Within QTreeView::setModel() the header might emit columnCountChanged
which then tries to update the geometries based on the old model which
is wrong.
Fix it by setting geometryRecursionBlock to true so
QTreeView::updateGeometries() will not ask the old model for it's data.
Fixes: QTBUG-75982
Change-Id: Ia0dd36cd7c6c5347fbc285deac43da6941accbe7
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Workaround for libpng bug in GCC 8.1.0.
Task-number: QTQAINFRA-3303
Change-Id: Id7668e795cb4ab16de3199fc3727d844aa31bfad
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
Task-number: QTBUG-74953
Change-Id: I9a630c9245d8b0afe40ade9199cf4f1d358275da
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
We used to have the assumption that moving the cursor to an out of
range position is valid and will result in a hidden cursor.
This is apparently not the case. For example, on an RPi4 with Mesa
V3D we get lots of funny artifacts after doing drmModeMoveCursor()
to invalid positions.
To remedy this, start hiding the cursor correctly when the position
is clearly out of the screen's bounds.
Task-number: QTBUG-79924
Change-Id: I3ef7ad0ce928546399443f21452f0b6deadf8036
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Adds instrumentation for fuzzing to the binaries but links to the usual
main function instead of a fuzzer's. The similar sanitizer "fuzzer"
should then be used only for building the test itself.
Requires clang 6 or higher.
Change-Id: I24ee1f018b0b97f2977dc86fbdc29a164d7c4e01
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Android uses its own time-zone naming, which includes a zone called
"Canada/East-Saskatchewan", whose second component is 17 characters
long. This violates a rule in the IANA naming scheme for zones, that
limits components to 14 characters each. So tweak the isValidId()
check to allow Android its long names.
Android has added Outer Mongolian time-zones, which are as borked as
many others in 1970, so blacklist those transitionEachZone() tests.
Fixes: QTBUG-69128
Change-Id: I46f674f095431335b16900860d83b624257ae3bb
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
The reader uses int variables, so use the same in the writer.
Change-Id: I1496b069cc534f1a838dfffd15c94c7cacd3dd93
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Since the backend is async, the settings will not be ready to read/write
instantly as on other platforms, but only be ready after the
filesystem has been synced to the sandbox. This takes at least 250 to
500 ms. The QSettings status() or isWritable() can be used to discern when the
settings are ready for use.
This also fixes a crash in threaded wasm
Task-number: QTBUG-70002
Fixes: QTBUG-63923
Fixes: QTBUG-79650
Change-Id: If24c6ada1b91b2a565ed6733da74972c3027f622
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
So we are in sync with QWaitCondition::wait().
Task-number: QTBUG-64266
Change-Id: I1d7487786513241cedd35d202c4ddee4937b08ec
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The JSON collection step has to be target_predeps in order to be
executed if the only place its output is referred to is INSTALLS.
Furthermore, some CONFIG options clear the INSTALLS variable.
Therefore, we need to add the metatypes CONFIG entries after
those.
Change-Id: I4694ab1d82c13cb4e3886c1722a03255d14b7f29
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Otherwise we cannot find them when building further projects within the
same module. For example, qmlmodels needs the metatypes of qml in order
to build its plugins.qmltypes file, but the QML metatypes would only be
available after installing.
Change-Id: Ic2cdfa5b57fd3e5977eea16f2a9b192386737a1f
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
The atomic modesetting support was not prepared for page flips being
issued from different (per-screen) threads.
This could be seen with the threaded render loop of Qt Quick: having a
QQuickWindow per screen means having a dedicated render thread for each
screen. QKmsDevice used simply instance variables to keep track of the
request. This leads to the commit failing with EBUSY sooner or later.
Make the atomic request and related variables thread local.
This prevents failing drmModeAtomicCommit() with 2 or more screens and
the threaded render loop. It does not fix other potential issues when
waiting for page flips to complete, that is to be tackled separately.
Task-number: QTBUG-74953
Change-Id: I2dac10d5e9bdc0cb556ac78c9643c96d40d692e4
Reviewed-by: Johan Helsing <johan.helsing@qt.io>
QObjectPrivate::threadData used to be a QThreadData *, and was
read and written from multiple threads without proper synchronization.
As an example, it was read from QCoreApplication::postEvent and
written from QObject::moveToThread, therefore causing UB.
Port threadData to a proper atomic, removing the races. Fix all usage
points.
In general, QObject is documented to be simply reentrant,
not thread-safe, and certain bits (e.g. timers, moveToThread)
are not even reentrant. The reasoning therefore is that a given
QObject's threadData is not supposed to be touched by multiple
threads without some synchronization happening elsewhere, and
therefore relaxed loads should be sufficient.
