Merge remote branch 'qt/master' into lighthouse-master

Conflicts:
	configure
	src/corelib/global/qglobal.h
	src/corelib/tools/qsimd.cpp

19 files changed, 488 insertions, 226 deletions

diff --git a/src/corelib/tools/qbitarray.h b/src/corelib/tools/qbitarray.h
index 60cdc9c..bd79904 100644
--- a/src/corelib/tools/qbitarray.h
+++ b/src/corelib/tools/qbitarray.h
@@ -63,6 +63,10 @@ public:
     explicit QBitArray(int size, bool val = false);
     QBitArray(const QBitArray &other) : d(other.d) {}
     inline QBitArray &operator=(const QBitArray &other) { d = other.d; return *this; }
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QBitArray &operator=(QBitArray &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
 
     inline int size() const { return (d.size() << 3) - *d.constData(); }
     inline int count() const { return (d.size() << 3) - *d.constData(); }
diff --git a/src/corelib/tools/qbytearray.cpp b/src/corelib/tools/qbytearray.cpp
index 3062c4a..0a4aad2 100644
--- a/src/corelib/tools/qbytearray.cpp
+++ b/src/corelib/tools/qbytearray.cpp
@@ -854,7 +854,7 @@ QByteArray::Data QByteArray::shared_empty = { Q_BASIC_ATOMIC_INITIALIZER(1),
     This operation takes \l{constant time}, because QByteArray is
     \l{implicitly shared}. This makes returning a QByteArray from a
     function very fast. If a shared instance is modified, it will be
-    copied (copy-on-write), and that takes \l{linear time}.
+    copied (copy-on-write), taking \l{linear time}.
 
     \sa operator=()
 */
@@ -1190,10 +1190,18 @@ void QByteArray::chop(int n)
     Example:
     \snippet doc/src/snippets/code/src_corelib_tools_qbytearray.cpp 12
 
-    This operation is typically very fast (\l{constant time}),
-    because QByteArray preallocates extra space at the end of the
-    character data so it can grow without reallocating the entire
-    data each time.
+    Note: QByteArray is an \l{implicitly shared} class. Consequently,
+    if \e this is an empty QByteArray, then \e this will just share
+    the data held in \a ba. In this case, no copying of data is done,
+    taking \l{constant time}. If a shared instance is modified, it will
+    be copied (copy-on-write), taking \l{linear time}.
+
+    If \e this is not an empty QByteArray, a deep copy of the data is
+    performed, taking \l{linear time}.
+
+    This operation typically does not suffer from allocation overhead,
+    because QByteArray preallocates extra space at the end of the data
+    so that it may grow without reallocating for each append operation.
 
     \sa append(), prepend()
 */
@@ -1478,7 +1486,12 @@ QByteArray QByteArray::nulTerminated() const
 
     Note: QByteArray is an \l{implicitly shared} class. Consequently,
     if \e this is an empty QByteArray, then \e this will just share
-    the data held in \a ba. In this case, no copying of data is done.
+    the data held in \a ba. In this case, no copying of data is done,
+    taking \l{constant time}. If a shared instance is modified, it will
+    be copied (copy-on-write), taking \l{linear time}.
+
+    If \e this is not an empty QByteArray, a deep copy of the data is
+    performed, taking \l{linear time}.
 
     \sa append(), insert()
 */
@@ -1551,14 +1564,18 @@ QByteArray &QByteArray::prepend(char ch)
 
     This is the same as insert(size(), \a ba).
 
-    This operation is typically very fast (\l{constant time}),
-    because QByteArray preallocates extra space at the end of the
-    character data so it can grow without reallocating the entire
-    data each time.
-
     Note: QByteArray is an \l{implicitly shared} class. Consequently,
     if \e this is an empty QByteArray, then \e this will just share
-    the data held in \a ba. In this case, no copying of data is done.
+    the data held in \a ba. In this case, no copying of data is done,
+    taking \l{constant time}. If a shared instance is modified, it will
+    be copied (copy-on-write), taking \l{linear time}.
+
+    If \e this is not an empty QByteArray, a deep copy of the data is
+    performed, taking \l{linear time}.
+
+    This operation typically does not suffer from allocation overhead,
+    because QByteArray preallocates extra space at the end of the data
+    so that it may grow without reallocating for each append operation.
 
     \sa operator+=(), prepend(), insert()
 */
diff --git a/src/corelib/tools/qbytearray.h b/src/corelib/tools/qbytearray.h
index a3fe3f5..3cdcaab 100644
--- a/src/corelib/tools/qbytearray.h
+++ b/src/corelib/tools/qbytearray.h
@@ -144,6 +144,10 @@ public:
 
     QByteArray &operator=(const QByteArray &);
     QByteArray &operator=(const char *str);
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QByteArray &operator=(QByteArray &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
 
     inline int size() const;
     bool isEmpty() const;
diff --git a/src/corelib/tools/qcontiguouscache.h b/src/corelib/tools/qcontiguouscache.h
index f767962..4c1a846 100644
--- a/src/corelib/tools/qcontiguouscache.h
+++ b/src/corelib/tools/qcontiguouscache.h
@@ -110,6 +110,10 @@ public:
     inline void setSharable(bool sharable) { if (!sharable) detach(); d->sharable = sharable; }
 
     QContiguousCache<T> &operator=(const QContiguousCache<T> &other);
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QContiguousCache<T> &operator=(QContiguousCache<T> &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
     bool operator==(const QContiguousCache<T> &other) const;
     inline bool operator!=(const QContiguousCache<T> &other) const { return !(*this == other); }
 
diff --git a/src/corelib/tools/qeasingcurve.cpp b/src/corelib/tools/qeasingcurve.cpp
index ee791e0..7fe9170 100644
--- a/src/corelib/tools/qeasingcurve.cpp
+++ b/src/corelib/tools/qeasingcurve.cpp
@@ -332,7 +332,7 @@ public:
     enum Type { In, Out, InOut, OutIn };
 
