When doing format conversion, the optimized inplace codepath did not
check if the image data was readonly, i.e. if the QImage had been
created by the constructor taking an existing external buffer.
Task-number: QTBUG-44610
Change-Id: I085ff8da427bc4ee392f548dffd2418b63148965
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
it's documented to fall back to Prefix if qt.conf is present but
Settings is not specified.
Task-number: QTBUG-44644
Change-Id: I8ef6659cbdad51b2fb3c1075ea6f0af4997117ed
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
If one asks for the global OpenGL context via the right attribute,
and then creates a QGLContext sharing with that context, Qt will
create anoter, dummy QGLContext owned by the global context.
At application shutdown, and specifically after ~QObject of QGuiApplication
has run, the QObjectPrivate dpointer gets destroyed. That destroys the
owning global QOpenGLContext, and therefore that dummy QGLContext as well.
However, QGLContext dtor triggers a code path that accesses qApp as
if it were still alive. That's not the case any more -- it has already
been destroyed. So, introduce a check and avoid dereferencing NULL.
Task-number: QTBUG-44621
Change-Id: Ic160ac99e9269db999e76229b1c7f8c53bd2be61
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
This fixes a division by 0 when processing a maliciously crafted BMP
file. No impact beyond DoS.
Task-number: QTBUG-44547
Change-Id: Ifcded2c0aa712e90d23e6b3969af0ec3add53973
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
before marking the backingstore as dirty. The QWidgetBackingstore has an
assert that the region or the rect passed in is not empty, hence its
programming error not to check before calling this function.
Task-number: QTBUG-43489
Change-Id: Ic67fb6ca7959466e1758ce91827cd4b8acdf73fc
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Those QTimeZones failed to convert to other timezones because the data()
virtual function was never overridden and reimplemented. That meant all
QUtcTimeZonePrivate objects were *really* UTC, with no offset.
Task-number: QTBUG-44600
Change-Id: Ia0aac2f09e9245339951ffff13c5294bb783c674
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
The c2m() function which converts a const_iterator into an iterator
is a broken concept for an implicitly shared container such as
QVector, because the act of calling begin() as the starting
point already detaches and invalidates the c2m argument.
This could be fixed in c2m, but the bug wasn't even in c2m,
but in removeAll(), which called end() before c2m, so the c2m
argument was already invalidated when entering c2m.
The solution is to store the positions as indices instead of
iterators before calling the first detaching function.
Task-number: QTBUG-44592
Change-Id: I66cf4f1277e71148a4d5b5bbfb6a3369ad02db68
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
The property INTERFACE_LINK_LIBRARIES has a single value
and is configuration independent.
Change-Id: I9130b1d0444563335825e3806f693a39f65ca67d
Reviewed-by: Stephen Kelly <steveire@gmail.com>
Move the optimized dash drawing into stroke() in order to apply the
optimization to polygons in addition to lines.
In the case of polygons/polylines, a vertex patch is redrawn using the
original dash brush in order to respect the joinStyle of the line.
As the line correction code flows through both the optimized dashed path
and the standard geometry code path, line adjustment is now also performed
for normal geometry.
Task-number: QTBUG-40604
Done-with: Louai Al-Khanji <louai.al-khanji@theqtcompany.com>
Done-with: Andrew Knight <andrew.knight@theqtcompany.com>
Change-Id: I668369b4aadb6a1bbbd4d621cb8ce1e3b19fbbc9
Reviewed-by: Louai Al-Khanji <louai.al-khanji@theqtcompany.com>
Byte is definitely wrong. GL_OES_depth_texture states that short or int
are the only options, GL_UNSIGNED_BYTE is thus rejected.
Let's ask for the highest (32 bits). There's no guarantee that the implementation
honors this, but at least we tried.
Change-Id: I14dd9d4ab56b0b69cdab341f95f47499786d174e
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
We where allowing conversion from jobject to QJNIObject without taking
ownership of the jobject. Since we are managing the JNI environment we
should not allow conversions without having the option of taking
ownership of the local ref. This is now done by making the conversions
explicit, i.e., local refs are converted through
QJNIObjectPrivate::fromLocalRef() and global refs through the
QJNIObjectPrivate's jobject constructor.
This change breaks SC, but the API is private and no usage have been
found outside QtBase.
Change-Id: I3175f171699ec3f8e65144aaebc6246bc6e5bb4d
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@theqtcompany.com>
wait() was always returning immediately, regardless of the timeout
value, due to a timespec comparison from different clock types.
On Android 5.0, qt_gettime() uses the monotonic clock but the wait
condition was using the real time clock.
__pthread_cond_timedwait_relative() is not exported anymore in
Android 5.0, we therefore fall back to pthread_cond_timedwait().
Since the monotonic clock is now available, qt_gettime returns
a time based on it. The wait condition consequently needs
to use the monotonic clock.