As drive-by change: refactor QCoreApplication::postEvent.
It was particularly subtle, because it had a loop using a volatile
to cope with the possibility of the receiver object switching thread
while we tried to lock its thread's event queue.
However, volatile does not achieve any synchronization, so drop it,
and refactor the algorithm using better locking primitives.
Put this algorithm in a common place, and also reuse it from
removePostedEvents, which was lacking any synchronization.
Change-Id: Icc755f7eb418ff54b33db4bdd87fd8eaf4e82c7a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The event manager has a family of the functions for registering
sockets notifiers, event notifiers, and timers. To ensure
efficient debugging, it would be useful to have one approach
regarding input parameters validation for the entire set of
that functions. Based on registerSocketNotifier() implementation,
this patch offers the same debugging principles for
QWinEventNotifier and QTimer. Some debug messages have also
been refined.
Change-Id: I1418ef43c51f7b794462b5e9c8a849633e0c60f9
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Since 65314b6ce8 there is a TextFormat
for MarkdownText, and QWidgetTextControl supports that, but QLabel
does it in its own way and sets plain text or rich text on the text
document itself. Add a code path for MarkdownText there.
[ChangeLog][QtWidgets][QLabel] Markdown is now a supported textFormat
for QLabel.
Fixes: QTBUG-79766
Change-Id: Ib9370ef300089af2c4d6070e545c5470f32833a8
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
With the qrcFiles entry in the deployment JSON for Android, it can now
pass this on to qmlimportscanner for scanning the qrc files for the
available imports. This enables qmake to populate the qrc files it has
referenced in the project, be it generated by qmake or added by the
user.
Task-number: QTBUG-55259
Change-Id: Ic512ce6f24508b3ea09ebdd07ac4446debfd9155
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
With logging enabled, all the output will slow down
execution and fill up your hard disc in about a day.
Task-number: QTBUG-79050
Change-Id: I5dcac2f349f7dbe471a5e6dd7006b89d312aeeaf
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
They were tucked away in the back-end of the isTimeZoneIdAvailable()
test, but a separate isValidId() test had been added more recently,
which made some (arguably all) of them redundant. Reworked this test
in the process, so that the QSKIP() happens in _data() once instead of
in the test that's never run because there are no data rows.
Change-Id: Icaa6227ace9a1aa944d085691cdcfb3adf4a51dc
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
In some cases of inheritance, timer deletion can be triggered from the
event dispatcher destructor where QThreadData::eventDispatcher is
already nullptr. Despite the fact that the application is in shutdown
phase, we should free the resource.
Change-Id: I61ed1d817fd7638953f7d629823f19d4f6f1ee00
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
And we'll get rid of them in Qt 6.
Task-number: QTBUG-75638
Change-Id: I34764f93bf579da0640a930d9160783ea9c8317d
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
QFontDatabase::systemFont(FixedFont) determines the font for inline code and
code blocks in a markdown document. Now we change the size of that font to the
same size as QTextDocument::defaultFont() so that the user has the ability to
customize the font size in each document instead of only system-wide.
Change-Id: Ief7367336f7613e88695dbb08bcb7e9f50db8961
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
This fixes a bug where app build would fail with AssertionError: SIMD is
used, but not supported in WASM mode yet
This patch was wiped out by the recent freetype update and is exactly
the same as
44b91a619d
Fixes: QTBUG-79938
Change-Id: Iaa8f23c83d0488ddd351454a674a6cad76e7cc8b
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Liang Qi <liang.qi@qt.io>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
In file included from ../../include/QtCore/qlist.h:1,
from ../../include/QtCore/../../../qtbase_dev_de_verdad/src/corelib/kernel/qobject.h:49,
from ../../include/QtCore/qobject.h:1,
from ../../include/QtCore/../../../qtbase_dev_de_verdad/src/corelib/kernel/qcoreapplication.h:46,
from ../../include/QtCore/qcoreapplication.h:1,
from /src/widgets/kernel/../../gui/kernel/../../corelib/global/qt_pch.h:66,
from /src/widgets/kernel/../../gui/kernel/qt_gui_pch.h:48,
from /src/widgets/kernel/qt_widgets_pch.h:48:
../../include/QtCore/../../../qtbase_dev_de_verdad/src/corelib/tools/qlist.h: In instantiation of ‘void QList<T>::node_construct(QList<T>::Node*, const T&) [with T = QTableWidgetSelectionRange]’:
../../include/QtCore/../../../qtbase_dev_de_verdad/src/corelib/tools/qlist.h:614:13: required from ‘void QList<T>::append(const T&) [with T = QTableWidgetSelectionRange]’
/src/widgets/itemviews/qtablewidget.cpp:2416:71: required from here
../../include/QtCore/../../../qtbase_dev_de_verdad/src/corelib/tools/qlist.h:471:35: warning: implicitly-declared ‘constexpr QTableWidgetSelectionRange& QTableWidgetSelectionRange::operator=(const QTableWidgetSelectionRange&)’ is deprecated [-Wdeprecated-copy]
471 | else *reinterpret_cast<T*>(n) = t;
| ~~~~~~~~~~~~~~~~~~~~~~~~~^~~
In file included from /src/widgets/itemviews/qtablewidget.cpp:40:
/src/widgets/itemviews/qtablewidget.h:52:24: note: because ‘QTableWidgetSelectionRange’ has user-provided ‘QTableWidgetSelectionRange::QTableWidgetSelectionRange(const QTableWidgetSelectionRange&)’
52 | class Q_WIDGETS_EXPORT QTableWidgetSelectionRange
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
Change-Id: Iad959315ad374ef288f5fffd15d6876cb63bce8e
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
(cherry picked from commit 127ed7e6e0)
Otherwise we end up with using the same plane for multiple crtcs, which
fails (with atomic - planes have no importance when atomic is not enabled).