     QEasingCurveFunction(QEasingCurveFunction::Type type = In, qreal period = 0.3, qreal amplitude = 1.0,
-        qreal overshoot = 1.70158f)
+        qreal overshoot = 1.70158)
         : _t(type), _p(period), _a(amplitude), _o(overshoot)
     { }
     virtual ~QEasingCurveFunction() {}
@@ -359,9 +359,9 @@ QEasingCurveFunction *QEasingCurveFunction::copy() const
 bool QEasingCurveFunction::operator==(const QEasingCurveFunction& other)
 {
     return _t == other._t &&
-           _p == other._p &&
-           _a == other._a &&
-           _o == other._o;
+           qFuzzyCompare(_p, other._p) &&
+           qFuzzyCompare(_a, other._a) &&
+           qFuzzyCompare(_o, other._o);
 }
 
 QT_BEGIN_INCLUDE_NAMESPACE
@@ -400,8 +400,8 @@ struct ElasticEase : public QEasingCurveFunction
 
     qreal value(qreal t)
     {
-        qreal p = (_p < 0) ? 0.3f : _p;
-        qreal a = (_a < 0) ? 1.0f : _a;
+        qreal p = (_p < 0) ? qreal(0.3) : _p;
+        qreal a = (_a < 0) ? qreal(1.0) : _a;
         switch(_t) {
         case In:
             return easeInElastic(t, a, p);
@@ -420,7 +420,7 @@ struct ElasticEase : public QEasingCurveFunction
 struct BounceEase : public QEasingCurveFunction
 {
     BounceEase(Type type)
-        : QEasingCurveFunction(type, 0.3f, 1.0f)
+        : QEasingCurveFunction(type, qreal(0.3), qreal(1.0))
     { }
 
     QEasingCurveFunction *copy() const
@@ -432,7 +432,7 @@ struct BounceEase : public QEasingCurveFunction
 
     qreal value(qreal t)
     {
-        qreal a = (_a < 0) ? 1.0f : _a;
+        qreal a = (_a < 0) ? qreal(1.0) : _a;
         switch(_t) {
         case In:
             return easeInBounce(t, a);
@@ -451,7 +451,7 @@ struct BounceEase : public QEasingCurveFunction
 struct BackEase : public QEasingCurveFunction
 {
     BackEase(Type type)
-        : QEasingCurveFunction(type, 0.3f, 1.0f, 1.70158f)
+        : QEasingCurveFunction(type, qreal(0.3), qreal(1.0), qreal(1.70158))
     { }
 
     QEasingCurveFunction *copy() const
@@ -463,7 +463,7 @@ struct BackEase : public QEasingCurveFunction
 
     qreal value(qreal t)
     {
-        qreal o = (_o < 0) ? 1.70158f : _o;
+        qreal o = (_o < 0) ? qreal(1.70158) : _o;
         switch(_t) {
         case In:
             return easeInBack(t, o);
@@ -595,7 +595,7 @@ static QEasingCurveFunction *curveToFunctionObject(QEasingCurve::Type type)
         curveFunc = new BackEase(BackEase::OutIn);
         break;
     default:
-        curveFunc = new QEasingCurveFunction(QEasingCurveFunction::In, 0.3f, 1.0f, 1.70158f);     // ###
+        curveFunc = new QEasingCurveFunction(QEasingCurveFunction::In, qreal(0.3), qreal(1.0), qreal(1.70158));
     }
 
     return curveFunc;
@@ -657,9 +657,17 @@ bool QEasingCurve::operator==(const QEasingCurve &other) const
 {
     bool res = d_ptr->func == other.d_ptr->func
             && d_ptr->type == other.d_ptr->type;
-    if (res && d_ptr->config && other.d_ptr->config) {
+    if (res) {
+        if (d_ptr->config && other.d_ptr->config) {
         // catch the config content
-        res = d_ptr->config->operator==(*(other.d_ptr->config));
+            res = d_ptr->config->operator==(*(other.d_ptr->config));
+
+        } else if (d_ptr->config || other.d_ptr->config) {
+        // one one has a config object, which could contain default values
+            res = qFuzzyCompare(amplitude(), other.amplitude()) &&
+                  qFuzzyCompare(period(), other.period()) &&
+                  qFuzzyCompare(overshoot(), other.overshoot());
+        }
     }
     return res;
 }
@@ -681,7 +689,7 @@ bool QEasingCurve::operator==(const QEasingCurve &other) const
  */
 qreal QEasingCurve::amplitude() const
 {
-    return d_ptr->config ? d_ptr->config->_a : 1.0;
+    return d_ptr->config ? d_ptr->config->_a : qreal(1.0);
 }
 
 /*!
@@ -705,7 +713,7 @@ void QEasingCurve::setAmplitude(qreal amplitude)
  */
 qreal QEasingCurve::period() const
 {
-    return d_ptr->config ? d_ptr->config->_p : 0.3;
+    return d_ptr->config ? d_ptr->config->_p : qreal(0.3);
 }
 
 /*!
@@ -729,7 +737,7 @@ void QEasingCurve::setPeriod(qreal period)
  */
 qreal QEasingCurve::overshoot() const
 {
-    return d_ptr->config ? d_ptr->config->_o : 1.70158f;
+    return d_ptr->config ? d_ptr->config->_o : qreal(1.70158) ;
 }
 
 /*!
diff --git a/src/corelib/tools/qelapsedtimer_unix.cpp b/src/corelib/tools/qelapsedtimer_unix.cpp
index 2c4ea58..633fa00 100644
--- a/src/corelib/tools/qelapsedtimer_unix.cpp
+++ b/src/corelib/tools/qelapsedtimer_unix.cpp
@@ -40,22 +40,58 @@
 ****************************************************************************/
 
 #include "qelapsedtimer.h"
-#include "qpair.h"
 #include <sys/time.h>
 #include <time.h>
 #include <unistd.h>
 
