Printing pools by calling [NSAutoreleasePool showPools] will now give
a more detailed view of where the various pools in the stack were
created, eg:
AUTORELEASE POOLS for thread 0x1000be5c0
17 releases pending.
[0x107802000] ................ PAGE (hot) (cold)
[0x107802038] ################ POOL 0x107802038
[0x107802040] 0x107791c70 ^-- allocated in function: main
[0x107802048] ################ POOL 0x107802048
[0x107802050] 0x1073b2e80 ^-- allocated in function: QCocoaWindow::initialize()
[0x107802058] 0x107111ed0 NSCompositeAppearance
[0x107802060] 0x107111ed0 NSCompositeAppearance
[0x107802068] 0x107111ed0 NSCompositeAppearance
[0x107802070] 0x1073bbe10 __NSCFString
[0x107802078] 0x1073bbde0 _NSViewBackingLayer
[0x107802080] 0x1073bc100 NSWeakObjectValue
[0x107802088] 0x1073bbe40 QNSView
[0x107802090] 0x1073bbe40 QNSView
[0x107802098] 0x107111ed0 NSCompositeAppearance
[0x1078020a0] 0x107111ed0 NSCompositeAppearance
[0x1078020a8] 0x1073bbe40 QNSView
[0x1078020b0] ################ POOL 0x1078020b0
[0x1078020b8] 0x1073bbe30 ^-- allocated in function: QCocoaWindow::recreateWindowIfNeeded()
Change-Id: I97faf30db5835fea2f05320435b1b8c334a478d1
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Constructing a QStringRef directly from the string, offset and a
length is UB if the offset + length exceeds the string's length.
Thanks to Robert Loehning and libFuzzer for finding this.
QString::midRef (as correctly used in both changed uses of QStringRef,
since 432d3b6962) takes care of that for us. Changed one UB case and
a matching but correct case, for consistency.
In the process, deduplicate a QStringList look-up.
Added tests to exercise the code (but the one that exercises the
formerly UB case doesn't crash before the fix, so isn't very useful;
the invalid read is only outside the array it's scanning, not outside
allocated memory).
Change-Id: I7051bbbc0267dd7ec0a8f75eee2034d0b7eb75a2
Reviewed-by: Anton Kudryavtsev <antkudr@mail.ru>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The timeout will never be larger than numeric_limits<quint64>::max(),
especially on platforms with 32-bit longs.
Instead, test if the timeout is exactly numeric_limits<unsigned long>::max(),
which matches the ULONG_MAX value which is documented
to indicate no timeout.
Change-Id: Ib663eddb5703797c50c04fd4eae60bd64f379d1c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Try to better describe what it is and what it does. Also mention
its strongest use case.
Change-Id: Ib5c3e8a3c9b96169c139c5d7e8995a6a49d7d5e1
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
QSharedDataPointer obeys the regular Qt container thread-safety rules:
it's thread-safe in const methods but not in mutating ones. QSDP::data()
is mutating, which causes a data race. For example, if the contained
QLocalePrivate has a refcount of 2 and two threads see that, both
threads will try to detach and then replace the pointer, but that
pointer replacement is not atomic.
Using QExplicitSharedDataPointer makes the race go away, since data() is
now non-mutating. QESDP is used only to destroy the QLocalePrivate on
program shutdown.
Note that there are still race conditions relating to *updating* the
locale private.
Fixes: QTBUG-73403
Change-Id: Id98140e1c2f0426cabbefffd157ed6ec30a3e08f
Reviewed-by: Thomas Sondergaard <thomas@sondergaard.cc>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Where non-format characters appear in the format string, they are not
"ignored" as claimed; they are passed through verbatim. Formats
without separators between fields are in fact supported, although
results may be ambiguous. Cleaned up phrasing in the process.
Fixes: QTBUG-51208
Change-Id: I7284a36c48aa0be29deaa16945ca0212e9e6f72c
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Add a note explaining that it should not be used in multi-monitor
environments.
Task-number: QTBUG-73232
Change-Id: I7b61cef74eac4082efe849b3d5bdbbe5f04fcdbb
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
The meta object system stores enums as signed int, probably for
performance reasons. This is good enough for about 99% of the use cases.
If you try to register larger types and then access them through the
metaobject system, you can get undefined behavior.