Change-Id: Ie7cf909b81107edd7207c3c039b3ec1f5422303f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@theqtcompany.com>
Most of the time the widget passed in to finish() will not yet have a
window handle so it would just call close() straight away. This
way it ensures has a window handle.
Task-number: QTBUG-35757
Change-Id: I2e212a2ea496cc4e4bba57a8c2d0d127afc43e48
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Reviewed-by: Thorbjørn Lund Martsum <tmartsum@gmail.com>
Reviewed-by: Alex Blasche <alexander.blasche@theqtcompany.com>
If we call them just in the constructor all the signals they sent
out can't be connected and will be lost, particularly this means
the QNetworkConfigurationManager doesn't see my ethernet connection
and thus thinks i'm not online
Change-Id: I1480f76338d6ae4fbed676f9fa40ada18ea431ad
Reviewed-by: Alex Blasche <alexander.blasche@theqtcompany.com>
the qt headers live in the source dir in this configuration.
instead of hard-coding the path in the project file, use the correct
module variable. this requires harfbuzz-ng to be built after corelib.
Change-Id: If1b64b59a0939d3b5190331fcf70da566d4eeedd
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Whenever the selection changes it will call panelSelectionDidChange
even if the panel is not visible which causes it to overwrite our
chosen initial file with "untitled". Therefore we only set the name
when the panel is actually visible.
Change-Id: I09f7ebabce5c2139a7f74b953391d39dfad65db2
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
We don't receive ButtonRelease event after closing a popup
by clicking outside of the client area. Thus the internal
state of mouse buttons in the xcb plugin becomes outdated
until we receive ButtonRelease event.
This commit updates the internal state of mouse buttons
from MotionNotify events. So when a user will move a mouse
on the client area, the xcb plugin will send a mouse event
with updated buttons to Qt Gui and QGuiApplication will
detect the following mouse events correctly.
Task-number: QTBUG-32609
Task-number: QTBUG-35065
Task-number: QTBUG-43776
Task-number: QTBUG-44166
Task-number: QTBUG-44231
Change-Id: Ica334dfbf04f7ef81db86b25262328fe5da11808
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@digia.com>
Reviewed-by: Gatis Paeglis <gatis.paeglis@digia.com>
The current SQL plugins path are still Qt 4 style. This patch adds
qtbase to ensure users will be directed to the right place.
Change-Id: I11bf933864f0a8955079bf7175eb1e33bb5bc07f
Reviewed-by: Sze Howe Koh <szehowe.koh@gmail.com>
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>
In some documents, "{QObject}s" was used which didn't show up as link. This is
fixed by using "\l{QObject}s" instead.
Change-Id: I90dbd543790842b242a11f3f94a32d4273ebb38d
Reviewed-by: Martin Smith <martin.smith@digia.com>
Task-number: QTBUG-44390
Change-Id: I1a800c709d3543699131ffff13c19789a12dac0b
Reviewed-by: Martin Smith <martin.smith@digia.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
The current attempt to fix the problem described in b8d0fac5a9
has shown to have unforeseen issues. The assert in the
patch has been triggered in user code (see bug report), which, when
looking more closely, shows that the assert cannot be guaranteed
for all cases. E.g if user code (or Qt drag'n'drop code) starts a new
event loop upon receiving the move event, this can deliver new mouse
events before the move returns, which will trigger the assert.
So it seems like the most correct solution is to *post* the second button
event after all (directly trailing the move event), to at least make sure
mouse events are being delivered in the correct order. The crash that caused
the original change will need to be fixed by other means.
Task-number: QTBUG-38597
Change-Id: I47633d1ea310b7c16e937ef0404d9954281452c3
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
Change a4e1c15b1 will cause setGeometry to take immediate effect
by doing a synchronuous layout and a flush. As a side-effect, the
flush will also send posted touch events. So if the current
setGeometry call comes from a touchMove, a touchEnd might
be delivered before the touchMove returns.
A crash in QToolBar drag-n-drop is seen from this. There
drag data is deleted on mouse release (during the flush), while
a mouse move is already handled on the stack. When the code returns
to the move handler, drag-data has been deleted, and we have a crash.
Since the root problem is that we flush events on a callback from
the app, we can be more careful and precise by only flushing non-user
input events from layoutSubviews.
Change-Id: I3f5e025d774d21509ef837de9d7e85a1ac6e48cc
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
This is a no-op change. It doesn't improve or worsen the code at all.
Change-Id: Ifd5273842370ca9bce0ed74f2f2d54d453797948
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Linux's socket(2) system call can take SOCK_NONBLOCK and let us avoid
the extra two fcntl(2) to change the setting after it's created.
Change-Id: Ie0d7a2d51acc21acd850cbee6d0f76e27b1ae859
Reviewed-by: Richard J. Moore <rich@kde.org>
QJsonObject requires that, since it does binary searches for the keys.