-#if !defined(QT_NO_CLOCK_MONOTONIC)
-# if defined(QT_BOOTSTRAPPED)
-#  define QT_NO_CLOCK_MONOTONIC
+#if defined(QT_NO_CLOCK_MONOTONIC) || defined(QT_BOOTSTRAPPED)
+// turn off the monotonic clock
+# ifdef _POSIX_MONOTONIC_CLOCK
+#  undef _POSIX_MONOTONIC_CLOCK
 # endif
+# define _POSIX_MONOTONIC_CLOCK -1
 #endif
 
 QT_BEGIN_NAMESPACE
 
-static qint64 fractionAdjustment()
+#if (_POSIX_MONOTONIC_CLOCK-0 != 0)
+static const bool monotonicClockChecked = true;
+static const bool monotonicClockAvailable = _POSIX_MONOTONIC_CLOCK > 0;
+#else
+static int monotonicClockChecked = false;
+static int monotonicClockAvailable = false;
+#endif
+
+#ifdef Q_CC_GNU
+# define is_likely(x) __builtin_expect((x), 1)
+#else
+# define is_likely(x) (x)
+#endif
+#define load_acquire(x) ((volatile const int&)(x))
+#define store_release(x,v) ((volatile int&)(x) = (v))
+
+static void unixCheckClockType()
+{
+#if (_POSIX_MONOTONIC_CLOCK-0 == 0)
+    if (is_likely(load_acquire(monotonicClockChecked)))
+        return;
+
+# if defined(_SC_MONOTONIC_CLOCK)
+    // detect if the system support monotonic timers
+    long x = sysconf(_SC_MONOTONIC_CLOCK);
+    store_release(monotonicClockAvailable, x >= 200112L);
+# endif
+
+    store_release(monotonicClockChecked, true);
+#endif
+}
+
+static inline qint64 fractionAdjustment()
 {
-    if (QElapsedTimer::isMonotonic()) {
+    // disabled, but otherwise indicates bad usage of QElapsedTimer
+    //Q_ASSERT(monotonicClockChecked);
+
+    if (monotonicClockAvailable) {
         // the monotonic timer is measured in nanoseconds
         // 1 ms = 1000000 ns
         return 1000*1000ull;
@@ -68,90 +104,73 @@ static qint64 fractionAdjustment()
 
 bool QElapsedTimer::isMonotonic()
 {
-#if (_POSIX_MONOTONIC_CLOCK-0 > 0)
-    return true;
-#else
-    static int returnValue = 0;
-
-    if (returnValue == 0) {
-#  if (_POSIX_MONOTONIC_CLOCK-0 < 0) || !defined(_SC_MONOTONIC_CLOCK)
-        returnValue = -1;
-#  elif (_POSIX_MONOTONIC_CLOCK == 0)
-        // detect if the system support monotonic timers
-        long x = sysconf(_SC_MONOTONIC_CLOCK);
-        returnValue = (x >= 200112L) ? 1 : -1;
-#  endif
-    }
-
-    return returnValue != -1;
-#endif
+    unixCheckClockType();
+    return monotonicClockAvailable;
 }
 
 QElapsedTimer::ClockType QElapsedTimer::clockType()
 {
-    return isMonotonic() ? MonotonicClock : SystemTime;
+    unixCheckClockType();
+    return monotonicClockAvailable ? MonotonicClock : SystemTime;
 }
 
-static inline QPair<long, long> do_gettime()
+static inline void do_gettime(qint64 *sec, qint64 *frac)
 {
-#if (_POSIX_MONOTONIC_CLOCK-0 > 0)
-    timespec ts;
-    clock_gettime(CLOCK_MONOTONIC, &ts);
-    return qMakePair<long,long>(ts.tv_sec, ts.tv_nsec);
-#else
-#  if !defined(QT_NO_CLOCK_MONOTONIC) && !defined(QT_BOOTSTRAPPED)
-    if (QElapsedTimer::isMonotonic()) {
+#if (_POSIX_MONOTONIC_CLOCK-0 >= 0)
+    unixCheckClockType();
+    if (is_likely(monotonicClockAvailable)) {
         timespec ts;
         clock_gettime(CLOCK_MONOTONIC, &ts);
-        return qMakePair<long,long>(ts.tv_sec, ts.tv_nsec);
+        *sec = ts.tv_sec;
+        *frac = ts.tv_nsec;
+        return;
     }
-#  endif
+#endif
     // use gettimeofday
     timeval tv;
     ::gettimeofday(&tv, 0);
-    return qMakePair<long,long>(tv.tv_sec, tv.tv_usec);
-#endif
+    *sec = tv.tv_sec;
+    *frac = tv.tv_usec;
 }
 
 // used in qcore_unix.cpp and qeventdispatcher_unix.cpp
 timeval qt_gettime()
 {
-    QPair<long, long> r = do_gettime();
+    qint64 sec, frac;
+    do_gettime(&sec, &frac);
 
     timeval tv;
-    tv.tv_sec = r.first;
-    tv.tv_usec = r.second;
-    if (QElapsedTimer::isMonotonic())
+    tv.tv_sec = sec;
+    tv.tv_usec = frac;
+    if (monotonicClockAvailable)
         tv.tv_usec /= 1000;
 
     return tv;
 }
 
+static qint64 elapsedAndRestart(qint64 sec, qint64 frac,
+                                qint64 *nowsec, qint64 *nowfrac)
+{
+    do_gettime(nowsec, nowfrac);
+    sec = *nowsec - sec;
+    frac = *nowfrac - frac;
+    return sec * Q_INT64_C(1000) + frac / fractionAdjustment();
+}
+
 void QElapsedTimer::start()
 {
-    QPair<long, long> r = do_gettime();
-    t1 = r.first;
-    t2 = r.second;
+    do_gettime(&t1, &t2);
 }
 
