Summary:
In 47343fb we made GBM buffer shared.
What we wanted to do was:
Unbox the shared_pointer<GBMSurface> to give us a GBMSurface* object
Call the gbm_surface*() on that operator
Then cast that to a void* for eglCreatePlatformWindowSurfaceEXT
What we did:
Cast the std::shared_ptr<GBMSurface> to a gbm_surface* then cast that
to void*.
This is just a garbage value and it crashes in Mesa when we do our first
paint.
I've replaced that with an explicit method then we can use shared_ptr's
-> operator rather than get() which does the right thing in a readable
way.
Test Plan:
It crashed after rebasing to master (for Aleix too)
No longer crashes
Reviewers: #plasma
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D8251
Summary:
The gbm_surface is owned by the EglGbmBackend, but it's not the only one
using it. The DrmSurfaceBuffer is also using it and needs it to destroy
the gbm_bo. Now this can become a problem in the following situation:
* a page flip is still pending
* the EglGbmBackend destroys the gbm_surface
-> when the page flip happens the DrmSurfaceBuffer will try to destroy
the gbm_bo and crash as the gbm_surface is no longer valid. This
situation can happen when switching screens or when switching compositing
backend (OpenGL 2 -> OpenGL 3).
To address this problem a class GbmSurface is added which wrapps the
gbm_surface pointer. The EglGbmBackend creates and holds a shared pointer
to the GbmSurface and passes that one to the DrmSurfaceBuffer. So when
cleaning up the gbm_surface only the shared pointer is reset and in case
the DrmSurfaceBuffer still needs it, it can access it without problems.
BUG: 385372
FIXED-IN: 5.11.0
Test Plan: Not yet
Reviewers: #kwin, #plasma, subdiff
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D8152
guix (like nix) used a separate prefix for each package. Thus the header
files for
qtdeclarative are in /gnu/store/…-qtdeclarative-5.9.1/include/qt5/ while
the ones for
qtbase are in /gnu/store/…-base-5.9.1/include/qt5/. This means that
*each* dependency's include directory must be specified. This is in
contrast to a "normal" Unix-system, where all includes end up in
/usr/include/qt5 and missing to defined some include-dependencies does
often not raise an error.
Differential Revision: https://phabricator.kde.org/D8112
Summary:
So far the method only operated on X11 clients. So when the last desktop
got removed Wayland clients were still on it. As the auto test showed:
this results in a crash.
Credits go to code coverage as it showed that area as red, which made me
look on it and realize this must be broken.
Test Plan: New test case added
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D8082
Summary:
It was possible that we performed an early exit when the caption changes
and this results in the visible name being wrong. E.g if we have two
windows called foo, the second one has the visible name "foo <2>". After
changing that to "bar" the bug resulted in it still being the foo
variant instead of getting cleared.
BUG: 384760
FIXED-IN: 5.11.0
Test Plan: Created new test case exposing the problem
Reviewers: #kwin, #plasma, hein
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7871
Summary:
The code in events.cpp was problematic as it was called in a Wayland
session. So KWin changed outputs, this gets mirrored to XWayland and
then KWin reacted on the XRandR event and might have even changed the
refresh rate due to that - bad idea.
This change moves the code into the already existing X11EventFilter for
XRandR events in XRandRScreens.
Test Plan: Run kwin_x11 in gdb on Xephyr, breakpoint in new code and triggered XRandR event
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7654
Summary:
The syncing of the window geometry to the internal geometry can
unfortunately cause a freeze in very special conditions:
1. create QML component
2. a Plasma::Dialog gets created
3. It creates the DialogShadows
4. This triggers QGlobalStatic creation which locks a non-recursive
mutex
5. The creation of DialogShadows creates a Registry and triggers a
roundtrip on the Wayland server
6. KWin processes all Wayland events
7. This triggers the creation of a ShellClient
8. The ShellClient has a PlasmaShellSurface which requested a position
9. The new geometry does not match the geometry of the Plasma::Dialog
10. ShellClient syncs the geometry to the Plasma::Dialog
11. Plasma::Dialog updates the theme because window geometry changed
12. This accesses the DialogShadows...
which is still in the non recursive mutex and we have a freeze.
By delaying the sync to the end of cycle we jump out of this deadly
sequence.
BUG: 384441
Test Plan:
The freeze doesn't hit any more. It's possible that some test
cases need adjustments.
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7712
Needed by testing of QPainter scene to access the back buffer. Exposed
as a virtual method in Scene, so that the test does not have to cast to
SceneQPainter.
Summary:
KWin::displayWidth and KWin::displayHeight are bound to X11 which
doesn't make much sense on X11. In addition KWin internally knows
the overall display dimensions through the Screens singleton class.
