Summary:
resourceClass is the more useful half of WM_CLASS. It's what modern X
clients are supposed to set to a .desktop file name, which is closer to
the semantics of appId than resourceName.
Reviewers: #plasma, graesslin
Subscribers: kwin, plasma-devel, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5815
Summary:
I keep having to re-add this patch locally to see which config is
actually in play.
Test Plan: Compiles
Reviewers: #plasma, graesslin
Reviewed By: #plasma, graesslin
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5795
Summary:
TODO no longer applies since ShellClient::init already connects
SurfaceInterface::sizeChanged to update m_clientSize
Test Plan
Subscribers: kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5787
This patch makes the AMS execution path work with the new DrmCrtc and
DrmBuffer structure and solves major issues about:
* VT switching
* DPMS
* Hot plugging
* Logout
* Memory leaks
Test Plan:
Tested with Gl and QPainter.
Reviewers: #kwin
Subscribers: kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5191
Split off GBM based buffers to a separate file, which gets only included,
when GBM is available.
Note, that this also removes the gbmCallback, since already before this
patch we did delete the buffers always without it.
The plan is to later use this file for via GBM directly imported Wayland
buffers as well.
Test Plan:
Tested with Gl and QPainter backends.
Reviewers: #kwin
Subscribers: kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5179
To get an image from KWin to the screen in the DRM pipeline we combine a CRTC,
an encoder and a connector. These objects are static in the sense, that they
represent real hardware on the graphics card, which doesn't change in a
session. See here for more details:
https://01.org/linuxgraphics/gfx-docs/drm/gpu/drm-kms.html
Until now we used DrmOutput as the main representation for such an active
rendering pipeline. I.e. it gets created and destroyed on hot plug events of
displays. On the other side we had no fixed representation of the static kernel
objects throughout the lifetime of KWin. This has several disadvantages:
* We always need to query all available static objects on an hot plug event.
* We can't manipulate the frame buffer of a CRTC after an output has been
disconnected
* Adding functionality for driving multiple displays on a single CRTC (i.e.
cloning) would be difficult
* We can't destroy the last frame buffer on display disconnect because the CRTC
still accesses it and have therefore a memory leak on every display disconnect
This patch solves these issues by storing representations of all available CRTC
and Connector objects in DrmBackend on init via DrmCrtc and DrmConnector
instances. On an hotplug event these vectors are looped for a fitting CRTC and
Connector combinations. Buffer handling is moved to the respective CRTC
instance. All changes in overview:
* Query all available CRTCs and Connectors and save for subsequent hotplug
events
* Fix logic errors in `queryResources()`
* Move framebuffers, buffer flip and blank logic in DrmCrtc
* Remove `restoreSaved()`. It isn't necessary and is dangerous if the old
framebuffer was deleted in the meantime. Also could reveal sensitive user
info from old session.
Test Plan:
Login, logout, VT switching, connect and disconnect external monitor, energy
saving mode.
Reviewers: #kwin
Subscribers: kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5118
This is what 7ce380497f tried to do (and got reverted by my previous commit).
The approach is basically the same, just temporarily fake that the window
is on all activities, thus it will be temporarily shown for the session
interact, but this time there seem to be no broken side-effects (and if there
are, they'll be at least limited to activities and not break virtual desktops).
Also, 'needsSessionInteract' was a misnomer - it certainly wasn't set for all
windows that needed session interaction. Just call it what it really is.
Differential Revision: https://phabricator.kde.org/D5614
The member variable is not what determines which desktop a window is on,
it's more like caching the state. Therefore trying to fake a new value
eventually leads to inconsistencies, e.g. having KWrite open with unsaved
changes on inactive desktop results in the window ending up on all desktops
after session save (both successful and cancelled).
This pretty much reverts the whole 7ce380497f that introduced this and also
a0a976885c that tried to fix some of the problems resulting from it.
The original problem of session saving of windows of inactive activities still
remains, to be fixed by another commit.
Differential Revision: https://phabricator.kde.org/D5613
Summary:
Making QActions with the Connection as a parent is dangerous as it gets
moved to a new thread, moving all child objects to that thread.
KGlobalAccel and QAction aren't thread safe and we shouldn't be using it
in two threads. This was notably seen when runnng invokeShortcut over
DBus which then invokes it on the main thread. Something my laptop was
doing when I closed the lid.
