| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
| |
On Windows, qt_error_string() returns the string corresponding to the
Win32 API, not an errno. Replace those uses for a function that works
for errno values.
Change-Id: I1eba2b016de74620bfc8fffd14ccce6162bafdca
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Compilation and link times in CPU seconds with GCC 7, using precompiled
headers (not including moc, rcc, uic, etc. steps or headersclean):
Before After
Debug -O0 198,1 180,3
Debug -Og 240,7 229,2
Release -O3 267,1 249,2
Release LTO 239,4 229,8
QtCore required a little manual adjusting because some files are
bootstrapped into moc itself and into qmake.
Change-Id: I84e363d735b443cb9beefffd14b8b57c10e7da36
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
For the windows file system engine, we add an extra macro to use
library loading if configured to do so, but avoid it on WinRT, as
none of the symbols would be found.
We also QT_REQUIRE_CONFIG(library) in the library headers and
exclude the sources from the build if library loading is disabled.
This, in turn, makes it necessary to clean up some header inclusions.
Change-Id: I2b152cb5b47a2658996b6f4702b038536a5704ec
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
|
|
|
|
|
|
|
|
|
| |
This is still not a great place for it, but this way we can
Q_REQUIRE_CONFIG(library) in qlibrary_p.h and qfactoryloader_p.h is the
only private header that doesn't require library support in plugin.pri.
Change-Id: Ia6c1ac8799822eca1e0cbeca98d1c1b8de74beae
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add proper support for 'so' and 'bundle' suffixes. Qt wrongly assumes
.so libraries are not versioned on Apple platforms, which is
wrong. Also, the shared library .bundle which is what Apple recommends
instead of .so, are also versioned (not to be confound with the
different Core Foundation bundles, which are directory hierarchy).
For more info, see http://docstore.mik.ua/orelly/unix3/mac/ch05_03.htm.
Especially the part that reads:
"Loadable modules, called bundles in Mac OS X, have the file type
MH_BUNDLE. Most Unix-based software ports usually produce bundles with a
.so extension, for the sake of consistency across platforms. Although
Apple recommends giving bundles a .bundle extension, it isn't
mandatory."
Task-number: QTBUG-50446
Change-Id: Iacd5136397a12d65d83821434f332eb602550b4b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
qmake/doc/src/qmake-manual.qdoc
src/corelib/global/qglobal.cpp
src/corelib/tools/qstring.cpp
src/network/socket/qabstractsocket.cpp
src/network/socket/qnativesocketengine_unix.cpp
src/plugins/platforms/eglfs/api/qeglfsglobal.h
Change-Id: Id5dfdbd30fa996f9b4b66a0b030b7d3b8c0ef288
|
| |\
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Conflicts:
src/corelib/global/qglobal.cpp
src/corelib/io/qsettings.cpp
src/corelib/itemmodels/qstringlistmodel.cpp
tests/auto/gui/image/qimagewriter/tst_qimagewriter.cpp
Change-Id: I1c6c306ef42c3c0234b19907914b19da706b4a03
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
As of version 10.12 (Sierra), the name of Apple's desktop operating
system will be macOS. Change the occurrences where the Mac platform
is discussed to use a macro \macos, which expands to 'macOS'. This
helps with adapting to future renaming.
Update the instructions on mac-specific Q_OS_* macro usage.
Add a \target for the old 'Qt for OS X' topic to keep links working
for other documentation modules that try to link with the old name.
Change-Id: Id33fb0cd985df702a4ae4efb4c5fd428e77d9b85
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
|
|\| |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Conflicts:
configure
5.7 now supports clang on android; but dev re-worked configure
src/gui/kernel/qevent.h
One side renamed a parameter of a constructor; the other added an
alternate constructor on the next line. Applied the rename to both
for consistency.
tests/auto/tools/moc/tst_moc.cpp
Each side added a new test at the end.
.qmake.conf
Ignored 5.7's change to MODULE_VERSION.
configure.json
No conflict noticed by git; but changes in 5.7 were needed for the
re-worked configure to accommodate 5.7's stricter handling of C++11.
Change-Id: I9cda53836a32d7bf83828212c7ea00b1de3e09d2
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
We were printing 'loaded library "xxx"' even when ret == false, which
was misleading. So instead print the error string.
Change-Id: Ib57b52598e2f452985e9fffd1459ea860ed2dfcf
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Replace substring functions that return QString with
corresponding functions that return QStringRef where
it's possible.
Create QString from QStringRef only where necessary.
