When a foreign event loop that does not enter an alertable wait state
is running (which is also the case when a native dialog window is
modal), pipe handlers would freeze temporarily due to their APC
callbacks not being invoked.
We address this problem by moving the I/O callbacks to the Windows
thread pool, and only posting completion events to the main loop
from there. That makes the actual I/O completely independent from
any main loop, while the signal delivery works also with foreign
loops (because Qt event delivery uses Windows messages, which foreign
loops typically handle correctly).
As a nice side effect, performance (and in particular scalability)
is improved.
Several other approaches have been tried:
1) Using QWinEventNotifier was about a quarter slower and scaled much
worse. Additionally, it also required a rather egregious hack to
handle the (pathological) case of a single thread talking to both
ends of a QLocalSocket synchronously.
2) Queuing APCs from the thread pool to the main thread and also
posting wake-up events to its event loop, and handling I/O on the
main thread; this performed roughly like this solution, but scaled
half as well, and the separate wake-up path was still deemed hacky.
3) Only posting wake-up events to the main thread from the thread pool,
and still handling I/O on the main thread; this still performed
comparably to 2), and the pathological case was not handled at all.
4) Using this approach for reads and that of 3) for writes was slightly
faster with big amounts of data, but scaled slightly worse, and the
diverging implementations were deemed not desirable.
Fixes: QTBUG-64443
Change-Id: I66443c3021d6ba98639a214c3e768be97d2cf14b
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Weird macro had inverted notion of type safety: instead of casting
parameters and return values (they are pointers) to work with generic
OPENSSL_sk_xxx functions, it was ... casting a function pointer to an invalid
type to get ... nothing, but UB. Home-brewed (un)'safestack'!!!
Change-Id: Ib91a7ba4cd472f370836797e422456f91a4385b0
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
(cherry picked from commit 9fc2f1f076b953ff0979fb73ed6e70ac9bde278c)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Also since we have to properly support 'no-ssl' configure option
(alas, we support QSslCertificate on such builds) - introduce
a minimal crippled QTlsBackendCertOnly, which depends on
X509CertificateGeneric.
Fixes: QTBUG-90954
Task-number: QTBUG-65922
Change-Id: Ib9d62903f16b7c0eaaa23e319a822c24a7631dc6
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit 41fc143635c25f937a557f09890601f6c7d38736)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
... which makes me wonder, why isn't this stuff bumped automatically
when a minor version branch is created?
Change-Id: Ia43f898163a4baa0896a09bd13d65cf534fe1df5
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Avoid painting errors with dpr < 1, also for the PassThrough
mode.
This limits the minimum high-dpi scale factor to 1,
for the code path which determines the scale factor
based on screen DPI.
Pick-to: 6.1
Task-number: QTBUG-89948
Change-Id: I14b3f130f0ae141d5f05c87437f926a9f76d1dec
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
If control is nullptr in the beginning, then it will be nullptr later as
well, and the function won't do anything. And if the effectiveTextFormat
is Qt::PlainText, then the linkToCopy will be empty, and the function
won't do anything, either.
So we can just handle these cases right away, making the code simplier.
Fixes static analyzer warning 43de3f3125108b4353afd94e94f59926.
Pick-to: 6.1
Change-Id: I5b8eb94a1e40c2725de6a168298d8f3bcde748eb
Reviewed-by: David Skoland <david.skoland@qt.io>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
The graphicsview doc snippet has an explicit
QT_REQUIRE_CONFIG(printdialog) check that will fail the
build if Qt is not configured with "printdialog".
This check should be mirrored in the cmake file, so that
we skip building the snippet in the first place if the
feature is not available.
As it stood, we would build the snippet if "printsupport"
was enabled. But this is too granular, as you can configure
Qt to have general print support, but at the same time, skip
the print dialog. This caused the build to fail on iOS.
This patch will check the correct feature in the cmake file.
Change-Id: I31acc1f7c257e08374ea7b84a7fc38c66f214271
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Move the cursor out of the way to prevent it from interfering.
