According to the Intel Optimization Manual section 3.5.1.1 Use of INC
and DEC Instructions, those instructions modify only part of the flags
register, so they mey introduce unnecessary data dependencies on
previous flag-setting operations so that the resulting flags are
computed. Preferring ADD and SUB (rule 33) is recommended.
However, we don't do it for 16-bit integers. The reason is that the
presence of the 0x66 prefix may trigger a slower decoding codepath in
the processor (up to 6 cycles, as opposed to 1). The same Intel manual
talks about Length-Changing Prefix, which applies in particular to
instructions with 16-bit immediates. The assembler generally produces
uses the 8-bit immediate variant of the ADD and SUB instructions, but
to be on the safe side, we prefer to use INC and DEC here.
Change-Id: Ic03236ac600a5b4e087614d21df5d3c666ae064e
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
The assembly output showed that GCC was generating some wrong code in
some conditions, so update the constraints so it will do the right
thing: the expectedValue constraint needs to be in/out with early
clobber. In/out because cmpxchg8b really does produce output and, even
if we don't care about it, GCC needs to be told that the registers
used (EAX:EDX) were modified. The early clobber is necessary so it
won't schedule EAX or EDX to be the same as the EBX_reg (the register
we'll xchg EBX with).
Since EAX and EDX are in/out and EBX can't be used, the only remaining
low register for the "sete" instruction is CL. So use it directly and
set ECX to be in/out too.
For whatever reason, it can't find enough registers in debug mode and
this expansion doesn't work. It looks like a bug though, since this
requires 4 registers and one memory operand and in debug mode it must
have EAX, ECX, EDX, ESI and EDI free for use. One of ESI or EDI is
used to xchg EBX with, which means there must be at least one more
free general register.
Change-Id: I1f11e68d776bf9ad216b34ca316a53129122fabe
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Lack of support for these types is not a real issue as endian
conversions on byte-sized types are no-ops. Still, the conversions are
useful as they facilitate writing of generic code. They can also be used
explicitly as a way to document in code an endian-specific binary
format:
uchar *data;
quint8 tag = qFromLittleEndian<quint8>(data++);
quint32 size = qFromLittleEndian<quint32>(data);
This commit also adds a test for functions documented in the QtEndian
header.
Change-Id: I2f6c876ce89d2adb8c03a1c8a25921d225bf6f92
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The only use of QUrl in qmake, moc, uic and rcc is due to QVariant's
internals, so let's disable it. This means those binaries are now
probably a lot smaller since the parsing and IDNA code don't need to
be present.
Change-Id: Ie156b0817d119b2ba5d3dcb9712a9fea2ee7d4a1
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
metaObject pointer will be used in future.
Change-Id: I1f335687ad1aa443def21efcb5d4a2eaf3583c44
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Intel CC 12.1 supports AVX2 but only with -march=core-avx2. The -mavx2
option produces a warning.
GCC 4.6 does not recognise any option.
GCC 4.7 recognises both -mavx2 and -march=core-avx2 so let's use the
latter for now. We may need to change to -mavx2 when there's an AMD
processor that supports AVX2 too.
Change-Id: I529240e6e6c2c0e3942d357e0320212d954fe4de
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
This also removes the check for SSE, but the check for SSE2 and
further technologies is kept. If SSE2 is present, then SSE is too. We
don't have any code that uses the original SSE instructions only.
Remove the CMOV detection, since we don't use that anywhere and we're
not likely to ever use them..
Change-Id: I3faf2c555ad1c007c52a54644138902f716c1fe1
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
For all practical purposes, the fallback introduced here returns the
desired value. Having a fallback enables unconditional use of Q_ALIGNOF.
For compilers that provide native support for it, Q_ALIGNOF is otherwise
#defined in qcompilerdetection.h.
Change-Id: Ie148ca8936cbbf8b80fe87771a14797c39a9d30c
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Internally we construct QByteArrays from QStaticByteArrays. For example
moc is generating QStaticByteArray structure for every string it saves.
