At the moment, we rely on placeholder output getting created, but there
are cases where it's desired to spin kwin with virtual outputs so you
could take screenshots and things as such.
The main motivation behind this change is to allow setting the cursor
per output, which eases up implementing things such as cursor output layers.
It also has another advantage - output related code is more
encapsulated. Furthermore, we could decouple output backends from Cursor.
The previous logic would set `defaults` to the last found group.
Since a touchpad is a pointer, it would set `defaults` to point
to `[Libinput][Defaults][Pointer]`, and thus the next clause would
try to find `[Libinput][Defaults][Pointer][Touchpad]`, instead of the
expected `[Libinput][Defaults][Touchpad]`. That is, only if a default
was set for Pointer.
Signed-off-by: Samuel Dionne-Riel <samuel@dionne-riel.com>
Currently, if the pointer is locked, the wayland backend will create a
subsurface. However, when the cursor moves to one of the screen edges,
it looks weird because the xdg window geometry is computed as bounding
geometry of the main surface and its subsurfaces. However, even after
calling xdg_surface.set_window_geometry, it still looks weird because
subsurfaces are stacked above window decoration.
With the proposed change, when the pointer is locked, the wayland
backend will hide the host compositor and use the software cursor. If
the pointer is unlocked, it will go back to using host cursor.
Long term goal is to make every output have its own cursor to make
output layer logic simpler.
Mesa's implementation of wl_egl_window_resize() is pretty smart. It
avoids immediately resizing the buffers. On the other hand, its API
doesn't enforce that behavior and other implementations of wayland-egl
may choose to use a different strategy.
This change moves wl_egl_window_resize() to beginFrame() to make buffer
resizing handling more nicer. Also, for the record, the drm backend
works in similar fashion.
These concepts are hard to map to InputDevice abstractions, so remove
them and let kwin use its own internal keymap instead (rather than
get it overridden by the host compositor's keymap) and track modifiers
based on the events it receives.
If the goal is to have these at any cost, I think the way out of it
would be to provide wayland backend specific signals. I would like to
avoid doing that though because it breaks encapsulation and we would
need some casts in kwin to wire in these events.
It ensures that connector type naming is consistent across compositors
and saves us some work adding string mappings.
drmModeGetConnectorTypeName() uses the same naming scheme as the drm
backend so the client side should be unaffected by this change.
CCBUG: 385135
Removing modifiers one by one creates a huge amount of combinations that
can make KWin hang for a long time in some cases. Instead of doing that,
leave it up to Mesa to pick a modifier when a setup doesn't work, which
will generally result in the least amount of bandwidth used and should be
able to power all output combinations.
When setting a device's output name externally we assign the relevant output to the device
If we pass an empty name we set a null output, which causes a crash the next time the touchscreen is touched
Since passing an empty name semantically means 'Detect output automatically' rerun the mapping heuristic so we get a valid output
Using OutputBackend::outputsQueried can cause problems because it's emitted
before output settings like enablement get adjusted to the new output configuration.
It also doesn't react to output changes initiated by KScreen and is in the
way of plans to use multiple output backends at the same time.
BUG: 461901
Currently, OutputBackend stores the initial state for windowed backends.
It's messy because the backends can have different needs. Instead, make
windowed backends take an Options struct with all possible options.
The server-decoration protocol is deprecated, and the clients should use
the xdg-decoration-v1 protocol instead.
kwin will indicate that it wants server side decoration. If the host
compositor insists on client-side decorations, the wayland backend will
render no csd. However, this is de-facto state at the moment too,
perhaps it can be improved by using libdecor?
Regardless, no csd is the current state too, except that both host and
nested compositors can enter a loop where one side says to use CSD and
the other side says that they want SSD.
OutputBackend has a concept of readiness. When the host compositor goes
down, the OutputBackend will be marked as not ready, and when it
reappears, the output backend will be marked ready again.
On the other hand, host compositor going down is a niche case, it's not
something that often happens and it's hard to justify adding more moving
parts to the startup code. It's easier to call initialize() and check
whether it fails rather than call initialize() and then monitor isReady.
Therefore, this change drops OutputBackend::isReady() to make startup
simpler.
This change removes async bits from the startup code in the wayland
backend to make it simpler. The main motivation behind this change is to
refactor the wayland backend so OutputBackend::isReady() can be removed.
If the platform doesn't support setting gamma ramps, let the request
fail. In long term, we need a software fallback too so this property is
not that relevant.
It's odd to manipulate host cursor position and it's not highly
important, for example we haven't had a need to warp the cursor on
wayland.
The main motivation behind this change is to slim down the Platform to
make it more output oriented.
Due to being a compositor, kwin has to conform to some certain
interfaces. It means a lot of virtual functions and function tables to
integrate with C APIs. Naturally, we not always want to use every
argument in such functions.
Since we get -Wunused-parameter from -Wall, we have to plumb those
unused arguments in order to suppress compiler warnings at the moment.
However, I don't think that extra work is worth it. We cannot change or
alter prototypes in any way to fix the warning the desired way. Q_UNUSED
and similar macros are not good indicators of whether an argument is
used too, we tend to overlook putting or removing those macros. I've
also noticed that Q_UNUSED are not used to guide us with the removal no
longer needed parameters.
