Summary:
xcb_put_image doesn't handle big images well. Contrary to XPutImage, the
caller must manually split the data in such a way that each xcb_put_image
request doesn't exceed the maximum request length. Not doing so will result
in libxcb shutting down the connection.
CCBUG: 338489
CCBUG: 388182
Test Plan: Take a screenshot of an active fullscreen client on a 4K monitor.
Reviewers: #kwin, davidedmundson
Reviewed By: #kwin, davidedmundson
Subscribers: davidedmundson, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D21370
Summary:
Sesison Manager stores all relevant clients. There's an assert if the
window type is outside of the standard client window types. It assumed
that all windows outside this would be Unmanaged windows rather than
Client objects, something probably true but not something enforced.
This particular crash was probably cased as we have a new window type in
Plasma OSD, which does not set BypassWindowManager in Qt window flags.
BUG: 395712
Test Plan:
Set to restore session
Logged out and back in
Saw some windows
Set to restore manually saved session
Hit save
No crash
Reviewers: #kwin, graesslin
Reviewed By: #kwin, graesslin
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D13715
Summary:
The NVIDIA implementation of glXSwapBuffers will, by default, queue up
to two frames for presentation before blocking. KWin's compositor,
however, assumes that calls to glXSwapBuffers will always block until
the next vblank when rendering double buffered. This assumption isn't
valid, as glXSwapBuffers is specified as being an implicit glFlush,
not an implicit glFinish, and so it isn't required to block. When this
assumption is violated, KWin's frame timing logic will
break. Specifically, there will be extraneous calls to
setCompositeTimer with a waitTime of 0 after the non-blocking buffer
swaps, dramatically reducing desktop responsiveness. To remedy this,
a call to glXWaitGL was added by Thomas Luebking after glXSwapBuffers
in 2015 (see bug 346275, commit
8bea96d701). That glXWaitGL call is
equivalent to a glFinish call in direct rendering, so it was a good
way to make glXSwapBuffers behave as though it implied a glFinish
call.
However, the NVIDIA driver will by default do a busy wait in glFinish,
for reduced latency. Therefore that change dramatically increased CPU
usage. GL_YIELD can be set to USLEEP (case insensitive) to change
the behavior and use usleep instead. When using the NVIDIA driver,
KWin will disable vsync entirely if GL_YIELD isn't set to USLEEP
(case sensitive, a bug in KWin).
However, the NVIDIA driver supports another environment variable,
__GL_MaxFramesAllowed, which can be used to control how many frames
may be queued by glXSwapBuffers. If this is set to 1 the function
will always block until retrace, in line with KWin's expectations.
This allows the now-unnecessary call to glXWaitGL to be removed along
with the logic to conditionally disable vsync, providing a better
experience on NVIDIA hardware.
Reviewers: #kwin, davidedmundson, zzag
Reviewed By: #kwin, davidedmundson, zzag
Subscribers: kwin, davidedmundson, zzag
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D19867
Summary:
VirtualDesktopManager is initialized in two places: Workspace::init and
Workspace::initWithX11. The former method loads virtual desktops from
the config file and the latter method synchronizes VirtualDesktopManager
with RootInfo.
Both methods do
if (!VirtualDesktopManager::self()->setCurrent(m_initialDesktop))
VirtualDesktopManager::self()->setCurrent(1);
which makes sense in Workspace::init, but not in Workspace::initWithX11.
When Workspace::initWithX11 is called, the current virtual desktop is
the same as m_initialDesktop. So that piece of code basically makes
the first virtual desktop current no matter what.
BUG: 390295
FIXED-IN: 5.15.3
Reviewers: #kwin, davidedmundson
Reviewed By: #kwin, davidedmundson
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D19520
Summary:
We can't use Client::noBorder while saving the desktop session because
the returned value is not necessarily set by the user.
BUG: 403948
Reviewers: #kwin, davidedmundson
Reviewed By: #kwin, davidedmundson
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D18756
Summary:
Mesa requires XESetWireToEvent xlib callbacks to be called
when DRI2 is used. This is done by the GLX integration in
the Qt's xcb plugin, but Qt 5.12 initializes the GLX integration
only when required, e.g. when a window with OpenGL support is
created or when availability of OpenGL is checked.
So force initialization of the GLX integration by calling
QOpenGLContext::supportsThreadedOpenGL().
https://codereview.qt-project.org/#/c/6557/https://bugzilla.opensuse.org/show_bug.cgi?id=1120090
Reviewers: #kwin, graesslin
Reviewed By: #kwin, graesslin
Subscribers: davidedmundson, graesslin, fvogt, filipf, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D18366
Summary:
When using the natural layout algorithm with the fill-gaps option, a small
error (less than one) is introduced in windows' aspect ratio each time they are
enlarged due to floating-point roundoff.
Currently, the algorithm computes the width and height enlargement factors and
then attempts to enlarge in each of the four possible directions, repeating
until it can't enlarge any windows any further. Hence, this aspect ratio error
can be multiplied by up to four. Especially for small, long, and narrow
windows, this can result in a total error of greater than one by the end of
that loop iteration. If this occurs, on subsequent iterations the height
enlargement factor might then be computed as negative violating some of the
core assumptions of the algorithm and resulting in the loop iterating endlessly
until one of the window dimensions overflows, freezing the program for up to
several minutes.