Change-Id: Id9ea11b16947220cd27787c0b529de62d10b6c26
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
|
|\|
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/angle/src/libGLESv2/libGLESv2.pro
src/plugins/platforms/eglfs/deviceintegration/eglfs_kms_egldevice/qeglfskmsegldeviceintegration.cpp
Change-Id: If8da4cfe8f57fea9f78e7239f378a6302c01674e
|
| |
| |
| |
| |
| | |
Change-Id: I280673e7fd582c5877e8e17ed00318cb10cd537c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
It's interesting that the HMODULE/HINSTANCE pointer points to something
in the actual module that got loaded, so we can use its value as
"address of something in the module" for GetModuleHandleEx. The "PIN"
flag tells the Windows DLL loader to never unload.
Change-Id: Ifea6e497f11a461db432ffff1449a2169eb6293e
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Not that we require it, but since The Qt Company did it for all files
they have copyright, even if they haven't touched the file in years
(especially not in 2016), I'm doing the same.
Change-Id: I7a9e11d7b64a4cc78e24ffff142b4c9d53039846
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
From Qt 5.7 -> LGPL v2.1 isn't an option anymore, see
http://blog.qt.io/blog/2016/01/13/new-agreement-with-the-kde-free-qt-foundation/
Updated license headers to use new LGPL header instead of LGPL21 one
(in those files which will be under LGPL v3)
Change-Id: I046ec3e47b1876cd7b4b0353a576b352e3a946d9
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Saves just shy of 4KiB in text size on optimized GCC 4.9 Linux
AMD64 builds, iow: ~0.07% of the total QtCore library size.
Change-Id: I87fdcc8ee25c6bb5dabddb9a694ab4496b1538fa
Reviewed-by: David Faure <david.faure@kdab.com>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
|
|\|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/corelib/kernel/qcoreapplication.cpp
src/corelib/kernel/qeventdispatcher_blackberry.cpp
src/network/bearer/qnetworkconfiguration.cpp
src/plugins/bearer/blackberry/qbbengine.cpp
src/plugins/platforms/android/androidjnimain.cpp
src/plugins/platforms/android/qandroidplatformtheme.cpp
src/plugins/platforms/qnx/qqnxbpseventfilter.cpp
src/plugins/platforms/qnx/qqnxfiledialoghelper_bb10.cpp
src/plugins/platforms/qnx/qqnxinputcontext_imf.cpp
src/plugins/platforms/qnx/qqnxintegration.cpp
src/plugins/platforms/qnx/qqnxnavigatorbps.cpp
src/plugins/platforms/qnx/qqnxvirtualkeyboardbps.cpp
src/plugins/platforms/qnx/qqnxwindow.cpp
src/widgets/kernel/qwidgetwindow.cpp
src/widgets/styles/qwindowsvistastyle.cpp
src/widgets/styles/qwindowsxpstyle.cpp
src/widgets/widgets/qtoolbararealayout.cpp
tests/auto/corelib/global/qflags/qflags.pro
tests/auto/corelib/itemmodels/qitemmodel/modelstotest.cpp
tests/auto/corelib/tools/qversionnumber/qversionnumber.pro
tests/auto/widgets/dialogs/qfilesystemmodel/tst_qfilesystemmodel.cpp
Change-Id: I37be88c6c185bb85404823353e027a0a6acdbce4
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This information is already registered by the QMessageLogger ctor.
Where, by dropping the << Q_FUNC_INFO in ostream-style qDebug(), only
a string literal remained, converted to printf-style qDebug() on the
go.
Change-Id: I3f261c98fd7bcfa1fead381a75a82713bb75e6f3
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We keep two symbols from QPluginLoader defined, even if QT_NO_LIBRARY
is set in order to be able to locate static plugins. This doesn't
constitute any loading of shared libraries or plugins from external
files at runtime.
Using these symbols we can enable most of QFactoryLoader even if
QT_NO_LIBRARY is set. Only update(), refreshAll(), library() and
the dtor make no sense then. This way QGenericPlugin also becomes
useful with QT_NO_LIBRARY.
Task-number: QTBUG-3045
Change-Id: Ib7842ce5799e8e2caa46431d95fddd1adda0fc41
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|\
| |
| |
| | |
Change-Id: I004854a25ebbf12b1fda88900162fe7878716c58
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Use the name "OS X" instead of "Mac OS X", "Mac OS" and "OSX",
and mention iOS. Replace "Carbon Preferences API" by
"CFPreferences API" in the QSettings documentation.