New test cases check if a QByteArray constructed from a QStaticByteArray
behaves as a not statically constructed one.
Change-Id: Ia4aa9a1a5bc0209507636c683a782dda00eae85c
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
QWorkspace had been called Q3Workspace before Qt4.0 finally released.
In a sense, it is a Qt3 support Widget. And QWorkspace has been
deprecated and replaced by QMdiArea at Qt4.3.
Change-Id: Iea1bf831c9960c23c2b21d51fdc7c13b303642ea
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
This allows me to keep the UTF-8 invalid data in one safe place. I
won't need to copy & paste it.
Change-Id: Icb909d08b7f8d0e1ffbc28e01a0ba0c1fa9dccf0
Reviewed-by: David Faure <faure@kde.org>
QSharedPointer is about to become final.
Instead of inheriting from it to add implicit
conversions to and from QSslSocket*, make
QSslSocketPtr a typedef, and make the
conversions explicit.
Change-Id: I4eebb262ab5aef348f4d676f9e839325d4ed13da
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QSharedPointer is about to be made final. Instead
of inheriting from it to gain access to the
d-pointer, cast it to a layout-compatible struct
and access the pointer from there.
Assert liberally to ensure layout compatibility.
Change-Id: Ifc0fa6a6608e861469286673844325663f4f7fcc
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Remove the inheritance hack used in
QDirIterator to gain access to QDir's
d-pointer by simply making QDirIterator
a friend of QDir.
This allows to turn QDir into a final class.
Change-Id: I97efef8714bb194d62b9fe5192ce240a90f2bf97
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QString and QVariant are about to be marked
Q_DECL_FINAL_CLASS, so change inheritance to
composition.
At least this was private inheritance...
Change-Id: I43caaa6c03041b8f0bd0f7987ddb4c6ff8309e50
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QSharedPointer isn't meant to be used as a
base class.
Instead of inheriting from it to add implicit
conversions to and from QNetworkReply*, make
QNetworkReplyPtr a typedef, overload two
oft-used functions to take a QNetworkReplyPtr
in addition to QNetworkReply*, and otherwise
make the conversions explicit.
Change-Id: I1eff1793a19f2d5bad1cce8de74c0786675a50f3
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The QLocalSocket is using a QLocalUnixSocket which is a QTcpSocket
that has buffering enabled. Because QLocalSocket is a QIODevice it
also has its own read buffer. By opening the QIODevice Unbuffered
we only use the buffer in the QLocalUnixSocket.
Change-Id: I4b9091522f73b0831b20efb4a2a00c2603fc748b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Key events were not sent to items on graphics view after the change
412dbdf410. This is because the change
only checks if QGuiApplication::focusObject() exists. Qt::ImEnabled
needs to be checked too.
Change-Id: I2a78af717a7a1a5d84fbc8b521253bdc25b43286
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
Use the heuristic that windows that are not an
NSPanel and has the MaximizedButtonHint get the
fullscreen button.
Change-Id: I9a3520b7e4cd1833421d5d2f5b73da7edbfc3e74
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Change-Id: Ibc3a95f3560af909a45d147098184c2194187932
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
QMap used to use a skiplist in Qt 4.x, which has variable
sized nodes and we can thus not optimise using custom
allocators.
The rewrite now uses a red-black tree, and all allocations
and tree operations happen in the cpp file. This will allow
us to introduce custom allocation schemes in later versions
of Qt.
Added some more tests and a benchmark. Memory consumption
of the new QMap implementation is pretty much the same as before.
Performance of insertion and lookup has increased by 10-30%. iteration
is slower, but still extremely fast and should not matter compared
to the work usually done when iterating.
Change-Id: I8796c0e4b207d01111e2ead7ae55afb464dd88f5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The regular expression should look for underscores in addition to
alphanumeric characters.
Change-Id: Idc3dbd67291ec1420f818d74fba8413b1e7cbcf1
Reviewed-by: Girish Ramakrishnan <girish.1.ramakrishnan@nokia.com>
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Free form tentative commit proved to require too much fiddling on
rendered text vs. logical content. Needs simpler mechanism.
