This patch was generated with tooling from patchset 31 of
https://codereview.qt-project.org/c/qt/qtqa/+/267034 in interactive
mode. General platform names were chosen if greater than 60% of the
currently active platforms of a given type in COIN recently failed.
Change-Id: Ia4bde7f0ec422bbb727dc9d7151295159094f146
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
It can happen that AppKit calls -mouseDown: on a popup's view, but
we consider the click to be outside of popup's area (happens on the
1-pixel edge of a 'geometry', QRect::contains() returns false).
If we send close event to essentially 'self', m_platformWindow
is becoming nullptr. So we bail out early, no further processing
is needed.
Fixes: QTBUG-77348
Change-Id: I224943e6bcf4ae052412ef7dc7b23a94f999aa19
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
It was added for Symbian almost 10 years ago (d7057e7c1f1a), for a somewhat
dubious use-case. The Symbian code is since long gone (ae30d7141), so the
remaining pieces are just adding complexity to the already intricate workings
of the QtWidgets backingstore/painting logic.
Task-number: QTBUG-8697
Change-Id: I82af610a8ac26719c588ac63f06b4501f59b400d
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
As exposed by tst_QObjectRace::destroyRace we would sometimes end up
with a double-free when destroying a QSlotObject in multi-threaded
scenarios. One free would be done in ~QObject as the receiver was being
destroyed while the other free was done when deleting a QMetaCallEvent
object after we realized it was not needed because the receiver was
destroyed.
Since we can be in a separate thread from the receiver we should lock
before referencing the connection object.
Amends b7d073e990.
Change-Id: Icb53862dc880ae9a4e5581a1a9ee693573f7d9c7
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
env var values might contain '=' char, so we can't use split.
Change-Id: Iedf3ea46a847acaaf02f51bc80586a519fe7a310
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Makes the code a little cleaner, avoiding an issue caused
by UB and/or optimization bug in msvc2019.
Fixes: QTBUG-77119
Fixes: QTBUG-77230
Change-Id: I9bc8f427a90e6fe32b3c26301bbb703a3c4ad846
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
The paths to the libraries and prl files should have the "_debug"
suffix for the debug configuration. This prefix is added to the TARGET
when by qt_module.prf when doing a debug build, but not during a
debug_and_release build.
Make sure to strip the _debug suffix if it's there, and re-add it later
always, to be consistent in both debug_and_release builds and in
debug builds.
Amends a12b6e7bf6.
Task-number: QTBUG-38913
Task-number: QTBUG-75520
Task-number: QTBUG-77092
Change-Id: I29e88f2b991e2be06b23652d64edc768fa35a5ae
(cherry picked from qt/78d67d17a6c108a419816b8bd47f78864ddbb07f)
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
While there is likely no example of it in Qt itself, applications can
use QOpenGLShaderProgram instances on different threads. These instances
have nothing to do with each other but they do share a global cache object.
This becomes problematic without proper synchronization.
Change-Id: I80faf73f34af7e67349eee916bb3f216e22c07fd
Fixes: QTBUG-77469
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
Use FLIP_DISCARD swapchains only on Win10, stick
with DISCARD otherwise. This may fix the swapchain
creation problems on Windows 7.
Add a QT_D3D_NO_FLIP env.var. to make it possible to
disable using FLIP_DISCARD even on Win10. This is there
for troubleshooting purposes.
Finally, fix the backbuffer handling. What we originally
ported from the D3D12 backend of Qt Quick is not quite how
DXGI used to work with D3D11 and earlier. GetBuffer() can
only be used to query index 0, and that's the backbuffer,
the rest is managed internally. Follow this model.
As an added bonus, disable Alt+Enter.
Change-Id: Ie5c7a1e813864e7f873d55bc72cb22fc09213a05
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: André de la Rocha <andre.rocha@qt.io>
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
Qt Quick apps feature an occasional flicker which seems to be caused
by updating the contents of a Static (or Immutable) QRhiBuffer in a frame
where the QRhiBuffer in question is read in the previous frame as well.
On macOS these types map to a Managed MTLBuffer and only one native buffer
object (MTLBuffer). It seems modifying such a buffer is not safe if the
previous frame has not completed. (this may be as expected, but hard to
tell due to Metal's underdocumented automatic hazard tracking which we
rely on atm)
So for now switch to having 2 native buffers, like we do for Dynamic
(on iOS/tvOS this would be the case anyway since there all buffers are
host visible and slotted regardless of the QRhiBuffer type).
This seems to solve the issue.
To be seen if we want to move to a more Vulkan-like setup where Immutable
and Static map to device local (Private).