This fixes systems where we run with atomic enabled and there are multiple
primary planes with their possible_crtcs matching multiple crtcs.
Task-number: QTBUG-74953
Change-Id: I8bcbdd389265d09f8851187881102fb5b9a83b5c
Reviewed-by: Johan Helsing <johan.helsing@qt.io>
With atomic enabled (QT_QPA_EGLFS_KMS_ATOMIC=1) the plane chosen for
a crtc becomes important. The current logic is pretty broken when there
are multiple planes available with many of them marked as being supported
by multiple CRTCs. Choosing the same plane for multiple crtcs results in
failing the atomic commit. This happens with a RPi4 with two screens
connected, for example. (because there are 2*3 planes: one primary, one
overlay, one cursor for each screen, but Qt is trying to use the same
primary plane with both CRTCs)
The issue does not surface in special environments where there is one
(dedicated) plane per crtc exposed by drm, but becomes apparent on a PC
or on the RPi where the setup described above is the common case.
As a temporary solution allow doing the following:
export QT_QPA_EGLFS_KMS_PLANES_FOR_CRTCS="49,28:78,57"
This would then force using plane id 28 for crtc id 49, and plane id 57
for crtc id 78. (the ids can be discovered either from the eglfs debug
logs, or by checking /sys/kernel/debug/dri/<N>/state)
This is to be complemented with a fix for picking a unique primary plane
for all CRTCs but that's going to be a separate patch. Allowing a manual
override is important regardless since it helps troubleshooting.
Task-number: QTBUG-74953
Change-Id: Ie03f80dac31813f2c4489235d435769dd3d3883e
Reviewed-by: Johan Helsing <johan.helsing@qt.io>
I see no reason why this would make sense to look for plugins in the current
directory. And when there are plugins there, it may actually be wrong
Change-Id: I5f5aa168021fedddafce90effde0d5762cd0c4c5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
qmake matches those targets in a purely lexical way, disregarding any
scopes. do_install is much too generic and clashes with another
do_install target in the qmltypes installation code in qtdeclarative.
Change-Id: I8edab329eacc548cee880a9182a1e5ae4dd095fb
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
If the qmltypes are generated by qmltyperegistrar, we should not offer a
rule to generate them using qmlplugindump. Otherwise we get warnings
from make.
Change-Id: Ifb88cbff8e6c577f9889ff3f071bf31f48f634ad
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
On NTFS, a junction point can be created and deleted by the mklink and
rmdir commands, respectively. If a directory is not identified
correctly as a junction, then applications will likely try to remove
it using recursive methods, leading to fatal data loss.
With this change, Qt can identify file system entries as junctions,
allowing applications to use the correct file system operation to
remove it.
The test needs to delay the cleaning up of junctions and files it
creates until the checks are complete; since they might fail and make
the test function return prematurely, use a scope guard.
[ChangeLog][QtCore][QFileInfo] Add QFileInfo::isJunction so that
applications can recognize NTFS file system entries as junctions
Task-number: QTBUG-75869
Change-Id: I3c208245afbd9fb7555515fb776ff63b133ca858
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
The dummy implementation of QWaitCondition which is used when the thread
feature is not available lacks some functions which were recently added.
So we have to add them now.