Amends 3f3d5e6716.
Task-number: QTBUG-89082
Pick-to: 5.15
Pick-to: 6.0
Change-Id: I0efbe2b5618a04e92839083a1bd3383e05d6ff93
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
Do not include the first parameter in the main tag anymore. That is
a left over from before unaligned access was added.
Pick-to: 6.1
Change-Id: I2caf027c6b9930d3e23fe5ec8bdcd45db34835a9
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Contrary to its name, this command was also setting a target property.
Since it was only called in one place and that caller can just as
easily set the property instead, rename the command to make clear its
internal nature and refactor it so that the caller is responsible for
setting that property instead.
Also make it an error rather than just a warning if the command is used
for a target that doesn't belong to any module. Since this is now
unambiguously an internal command, we should always expect the target
to belong to a module.
Pick-to: 6.1
Change-Id: I929a652ddd482653868fc9df887f38f4bc7f35d9
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Some popular fonts on linux systems follow a naming convention,
where font styles are named 'Regular', 'Italic', 'Bold' and
'Bold Italic'. Qt so far only accepts the words Italic and Oblique
interchangably, but not with Regular and Normal. This change will
make sure that Regular is interpreted the same as the Normal font
style.
Fixes: QTBUG-90396
Pick-to: 6.0 6.1 5.15
Change-Id: Ibbaf086de742c91f4d380c937ca80e846aa32a2e
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
- Fix the save as file dialog being in "Open" mode by setting
acceptMode
- Fix clazy warnings about detaching QList by using constFirst()
- Fix clazy warning about passing a context to slot connection
Pick-to: 6.0 6.1 5.15
Change-Id: I0c800e9829e118fcec477322aa2a13660e3b51d2
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
QMakeLibraryInfo uses external data to produce paths. This causes
issues when trying to use it out of the existing qmake environment.
Add data fields that contain the path to the binary that uses
QMakeLibraryInfo and manually specified qtconf.
Task-number: QTBUG-75870
Change-Id: Ic6fa274ede3a9287826ff66c79f155b10d0d455c
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Since the QT_CONFIGURE_HOSTBINDIR_TO_HOSTPREFIX_PATH and
QT_CONFIGURE_HOSTBINDIR_TO_EXTPREFIX_PATH definitions keep the same
value in modern CMake build, no need to have special handling in cases
where these values are used in qmake. Also it will be useful to
specify the relative path to the prefix directory from the directories
different of 'bin' when use QMakeLibraryInfo.
Task-number: QTBUG-75870
Change-Id: I5a777001eb334dcf05e22853a514d4257352d59b
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
QT_CONFIGURE_CROSSBUILD is always '0', so it's not necessary to keep
conditional compilation in QMakeLibraryInfo.
Also the QT_CONFIGURE_SYSROOTIFY_PREFIX definition is never used
in the project.
Task-number: QTBUG-75870
Change-Id: I1fe42dce40fddc6a72254736883f93aa4727f6b3
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Including one entry of the value in the header is pointless after
the unaligned access rewrite, and a potentially dangerous pattern,
though safe here due to overchecking.
Pick-to: 6.1
Change-Id: I4c0380040f89920467c309503408f1df6f88423f
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Previously, *datetime was only written to if the parse was a success.
When parsing a date-time that's invalid by virtue of falling in a
spring-forward gap, the parser returns a date-time that is invalid but
has a toMSecsSinceEpoch() suitable for use in creating a sensible
interpretation of the parsed string (in offset by the width of the gap
from the specified position in the gap). It is more useful to return
this value than a default-created QDateTime.
Change-Id: I89f39e729b1f9fede1532d8b82f6f676477ddadb
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
When refreshing a QDateTime(,, Qt::LocalTime) we call mktime on data
obtained from it, passing in the DST status (when known; this keeps
two otherwise identical times in a fall-back distinct). One of the
tests relies on changing zone under the feet of such a date-time,
created in Hawaiian standard time; it serializes it, the reads it back
in Western Australian Daylight-saving time and expects the results to
be equal. However, the two differ in DST-ness, which leads to mktime()
failing for the Hawaiian original, with unwelcome results.