Change-Id: I76013f58a2e183ad8eab0705b28a03b395c4530c
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
create_cmake.prf populates the values of CMAKE_RELEASE_TYPE and
CMAKE_DEBUG_TYPE depending on if Qt was configured with debug, or
release, or the build_all feature was set (which implies
debug_and_release).
simulator_and_device also implies build_all. This
is a problem when configuring a Qt simulator_and_device build with
only a "debug" configuration, or only a "release" configuration.
In that case we would try to parse prl files for both configurations,
even though only one configuration exists.
Switch to checking for debug_and_release scope explicitly instead of
build_all. This allows configuring and building a Qt iOS
device_and_simulator debug configuration which is usable from CMake.
Task-number: QTBUG-38913
Change-Id: Ife6d5d34d2b6bb1ac787d901a166e41c6e0c844b
Reviewed-by: Jörg Bornemann <joerg.bornemann@qt.io>
Cleanup the QtWidgets animation examples:
- use nullptr
- use normalized includes, remove unused includes
- fix style
- fix crash of sub-attaq when the game ended (error during range-based
for loop porting)
- don't use keyword 'final' for a variable name
Change-Id: Id23be8ff8b1b310da005d13c052fe547f6a0d63a
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
The QVector dirtyOnScreenWidgets was aggregated by pointer, which
makes no sense, as a QVector is just as large as a pointer (and even
in Qt 6, when it will be larger, it's not going to be horrible). But
this complicated the code quite a bit.
Aggregate by value instead (it's just one of three such vectors now).
Drive-by fixes:
- use QVector::removeAll() instead of rolling your own
- port two indexed loops to ranged ones. In the first case, it's safe,
as the loop body clearly doesn't touch the iteratee (it's just a
std::accumulate). In the second, the question no longer applies, as
we're now using a consume loop.
Change-Id: Icd4ac13bb4a6f9a783f0adf2fb6a5bdfacd1f91a
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
There's really no reason for it to be out-of-line, and we're going to
use it in QBezier::split(), which is inline, and we want the optimizer
to have a field day with the source, without a compiler firewall in
the way.
Change-Id: I49ae3a87fcce1e2dc87a9081f567503e5a98ef6b
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Qt has traditionally considered Windows shortcut files equivalent to
symlinks on Unix file systems. Because of NTFS symlinks, the
interpretation of shotcut files as symlinks is confusing.
In this change, QFileInfo treats shortcut (.lnk) files as regular files
but can follow the pointed object.
In addition, QFileInfo introduces a more comprehensive file type. So
that applications can make well-informed decisions about how to treat a
file system entry.
Based on the implementation of QFileInfo::type(), two inline helper
functions are introduced to QFileInfo.
1. isSymbolicLink, returns true if it points to a symbolic link.
2. isShortcut, returns true if it points to a shortcut.
[ChangeLog][QtCore][QFileInfo] Introduce QFileInfo::type() to replace
the isSymLink method.
Task-number: QTBUG-75869
Change-Id: Icc0dd52f9ad0ea50b0265d77ee0d0a3d25054e39
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Amends 94d7603d51.
The port from QVector<QPlatformTextureList*> to a container of
unique_ptr<QPlatformTextureList> uncovered that QPlatformTextureList
isn't defined for QT_NO_OPENGL builds.
Some unguarded forward-declarations made the old declaration compile
by accident. The new code caught this, so add the #ifdef that had been
missing all along.
Change-Id: If3b14fc24007b1c917a41ab83343c2e5e65fc643
Reviewed-by: Martin Storsjö <martin@martin.st>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Tasuku Suzuki <tasuku.suzuki@qbc.io>
Observed infrequently in the QDBus tests, it would deadlock when
destroying QDBusServer at the same time as qDBusNewConnection was being
executed as they were locking the same locks, but in opposite order.
QDBusServer locks d->lock, then QDBusConnectionManager::instance()->mutex.
While qDBusNewConnection locks QDBusConnectionManager::instance()->mutex,
then serverConnection->lock (and serverConnection here
is QDBusServer's d-pointer).
QOrderedMutexLocker cannot be used in this situation because it
operates on QMutex*, which d->lock (QReadWriteLock) is not.
Change the code to lock QDBusConnectionManager's mutex before d->lock
and then unlock the QMutexLocker where it would previously destruct.
If QDBusConnectionManager has already been destroyed then we pass a
nullptr to the QMutexLocker which is fine and will not do anything.
Fixes: QTBUG-74635
Change-Id: I7f02d7759da67377996ef042c81b0969ccb8aadb
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
As is said in RFC7301 in section 3.1 [1]:
Protocols are named by IANA-registered, opaque, non-empty byte strings
[...]. Empty strings MUST NOT be included and byte strings MUST NOT be
truncated.