Task-number: QTBUG-71947
Change-Id: I16b395547c22fad10b476c2c2a0768538db0a20e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Current tip-of-tree clang (after Clang 8 was branched) added an
intrinsic function __builtin_ia32_xgetbv, and added the following
define that provides _xgetbv:
#define _xgetbv(A) __builtin_ia32_xgetbv((long long)(A))
This fallback declaration of the _xgetbv function only is used
in case the Q_OS_WIN branch of the #if/#elif below is used, if
the #if (defined(Q_CC_GNU) && !defined(Q_CC_EMSCRIPTEN)) ||
defined(Q_CC_GHS) wasn't taken. I left out the
!defined(Q_CC_EMSCRIPTEN) part as I believe Q_OS_WIN and
Q_CC_EMSCRIPTEN are mutually exclusive.
Change-Id: I257fc4283ff9f0845df51ab764cf58acdf285c66
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Checking the known flags prevented us from correctly setting the
LinkType flag if we had previously set LinkType as a known flag since
the flag is not reset between updates.
Manifested itself in situations where the file info is loaded and then
the LegacyLinkType flag is checked through
QFileInfoPrivate::checkAttribute. Since the LegacyLinkType is not
set for Windows it will update the metadata and exclude the LinkType
flag.
Change-Id: Iea27f42fe11f36ba2247e52fa9c82b4639666a64
Fixes: QTBUG-72644
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Q_DECLARE_PRIVATE gets used in the declaration of the public class,
where the private class is typically visible only as a forward-decl,
with no knowledge of what it's based on; consequently, the macro is
obliged to use reinterpret_cast<>, which is subject to warnings when
the compiler *can* see both types and their alignments differ. The
same applies to Q_DECLARE_PRIVATE_D.
So suppress gcc's -Wcast-align around the d_func() return statements.
(If we get similar problems with other compilers we can add their
suppressions likewise; but, for now, we've only seen this on MIPS64,
where we use gcc.) This tripped over one use of Q_DECLARE_PRIVATE in
a private Q_SLOTS: section; for some reason, gcc didn't like the
semicolon on the friend declaration. Changing the context to plain
private fixed that.
Fixes: QTBUG-72885
Change-Id: I5edc11d46bd4eb820713adede79d53191a7e2736
Reviewed-by: Liang Qi <liang.qi@qt.io>
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
Reviewed-by: Boxiang Sun <daetalusun@gmail.com>
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>
`kdevelop -s doesnotexist` led to
ASSERT failure in Q_GLOBAL_STATIC: "The global static was used after being destroyed
because of a qDebug() statement in some "unregister" method called by a
global object's destructor. This is normally fine, but with
%{time} in QT_MESSAGE_PATTERN, qDebug() ends up using QLocale after its
global objects (systemLocalePrivate and defaultLocalePrivate) were
destroyed.
Change-Id: I8d8b34e0197ad1eda8283fcf36d2c250385bb1a3
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Its filePath() and absoluteFilePath() don't trust its own
isAbsolute(), due to some infelicities on MS-Win; and kludged round a
consequent problem with resource paths; but other virtual file systems
weren't catered for. Replace the convoluted test there with a static
bool function (so that future kludges in this area shall only need to
edit one place; and can document why they're needed) and use a more
robust test that handles all virtual file systems (by asking
QFileInfo) but falls back to QFileSystemEntry to work round the known
infelicities on MS-Win. Add regression test for asset library paths
issue on iOS. Ammends 27f1f84c1c.
Moved a couple of local variables to after the early return, since it
doesn't need them, in the process.
Task-number: QTBUG-70237
Change-Id: Ib3954826df40ccf816beebe5c3751497e3bf6433
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
QKeyEvent's detailed description already documents this:
"A key event contains a special accept flag that indicates whether the
receiver will handle the key event. This flag is set by default for
QEvent::KeyPress and QEvent::KeyRelease, so there is no need to call
accept() when acting on a key event. For QEvent::ShortcutOverride the
receiver needs to explicitly accept the event to trigger the override.
Calling ignore() on a key event will propagate it to the parent widget.
The event is propagated up the parent widget chain until a widget
accepts it or an event filter consumes it."
However, someone looking at eventFilter() won't see this, resulting in
the shortcut not being consumed and propagating elsewhere, so mention
it in the docs for eventFilter() too.
Change-Id: I14fece52133be641ccdabd81f75706cd10b64669
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Venugopal Shivashankar <Venugopal.Shivashankar@qt.io>
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
When using the overload of QCoreApplication::processEvents that takes a
maxtime argument, the function will keep processing events until there
are no more events, or until it times out.
The problem is that the function doesn't distinguish between events that
were on the event queue when the function was called, and events generated
by processing events as part of its own execution. If for example a widget
calls update() in its paintEvent, the function will spin for the entire
duration of maxtime.
That doesn't work for qWaitFor, where we need to check the predicate
between each pass, so we use the overload of processEvents that doesn't
take a maxtime. That's fine, as we have our own timeout logic.