This patch simply moves the code to the Input class, where we set up the
libinput connection.
Test Plan:
Closed lid, kwin_wayland was still there when I resumed
Set manual shortcut for toggling touchpad, that still worked
Reviewers: #plasma
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5733
Summary:
Change 0bec9ad733 introduced a regrssion
on Wayland. On Wayland getTimestamp always returns 0, thus the X11
timestamp gets reset again and again. This results in broken interaction
for Xwayland windows as the broken unit tests show.
This change addresses the regression by ignoring a value of 0.
It means the addressed bug is still present on Wayland for X11 windows
as the timestamp doesn't get updated properly. This requires further
changes.
Test Plan: TestShade works again, testX11TimestampUpdate still works
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5731
Summary:
Normally the xclipboardsyncer should terminate because the socket
becomes unusable. But we have reports of it not really going down and
running amok.
In order to prevent such situations this change registers SIGTERM to be
sent to xclipboardsyncer when the parent process (that is kwin_wayland)
dies in whatever way. This ensures that xclipboardsyncer cannot become
an orphan.
BUG: 371862
Test Plan: Only compile tested, no way to get into the problematic situation
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5589
Summary:
So far KWin only updated the x11 timestamp if the new timestamp is larger
than the existing one. While this is a useful thing it creates problems
when the 32 bit msec based time stamp wraps around which happens after
running an X server for 49 days. After the timestamp wrapped around KWin
would not update the timestamp any more and thus some calls might fail.
Most prominent victims are keyboard and pointer grab which fails as the
timestamp is either larger than the server timestamp or smaller than the
last grab timestamp.
Another problem related to timestamp handling is KWin getting broken by
wrong timestamps sent by applications. A prominent example is clusterssh
which used to send a timestamp as unix time which is larger than the
x timestamp and thus our timestamp gets too large.
This change addresses these problems by allowing to reset the timestamp.
This is only used from updateXTime (which is normally invoked before we
do things like grabKeyboard). Thus we make QX11Info::getTimestamp the
ultimate trusted source for timestamps.
BUG: 377901
BUG: 348569
FIXED-IN: 5.8.7
Test Plan: As I cannot wait 50 days: unit tests for the two conditions added.
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5704
Summary:
Based on the feedback from discussion on plasma-devel the currently only
default touch action on a screen edge should be window switching. Given
that our default window switcher is located on the left and works very
nicely for touch event it becomes the default action on the left screen.
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5551
Summary:
To have touch events working we need to do memory layout changes on the
xcb events. This is fine for identifying the touch events which should
trigger the screen edge. But when passed on to Qt to have QtQuick windows
(e.g. Alt+Tab) handle the touch events, this results in a problem:
Qt itself does also the memory movement and then the movement is double
and touch events break.
To prevent this problem an RAII class is added which moves the memory in
the ctor and moves it back in the dtor. So during KWin's processing it
has the right memory layout and later on in Qt's processing it has the
proper "wrong" layout which Qt can fix again.
Test Plan: Touch events in Alt+Tab work
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5549
Summary: Basically just a copy and paste from the relevant Qt Wayland
parts.
Reviewers: #plasma
Subscribers: plasma-devel
Tags: #plasma
Differential Revision: https://phabricator.kde.org/D5172
Summary:
This patch caputres from the framebuffer using the framebuffer's
geometry, factoring in scale. We then keep the current normal DPI
framebuffer causing it to downsample there.
This is good because:
- it keeps the code very simple
- it's a performance optimisation. Blurring on 4k is naturally more
expensive than at regular DPI. Downsampling keeps it the same - and you
can't see a difference given it's high DPI and you're going to blur it
anwyay.
- it keeps kernel sizes somewhat resolution independent so it will look
just as blurry across multiple screens.
::doCachedBlur still needs doing.
Test Plan:
Ran an app
Ran the kwindowsystem blur test
Observed the right part of the window being blurred
Reviewers: #plasma
Subscribers: plasma-devel
Tags: #plasma
Differential Revision: https://phabricator.kde.org/D4963
Summary:
The backend is responsible for setting the viewport, the GLRenderTarget
should save and restore the viewport actually set by the backend.
Test Plan:
Ran Blur effect which uses this method, under scaling the viewport "restored"
differed from the one set by the backend.
Reviewers: #plasma
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D4952
Summary:
This method already translates to s_virtualScreenGeometry as the source
co-ordinates are in compostior space we should take scaling into account here too.