 qint64 QElapsedTimer::restart()
 {
-    QPair<long, long> r = do_gettime();
-    qint64 oldt1 = t1;
-    qint64 oldt2 = t2;
-    t1 = r.first;
-    t2 = r.second;
-
-    r.first -= oldt1;
-    r.second -= oldt2;
-    return r.first * Q_INT64_C(1000) + r.second / fractionAdjustment();
+    return elapsedAndRestart(t1, t2, &t1, &t2);
 }
 
 qint64 QElapsedTimer::elapsed() const
 {
-    QElapsedTimer now;
-    now.start();
-    return msecsTo(now);
+    qint64 sec, frac;
+    return elapsedAndRestart(t1, t2, &sec, &frac);
 }
 
 qint64 QElapsedTimer::msecsSinceReference() const
diff --git a/src/corelib/tools/qhash.h b/src/corelib/tools/qhash.h
index 14ed514..992ff33 100644
--- a/src/corelib/tools/qhash.h
+++ b/src/corelib/tools/qhash.h
@@ -283,6 +283,10 @@ public:
     inline ~QHash() { if (!d->ref.deref()) freeData(d); }
 
     QHash<Key, T> &operator=(const QHash<Key, T> &other);
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QHash<Key, T> &operator=(QHash<Key, T> &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
 
     bool operator==(const QHash<Key, T> &other) const;
     inline bool operator!=(const QHash<Key, T> &other) const { return !(*this == other); }
diff --git a/src/corelib/tools/qlinkedlist.h b/src/corelib/tools/qlinkedlist.h
index 9b3efa3..c087944 100644
--- a/src/corelib/tools/qlinkedlist.h
+++ b/src/corelib/tools/qlinkedlist.h
@@ -85,6 +85,10 @@ public:
     inline QLinkedList(const QLinkedList<T> &l) : d(l.d) { d->ref.ref(); if (!d->sharable) detach(); }
     ~QLinkedList();
     QLinkedList<T> &operator=(const QLinkedList<T> &);
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QLinkedList<T> &operator=(QLinkedList<T> &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
     bool operator==(const QLinkedList<T> &l) const;
     inline bool operator!=(const QLinkedList<T> &l) const { return !(*this == l); }
 
diff --git a/src/corelib/tools/qlist.h b/src/corelib/tools/qlist.h
index 1282bca..8f988d6 100644
--- a/src/corelib/tools/qlist.h
+++ b/src/corelib/tools/qlist.h
@@ -50,6 +50,10 @@
 #include <iterator>
 #include <list>
 #endif
+#ifdef Q_COMPILER_INITIALIZER_LISTS
+#include <iterator>
+#include <initializer_list>
+#endif
 
 #include <new>
 #include <limits.h>
@@ -118,6 +122,14 @@ public:
     inline QList(const QList<T> &l) : d(l.d) { d->ref.ref(); if (!d->sharable) detach_helper(); }
     ~QList();
     QList<T> &operator=(const QList<T> &l);
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QList &operator=(QList &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
+#ifdef Q_COMPILER_INITIALIZER_LISTS
+    inline QList(std::initializer_list<T> args) : d(&QListData::shared_null)
+    { d->ref.ref(); qCopy(args.begin(), args.end(), std::back_inserter(*this)); }
+#endif
     bool operator==(const QList<T> &l) const;
     inline bool operator!=(const QList<T> &l) const { return !(*this == l); }
 
diff --git a/src/corelib/tools/qmap.h b/src/corelib/tools/qmap.h
index 08f5a35..ce8fd75 100644
--- a/src/corelib/tools/qmap.h
+++ b/src/corelib/tools/qmap.h
@@ -185,6 +185,10 @@ public:
     inline ~QMap() { if (!d) return; if (!d->ref.deref()) freeData(d); }
 
     QMap<Key, T> &operator=(const QMap<Key, T> &other);
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QMap<Key, T> &operator=(QMap<Key, T> &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
 #ifndef QT_NO_STL
     explicit QMap(const typename std::map<Key, T> &other);
     std::map<Key, T> toStdMap() const;
diff --git a/src/corelib/tools/qregexp.h b/src/corelib/tools/qregexp.h
index e19c130..0b4a702 100644
--- a/src/corelib/tools/qregexp.h
+++ b/src/corelib/tools/qregexp.h
@@ -76,6 +76,10 @@ public:
     QRegExp(const QRegExp &rx);
     ~QRegExp();
     QRegExp &operator=(const QRegExp &rx);
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QRegExp &operator=(QRegExp &&other)
+    { qSwap(priv,other.priv); return *this; }
+#endif
 
     bool operator==(const QRegExp &rx) const;
     inline bool operator!=(const QRegExp &rx) const { return !operator==(rx); }
diff --git a/src/corelib/tools/qset.h b/src/corelib/tools/qset.h
index fe50d4d..dc3c45a 100644
--- a/src/corelib/tools/qset.h
+++ b/src/corelib/tools/qset.h
@@ -61,6 +61,10 @@ public:
 
     inline QSet<T> &operator=(const QSet<T> &other)
         { q_hash = other.q_hash; return *this; }
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QSet<T> &operator=(QSet<T> &&other)
+        { qSwap(q_hash, other.q_hash); return *this; }
+#endif
 
     inline bool operator==(const QSet<T> &other) const
         { return q_hash == other.q_hash; }
diff --git a/src/corelib/tools/qshareddata.h b/src/corelib/tools/qshareddata.h
index 1ae2fa9..b646a9d 100644
--- a/src/corelib/tools/qshareddata.h
+++ b/src/corelib/tools/qshareddata.h
@@ -113,6 +113,10 @@ public:
         }
         return *this;
     }
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QSharedDataPointer<T> &operator=(QSharedDataPointer<T> &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
 
     inline bool operator!() const { return !d; }
 
@@ -192,6 +196,10 @@ public:
         }
         return *this;
     }
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QSharedDataPointer<T> &operator=(QSharedDataPointer<T> &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
 