Change-Id: I9738d7d0187c36d4a5ddfcd3fd075b0bd84583c4
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
An UUID is 36 characters long, but sysctl and sysctlbyname return a
null terminated string with 37 characters. That was too long for the
provided buffer. Surprisingly the return code was still 0 instead of -1.
The returned buffer was empty though.
Change-Id: Ic4d20ecc1b2b3a3e98468d31ac304957d56deee9
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When running a 32bit application on a 64bit Windows, the call to open
the key
"HKLM/Software/Microsoft/Cryptography/MachineGuid"
will by default be redirect to
"HKLM/Software/WOW6432Node/Microsoft/Cryptography/MachineGuid"
which does not exist.
Instead access the 64bit key from either a 32bit or 64bit application.
KEY_WOW64_64KEY has no effect on 32bit Windows.
Change-Id: Ic5e13f99d08aef2658d58a52cffe66dbab0510b8
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
qtbase/src/corelib/kernel/qmetatype.cpp: In static member function ‘static void QMetaType::destroy(int, void*)’:
qtbase/src/corelib/kernel/qmetatype.cpp:2599:27: error: ‘info.QMetaType::m_destructor’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
if (m_typedDestructor && !m_destructor)
~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
qtbase/src/corelib/kernel/qmetatype.cpp:1868:15: note: ‘info.QMetaType::m_destructor’ was declared here
QMetaType info(type);
^~~~
qtbase/src/corelib/kernel/qmetatype.cpp:2600:26: error: ‘info.QMetaType::m_typedDestructor’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
m_typedDestructor(m_typeId, data);
~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
qtbase/src/corelib/kernel/qmetatype.cpp:1868:15: note: ‘info.QMetaType::m_typedDestructor’ was declared here
QMetaType info(type);
^~~~
The extended (not inlined) function may be called on a half
initialized invalid instance.
Change-Id: I26d677a8ad2bd0c5846233f06393e774d377936d
Reviewed-by: Liang Qi <liang.qi@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Qt Test library sources specific to Core, GUI and Widgets modules
were moved around in commit 88867e39b.
The new source locations must be referenced in Qt Test documentation
configuration. The same sources are excluded in their original doc
projects, and the related snippet file is moved over to qttestlib.
The commit also fixes the remaining documentation issues for Qt Test.
Change-Id: Ibe011aa83639e574d647f12bc9e53e618781bce6
Reviewed-by: Martin Smith <martin.smith@qt.io>
QStringListModel::setData documentation states that
"The dataChanged() signal is emitted if the item is changed."
This patch actually respects the doc. setData will check that the data
actually changed before sending the dataChanged signal.
[ChangeLog][QtCore][QStringListModel] setData will now emit the
dataChanged() signal only if the string set is different from
the one already contained in the model
Change-Id: I4308a6f3b4851203fb899c5e29a36076e0c32f2f
Reviewed-by: David Faure <david.faure@kdab.com>
The code path that this refers to was removed already for Qt 5.11 in
commit 53fb2c48ef.
Change-Id: I4a7ae1b89b24c0ab7ceaa43f763c7ef422ca4900
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Otherwise it gets linkified, which looks inconsistent. Instead,
use \sa for functions where QT_NO_CAST_FROM_ASCII
or QT_RESTRICTED_CAST_FROM_ASCII is referenced.
Change-Id: Ic3933d8c4c81c963215de7f3aac4d0a11e61cbc2
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Reviewed-by: Martin Smith <martin.smith@qt.io>
There's no 64-bit ADD instruction, so we make do with ADD+ADC. This is
what Clang generates. ICC uses the two as well, but then performs some
subtractions to find out if it overflowed. GCC for some inexplicable
reason attempts to use SSE2 if that's enabled, otherwise it performs the
subtractions like ICC.
Alternative implementation which generates better code, but violates
strict aliasing:
uint *low = reinterpret_cast<uint *>(r);
uint *high = low + 1;
return _addcarry_u32(_addcarry_u32(0, unsigned(v1), unsigned(v2), low),
v1 >> 32, v2 >> 32, high);
Manual testing shows this works. tst_qnumeric passes in debug mode. MSVC
2017 15.9 still miscompiles in release mode (reported to MS as [1]).
[1] https://developercommunity.visualstudio.com/content/problem/409039/-addcarry-u32-wrong-results-with-constant-inputs.html
Change-Id: I61ce366d57bc46c89db5fffd15704d53ebd4af3c
Reviewed-by: Thomas Miller <thomaslmiller91@gmail.com>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
This fixes qtdoc failing to build on i586 because of an assertion in
libclang since Q_QDOC is defined and thus the declaration of the
qfloat16(float) constructor and operator float() are removed, thus
their definitions should be removed too, which is what this patch
does.