Change-Id: Ia4e341abf342d25675fd1129efb11094dde410b2
Reviewed-by: Joona Petrell <joona.t.petrell@nokia.com>
Just so I have a similar workflow all over the place.
Change-Id: I99df5042eac1b4fbf3613f339e8a4522f94ed1f5
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Was not emitted when removed by input method event.
Change-Id: Ia2c0dcb09d42826188d4612f4c1705a41874a31d
Reviewed-by: Andrew den Exter <andrew.den-exter@nokia.com>
Fixed Q_WS_WIN flagging to Q_OS_WIN in QPrinter API and related
implementation to make API match the documentation and Qt 4.8.
Also Removed the unused internal HDC related functions from the API,
that were previously behind Q_WS_WIN flag.
Some of the properties tested are documented to be valid for native
print engine only in X11 environment, so skipped testing those in
non-xcb environments.
Copy collation is also apparently not supported in Windows native print
engine, though this seems to be undocumented, so skipped that only in
Windows.
At least one of the test blocks in tst_QPrinter::valuePreservation()
failed due to default printer not getting set properly, so fixed that,
too.
Task-number: QTBUG-24191
Task-number: QTBUG-22927
Change-Id: I44a5e3d647a1279fcc7f1e99de6881f9be330246
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
The QApplication::setGraphicsSystem() call is entirely vestigial after the
QPA rearchitecting of Qt. Deprecate it to give people a heads up that their
original intent no longer has any impact on their application.
Change-Id: I21838431db1057f7f7858b4a434ae0eebd2317de
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
Reviewed-by: Girish Ramakrishnan <girish.1.ramakrishnan@nokia.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reread the environment, and update the locale data on request.
Also fix a small bug where the time format wasn't inherited from
the LANG environment variable if LC_TIME wasn't set.
Change-Id: I619447603ade19d5587b427491cfd9c8fe814858
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Don't assume that the global static is a function. Instead, create a
class that is friends.
Change-Id: I992f4d819b367899b19d95b4983e6b243f0ed932
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
GCC 3.3 is no longer supported. IRIX is no longer supported.
Change-Id: I1656a6fdcf2b244f0f6c812e71b0e793d37fb98b
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Any message/error in mkspecs or qmake feature files ends up confusing
the current arch detection logic. Instead, search for
"Project MESSAGE: .* Architecture: <arch>".
Change-Id: I308932a5b75f3a1fcbc4fe30c74faf2e83b2d752
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Cannot have double-to-float conversions inside { } in C++11.
Change-Id: I13c27307efd703420b6667d919bb42c2ee82e6cc
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
The static library needs to link to zlib too, so that the plugins
using this library link to zlib.
At runtime this doesn't make a difference because -system-zlib means
QtCore is linked to zlib and the system zlib shared library was
loaded, so the symbols are resolved. However, since -Wl,-no-undefined
was added to the build, the plugin must link to zlib explicitly too.
Change-Id: I10d6df8814e823079fee76ae6a7d5b55057c9daf
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
The symbol must be exported from QtGui for the plugins to be able to
see it.
Also, fix the build in namespaced case: cannot use extern in
QtPlatformSupport because it's not compiled into the Qt namespace.
Change-Id: I029533f3524e3cbf6c87aed79c1f2e7b55aebb9b
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
The base class has a virtual destructor, so the destructor is already
virtual. Make it non-implicit so that the virtual table and other
virtual inline methods don't get emitted everywhere.
Change-Id: I15296c1114086ff0b1da701ccd51525bec99d76b
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
This must have been a mistake. IA-64 processors used to contain an
IA-32 core, which supported SSE2. However, the IA-64 mode supported no
such thing.
Change-Id: I731c6018166af6b897cb697c68aad4ef3ae93132
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
permits relocatable qmake installations in, say, $sysroot/hosttools/.
note that it makes patently no sense to have a relative Prefix when
using a relative Sysroot.
Change-Id: I36e5149acb20a885c3cd244bc9d67aa48b01a56f
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