     inline bool operator!() const { return !d; }
 
diff --git a/src/corelib/tools/qsharedpointer_impl.h b/src/corelib/tools/qsharedpointer_impl.h
index bb06f3a..20dda12 100644
--- a/src/corelib/tools/qsharedpointer_impl.h
+++ b/src/corelib/tools/qsharedpointer_impl.h
@@ -465,6 +465,13 @@ public:
         BaseClass::internalCopy(other);
         return *this;
     }
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QSharedPointer<T> &operator=(QSharedPointer<T> &&other)
+    {
+        QSharedPointer<T>::internalSwap(other);
+        return *this;
+    }
+#endif
 
     template <class X>
     inline QSharedPointer(const QSharedPointer<X> &other) : BaseClass(other)
diff --git a/src/corelib/tools/qsimd.cpp b/src/corelib/tools/qsimd.cpp
index fcc8dd7..9b431a2 100644
--- a/src/corelib/tools/qsimd.cpp
+++ b/src/corelib/tools/qsimd.cpp
@@ -41,6 +41,7 @@
 
 #include "qsimd_p.h"
 #include <QByteArray>
+#include <stdio.h>
 
 #if defined(Q_OS_WINCE)
 #include <windows.h>
@@ -50,20 +51,37 @@
 #include <intrin.h>
 #endif
 
-QT_BEGIN_NAMESPACE
+#if defined(Q_OS_LINUX) && defined(__arm__)
+#include "private/qcore_unix_p.h"
 
-uint qDetectCPUFeatures()
-{
-    static uint features = 0xffffffff;
-    if (features != 0xffffffff)
-        return features;
+// the kernel header definitions for HWCAP_*
+// (the ones we need/may need anyway)
+
+// copied from <asm/hwcap.h> (ARM)
+#define HWCAP_IWMMXT    512
+#define HWCAP_CRUNCH    1024
+#define HWCAP_THUMBEE   2048
+#define HWCAP_NEON      4096
+#define HWCAP_VFPv3     8192
+#define HWCAP_VFPv3D16  16384
+
+// copied from <linux/auxvec.h>
+#define AT_HWCAP  16    /* arch dependent hints at CPU capabilities */
+
+#endif
+
+QT_BEGIN_NAMESPACE
 
 #if defined (Q_OS_NACL)
-    // The qDetectCPUFeatures assembler fails verification in the
-    // Native Client sandbox due to use of pushf/popf.
-    features = 0;
-    return features;
+static inline uint detectProcessorFeatures()
+{
+    return 0;
+}
 #elif defined (Q_OS_WINCE)
+static inline uint detectProcessorFeatures()
+{
+    uint features = 0;
+
 #if defined (ARM)
     if (IsProcessorFeaturePresent(PF_ARM_INTEL_WMMX)) {
         features = IWMMXT;
@@ -83,77 +101,98 @@ uint qDetectCPUFeatures()
 #endif
     features = 0;
     return features;
-#elif defined(QT_HAVE_IWMMXT)
+}
+
+#elif defined(__arm__) || defined(__arm) || defined(QT_HAVE_IWMMXT) || defined(QT_HAVE_NEON)
+static inline uint detectProcessorFeatures()
+{
+    uint features = 0;
+
+#if defined(Q_OS_LINUX)
+    int auxv = ::qt_safe_open("/proc/self/auxv", O_RDONLY);
+    if (auxv != -1) {
+        unsigned long vector[64];
+        int nread;
+        while (features == 0) {
+            nread = ::qt_safe_read(auxv, (char *)vector, sizeof vector);
+            if (nread <= 0) {
+                // EOF or error
+                break;
+            }
+
+            int max = nread / (sizeof vector[0]);
+            for (int i = 0; i < max; i += 2)
+                if (vector[i] == AT_HWCAP) {
+                    if (vector[i+1] & HWCAP_IWMMXT)
+                        features |= IWMMXT;
+                    if (vector[i+1] & HWCAP_NEON)
+                        features |= NEON;
+                    break;
+                }
+        }
+
+        ::qt_safe_close(auxv);
+        return features;
+    }
+    // fall back if /proc/self/auxv wasn't found
+#endif
+
+#if defined(QT_HAVE_IWMMXT)
     // runtime detection only available when running as a previlegied process
-    static const bool doIWMMXT = !qgetenv("QT_NO_IWMMXT").toInt();
-    features = doIWMMXT ? IWMMXT : 0;
-    return features;
+    features = IWMMXT;
 #elif defined(QT_HAVE_NEON)
-    static const bool doNEON = !qgetenv("QT_NO_NEON").toInt();
-    features = doNEON ? NEON : 0;
+    features = NEON;
+#endif
+
     return features;
-#else
-    features = 0;
-#if defined(__x86_64__) || defined(Q_OS_WIN64)
-    features = MMX|SSE|SSE2|CMOV;
-#elif defined(__ia64__)
-    features = MMX|SSE|SSE2;
+}
+
 #elif defined(__i386__) || defined(_M_IX86)
+static inline uint detectProcessorFeatures()
+{
+    uint features = 0;
+
     unsigned int extended_result = 0;
     unsigned int feature_result = 0;
     uint result = 0;
     /* see p. 118 of amd64 instruction set manual Vol3 */
 #if defined(Q_CC_GNU)
-    asm ("push %%ebx\n"
-         "pushf\n"
-         "pop %%eax\n"
-         "mov %%eax, %%ebx\n"
-         "xor $0x00200000, %%eax\n"
-         "push %%eax\n"
+    long cpuid_supported, tmp1;
+    asm ("pushf\n"
+         "pop %0\n"
+         "mov %0, %1\n"
+         "xor $0x00200000, %0\n"
+         "push %0\n"
          "popf\n"
          "pushf\n"
-         "pop %%eax\n"
-         "xor %%edx, %%edx\n"
-         "xor %%ebx, %%eax\n"
-         "jz 1f\n"
+         "pop %0\n"
+         "xor %1, %0\n" // %eax is now 0 if CPUID is not supported
+         : "=a" (cpuid_supported), "=r" (tmp1)
+         );
+    if (cpuid_supported) {
+        asm ("xchg %%ebx, %2\n"
+             "cpuid\n"
+             "xchg %%ebx, %2\n"
+            : "=c" (feature_result), "=d" (result), "=&r" (tmp1)
+            : "a" (1));
 