Change-Id: Ia7f9fb874276c7c445a1649df521b96ff43daa0c
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reviewed-by: Venugopal Shivashankar <venugopal.shivashankar@digia.com>
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Qt copyrights are now in The Qt Company, so we could update the source
code headers accordingly. In the same go we should also fix the links to
point to qt.io.
Outdated header.LGPL removed (use header.LGPL21 instead)
Old header.LGPL3 renamed to header.LGPL3-COMM to match actual licensing
combination. New header.LGPL-COMM taken in the use file which were
using old header.LGPL3 (src/plugins/platforms/android/extract.cpp)
Added new header.LGPL3 containing Commercial + LGPLv3 + GPLv2 license
combination
Change-Id: I6f49b819a8a20cc4f88b794a8f6726d975e8ffbe
Reviewed-by: Matti Paaso <matti.paaso@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Q_OS_WINRT was missing in the Windows-condition.
Condition was simplified through use of Q_OS_WIN.
Change-Id: I1a49d2d9c413dc2156930b6915e1675abcdde36f
Reviewed-by: Oliver Wolff <oliver.wolff@theqtcompany.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Adds an option to request the RTLD_DEEPBIND flag to dlopen. On Linux
this can be used to force a library to resolve global symbols locally
instead of using the similarly named symbols already loaded.
This makes it possible to load and use plugins linked against Qt 4
without crashing.
[ChangeLog][QtCore][QLibrary] Added DeepBindHint which maps to RTLD_DEEPBIND
on Linux making it possible to load libraries with external symbols that
clash with already loaded ones, such as plugins linked to Qt4.
Change-Id: I4edb4af68e4a47e932a87d108360dba8d91dc34a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|\|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/corelib/tools/qbytearray.cpp
src/gui/image/qimage.cpp
src/gui/image/qppmhandler.cpp
src/gui/kernel/qguiapplication.cpp
src/gui/painting/qpaintengine_raster.cpp
Change-Id: I7c1a8e7ebdfd7f7ae767fdb932823498a7660765
|
| |
| |
| |
| |
| |
| | |
Task-number: QTBUG-40362
Change-Id: I4fa4a69c13a60f8c7915b6c9659a4daaae4ecc1a
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
- Renamed LICENSE.LGPL to LICENSE.LGPLv21
- Added LICENSE.LGPLv3
- Removed LICENSE.GPL
Change-Id: Iec3406e3eb3f133be549092015cefe33d259a3f2
Reviewed-by: Iikka Eklund <iikka.eklund@digia.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Instead of initializing debug_env to -1 (thus forcing the variable
into the data segment), and then overwriting the -1 with a read
from the env-var, dynamically initialize the variable from the
env-var directly, thus allowing the variable back into the bss
segment (which doesn't occupy storage in the executable).
There may have been a reason to do it this way when the old code
could fail due to the memory allocation involved, but now with
qEnvironmentVariableIntValue(), that is no longer a reason.
Change-Id: I9f21b0783ff348f50b574395fc07f2869a14102e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|/
|
|
|
|
|
|
| |
It doesn't allocate memory, so cannot throw and is a lot faster
than qgetenv().
Change-Id: I8d440619edfbd90045564e1f92676f1e1f87e136
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
They shouldn't exist in the first place. They exist in two cases only:
1) mistake by the user in the QLibrary or QPluginLoader constructors or
setFileName
2) as a kludge for setLoadHints before a file name is set (we need to
store the user's requested hints somewhere)
This is important for the second case, as otherwise all QLibrary and
QPluginLoader objects without a file name would share the setting.
Task-number: QTBUG-39642
Change-Id: Iebff0252fd4d95a1d54caf338d4e2fff4de3b189
Reviewed-by: David Faure <david.faure@kdab.com>
|
|
|
|
|
|
|
|
|
|
| |
This commit makes replaces the loadHints member with a setter, a getter
and an atomic variable. The setter will not set anything if the library
has already been loaded.
Task-number: QTBUG-39642
Change-Id: Ibb7692f16d80211b52aaf4dc88db1a989738a24d
Reviewed-by: David Faure <david.faure@kdab.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit moves the setting of the loadHints to inside the constructor
or to QLibraryStore::findOrCreate (which is under a mutex). This avoids
data race conditions with two threads asking for the same plugin at the
same time, with different load hints.
This also opportunistically moves the setting of the error message for
empty file names.