Notice this case, failure with DST-ness claimed known, and retry
without the claim, so as to correct the DST-ness.
Change-Id: Id0278df53130f76fc587769efe946ca9af1adc26
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When computing a recent and imminent time, to bracket the time for
which we want data, take care not to cycle round to the other end of
the range of representable times.
Rephrased comments on this function, in the process, to more
accurately reflect what we're doing.
Change-Id: Iacd36186abc6b19d0ca03981aec80b2c5af077b3
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The addition of a sanity-assertion revealed that, for an invalid
time-zone, refreshZonedDateTime() left epochMSecs unset but computed
offsetFromUtc from it none the less. Leave it as zero in that case, or
any other where the conversion to UTC didn't give valid date and time.
Change-Id: I0ebd955798532e91e7e211bf065667e313ee5c2d
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This take more lines but makes the condition clearer and the lines
shorter, even after converting to use the names for constants in the
condition.
Change-Id: I9e5b7b79ff62095ed11b8723be238444fd32d9c1
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Various constants used in qdatetime.cpp were cast to qint64() where
used. There ware also some Q_INT64_C()s, two of which should have used
members of this enum; the third suggested a new addition to the enum.
Adding that and basing the enum on qint64 eliminates the need for
casting, although one asprintf() does now require a cast back to int.
There were also some redundant casts to qint64(), so I removed those.
Change-Id: Ia51ad8020f037badb1506ca379da19098a8655f8
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QPointF operator== is fuzzy, hence it can't be hashed efficiently.
Prevent the erroneous addition of the overload by client code
by providing it as deleted.
Change-Id: I2cfaaf5c2c8896ec9a7929f7c7bf52a912d0450f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
In QGraphicsScene's function draw, don't copy the children into a
const copy (to prevent detaching in the loops), as they might get
sorted. Instead, use a const reference.
Amends e349f787d5, which introduced
the bug by making a copy of the unsorted list.
Fixes: QTBUG-81316
Pick-to: 6.0 6.1
Change-Id: Iab9b87e75178a883806088a9db234c7a2aaa9301
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
After the unaligned ICC profile fix, we could read 2 bytes past the end
of the buffer for curv elements with 0 parameters, though we wouldn't
use or return the data.
Pick-to: 6.1 6.0 5.15
Change-Id: Ibea9fa2840c01d690716d8f48e301fbfa67aa163
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Before this change, the QMetaType for T const* where T is derived from
QObject would not store the static QMetaObject of T. This commit changes
this. As a consequence, the metatype system can now convert between
const and non-const pointers to QObject. Note that this allows casting
const away; but so does C++ with const_cast.
In addition, a new flag, QMetaType::IsImmutable is introduced, and used
to tag the metatypes of pointer to const types. This allows code to
discern between pointers to mutable and const QObjects, which is
relevant for the QML engine.
Task-number: QTBUG-82354
Change-Id: I3e4e4f39f565bd99a65e161528ce5304df73d6d6
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Getters and setters of bindable properties have to be carefully
written to avoid several problems. This patch adds documentation
for this.
Task-number: QTBUG-89505
Task-number: QTBUG-90511
Change-Id: Ib25590b3d8d95c490d9555c0f258f48cb6cfe4a9
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Karsten Heimrich <karsten.heimrich@qt.io>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Assert the expected side effect of createTLSysExtra, which might not
allocate a window, but must do so in this case (as we have already
returned if the QWidget is not a window).
Fixes static analyzer warning 2f3bbfe8addb586445e96f8906d6769e
Pick-to: 6.1
Change-Id: I4d5b8651b3510eff8e4a7b25889c0521ba6a4247
Reviewed-by: David Skoland <david.skoland@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Messages about Qt deficiencies reduced to info, and info set as the
default message level.