This method already supports resizing if source and target sizes do not match.
Calling funcitons can either double the size of the target if they want native resolution
or leave as-is if they want the result in standard DPI.
This implements scaling in the screenshot, magnifier and simple blur effect.
Test Plan:
Tested screenshot effect using spectacle.
I hit an (unrelated, was reproducible on master) crash whilst using this, so it is not fully tested
Ran magnifier effect manually
Reviewers: #plasma
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D4951
Summary:
glReadPixels needs the correct location relative to the
framebuffer so we need to factor in the scale as well as translation
when going from compositor space.
Test Plan:
Ran the plasma colour picker plasmoid in windowed mode
Clicked on multiple parts of a window
It was right every time
Reviewers: #plasma
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D4950
Summary:
Background contrast reads pixels from the framebuffer
we need to convert from compositor to framebuffer co-ordinates
when an output is scaled
Test Plan:
Ran the manual test in kwindowsystem. Moved window over dolphin.
Visually checked output
Reviewers: #plasma
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D4949
Summary:
GLRenderTarget already stores the translation from global compositor
space to the render target co-ordinates, we also need a scale factor for
the exact same reasons.
Test Plan: Used in various effects after this patch and it works there.
Reviewers: #plasma
Subscribers: plasma-devel, kwin, #kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D4948
Summary: We need to set the viewport so that we scale from device pixels to global compositor space.
Test Plan:
Ran kwin_wayland properly on my laptop without setting KWIN_COMPOSE.
Most things worked.
Reviewers: #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D3504
When quick tiling animate the outline from the window geometry to the position it would have afterwards.
This provides a visual hint to from where to where the window will go.
Differential Revision: https://phabricator.kde.org/D5262
Summary:
There are a few places where KWin needs to read values from kcminputrc.
As I need to add yet another one it makes more sense to properly
structure it like in other cases and have only one kcminputrc hold in
the application. This also allows to better mock the config values in
the integration tests.
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5540
Summary:
Consider the case that capslock gets pressed and released.
In the case of Weston we have a sequence of:
1. Key press event
2. Modifier changed event
3. Key release event
4. Modifier changed event
KWin however used to send the events in the following sequence:
1. Modifier changed event (on key press)
2. Key press event
3. Modifier changed event (on key release)
4. Key release event
It looks like Xwayland is not able to properly process the sequence
sent by KWin. And in fact KWin's sequence is wrong as it sends a state
which does not match. We report that the caps lock is pressed in the
modifiers prior to the application getting informed about the key press
of caps lock.
This change aligns KWin's implementation to the behavior of Weston. The
main difference is that when modifiers change Xkb internally caches the
serialized modifier states. And KeyboardInputRedirection just forwards
the modifiers to KWayland::Server::SeatInterface once the processing has
finished. SeatInterface ignores the forwarding if no states changes, so
it is fine to do it that way.
BUG: 377155
Test Plan:
Not yet tested with an affected Xwayland as I only have 1.18 and the
problem started with 1.19. But verified the sequence of events with WAYLAND_DEBUG
and caps lock stil working in QtWayland clients and Xwayland 1.18
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5452
Summary:
Another regression from enabling touch support on X11.
The approach window for pointer input gets unmapped as soon as the mouse
enters the window. This ensures that mouse motion events are not stolen
from other applications. But with the touch events we did not even react
on the enter event if it's not activated for pointer. The result was an
area around the screenedge being blocked for pointer input.
This change only creates and maps the approach window if the edge is
activated for pointer input.
BUG: 378951
Test Plan:
Activated edges through pointer and touch, reconfigured and tested
motion events
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5528
Summary:
With the TabBox I observed the following issue:
1: trigger TabBox through screen edge
2: exit TabBox
3: trigger TabBox again through screen edge
Actual behavior: in step 3 the GUI does not show. If now using Alt+Tab
to show the TabBox the gui shows, but through screen edge it stays
broken.
Investigation showed that the difference is a direct show in screen edge
case and a delayed show in Alt+Tab case. Futher investigation shows that
an invalid geometry gets requested in the broken case. While this might
indicate an issue in another area it makes sense to protect KWin
internally against it and not to break rendering.
Thus this change ensures that a valid FBO does not get replaced by an
invalid sized FBO.
Test Plan:
Tested that Alt+Tab works correctly when triggered through
edge
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D5438