Task-number: QTBUG-39642
Change-Id: I497a41781d10e407d6420116a0b05fdfe2b548de
Reviewed-by: David Faure <david.faure@kdab.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When QtCore's global destructors are run, they delete the global
QLibraryStore qt_library_data and set the pointer to null. If something
happened to call QLibraryStore::instance() later, it would be recreated
and then weird things might happen. So prevent that from happening.
That usually cannot happen, since the only thing that can run after
QtCore global destructors are other QtCore global destructors or global
destructors from libraries that do not depend on QtCore. So we're
reasonably safe. There are two conditions in which something could run
after QLibraryStore::cleanup() and still try to access QLibraryStore:
1) indirect dependency, like a global destructor from a library that
doesn't depend on QtCore running code from another module that does.
2) static builds of Qt modules. In that case, the order of the global
destructors is totally arbitrary and we could get one from a module that
depends on QtCore running after QtCore's. That is the case from the bug
report.
Task-number: QTBUG-36294
Change-Id: Id199671275fd2535acf2d158857ce46b474e579b
Reviewed-by: Kai Koehne <kai.koehne@digia.com>
Reviewed-by: Tim Jenssen <tim.jenssen@digia.com>
|
|
|
|
|
|
|
|
|
|
|
| |
In the event the plugin loader encounters it due to some configuration
problem, it is otherwise detected as a plugin causing a crash
when trying to read the meta data.
Task-number: QTBUG-35970
Change-Id: I34182b3a61125e3b192dfd4b1dc03bd1e98c693e
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Default values should have mark-up to denote that they are code.
This commit changes:
-"property is true" to "property is \c true".
-"Returns true" to "Returns \c true".
-"property is false" to "property is \c false".
-"returns true" to "returns \c true".
-"returns false" to "returns \c false".
src/3rdparty and non-documentation instances were ignored.
Task-number: QTBUG-33360
Change-Id: Ie87eaa57af947caa1230602b61c5c46292a4cf4e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The reason why the scanning of files was restricted to Unix is lost in
history. My guess is that it was added *because* on Unix you couldn't
do the LoadLibrary trick to load without dependencies. That is, it
makes no comparison to whether scanning is faster than LoadLibrary.
So assume that scanning is faster, especially now that we have memory
mapping in QFile (it wasn't there in Qt 4.0).
This simplifies greatly the codebase because we no longer need to try
to load and later unload the library to check whether it is a Qt
plugin or not, nor deal with Windows non-C++ exceptions. Just scan on
all platforms.
Change-Id: If269e009bfb9920805ca1aa6c2b66aacdd5b26e9
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We already had an ELF decoder, which helped us greatly to find the
metadata and that catches most Unix systems (Solaris, QNX, HP-UXi, and
all of the free Unixes). On other Unix systems, aside from Mac OS X,
we simply scanned the entire file for the signature. On Windows, even
without a COFF-PE decoder, we use a LoadLibrary trick to load the
plugin without loading the dependent libraries. In most cases, that
works.
Unfortunately, on Mac OS X we didn't have a decoder and nor could we
do the file scan: because Mac OS X binaries could be fat binaries, we
wouldn't know which architecture's signature we had found.
No more. This adds a full Mach-O decoder to QtCore. It is also capable
of finding the boundaries of the architecture's binary, but that
functionality is disabled since all Qt 5 plugins have plugin metadata
sections.
Change-Id: I2d5c04c5ecf024864b8a43f31ab6b7e6c5eae9ce
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
For systems where the Unix signature checker isn't enabled (read: Mac
and Windows), QPluginLoader must actually load the plugin to query for
the metadata. On Mac it even tried to keep the library loaded to avoid
unloading and reloading again when the user calls load().
However, that plus the fact that it was calling load_sys() (on Mac)
meant that it would bypass the reference count checking. And on all
Unix, if a library-that-wasnt-a-plugin was already loaded by way of a
QLibrary, it would have an effect of unloading said library.
So remove the "caching" of the library. We should instead invest time to
write a proper Mach-O binary decoder.
Task-number: QTBUG-29776
Change-Id: Iebbddabe60047aafedeced21f26a170f59656757
Reviewed-by: Liang Qi <liang.qi@digia.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Both QPluginLoader::unload() and QLibrary::unload() protect against
that (they have a "did_load" member), but QFactoryLoaderPrivate's
destructor doesn't. In the past (Qt4) all plugins had to be loaded
anyway, so there was no mistake in the reference counting. With Qt 5,
we don't load plugins unless they're actually used (in
QFactoryLoader::instance).