Pick-to: 6.1 6.0 5.15
Fixes: QTBUG-91401
Change-Id: Ia97438f08de5e0383e77631dca9b253b47ef8167
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Instead of intercepting the WM_QT_SENDPOSTEDEVENTS message in the hook
procedure, we can handle it at receive points, making
qt_GetMessageHook() unnecessary.
Including general performance improvements, this patch fixes the issue
where some applications (e.g. Ableton Live) do not call a chain of
nested hooks for plugins.
Pick-to: 6.1 5.15
Fixes: QTBUG-90949
Change-Id: If8e96848392c6f10d45af2aac0567707d16af673
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Nuno Santos <nunosantos@imaginando.net>
White-point was calculated wrongly and some tables could cause bad
behavior in the tables.
Pick-to: 6.1 6.0 5.15
Change-Id: I24e8f5f3cc1306f5f898a4acbf7b008e26bd04e2
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
After QPoint(F), it's now the time of QSize(F) and QVectorND,
which can be unambiguously decomposed.
[ChangeLog][QtCore][QSize] QSize is now usable in a structured
binding declaration.
[ChangeLog][QtCore][QSizeF] QSizeF is now usable in a structured
binding declaration.
[ChangeLog][QtGui][QVector2D] QVector2D is now usable in a
structured binding declaration.
[ChangeLog][QtGui][QVector3D] QVector3D is now usable in a
structured binding declaration.
[ChangeLog][QtGui][QVector4D] QVector4D is now usable in a
structured binding declaration.
Change-Id: I67bb152f4210f2be27607179cd2ec522174cc483
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
If we are connected or bound, then we must have an engine.
Otherwise in code we frequently test for the d->socketEngine pointer,
including in waitForConnected.
Fixes static anlyzer warning 3b37ce4102843c97ceef2b7bb68a409d.
Pick-to: 6.1
Change-Id: Ib9818282ad2813db97859cbe1f184bcfb5ea4ab4
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
QIcon::detach might result in the d pointer being nullptr, and other
code tests for that condition and allocates the d pointer explicitly.
Change the order of the tests in setIsMask as well, fix static analzyer
warning 0a4723ca2d216ec8883a3d412aeded4c.
Pick-to: 6.1
Change-Id: Idca12f230036362f13cbaee1ae4f0e5ce6726015
Reviewed-by: David Skoland <david.skoland@qt.io>
Reviewed-by: Oliver Eftevaag <oliver.eftevaag@qt.io>
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Move the qmake-specific logic of the QLibraryInfo class to
qmake internals. 'qconfig.cpp.in' now stores information about
the library info entries to keep them consistent between qmake
and the Core library. qmake requires specific features enabled
in the Core library, so building qmake will be skipped if the
features are not enabled.
All flags directly related to the qmake have been removed from
Core lib.
Remove all bootstrap related sections from qmake CMakeLists.txt
Task-number: QTBUG-89369
Change-Id: I26de157d3bfd4a5526699296e9d46e1c180b89ae
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
When attaching a continuation with the default (QtFuture::Launch::Sync)
policy, it will be launched in the same thread where the parent has
been executing, or in the thread where the parent lives, if the
continuation is attached after the parent has already finished. Fixed
the test-case to make sure the continuations are attached while the
parent is still running, so that they can be resolved in the parent's
context.
Fixes: QTBUG-91373
Pick-to: 6.1
Change-Id: I82b3b0c47b76d121b44bd512659c08b3b474e351
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Reviewed-by: Jarek Kobus <jaroslaw.kobus@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Be more precise about attaching a continuation with the default
(QtFuture::Launch::Sync) launch policy after a continuation with
context.
Pick-to: 6.1
Change-Id: I5b80063df2443e5742033864ba012bf34ed4cdf7
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Reviewed-by: Jarek Kobus <jaroslaw.kobus@qt.io>
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