-         "mov $0x00000001, %%eax\n"
-         "cpuid\n"
-         "1:\n"
-         "pop %%ebx\n"
-         "mov %%edx, %0\n"
-         "mov %%ecx, %1\n"
-        : "=r" (result), "=r" (feature_result)
-        :
-        : "%eax", "%ecx", "%edx"
-        );
-
-    asm ("push %%ebx\n"
-         "pushf\n"
-         "pop %%eax\n"
-         "mov %%eax, %%ebx\n"
-         "xor $0x00200000, %%eax\n"
-         "push %%eax\n"
-         "popf\n"
-         "pushf\n"
-         "pop %%eax\n"
-         "xor %%edx, %%edx\n"
-         "xor %%ebx, %%eax\n"
-         "jz 2f\n"
+        asm ("xchg %%ebx, %1\n"
+             "cpuid\n"
+             "cmp $0x80000000, %%eax\n"
+             "jnbe 1f\n"
+             "xor %0, %0\n"
+             "jmp 2f\n"
+             "1:\n"
+             "mov $0x80000001, %%eax\n"
+             "cpuid\n"
+             "2:\n"
+             "xchg %%ebx, %1\n"
+            : "=d" (extended_result), "=&r" (tmp1)
+            : "a" (0x80000000)
+            : "%ecx"
+            );
+    }
 
-         "mov $0x80000000, %%eax\n"
-         "cpuid\n"
-         "cmp $0x80000000, %%eax\n"
-         "jbe 2f\n"
-         "mov $0x80000001, %%eax\n"
-         "cpuid\n"
-         "2:\n"
-         "pop %%ebx\n"
-         "mov %%edx, %0\n"
-        : "=r" (extended_result)
-        :
-        : "%eax", "%ecx", "%edx"
-        );
 #elif defined (Q_OS_WIN)
     _asm {
         push eax
@@ -215,6 +254,7 @@ uint qDetectCPUFeatures()
     }
 #endif
 
+
     // result now contains the standard feature bits
     if (result & (1u << 15))
         features |= CMOV;
@@ -241,33 +281,23 @@ uint qDetectCPUFeatures()
     if (feature_result & (1u << 28))
         features |= AVX;
 
-#endif // i386
+    return features;
+}
 
-#if defined(__x86_64__) || defined(Q_OS_WIN64)
+#elif defined(__x86_64) || defined(Q_OS_WIN64)
+static inline uint detectProcessorFeatures()
+{
+    uint features = MMX|SSE|SSE2|CMOV;
     uint feature_result = 0;
 
 #if defined(Q_CC_GNU)
-    asm ("push %%rbx\n"
-         "pushf\n"
-         "pop %%rax\n"
-         "mov %%eax, %%ebx\n"
-         "xor $0x00200000, %%eax\n"
-         "push %%rax\n"
-         "popf\n"
-         "pushf\n"
-         "pop %%rax\n"
-         "xor %%edx, %%edx\n"
-         "xor %%ebx, %%eax\n"
-         "jz 1f\n"
-
-         "mov $0x00000001, %%eax\n"
+    long tmp;
+    asm ("xchg %%rbx, %1\n"
          "cpuid\n"
-         "1:\n"
-         "pop %%rbx\n"
-         "mov %%ecx, %0\n"
-        : "=r" (feature_result)
-        :
-        : "%eax", "%ecx", "%edx"
+         "xchg %%rbx, %1\n"
+        : "=c" (feature_result), "=&r" (tmp)
+        : "a" (1)
+        : "%edx"
         );
 #elif defined (Q_OS_WIN64)
     {
@@ -287,33 +317,97 @@ uint qDetectCPUFeatures()
         features |= SSE4_2;
     if (feature_result & (1u << 28))
         features |= AVX;
-#endif // x86_64
 
-#if defined(QT_HAVE_MMX)
-    if (qgetenv("QT_NO_MMX").toInt())
-        features ^= MMX;
-#endif
-    if (qgetenv("QT_NO_MMXEXT").toInt())
-        features ^= MMXEXT;
+    return features;
+}
 
-#if defined(QT_HAVE_3DNOW)
-    if (qgetenv("QT_NO_3DNOW").toInt())
-        features ^= MMX3DNOW;
-#endif
-    if (qgetenv("QT_NO_3DNOWEXT").toInt())
-        features ^= MMX3DNOWEXT;
+#elif defined(__ia64__)
+static inline uint detectProcessorFeatures()
+{
+    return MMX|SSE|SSE2;
+}
 
-#if defined(QT_HAVE_SSE)
-    if (qgetenv("QT_NO_SSE").toInt())
-        features ^= SSE;
-#endif
-#if defined(QT_HAVE_SSE2)
-    if (qgetenv("QT_NO_SSE2").toInt())
-        features ^= SSE2;
+#else
+static inline uint detectProcessorFeatures()
+{
+    return 0;
+}
 #endif
 