To fix this, the height enlargement factor should be re-computed based on the
new width each time the window is enlarged, ensuring the error introduced in
the aspect ratio never exceeds one.
BUG: 364709
BUG: 380865
BUG: 368811
FIXED-IN: 5.15.0
Test Plan:
The most reliable way to reproduce the freeze seems to be to activate the
desktop-grid effect while a tool-tip window is fading in.
Ensure desktop-grid is configured to use present windows, and that present
windows is configured to use the natural layout algorithm with the fill gaps
option selected.
The freeze is still intermittent, but using this method should be able to be
triggered within about 10 tries without this fix.
After applying the fix, the freeze has never been observed.
Reviewers: #kwin, zzag
Reviewed By: #kwin, zzag
Subscribers: graesslin, kwin, zzag
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D16278
Summary:
Currently, when the user is moving a window in the desktop grid it is drawn from
the paintScreen function to ensure it appears above all other windows. However,
when drawing the window, the screenProjectionMatrix for the WindowPaintData structure
is never specified. Because of this, the window will not be visible if OpenGL
compositing along with Lanczos scaling are in use since its coordinates will never
be projected into screen-space (unless the window is maximized, in which case the
scaling code is bypassed).
BUG: 361371
BUG: 364509
FIXED-IN: 5.14.3
Test Plan:
Ensure OpenGL compositing is enabled and the scaling method is set to "Accurate" on
hardware supporting this option. Additionally, ensure the desktop grid effect is
active and is configured to use Present Windows. Open a window and ensure is it not
maximized. Click on the window and drag it around - its contents should remain visible
the entire time and follow the mouse.
Reviewers: #kwin, zzag, graesslin, davidedmundson
Reviewed By: #kwin, zzag, davidedmundson
Subscribers: broulik, davidedmundson, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D16430
Avoids triggering an assert and communicates that the timeout is running.
Ideally, there was a countdown label of some sort but this is better than crashing.
BUG: 399644
FIXED-IN: 5.12.8
Differential Revision: https://phabricator.kde.org/D16124
Summary:
While BuiltInEffects has effectData() function, many functions repeat
s_effectData.at(index(effect)), which is what effectData() is doing.
By using effectData(), we'll get rid of those repetitions and maybe make
easier transition to other underlying data structure that stores metadata
for builtin effects.
Test Plan: Compiles, all enabled builtin effects are loaded and working.
Reviewers: #kwin, davidedmundson
Reviewed By: #kwin, davidedmundson
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D13587
(cherry picked from commit 7bfaa6e913)
Summary:
Using QtQuick.Controls for Label and TextField can result in blurry font rendering for a fractional scaling (e.g. 1,5). There is a work around for QtQuick.Controls 2.x therefore using QtQuick.Controls 2.0 for Label and TextField resolves the problem
BUG: 366451
Reviewers: #kwin, davidedmundson
Reviewed By: #kwin, davidedmundson
Subscribers: ngraham, zzag, davidedmundson, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D14495
Summary:
Our rule handling has had a grave error for years. Whenever a window
with a rule was openend or closed the kwinrulesrc was written back to
disk.
The reason for this behavior is that temporary rules need to be discarded
once they were used. For that there is a method discardUsed which invokes
requestDiskStorage whenever a rule for the window was found. But it did
not check whether there was a rule requiring this.
This change modifies the discardUsed to track whether it changed a rule
and only writes back in case there was a change.
BUG: 393911
FIXED-IN: 5.12.6
Test Plan: Only compile tested
Reviewers: #kwin, #plasma
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D12749
Summary:
The drm platform has a special handling for touch events while the
screen is in dpms. All touch events are filtered out, but a double
tap wakes up the screen.
Due to the filtering out of the events the idle timers were not reset.
So if you only double tapped, the screen turned on, but not off again.
This change updates the timestamp after double tap, so that the idle
timers are restarted. All other events are still filtered out, to not
have "fake" events (e.g. smart phone in pocket) wake up the device.
BUG: 392754
FIXED-IN: 5.12.6
Test Plan: only compile tested
Reviewers: #kwin, #plasma
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D12668
Summary:
Changing the creation of the nested window from unstable v5 to v6 in
order to be supported on more compositors and to use KWin's better
supported platform.
Targeting 5.12 as this will help to remove support for unstable v5
from KWayland.
Test Plan: Run a nested KWin/Wayland on KWin/Wayland, everything looked fine
Reviewers: #kwin, #plasma
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D11806
Summary:
From Wayland documentation:
"When a seat's focus enters a surface, the pointer image is undefined and
a client should respond to this event by setting an appropriate pointer
image with the set_cursor request."
KWin's interpretation so far for the undefined pointer image was to
remove the pointer image when entering a surface waiting for the client
to set a cursor image. This can result in a short flicker as there might
be a frame without a cursor image.
This patch changes the behavior by keeping the previous image till the
application set a new one. This brings some advantages:
* if the application is not responding a cursor is still shown
* if the same cursor is used as in the previous window we don't have a
flicker
CCBUG: 393639
Test Plan: I cannot see the flicker, so only tested with the adjusted tests
Reviewers: #kwin, #plasma
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D12631