Change-Id: I8a7a116f51864cecb52fffff13bc24660c1cc1ac
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
backtrace_symbols is very slow because it tries to find the function
name from the address. In order to do that it needs to do a linear search
over all symbols. (Because the hash table goes the other way to find the
address from the symbol name)
The code is going to skip a few frames from QtCore. Since we cannot know
how many, we take a few more than necessary.
This patch changes the additional number of frames from 15 to 7
(Usually, there are about 5 suppressed frames).
We call backtrace_symbols several times for only one frame at the time.
So we are not looking up addresses we don't need after we printed the
right number of frames.
Calling many times backtrace_symbols means we do more malloc, but that's
negligible compared to the time we save. We anyway do a lot of other
allocations because of the regexp operations and such
So this patch is then saving about 10 frames lookups which allow to print
about 6 qDebug per miliseconds instead of only 2 when using %{backtrace depth=2}
Change-Id: Ic6ece2145d53dc570c80fcb0e4455dcef6bc40cb
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
also update some broken links while we are at it ...
Change-Id: I9707a01b01be82e229c8f135c9ae093d73fd1204
Reviewed-by: Martin Smith <martin.smith@digia.com>
Consider the following code:
QFileDialog dlg;
dlg.setFilter(QDir::Hidden | ...);
dlg.setOption(QFileDialog::DontUseNativeDialog)
setFilter stores the filter value in QDialogPrivate::options.
setOption will overwrite this value with the default value from
a newly created QFileSystemModel.
The intention was to set the filter in the model to the value in
options.
Change-Id: I561c5cf1ad4d9d729a56620d86ec549eb3105a07
Task-number: QTBUG-37085
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Specifically when rendering svg to a pdf, when painter
opacity is set, path strokes were transformed incorrectly
due to missed setting of dirtyTransform flag.
Task-number: QTBUG-38675
Change-Id: I861353822ccddd394910b8612687a244d195a41e
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Recent code in font support started to use the platformNativeInterface()
return value without checking for null. Most platform plugins, with the
notable exception of linuxfb and the minimal ones, provide an object.
Start providing a dummy object from linuxfb too. This is more sensible
then adding null checks everywhere.
This will prevent crashing with linuxfb on startup.
Task-number: QTBUG-44414
Change-Id: I48912132e1f8ad52e2a94c3d765dacc7b16f309a
Reviewed-by: Andy Nichols <andy.nichols@theqtcompany.com>
For most cases, the GCC/Intel atomics (__sync functions) are fine, but
there are some systems for which libgcc is incorrectly built (QNX
6.5.0). Additionally, this will allow Harfbuzz-NG to be supported in
exactly the same systems as Qt itself.
Task-number: QTBUG-43850
Change-Id: Ib53f57f70d4ad46863c45e74d60b0eb45ba9bd02
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
This code is C++ only. There isn't a single C++ compiler that fails to
understand the "inline" keyword, since it's required by C++98. Any
compiler older than C++98 is likely to choke on the template usage
further down, so this isn't necessary.
Moreover, the C++ standard says you cannot define macros.
[lib.macro.names] says "Nor shall such a translation unit define macros
for names lexically identical to keywords." -- technically, it's a
promise that the Standard Library headers won't do it, the wording means
that the entire translation unit won't do it, which implies no source
can do it.
MSVC complains about it:
fatal error C1189: #error : The C++ Standard Library forbids macroizing keywords. Enable warning C4005 to find the forbidden macro.
Change-Id: Ic2a0a03a0af47386e34bb698454a2040ef3f6a9d
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
execinfo.h is optional in uClibc. We need to check
__UCLIBC_HAS_BACKTRACE__ if uClibc is used.
Change-Id: Ie28be85b0b70472df1fc4a208581bb66ad34229e
Reviewed-by: Kai Koehne <kai.koehne@theqtcompany.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This only occurred if the item added was the only item in the last
row/column, since then the the internal grid failed to expand
(in this case it would always call expand(0, 0), which would not
create the extra row/column).
[ChangeLog][QtWidgets][QGridLayout] Fixed crash that sometimes happened
when adding items with spans that spanned to the bottom/right edge.
Task-number: QTBUG-38052
Change-Id: Iba95f6d9d9356b4d1c84c7b93f4af9b4ea0cf714
Reviewed-by: Paul Olav Tvete <paul.tvete@theqtcompany.com>
Some drivers appears to reset their scroll positions when make the
primary point. This patch listens for the device changed event and uses
the information in it to update the position.
Task-number: QTBUG-40829
Change-Id: If354dfde6300ceb2b81be14ea24594ab0658a6aa
Reviewed-by: Alex Blasche <alexander.blasche@theqtcompany.com>