+/*
+ * Use kdesdk/scripts/generate_string_table.pl to update the table below.
+ * Here's the data (don't forget the ONE leading space):
+ mmx
+ mmxext
+ mmx3dnow
+ mmx3dnowext
+ sse
+ sse2
+ cmov
+ iwmmxt
+ neon
+ sse3
+ ssse3
+ sse4.1
+ sse4.2
+ avx
+  */
+
+// begin generated
+static const char features_string[] =
+    " mmx\0"
+    " mmxext\0"
+    " mmx3dnow\0"
+    " mmx3dnowext\0"
+    " sse\0"
+    " sse2\0"
+    " cmov\0"
+    " iwmmxt\0"
+    " neon\0"
+    " sse3\0"
+    " ssse3\0"
+    " sse4.1\0"
+    " sse4.2\0"
+    " avx\0"
+    "\0";
+
+static const int features_indices[] = {
+       0,    5,   13,   23,   36,   41,   47,   53,
+      61,   67,   73,   80,   88,   96,   -1
+};
+// end generated
+
+const int features_count = (sizeof features_indices - 1) / (sizeof features_indices[0]);
+
+uint qDetectCPUFeatures()
+{
+    static QBasicAtomicInt features = Q_BASIC_ATOMIC_INITIALIZER(-1);
+    if (features != -1)
+        return features;
+
+    uint f = detectProcessorFeatures();
+    QByteArray disable = qgetenv("QT_NO_CPU_FEATURE");
+    if (!disable.isEmpty()) {
+        disable.prepend(' ');
+        for (int i = 0; i < features_count; ++i) {
+            if (disable.contains(features_string + features_indices[i]))
+                f &= ~(1 << i);
+        }
+    }
+
+    features = f;
     return features;
-#endif
+}
+
+void qDumpCPUFeatures()
+{
+    uint features = qDetectCPUFeatures();
+    printf("Processor features: ");
+    for (int i = 0; i < features_count; ++i) {
+        if (features & (1 << i))
+            printf("%s", features_string + features_indices[i]);
+    }
+    puts("");
 }
 
 QT_END_NAMESPACE
diff --git a/src/corelib/tools/qsimd_p.h b/src/corelib/tools/qsimd_p.h
index a3148fb..664543b 100644
--- a/src/corelib/tools/qsimd_p.h
+++ b/src/corelib/tools/qsimd_p.h
@@ -51,8 +51,13 @@ QT_BEGIN_HEADER
 #if defined(QT_NO_MAC_XARCH) || (defined(Q_OS_DARWIN) && (defined(__ppc__) || defined(__ppc64__)))
 // Disable MMX and SSE on Mac/PPC builds, or if the compiler
 // does not support -Xarch argument passing
-#undef QT_HAVE_SSE2
 #undef QT_HAVE_SSE
+#undef QT_HAVE_SSE2
+#undef QT_HAVE_SSE3
+#undef QT_HAVE_SSSE3
+#undef QT_HAVE_SSE4_1
+#undef QT_HAVE_SSE4_2
+#undef QT_HAVE_AVX
 #undef QT_HAVE_3DNOW
 #undef QT_HAVE_MMX
 #endif
@@ -85,6 +90,7 @@ QT_BEGIN_HEADER
 // SSE4.1 and SSE4.2 intrinsics
 #if (defined(QT_HAVE_SSE4_1) || defined(QT_HAVE_SSE4_2)) && (defined(__SSE4_1__) || defined(Q_CC_MSVC))
 #include <smmintrin.h>
+#include <nmmintrin.h>
 #endif
 
 // AVX intrinsics
diff --git a/src/corelib/tools/qstring.cpp b/src/corelib/tools/qstring.cpp
index de45a63..cc90b3a 100644
--- a/src/corelib/tools/qstring.cpp
+++ b/src/corelib/tools/qstring.cpp
@@ -189,19 +189,6 @@ static int ucstricmp(const ushort *a, const ushort *ae, const uchar *b)
     return 1;
 }
 
-// Unicode case-insensitive comparison
-static int ucstrcmp(const QChar *a, int alen, const QChar *b, int blen)
-{
-    if (a == b && alen == blen)
-        return 0;
-    int l = qMin(alen, blen);
-    while (l-- && *a == *b)
-        a++,b++;
-    if (l == -1)
-        return (alen-blen);
-    return a->unicode() - b->unicode();
-}
-
 // Unicode case-sensitive compare two same-sized strings
 static int ucstrncmp(const QChar *a, const QChar *b, int l)
 {
@@ -212,28 +199,71 @@ static int ucstrncmp(const QChar *a, const QChar *b, int l)
     return a->unicode() - b->unicode();
 }
 
+// Unicode case-sensitive comparison
+static int ucstrcmp(const QChar *a, int alen, const QChar *b, int blen)
+{
+    if (a == b && alen == blen)
+        return 0;
+    int l = qMin(alen, blen);
+    int cmp = ucstrncmp(a, b, l);
+    return cmp ? cmp : (alen-blen);
+}
+
 // Unicode case-insensitive compare two same-sized strings
 static int ucstrnicmp(const ushort *a, const ushort *b, int l)
 {
     return ucstricmp(a, a + l, b, b + l);
 }
 