Task-number: QTBUG-29773
Change-Id: I3278fa14bac7e26a9faaf999b4e42e950654ac9a
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Don't assume that all entries in the QLibrary global data refer to
libraries have have been loaded. There are three (not two) ways of
getting entries there:
1) creating a QLibrary
2) using the static QLibrary::resolve
3) via plugins
The unload code was meant to handle the first two cases only: libraries
are still loaded at the end of the execution if the static methods were
called or if QLibrary objects were leaked. It didn't handle the case of
plugins being found by the directory scanner in QFactoryLoader but never
loaded.
Note it's possible that this assertion also happened with leaked
QLibrary.
Change-Id: Idcd7a551f96d8fe500cbca682f8014f5122b7584
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
During exit, libraries are unloaded and global destructors are run. If
we call dlclose(3) from inside the global destructors, we might be
telling libdl to unload a module it has already unloaded.
I cannot reproduce the issue on my Fedora 17 machine with glibc 2.15,
but it could be reliably be reproduced on an Ubuntu 11.10. The assertion
is identical to the one reported upstream at
http://sourceware.org/bugzilla/show_bug.cgi?id=11941 (see better
explanation at https://bugzilla.novell.com/show_bug.cgi?id=622977). I
cannot find any evidence in glibc's source code that the bug has been
fixed.
Change-Id: I97745f89e8c5481196e645dada8762d607a9fb2c
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The previous solution was a global static, which got deleted at the end
of the execution of the application or at QtCore unload, whichever came
first. Unfortunately, the end of the execution often came first, which
is inconvenient: it means the global was deleted before all atexit
functions were run, including some QLibrary destructors.
Consequently, some QLibrary destructors did not reach the global data
and were thus unable to unload their libraries or delete their data
properly. The previous solution leaked.
This solution instead uses a Q_DESTRUCTOR_FUNCTION, which makes a
requirement to destroy only at QtCore unload time. Thus, we're sure that
all references have been dropped.
Additionally, during the cleanup, do try to unload the libraries that
have a single reference count left. That means either a QLibrary that
was destroyed without unload(), or a use of the static QLibrary::resolve
functions.
Change-Id: I12e0943b0c6edc27390c103b368d1b04bfe7e302
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
| |
Change-Id: Ic804938fc352291d011800d21e549c10acac66fb
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Loading both the debug and release version of the cocoa
plugins causes the objective-c runtime to print "duplicate
class definitions" warnings.
Fix this by directing the plugin loader to only load
one of the cocoa plugins if both are available. Implement
this as a special case directly in QFactoryLoader.
Define QT_NO_DEBUG_PLUGIN_CHECK to allow mixing
debug and release builds.
Task-number: QTBUG-28155
Change-Id: Ie1927b219cc501a821f91b7e4b56f0589e0acbf5
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
| |
Change copyrights and license headers from Nokia to Digia
Change-Id: If1cc974286d29fd01ec6c19dd4719a67f4c3f00e
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Plugin metadata has been updated in load(), with the
side-effect of metadata not beeing available until
plugin has been loaded - and which the new metadata
system tries to prevent in the first place. The
metadata is now updated (and avaiable) as soon as
a valid filename is set.
Change-Id: Ia5aedc67d8115e71c2ecbcbcadf786ba1c2893d8
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
|
|
|
|
|
| |
Change-Id: I19100755c97cc155c76a859e19940e9f9222d34e
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
|
|
|
|
|
|
|
|
|
| |
Unloading and reloading a plugin didn't work correctly,
because we didn't reset instance to 0 on unload.
Task-number: QTBUG-26098
Change-Id: Ic3e4497f359b1ca455be949dce9cafa9d67d8039
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If an absolute path is specified we try that first. Otherwise we first
try the most likely system-specific format (e.g. libfoo.so) on Unix.
This improves performance especially on systems with slow flash devices.
For example, prior to this commit loading the Xcursor library (in the
xcb plugin) results in attempts to dlopen:
"Xcursor"
"Xcursor.so.1"
"libXcursor"
"libXcursor.so.1"
With this commit this is reduced to a single attempt of:
"libXcursor.so.1"
Plugin loading uses absolute paths with QLibrary so there is no
performance penalty for plugins with this commit.
This is however a behavioural change with respect to Qt4 but one
that I believe is justified and wanted.
Change-Id: I7813afa335f9bf515e87934c2f8f97888818c69c
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
|