Reviewers: #kwin
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D1798
Summary:
Bringing another caption feature from X11 to Wayland. If we have
multiple windows with the same caption, starting from the second window
a suffix <number> is added.
E.g. if we have three windows with caption "foo", the naming is:
* foo
* foo <2>
* foo <3>
The change tries to use as much shared code between the X11 and Wayland
implementation. Unfortunately it's not possible to share completely as
the X11 implementation does X11 specific things like editing the visible
name.
By sharing the code the numbering also works cross windowing system.
That is if a window is called "foo" on X11, a new window on Wayland with
caption "foo" will get adjusted to "foo <2>" and vice versa.
The change also eliminates a duplicated signal for captionChanged in
ShellClient (found by test case).
By using the shared implementation on X11 side a bug gets fixed which
got introduced with the support of "unresponsive", this is no longer
considered and the numbering still works even if there is a window which
is unresponsive.
Test Plan: New test case and manual testing
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7425
Summary:
This change splits out the X11 specific event filtering into a dedicated
X11EventFilter. It is created in the x11 standalone platform plugin when
the first Edge is being created.
Some of the X11 specific code is removed from ScreenEdges, though more
refactoring is possible in ScreenEdges to share more code between X11
specific and generic implementation.
Test Plan: Run KWin on Xephyr, screen edge approach effect still shows
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7406
D7460 in kwayland assumes all WL_SHM_FORMAT_ARGB8888 buffers have
opacity premultipied RGB values.
Kwin tests need updating to do the same.
Rendering code did not need changing.
Test Plan: Tests now pass
Reviewers: #plasma, graesslin
Reviewed By: #plasma, graesslin
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7607
Summary:
A surface has an API to know which screen(s) it's on, this is useful
especially for knowing the scale it should render at.
In practice Qt currently doesn't do anything with this information; but
that's set to change.
Test_helpers is changed as we need to create output objects in order for
wl_surface to map them.
Closes task T4467
Test Plan:
Attached unit test.
As mentioned above, it doesn't have any real world impact currently, so not a lot
to test.
Reviewers: #plasma, graesslin
Reviewed By: #plasma, graesslin
Subscribers: johanhelsing, graesslin, plasma-devel, kwin, #kwin
Tags: #kwin
Maniphest Tasks: T4467
Differential Revision: https://phabricator.kde.org/D7359
Summary:
The KWindowSystem call which we used doesn't work on Wayland as it's only
implemented in the xcb variant and cannot be made available for Wayland
in an easy way as it is still XLib based.
This change turns the optional XCB-ICCCM dependency in a required one
and thus can use the functionality provided by said library to implement
what KWindowSystem provided.
BUG: 382789
Test Plan: New test case which failed with old code
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7046
Summary:
KWin passes the current keysym converted to a Qt::Key in the QKeyEvent.
The current keysym does not always change when a key gets released, so
when pressing a shortcut the release carry a Qt::Key which could be
considered as wrong.
QtWayland transforms the actual pressed/released key into a keysym and
passes that through the QKeyEvent. This change does the same for the
internal windows. A new QKeyEvent is created and adjusted in a way that
it matches what Qt expects.
Why not change everything to how Qt expects it? The key is used at
various places and in KWin internally we expect the behavior how it is
currently implemented. So it's better to use Qt's expectation only when
interacting with Qt.
Also the change carries a workaround for a bug in QKeySequenceEdit
(see QTBUG-62102) and transforms Super to Meta. As this adjustment only
makes sense for the internal windows we need to send in an adjusted
QKeyEvent anyway, so another argument for using the Qt behavior only in
this place.
Test Plan:
Can set a shortcut on Wayland and it can be used to activate
the window.
Reviewers: #kwin, #plasma
Subscribers: plasma-devel
Tags: #plasma
Differential Revision: https://phabricator.kde.org/D6828
Summary:
Based on the work of 3f4995fb9b this change
introduces a GlxContextAttributeBuilder to make the requesting of context
attributes cleaner, more verbose and less error prone copy and paste.
Test Plan:
Switched between Core and legacy and verified the output;
extended auto test
Reviewers: #kwin, #plasma
Subscribers: plasma-devel
Tags: #plasma
Differential Revision: https://phabricator.kde.org/D6411
Summary:
Split from the last review, as in order to track output's removed (in
any way that's useful) the registry needs to be on the heap and the change grew.
Test Plan:
Every test that previously passed on my system still does
Though currently that's not every test (for some reason) which was
my main motivation for splitting.
Reviewers: #plasma, graesslin
Reviewed By: #plasma, graesslin
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7377
Summary:
Turn Xkb into a QObject
Reduced dependencies on other classes and allows to emit signal directly
instead of emitting a signal on another class.
Dependency inject KWayland::Server::SeatInterface into Xkb
No need to depend on WaylandServer from within Xkb.