+// Benchmarking indicates that doing memcmp is much slower than
+// executing the comparison ourselves.
+//
+// The profiling was done on a population of calls to qMemEquals, generated
+// during a run of the demo browser. The profile of the data (32-bit x86
+// Linux) was:
+//
+//  total number of comparisons: 21353
+//  longest string compared: 95
+//  average comparison length: 14.8786
+//  cache-line crosses: 5661 (13.3%)
+//  alignment histogram:
+//   0xXXX0 = 512 (1.2%) strings, 0 (0.0%) of which same-aligned
+//   0xXXX2 = 15087 (35.3%) strings, 5145 (34.1%) of which same-aligned
+//   0xXXX4 = 525 (1.2%) strings, 0 (0.0%) of which same-aligned
+//   0xXXX6 = 557 (1.3%) strings, 6 (1.1%) of which same-aligned
+//   0xXXX8 = 509 (1.2%) strings, 0 (0.0%) of which same-aligned
+//   0xXXXa = 24358 (57.0%) strings, 9901 (40.6%) of which same-aligned
+//   0xXXXc = 557 (1.3%) strings, 0 (0.0%) of which same-aligned
+//   0xXXXe = 601 (1.4%) strings, 15 (2.5%) of which same-aligned
+//   total  = 42706 (100%) strings, 15067 (35.3%) of which same-aligned
+//
+// 92% of the strings have alignment of 2 or 10, which is due to malloc on
+// 32-bit Linux returning values aligned to 8 bytes, and offsetof(array, QString::Data) == 18.
+//
+// The profile on 64-bit will be different since offsetof(array, QString::Data) == 26.
+//
+// The benchmark results were, for a Core-i7 @ 2.67 GHz 32-bit, compiled with -O3 -funroll-loops:
+//   16-bit loads only:           872,301 CPU ticks [Qt 4.5 / memcmp]
+//   32- and 16-bit loads:        773,362 CPU ticks [Qt 4.6]
+//   SSE2 "movdqu" 128-bit loads: 618,736 CPU ticks
+//   SSE3 "lddqu" 128-bit loads:  619,954 CPU ticks
+//   SSSE3 "palignr" corrections: 852,147 CPU ticks
+//   SSE4.2 "pcmpestrm":          738,702 CPU ticks
+//
+// The same benchmark on an Atom N450 @ 1.66 GHz, is:
+//  16-bit loads only:            2,185,882 CPU ticks
+//  32- and 16-bit loads:         1,805,060 CPU ticks
+//  SSE2 "movdqu" 128-bit loads:  2,529,843 CPU ticks
+//  SSE3 "lddqu" 128-bit loads:   2,514,858 CPU ticks
+//  SSSE3 "palignr" corrections:  2,160,325 CPU ticks
+//  SSE4.2 not available
+//
+// The conclusion we reach is that alignment the SSE2 unaligned code can gain
+// 20% improvement in performance in some systems, but suffers a penalty due
+// to the unaligned loads on others.
+
 static bool qMemEquals(const quint16 *a, const quint16 *b, int length)
 {
-    // Benchmarking indicates that doing memcmp is much slower than
-    // executing the comparison ourselves.
-    // To make it even faster, we do a 32-bit comparison, comparing
-    // twice the amount of data as a normal word-by-word comparison.
-    //
-    // Benchmarking results on a 2.33 GHz Core2 Duo, with a 64-QChar
-    // block of data, with 4194304 iterations (per iteration):
-    //    operation             usec            cpu ticks
-    //     memcmp                330               710
-    //     16-bit                 79             167-171
-    //  32-bit aligned            49             105-109
-    //
-    // Testing also indicates that unaligned 32-bit loads are as
-    // performant as 32-bit aligned.
     if (a == b || !length)
         return true;
 
diff --git a/src/corelib/tools/qstring.h b/src/corelib/tools/qstring.h
index 2252353..589fdc2 100644
--- a/src/corelib/tools/qstring.h
+++ b/src/corelib/tools/qstring.h
@@ -104,7 +104,10 @@ public:
     QString &operator=(QChar c);
     QString &operator=(const QString &);
     inline QString &operator=(const QLatin1String &);
-
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QString &operator=(QString &&other)
+    { qSwap(d, other.d); return *this; }
+#endif
     inline int size() const { return d->size; }
     inline int count() const { return d->size; }
     inline int length() const;
@@ -613,6 +616,7 @@ private:
         ushort asciiCache : 1;
         ushort capacity : 1;
         ushort reserved : 11;
+        // ### Qt5: try to ensure that "array" is aligned to 16 bytes on both 32- and 64-bit
         ushort array[1];
     };
     static Data shared_null;
diff --git a/src/corelib/tools/qvector.h b/src/corelib/tools/qvector.h
index 2f0ad42..7eee33d 100644
--- a/src/corelib/tools/qvector.h
+++ b/src/corelib/tools/qvector.h
@@ -53,6 +53,9 @@
 #endif
 #include <stdlib.h>
 #include <string.h>
+#ifdef Q_COMPILER_INITIALIZER_LISTS
+#include <initializer_list>
+#endif
 
 QT_BEGIN_HEADER
 
@@ -118,6 +121,13 @@ public:
     inline QVector(const QVector<T> &v) : d(v.d) { d->ref.ref(); if (!d->sharable) detach_helper(); }
     inline ~QVector() { if (!d) return; if (!d->ref.deref()) free(p); }
     QVector<T> &operator=(const QVector<T> &v);
+#ifdef Q_COMPILER_RVALUE_REFS
+    inline QVector<T> operator=(QVector<T> &&other)
+    { qSwap(p, other.p); return *this; }
+#endif
+#ifdef Q_COMPILER_INITIALIZER_LISTS
+    inline QVector(std::initializer_list<T> args);
+#endif
     bool operator==(const QVector<T> &v) const;
     inline bool operator!=(const QVector<T> &v) const { return !(*this == v); }
 
@@ -293,11 +303,10 @@ public:
 
 #ifndef QT_NO_STL
     static inline QVector<T> fromStdVector(const std::vector<T> &vector)
-    { QVector<T> tmp; qCopy(vector.begin(), vector.end(), std::back_inserter(tmp)); return tmp; }
+    { QVector<T> tmp; tmp.reserve(vector.size()); qCopy(vector.begin(), vector.end(), std::back_inserter(tmp)); return tmp; }
     inline std::vector<T> toStdVector() const
-    { std::vector<T> tmp; qCopy(constBegin(), constEnd(), std::back_inserter(tmp)); return tmp; }
+    { std::vector<T> tmp; tmp.reserve(size()); qCopy(constBegin(), constEnd(), std::back_inserter(tmp)); return tmp; }
 #endif
-
 private:
     friend class QRegion; // Optimization for QRegion::rects()
 
@@ -426,6 +435,22 @@ QVector<T>::QVector(int asize, const T &t)
         new (--i) T(t);
 }
 
+#ifdef Q_COMPILER_INITIALIZER_LISTS
+template <typename T>
+QVector<T>::QVector(std::initializer_list<T> args)
+{
+    p = malloc(args.size());
+    d->ref = 1;
+    d->alloc = d->size = args.size();
+    d->sharable = true;
+    d->capacity = false;
+    T* i = p->array + d->size;
+    auto it = args.end();
+    while (i != p->array)
+        new (--i) T(*(--it));
+}
+#endif
+
 template <typename T>
 void QVector<T>::free(Data *x)
 {