Therefore, I think it's worth adding -Wno-unused-parameter compiler
option to stop the compiler producing warnings about unused parameters.
It changes nothing except that we don't need to put Q_UNUSED anymore,
which can be really cumbersome sometimes. Note that it doesn't affect
unused variables, you'll still get a -Wunused-variable compiler warning
if a variable is unused.
There were overflows in the bandwidth estimation calculation. To simplify things,
also only generate common modes if the display doesn't advertise them on its own
already.
This rewrite ensures that
- formats are tested for multi-gpu transfer
- dmabuf with either 10bpc or 8bpc buffers are preferred to cpu copy
- formats where drmModeAddFB2 fails are skipped
As the drm format number of 10 bit colors is lower, the fallback sorting
preferred those formats, even if KWIN_DRM_PREFER_COLOR_DEPTH=24 was set.
To fix that, prefer 8 bit colors over something random.
The X11 standalone backend has an obscure fallback to hw screen
inversion if compositing is disabled. While it's great, it conflicts
with existing features such as night color.
In order to make code cleaner and clean up dependencies, this change
moves screen inversion shortcut to the invert effect. While it can
potentially affect users that don't use compositing, I think the number
of such affected people is very small.
In hindsight, preferring hw over sw transform is nice, but I think we
need to analyze it first and then come up with a proper solution that
integrates with color management and works great both on x11 and wayland.
The screen needs to stay black until it gets enabled again, otherwise
slight timing differences can cause the screen to turn on full brightness
again immediately before turning off.
BUG: 460902
With the addition of KGlobalAccel::globalShortcutActiveChanged(),
there's a way to monitor all triggered actions without adding any
boilerplate code.
This change removes the KGlobalAccel bits from the Platform class. This
reduces the amount of boilerplate code; currently we need to be careful
about ensuring that Platform::setupActionForKGlobalAccel() is called.
Another reason behind this change is to simplify the api of Platform and
remove things that don't have anything to do with outputs.
Use isNull on QSizeF to check for a zero delta instead of comparing it
with a default-constructed QSizeF, which in practice initializes to
(-1.0,-1.0). This caused relative motion events to be omitted if the
delta happened to be equal to (-1.0,-1.0), causing mouse jumping in some
applications.
BUG: 444510
Signed-off-by: John Brooks <john@fastquake.com>
The CI now has the ability to limit the passing unit test requirements
to more specific platform configurations, so the not yet working Qt 6
unit tests shouldn't block this anymore.
If the output is turned back on before m_turnOffTimer expires, the
dpms input event filter won't be reset and all input events will go to
the void.
BUG: 460322
In libinput 1.19, three new pointer axis events were added in order to
provide support for high-resolution scrolling.
LIBINPUT_EVENT_POINTER_AXIS is de-facto deprecated and new users of
libinput should use instead SCROLL_WHEEL, SCROLL_FINGER, and
SCROLL_CONTINUOUS.
Discrete deltas were replaced with v120 delta values. 120 corresponds to
a single discrete delta. Smaller values correspond to "partial" wheel
ticks.
https://gitlab.freedesktop.org/wayland/wayland/-/merge_requests/72
At the moment, the keyboard interception code in the effects system
relies on Qt code processing key events. However, since QDesktopWidget
is removed in Qt 6, this is a blocker for Qt 6 port.
This change ports the X11 backend to private xkb keymap as indicates in
the todo comment. It allows us to drop the last QDesktopWidget usage.
With the virtual backend, it makes more sense to create outputs on
demand. It makes code clear, it would be also nice to use the virtual
backend for virtual outputs rather than have every backend implement its
own virtual outputs.
This removes the "OpenGLIsUnsafe" flag and replaces it with a timestamp
that we can check to see how long ago we tried enabling compositing, so
we can retry it if a certain amount of time has elapsed.
BUG: 452344
The drm fd can get stuck in readable state, in which case the
QSocketNotifier will fire the activated signal as often as it can
leading to high cpu usage.
We need to read() the drm fd in order to make the socket notifier stop
firing QSocketNotifier::activated.
This change removes the m_platform->isActive() check to ensure that
drmHandleEvent() gets called, in general, it should be safe as we only
notify the outputs about completed pageflips.
BUG: 452726
With this change, when the specified output is disconnected, it will
just behave as if none were selected but when it's connected again it
will pick it back up.
BUG: 456857
Remember to release all resources before the wayland socket goes:
- release the seat as we give up a seat
- release the ssdManager
- release the output decoration with the output itself
Otherwise they leak and stay dangling.
Currently the Workspace processes output updates as they occur, e.g.
when the drm backend scans connectors, the Workspace will handle
hotplugged outputs one by one or if an output configuration changes the
mode of several outputs, the workspace will process output layout
updates one by one instead of handling it in one pass. The main reason
for the current behavior is simplicity.
However, that can create issues because it's possible that the output
layout will be temporarily in degenerate state and features such as
sticking windows to their outputs will be broken.
In order to fix that, this change makes the Workspace process batched
output updates. There are several challenges - disconnected outputs have
to be alive when the outputsQueried signal is emitted, the workspace
needs to determine what outputs have been added or removed on its own.