[1]: https://tools.ietf.org/html/rfc7301#section-3.1
Change-Id: I38168ac570a433807e16121d5dec46d4ac73c4bf
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
As is said in RFC7301 in section 3.1 [1]:
Protocols are named by IANA-registered, opaque, non-empty byte strings
[...]. Empty strings MUST NOT be included and byte strings MUST NOT be
truncated.
[1]: https://tools.ietf.org/html/rfc7301#section-3.1
Change-Id: I2c41fa99984a53cc58803e5a264d06edac964cc6
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
It is a bit frustrating that all the initialization and cleanup code
are not in the QTLWExtra ctor and dtor. But that is for another patch.
Change-Id: I0e45f89c1a53eb2f9a5699d3fbbef1a628b55432
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Despite the name, it's fully owned by an individual QWidget object.
Also make the member mutable, so we can remove the const_cast hack in
QWidgetPrivate::shareContext(), and protect QT_NO_OPENGL builds, since
the naked pointer compiled by chance due to some unguarded forward
declarations while a unique_ptr will somewhere want to call the dtor,
which doesn't compile on an object of merely forward-declared type.
Change-Id: If8027b55d303822236fcdc1a79e4f3010967b4d2
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Qt for QNX is, by default, built without the clipboard feature.
Change-Id: Ie8a36ceb0c0f0a695ae7d0fcf6f0bd70d2a43e0c
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Reviewed-by: Dan Cape <dcape@qnx.com>
Reviewed-by: Rafael Roquetto <rafael@roquetto.com>
features.animation and features.scroller depend on the feature.
In total, this saves around 180KB from QtCore and 75KB from QtWidgets.
Change-Id: I65aac3ec4d50d62424ee33f44b99f3cfb91121d6
Reviewed-by: Thomas Hartmann <thomas.hartmann@qt.io>
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Follow up to 091a386eaf
Defaulting to querying from the egl config is fine, but dropping support
for the "format" key in the output list in the json config file is not
ideal.
Task-number: QTBUG-76748
Change-Id: I25dc99369d118c300cdef25b464426f6be85453b
Reviewed-by: Johan Helsing <johan.helsing@qt.io>
Add support for astc format files as an experimental feature.
To enable, configure with "-feature-texture_format_astc_experimental"
(Backported from commit 5a4db421bd94acd12a4ac1f77031996b95f85dbf)
Change-Id: I9a2f7b1fa20ba344b79637bafb50ff2bd0596747
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
In Windows-msys syncqt.pl expects CRLF line endings, and does not
work correctly with LF. syncqt.pl was fixed to be line-ending-agnostic.
Task-number: QTBUG-77192
Change-Id: Ie8029238bdd580bcf042ede0d0f64d5f01488406
Reviewed-by: Jörg Bornemann <joerg.bornemann@qt.io>
Also sanitize the initial WebAssembly hack. Both eglfs and wasm lack the concept
of true raster windows. A QWindow with RasterSurface is rendered with OpenGL
no matter what. The two platforms took two different approaches to work around
the rest of the machinery:
- wasm disabled the QOpenGLContext warning for non-OpenGL QWindows,
- eglfs forced the QWindow surfaceType to OpenGLSurface whenever it was
originally set to RasterSurface.
Now, the latter breaks since c4e9eabc30, leaving
all raster window applications failing on eglfs, because flush in the backingstore
is now checking the surface type and disallows OpenGLSurface windows. (just like
how QOpenGLContext disallows RasterSurface windows)
To solve all this correctly, introduce a new platform capability,
OpenGLOnRasterSurface, and remove the special handling in the platform plugins.
Change-Id: I7785dfb1c955577bbdccdc14ebaaac5babdec57c
Fixes: QTBUG-77100
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
(cherry picked from commit 53a6f7b783)
Reviewed-by: Jukka Jokiniva <jukka.jokiniva@qt.io>
I wonder whether a QIcon could be aggregated here by value,
as it has a null state and its default ctor sets d = nullptr.
Change-Id: I7a0f46e9fdd51a93afb5db768d46d93b08f307ec
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Use a vector<unique_ptr> (QVector cannot hold move-only classes), adapt
to different API.
Change-Id: Iece4b1bfcb35a02aac05935963e1e7f8c986b18d
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Turns out that kern.uuid is not as unique as we thought. Googling for
mine finds other instances of the same being used.
Fixes: QTBUG-75371
Change-Id: I95ecabe2f50e450c991afffd159850cc975ec0da
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
