
The comment says it all: update all settings which can be done through
the compositing KCM. Years ago screen edges was in the composite KCM, but
it no longer is. So there is no need to update the edges when the
compositing settings changes.
Use WindowAttributes and WindowGeometry everywhere where the xcb commands
had already been used.
Introduces another wrapper for overlay window and a subclass for query
tree which also wrapps the children command.
The ownership for virtual desktops is moved from Workspace into a new
VirtualDesktopManager. The manager is responsible for providing the count
of virtual desktops and keeping track of the currently used virtual
desktop.
All methods related to moving between desktops are also moved from
Workspace to the new manager, though all methods related to Clients on
Virtual Desktops remain in Workspace for the time being. This is to have
the new manager as independent from KWin core as possible.
An rather important change for the handling of virtual desktops is that
the count and the id of a desktop is now an unsinged integer instead of
an integer. The reason for that is that we cannot have a negative count
of desktops as well as it is not possible to be on a desktop with a
negative identifier.
In that regard it is important to remember that a Client can be on a
desktop with a negative identifier. The special value for a Client being
on all desktops is handled by using -1 as a desktop. For the time being
this is not adjusted but instead of comparing the virtual desktop ids one
should prefer to use the convenient methods like isOnDesktop and
isOnAllDesktops. This would allow in future to internally change the
representation for on all desktops.
If a section of comments consists of a list of links and all are broken
it's a sign that nobody has used these comments for a long time...
REVIEW: 107933
Use XDamageReportNonEmpty instead of XDamageReportRawRectangles.
In XDamageReportNonEmpty mode the server generates a single damage
event when the damage state transitions from not-damaged to damaged.
When the compositor is ready to paint the screen, it requests the
damage region for each window and resets the state to not-damaged.
With XCB we can request the damage regions for all windows in a
single roundtrip, making this the preferred mode.
This should reduce the number of wakeups and the time spent
processing damage events between repaints.
When turning off the compositor do not release the compositor selection
directly but delay it through a timer. The idea is that the internal
change when e.g. restarting the compositor or switching from XRender to
OpenGL should not be visible to the outside world.
This hopefully makes restarting the compositor more robust in Plasma due
to the SelectionWatcher sometimes reporting incorrect results.
When restarting KWin the change does not matter as the selection gets
force claimed by the new instance anyway.
CCBUG: 179042
REVIEW: 106844
The CompositingType enum turns into flags and two new values are
introduced: OpenGL1Compositing and OpenGL2Compositing.
Those new values are or-ed to OpenGLCompositing so that a simple check
for the flag OpenGLCompositing works in case of one of those two new
values. To make the generic check for OpenGL compositing easier a method
in EffectsHandler is introduced to just check for this.
The scenes now return either OpenGL1Compositing or OpenGL2Compositing
depending on which Scene implementation. None returns OpenGLCompositing.
SceneOpenGL turns into an abstract class with two concrete subclasses:
* SceneOpenGL1
* SceneOpenGL2
It provides a factory method which first creates either the GLX or EGL
backend which is passed to a static supported() method in the concrete
sub classes. These method can test whether the backend is sufficient to
be used for the OpenGL version in question. E.g. the OpenGL 2 scene
checks whether the context is direct.
The actual rendering is moved into the subclasses with specific OpenGL 1
and OpenGL 2 code. This should make the code more readable and requires
less checks whether a Shader is bound. This is now known through the
Scene: the OpenGL1 scene will never have a shader bound, the OpenGL2 scene
will always have a shader bound.
To make this more reliable the ShaderManager is extended by a disable
method used by SceneOpenGL1 to ensure that the ShaderManager will never
be used. This also obsoletes the need to read the KWin configuration
whether legacy GL is enabled. The check is moved into the supported
method of the OpenGL2 scene.
REVIEW: 106357
The new methods suspend and resume are meant to provide a better way to
influence the current compositing state than toggleCompositing. In
addition an overload setCompositing(bool) is added. The resume method is
implemented in a way that it can be used to try to start the compositor
again in case it failed.
Internally the method suspendResume is dropped as it does the same as
setCompositing just with inverted binary logic and worse name. The
compositingToggled signal is now emitted from within setup and finish to
ensure that especially the compositingToggled(true) signal is only
emitted if the Compositor could start.
Also the updateCompositingBlocking is adjusted to use the new dedicated
suspend and resume methods instead of the toggle method.
REVIEW: 106273
