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>
In 5bb178c479, the Android platform plugin was moved from
platforms/android to platforms/. The unforeseen consequence of this was that the plugin
loader for plugins/platforms would now find it, whereas before it would be ignored. It
would therefore be detected as the appropriate plugin, but since it was intended to be
loaded as a static plugin, loading it dynamically would fail.
Instead of fixing the static plugin loading, we remove this hack.
Fixes: QTBUG-78440
Change-Id: Idcb6c075fdebaf67644f32a59d7aaf0d1c0bbe20
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
At first glance, libraryPathMutex is only recursive because
setLibraryPaths(), addLibraryPath() and removeLibraryPath(), all of
which lock libraryPathMutex, may call libraryPaths(), which does, too.
This is easily fixed by splitting libraryPaths() into public
libraryPaths() and private libraryPathsLocked(), the latter expecting
to be called with the libraryPathMutex already held. And this is what
this patch does.
However, on second glance, the building of the initial app_libpaths
calls a monstrous amount of code, incl. QLibraryInfo, and some of
that code probably re-enters one of the library-path functions.
So while this patch is a step towards making libraryPathMutex
non-recursive, it's probably not the end.
Change-Id: I3ed83272ace6966980cf8e1db877f24c89789da3
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
... except four instances in QCoreApplication that would conflict with
another change.
Replace a locally-defined MutexUnlocker with a call to unlock() +
qScopedGuard'ed lock() to avoid having to spell out the locker type
while we can't depend on C++17 CTAD, yet.
In QSettings, move the new mutex locker into and out of
initDefaultPaths(), such as is idiomatic for std::unique_lock, but
wasn't possible with QMutexLocker (which is not movable).
Change-Id: I23056e13ecaa76159db583c7dccc6e05715e0788
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
In all of these cases, the effect of the change is local to one file.
Change-Id: I3bda3aadee3b42e7797183c2330183390b92d1f2
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The old code used the implicit conversions from QAtomicPointer<T> to T*
and vice versa. The semantics of these differ from the ones std::atomic
uses, so we're going to deprecate these, like we did for load() and
store(), too.
This patch fixex some users of these APIs before we deprecate them.
Change-Id: I0a88bb1c359392538bb64b511bfc62381a56a468
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Since e9e16c7464 running webengine application you can get
warning "Attribute Qt::AA_ShareOpenGLContexts must be set before
QCoreApplication is created."
WebEngine set shared open gl context on qt_call_pre_routines, so
when QCoreApplicationPrivate init() runs.
Fixes: QTBUG-76391
Change-Id: I5fc146ed70054b0c1597fe06615cea2d7a8969d8
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
Semi-automated, just needed ~20 manual fixes:
$ find \( -iname \*.cpp -or -iname \*.h \) -exec perl -pe 's/(\.|->)load\(\)/$1loadRelaxed\(\)/g' -i \{\} +
$ find \( -iname \*.cpp -or -iname \*.h \) -exec perl -pe 's/(\.|->)store\(/$1storeRelaxed\(/g' -i \{\} +
It can be easily improved (e.g. for store check that there are no commas
after the opening parens). The most common offender is QLibrary::load,
and some code using std::atomic directly.
Change-Id: I07c38a3c8ed32c924ef4999e85c7e45cf48f0f6c
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
It's a recurring bug seen in user code and a warning will help
reduce it.
Warns only for the attributes that have such requirement in the docs,
but maybe we should be more strict and warn for any attribute.
Change-Id: I68148521953221ad0e8be1028181f52a7f22d2cc
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Encode the consumed/filtered state in the _exit tracepoint and remove
the separate tracking of receiver event handling. Combined, this
reduces the size of the trace file.
Change-Id: Icb3cb2dd47798543905cea450046d6fad559a15b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This allows tools that look for matching `foo_entry/exit` pairs in the
trace data to work properly. An unmatched `_entry` would otherwise
confuse them, making them think that the call stack is continuously
increasing.
Change-Id: Idff7f587ea25c46ec86ad623cc82d503db34a194
Reviewed-by: Christoph Sterz <christoph.sterz@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Additionally, we also add a Q_TRACE_EXIT which runs a trace point when
the scope is exited, leveraging qScopeGuard behind the scenes.
Q_TRACE_SCOPE uses Q_TRACE_EXIT internally - the difference is that
the _SCOPE version enforces the naming scheme of _entry / _exit for the
tracepoints, whereas Q_TRACE_EXIT can be used generically.
Change-Id: I4a2f5ea09f451fcf664d07fd493b679f7527ac06
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This enables overriding the macro so that it translates
to 'None' in the Qt for Python context.
Change-Id: Ib3cecf57eeb0405a1929309b71e9f012a07f11cf
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
Cleans up most of corelib to use nullptr or default enums
where appropriate.
Change-Id: Ifcaac14ecdaaee730f87f10941db3ce407d71ef9
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
navigator.hardwareConcurrency can be accessed from
the main thread only. Read and cache the value on
QCoreApplication initialization.
Change-Id: I731f7f356ce106c7107977783d4b763326af06b6
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
Try to replace all wordings like '.. to 0' with '.. to \nullptr'. Also
checked for 'null pointer' and similar.