[autotests] Add test case for Xkb::toQtKey
Tests all key codes KKeyServer is able to map to Qt.
Migrate Xkb::toQtKey away from KKeyServer
The regression in kwindowsystem in
32526718eae99ccb594360627586eebdf793372b showed once more that it's time
to migrate the xkb keysym to Qt::Key mapping away from KKeyServer. The
main problem with KKeyServer is that it's made for X11 and we don't have
X11 here. This causes warnings printed at runtime and the code only
works by accident at all because KKeyServer fails to initialise
modifiers. That is normally KKeyServer would include the modifiers, but
that is nothing we are interested in for mapping xkb keysyms to Qt::Key.
So to address this, KWin now implements the mapping itself. It's based
on the implementation in KKeyServer, but also improved by providing
faster xkb to Qt conversion (which matters for KWin more than the other
direction).
Long term plan is to extend this so that also on X11 it can be used.
Then it should be split out and be used in KKeyServer as a replacement
for the current implementation which could support both X11 and Wayland.
Test Plan:
New test case added which covers all keysyms supported by
existing KKeyserver mapping
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7336
Summary:
With ddf3536f19 a regression got
introduced resulting in Client::captionChanged not being emitted if
Client::setCaption was called without the force parameter.
The regression is only in master branch.
BUG: 383444
Test Plan: Added test case passes
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D7290
Another random failure on build.kde.org (see T6546). This tries to use
QTRY_COMPARE on the SignalSpy's count instead of a wait in the hope that
this can handle the multi-threaded situation better. Though it might
also be possible that the other thread just didn't run in the 5 sec we
waited.
Summary:
The generation of the shortcut caption part is moved from Client to
AbstractClient. The ShellClient also has a captionSuffix and implements
the full part in caption.
Overall this needs more refactoring to support more sharing between the
two implementations. But one step at a time.
Reviewers: #kwin, #plasma
Subscribers: plasma-devel
Tags: #plasma
Differential Revision: https://phabricator.kde.org/D7093
It failed on CI. A possibility is that the wait for the quick tile
changed took to long, so that the next wait already passed and thus
failed. Instead of wait a QTRY_COMPARE on the size of the signalspy.
The StrutsTest is constantly aborting on build.kde.org and lots of
OpenGL stuff in output. This could indicate a problem with the Outline
as the failing test could trigger the Outline. So let's try to disable
it.
Summary:
The buffer for a XWayland window is larger than the actual size. Thus
we need to use the clientSize as reference, not the buffer size.
BUG: 382748
Test Plan: Test passes
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D6952
Summary:
Comparing a pure blue square isn't a very effective test as someone
cropping the image will still pass.
Reviewers: #plasma, graesslin
Reviewed By: #plasma, graesslin
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D6982
Summary:
Thanks to std::bind we don't need that many different slots to setup the
global shortcut connections. Instead we can have one shared
implementation which takes the argument and passes it to the window.
To support std::bind arguments in kwinbindings the initShortcut method
and dependencies are adjusted as well as a new macro is added.
As I don't want to include abstract_client.h in workspace.h a new enum
is created for the quick tiling flags used in Workspace. This caused a
larger refactoring as the change to an enum class also caused quite some
changes.
Test Plan: Affected test cases still pass
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D6783
The test was racing on build.kde.org and failing due to that. We have
a condition there where we wait for some parts being transmitted through
X and some through Wayland. The test assumed X is faster and failed due
to that.
Summary:
Moves most of the implementation from Client to AbstractClient, so that
it can be used for both Client and ShellClient. Only the X11 specific
code is kept in Client.
Not yet implemented is updating the window caption.
Unfortunately the testing of this feature showed that setting a window
shortcut is not working on Wayland at all (the Qt widget doesn't properly
catch the shortcut). So this feature is currently only of erm theoretical
use.
Test Plan: Added new test case. No testing in real world as explained.
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D6818
Summary:
The last_active_client is set when an AbstractClient gets activated. For
the X11 case the last_active_client was getting reset to nullptr when
the last_active_client gets destroyed. But for the ShellClient that did
not yet happen. This could result in a crash.
This change addresses the problem and adds a test case which triggered
the crash. The condition of the crash are difficult to generate though -
it took me about an hour to write the test for the crash.
1. Wayland client must be active
2. Explicit focus to null (no active client)
3. destroy Wayland window
4. X11 client which sets focus on itself without interaction with window
manager
Test Plan: test case no longer crashes
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D6852
KWin cannot handle the layering correctly when a Wayland window is active
and there is a fullscreen X11 window. This test tries to simulate the
situation, but it does not show it.
Nevertheless it's a good test, so let's add it.
CCBUG: 375759
David Edmundson