Change-Id: I32720e0857a1bd3fcf0e70078404a38dd8a8ca88
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
The linker must not throw away the lto.o file. We now instruct the
linker to create a non-temporary lto.o, dependent on the target name.
In order to do that we introduce a new mkspec variable
QMAKE_LFLAGS_LTCG_SEPARATE_DEBUG_INFO. This variable can contain
single-$ variable references that get evaluated when loading ltcg.prf.
Fixes: QTBUG-72846
Change-Id: I0ea882628d63e5406ba0ee68c7435af597364b0f
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
When a fixed size the window is moved to another screen
by QPlatformWindow::initialGeometry(), the size constraints
would be incorrectly scaled using the initial screen in the handling of
WM_GETMINMAXINFO.
To fix this, pass the resulting screen out of QPlatformWindow::initialGeometry()
and use it during the window creation phase.
Fixes: QTBUG-77307
Change-Id: I149a2a65e816da841a32abc14a495925bf9cc6f6
Reviewed-by: André de la Rocha <andre.rocha@qt.io>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Mark the class as [[nodiscard]] to ensure that instances are given names
and do not destruct (roll back) immediately. This avoids accidental code
like this:
QScopedValueRollback<Foo>(bar, baz);
which rolls back instantly, when it should be
QScopedValueRollback<Foo> blah(bar, baz);
which rolls back at the end of the scope.
Change-Id: I00269fe325b804078bd0a9d5058c941af7ba5597
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Obtain the client area and output frame and position relative to the
parent window.
Task-number: QTBUG-79861
Change-Id: I193dfbcdec7e27d32a70ada08ba271260eedc969
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
This is how the fake screen currently behaves in the xcb plugin. So if we are
to deduplicate, it's probably best to match current behavior first.
I still left an option for not being a virtual sibling in case other platforms
don't support showing a window on a placeholder screen.
Task-number: QTBUG-79711
Change-Id: I4e8b44d892efb85fdb003f1d473d0867442d7e4e
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
This flag tells the app to use as much of the screen as possible
while still keeping system UI visible, and can be supported on Android
by using translucent system UI, similar to iOS.
What this does:
1. It changes the current fullscreen/not-fullscreen logic to
allow three states: fullscreen, fullscreen with translucent
decorations and not-fullscreen.
2. In order for it to work, we have to send the actual
screen geometry and available geometry, at least in the case
where the user needs to know the available geometry to
know the safe area of the window. So we get the real screen
metrics and pass these to the QPA plugin (API level 17, so
we can do that now that the minimum version is 21.)
3. Note that getting the insets and calculating the useable
area does not work for non-fullscreen windows, since Android
is quite inconsistent in this respect. So in this case we
just use the window size and origin of 0,0 for the available
geometry.
4. Since we are touching this code anyway, this patch also tries to use
more consistent wording (calling it "available geometry" everywhere
instead of desktop geometry in some places and just geometry in
others, etc.)
[ChangeLog][Android] Qt::MaximizeUsingFullscreenGeometryHint window
flag is now supported, and will make the window fullscreen, but keep
the system UI on-screen, with a translucent background color.
Fixes: QTBUG-74202
Change-Id: I7a59a6c6fb51ebbdb86e7149e794726e67001279
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: BogDan Vatra <bogdan@kdab.com>
Static text controls should not return a ValueProvider. Otherwise,
Narrator reads out static texts and announces whether they are
editable or not, which is confusing to users. It should only read
out the text itself.
Fixes: QTBUG-79613
Change-Id: I080cd6a5db10f6f673b50c40ac7d87c3737d9b55
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
In QPlatformScreen there is a convenience function which resizes windows
when the screen changes. This did not have support for
MaximizeUsingFullscreenGeometryHint and would mistakenly resize these windows
to the available geometry.
Since not all QPA plugins support this hint, we have to add a capability
flag to avoid changing behavior on platforms where it works as intended.
Task-number: QTBUG-74202
Change-Id: Ife88f597fbb3affa722f63ac18fb5719ffa8ed33
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
QMenuBar is only forward declared in QMainWindow. I don't think this is
a recent change or breakage, just noticed then when running a complete
build of the 5.15 branch.
Change-Id: Ia1a072044bd26c5f3f51ec5f5b0f72f065a0576f
Reviewed-by: Johanna Äijälä <johanna.aijala@qt.io>
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
If an item was updated while the display was in sleep mode then when the
display came back from sleep it would not be updated. This change detects
when the display returns from sleep and repaints the windows to ensure
the updated items are presented.
Fixes: QTBUG-76307
Change-Id: I5fff5209e8a5c359d06ba1df61944690e9475ea6
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