Change-Id: I73341f59ba51e0798e816a8b1a532c7c7374b74a
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Don't call or implement functions which are not available when compiling
with QT_DISABLE_DEPRECATED_BEFORE=0x050d00
Change-Id: I949b12bba880c516391f312f58c8748303a1790d
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This is required for non-ANSI paths on Windows.
Change-Id: Id98140e1c2f0426cabbefffd157c4065c3bdfd40
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Replace null and '\c nullptr' with \nullptr in the documentation.
Change-Id: Ib9e0cfc2eb2830b213e6523773603d56180b0998
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
tr() recognizes %n and %Ln. it offers no way to escape lone percent
signs, which implies that they must be interpreted verbatim, which is
what the code actually does. except that it would run off the end if the
% appeared at the end of the string.
Fixes: QTBUG-57171
Done-with: Mateusz Starzycki <mstarzycki@gmail.com>
Change-Id: Icf81925c482be1ea66ec8daafb3e92ad17ea7fab
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Also clean up QTextCodec usage in qmake build and some includes
of qtextcodec.h.
Change-Id: I0475b82690024054add4e85a8724c8ea3adcf62a
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
The overload that takes a maxtime (now duration) was buggy, but we can't
change its behavior, so instead we clarify what it actually does.
Change-Id: I8a04fbaea5847c95b6ec6e73396304ab4debd35b
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
For winrt we cannot rely on the fact, that QThread::current will be called
from the correct thread for the first time. The application's main entry
point creates a suspended thread and starts it right afterwards. At that
moment, other functionality (QLoggingRegistry for example) might have
called QThread::current, which set the wrong thread as the main thread. In
order to avoid this situation, the main thread is explicitly set in
QCoreApplication's constructor.
Task-number: QTBUG-66418
Change-Id: I8b6347357a80eb395ae758bd3d420adef0826751
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
This is the squashed diff from wip/webassembly to dev.
Done-with: Peng Wu <peng.wu@intopalo.com>
Done-with: Sami Enne <sami.enne@intopalo.com>
Done-with: Morten Johan Sørvig <morten.sorvig@qt.io>
Started-by: Andrew Knight <andrew.knight@intopalo.com>
Change-Id: I6562433c0a38d6ec49ab675e0f104f2665f3392d
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
The application name wasn't always printed, so we try try a few
more possibilities before falling back to the process name.
We also run the check as early as possible, instead of relying
on a QCoreApplication.
We do not have to provide a dialog to the user, as macOS will do
this for us if the application is launched from Finder.
Change-Id: Ifbec86946d60294806364e08964852fd4b74ff56
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Add it to configure.json and replace all occurrences of QT_NO_THREAD
with QT_CONFIG(thread). Add conditions for other features that depend
on thread support. Remove conditions where we can use the QMutex and
QThreadStorage stubs.
Change-Id: I284e5d794fda9a4c6f4a1ab29e55aa686272a0eb
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
In some places we call startingUp(), in others we don't. It's probably
ok for those that have just created an object of a given class, which
knows whether the virtual call is necessary or not. But for the generic
case, we do call it.
Change-Id: If48c5c2e920c433298f1fffd153ee1cc75703204
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Use CamelCase, like the API they're tracing.
Change-Id: Ie718ab624d17c9186bcf05cc1276c8eccad7f454
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
Add tracepoints in all the main codepaths for event handling:
* QEvent ctors/dtor
* QCoreApplication::postEvent, sendEvent and sendSpontaneousEvent
* QCoreApplication / QApplication::notify, and around the handling
of event filters as well
I'm switching the name of the tracepoints themselves to have the
very same casing of the functions in Qt's own source code, this
improves readability a lot. The pre-existing ones will be changed
in an upcoming patch.
Change-Id: Iae2ba2bfdd76a82c85445bb5b86434e910427a70
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
There's no clear reason for these functions to be inline;
this prepares a tracepoint hook.
Change-Id: I3a6110a9333db4850c1d97038d5bfae8ab25d5d8
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When a value is omitted with \omitvalue, the comment must be either
deleted or omitted with \omit .. \endomit. Otherwise the comment is
merged into the comment for the previous value
Change-Id: Ic5cfb20d36853526ae939529b4609712cd0eb341
Reviewed-by: Martin Smith <martin.smith@qt.io>
Reviewed-by: Topi Reiniö <topi.reinio@qt.io>
A step towards having the application do its event dispatching though the
thread data's dispatcher, like QEventLoop, instead of keeping two references
to the same dispatcher, one in QCoreApplicationPrivate and one in QThreadData.
Change-Id: I7b215e7e99869d25638ec67f0666f632a508cc0f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QCoreApplication::setEventDispatcher sets the dispatcher on the current
thread, not on QCoreApplictionPrivate, so when running init() we don't
have an event dispatcher set.
Change-Id: Ia008e68b70777779ab14f1f7b9eeadac9adbcf7b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Giuseppe D'Angelo
BogDan Vatra
Marc Mutz
Mikhail Svetkin
Liang Qi
Michal Klocek
Sergio Martins
Milian Wolff
Venugopal Shivashankar
Allan Sandfeld Jensen
Morten Johan Sørvig
Qt Forward Merge Bot
Christian Ehrlicher
Thiago Macieira
Oswald Buddenhagen
David Faure
Tor Arne Vestbø
Oliver Wolff
Ulf Hermann