Fixes: QTBUG-72725
Done-with: Michal Srb <msrb@suse.com>
Change-Id: I6424873425d46345e09f411f9ce88f2520825da4
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Jesus Fernandez <Jesus.Fernandez@qt.io>
If empty paths is passed to the unix filesystem engine, we get a warning
about empty filename passed to function, before returning false. Fix
this by testing for empty string first, and don't send empty string to
file engine.
The current warning leads to code like
if (!filename.isEmpty() && QFile::exists(filename))
{
//
}
rather than the slightly cleaner
if (QFile::exists(filename))
{
//
}
Change-Id: I0207324889ec22e5a072c28d58337d117b0153b1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
use library objects for all variants, and inline the tests.
Change-Id: I029f9a6655a783dab4a22abf601aadbb484c03af
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
... and use that to inline the xlocalescanprint test.
Change-Id: I0973133d7f9ecc9a38b70dc4b83df174a35b2b1f
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
The new loop does 32 bytes (16 code units) at a time
Change-Id: I8f261579aad648fdb4f0fffd155412a4d77428e9
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
these were new on dev while the original migration happened on 5.9, or
came from new changes which hadn't adapted yet.
Change-Id: I5e48437061a97e6df6e93881c98471455e177631
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
The documentation is not explicit enough about the content of the
returned list of capturedList nor the element 0 when calling captured or
its friends.
The first element (aka element 0) is the whole string captured when one
or more groups are used.
Change-Id: I3c59ebfc9f6d762dd4d8aaf8f5c0de24359f53d7
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Sze Howe Koh <szehowe.koh@gmail.com>
The current implementation of wildcardToRegularExpression doesn't
anchor the pattern which makes it not narrow enough for globbing
patterns. This patch fixes that by applying anchoredPattern before
returning the wildcard pattern.
[ChangeLog][QtCore][QRegularExpression] The wildcardToRegularExpression
method now returns a properly anchored pattern.
Change-Id: I7bee73389d408cf42499652e4fb854517a8125b5
Fixes: QTBUG-72539
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
in addition to the actual library resolution, also resolve the headers
belonging to the library, to validate the include path, and possibly
ensure that the right version of the library is present.
the "include" entries were moved out of the "test" objects, and renamed
to "headers". this cleanly permits libraries without compile tests.
the headers were not put into the sources, because the variance among
the includes is generally orthogonal to the variance among the
libraries.
note that this - like the library resolution - provides no support for
darwin frameworks. consequently, the opengl libraries are excluded from
the conversion on darwin.
similarly, wasm is excluded (centrally), because emcc is magic and would
need advanced wizardry to be dealt with.
Change-Id: Ib390c75371efa2badcfec9b74274047ce67c3e5a
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Silence warnings about signed constants being shifted out of int range.
Change-Id: I5dc397de71f4de09e54ce3cbc0f8e3a1cf977b03
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Commit 6b875f0625 created a macro for it,
but in the multiple updates to find the best solution, we forgot to use
it.
Fixes: QTBUG-72429
Change-Id: I4ac1156702324f0fb814fffd156f27c1789d1409
Reviewed-by: Thomas Miller <thomaslmiller91@gmail.com>
Reviewed-by: Janne Koskinen <janne.p.koskinen@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
When these docs were written, the context.file and context.function
pointers were never null. But in commit d78fb442d7
(Qt 5.4), we made the logging pass null pointers in release builds. The
C standard does not say that passing null pointers is permitted. In
fact, it says "If no l length modifier is present, the argument shall be
a pointer to the initial element of an array of character type." and
that's pretty explicit that it needs to point to the initial element of
a string.
Fixes: QTBUG-72478
Change-Id: I4ac1156702324f0fb814fffd156f624ffefa1445
Reviewed-by: Topi Reiniö <topi.reinio@qt.io>
Tor Arne Vestbø
Edward Welbourne
Morten Johan Sørvig
Andreas Hartmetz
Thiago Macieira
Jesus Fernandez
Mikhail Svetkin
Friedemann Kleint
Qt Forward Merge Bot
Ulf Hermann
Martin Storsjö
Mårten Nordheim
Kai Koehne
Oswald Buddenhagen
David Faure
Mitch Curtis
Roland Rossgotterer
Liang Qi
Jędrzej Nowacki
Ville Voutilainen
Topi Reinio
Nico Vertriest
Luca Beldi
Antonio Larrosa
Sune Vuorela
Samuel Gaist
Allan Sandfeld Jensen