Merge remote-tracking branch 'origin/5.15' into dev

Change-Id: Ibee5acec72a1a1769d4bc5f23f56c7dc8d4cf3cb

24 files changed, 1434 insertions, 837 deletions

diff --git a/src/gui/doc/snippets/code/src_gui_image_qpixmapcache.cpp b/src/gui/doc/snippets/code/src_gui_image_qpixmapcache.cpp
index 3870237ac3..9043ee6361 100644
--- a/src/gui/doc/snippets/code/src_gui_image_qpixmapcache.cpp
+++ b/src/gui/doc/snippets/code/src_gui_image_qpixmapcache.cpp
@@ -49,9 +49,8 @@
 ****************************************************************************/
 
 //! [0]
-QPixmap* pp;
 QPixmap p;
-if ((pp=QPixmapCache::find("my_big_image", pm))) {
+if (QPixmap *pp = QPixmapCache::find("my_big_image"))) {
     p = *pp;
 } else {
     p.load("bigimage.png");
diff --git a/src/gui/image/qimage.cpp b/src/gui/image/qimage.cpp
index ce0c6170c1..189f12fd5c 100644
--- a/src/gui/image/qimage.cpp
+++ b/src/gui/image/qimage.cpp
@@ -68,6 +68,11 @@
 #include <private/qimage_p.h>
 #include <private/qfont_p.h>
 
+#if QT_CONFIG(thread)
+#include "qsemaphore.h"
+#include "qthreadpool.h"
+#endif
+
 QT_BEGIN_NAMESPACE
 
 static inline bool isLocked(QImageData *data)
@@ -4861,18 +4866,43 @@ void QImage::applyColorTransform(const QColorTransform &transform)
         Q_UNREACHABLE();
     }
 
+    std::function<void(int,int)> transformSegment;
+
     if (depth() > 32) {
-        for (int i = 0; i < height(); ++i) {
-            QRgba64 *scanline = reinterpret_cast<QRgba64 *>(scanLine(i));
-            transform.d->apply(scanline, scanline, width(), flags);
-        }
+        transformSegment = [&](int yStart, int yEnd) {
+            for (int y = yStart; y < yEnd; ++y) {
+                QRgba64 *scanline = reinterpret_cast<QRgba64 *>(scanLine(y));
+                transform.d->apply(scanline, scanline, width(), flags);
+            }
+        };
     } else {
-        for (int i = 0; i < height(); ++i) {
-            QRgb *scanline = reinterpret_cast<QRgb *>(scanLine(i));
-            transform.d->apply(scanline, scanline, width(), flags);
-        }
+        transformSegment = [&](int yStart, int yEnd) {
+            for (int y = yStart; y < yEnd; ++y) {
+                QRgb *scanline = reinterpret_cast<QRgb *>(scanLine(y));
+                transform.d->apply(scanline, scanline, width(), flags);
+            }
+        };
     }
 
+#if QT_CONFIG(thread)
+    int segments = sizeInBytes() / (1<<16);
+    segments = std::min(segments, height());
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (height() - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                transformSegment(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+    } else
+#endif
+        transformSegment(0, height());
+
     if (oldFormat != format())
         *this = std::move(*this).convertToFormat(oldFormat);
 }
diff --git a/src/gui/image/qimage_conversions.cpp b/src/gui/image/qimage_conversions.cpp
index 27088698ec..4f570d8684 100644
--- a/src/gui/image/qimage_conversions.cpp
+++ b/src/gui/image/qimage_conversions.cpp
@@ -43,7 +43,12 @@
 #include <private/qendian_p.h>
 #include <private/qsimd_p.h>
 #include <private/qimage_p.h>
+
 #include <qendian.h>
+#if QT_CONFIG(thread)
+#include <qsemaphore.h>
+#include <qthreadpool.h>
+#endif
 
 QT_BEGIN_NAMESPACE
 
@@ -159,12 +164,8 @@ void convert_generic(QImageData *dest, const QImageData *src, Qt::ImageConversio
     // Cannot be used with indexed formats.
     Q_ASSERT(dest->format > QImage::Format_Indexed8);
     Q_ASSERT(src->format > QImage::Format_Indexed8);
-    uint buf[BufferSize];
-    uint *buffer = buf;
     const QPixelLayout *srcLayout = &qPixelLayouts[src->format];
     const QPixelLayout *destLayout = &qPixelLayouts[dest->format];
-    const uchar *srcData = src->data;
-    uchar *destData = dest->data;
 
     FetchAndConvertPixelsFunc fetch = srcLayout->fetchToARGB32PM;
     ConvertAndStorePixelsFunc store = destLayout->storeFromARGB32PM;
@@ -197,59 +198,110 @@ void convert_generic(QImageData *dest, const QImageData *src, Qt::ImageConversio
         else
             store = destLayout->storeFromRGB32;
     }
-    QDitherInfo dither;
-    QDitherInfo *ditherPtr = nullptr;
-    if ((flags & Qt::PreferDither) && (flags & Qt::Dither_Mask) != Qt::ThresholdDither)
-        ditherPtr = &dither;
-
-    for (int y = 0; y < src->height; ++y) {
-        dither.y = y;
-        int x = 0;
-        while (x < src->width) {
-            dither.x = x;
-            int l = src->width - x;
-            if (destLayout->bpp == QPixelLayout::BPP32)
-                buffer = reinterpret_cast<uint *>(destData) + x;
-            else
-                l = qMin(l, BufferSize);
-            const uint *ptr = fetch(buffer, srcData, x, l, nullptr, ditherPtr);
-            store(destData, ptr, x, l, nullptr, ditherPtr);
-            x += l;
+
+    auto convertSegment = [=](int yStart, int yEnd) {
+        uint buf[BufferSize];
+        uint *buffer = buf;
+        const uchar *srcData = src->data + src->bytes_per_line * yStart;
+        uchar *destData = dest->data + dest->bytes_per_line * yStart;
+        QDitherInfo dither;
+        QDitherInfo *ditherPtr = nullptr;
+        if ((flags & Qt::PreferDither) && (flags & Qt::Dither_Mask) != Qt::ThresholdDither)
+            ditherPtr = &dither;
+        for (int y = yStart; y < yEnd; ++y) {
+            dither.y = y;
+            int x = 0;
+            while (x < src->width) {
+                dither.x = x;
+                int l = src->width - x;
+                if (destLayout->bpp == QPixelLayout::BPP32)
+                    buffer = reinterpret_cast<uint *>(destData) + x;
+                else
+                    l = qMin(l, BufferSize);
+                const uint *ptr = fetch(buffer, srcData, x, l, 0, ditherPtr);
+                store(destData, ptr, x, l, 0, ditherPtr);
+                x += l;
+            }
+            srcData += src->bytes_per_line;
+            destData += dest->bytes_per_line;
         }
-        srcData += src->bytes_per_line;
-        destData += dest->bytes_per_line;
-    }
+    };
+
+#if QT_CONFIG(thread)
+    int segments = src->nbytes / (1<<16);
+    segments = std::min(segments, src->height);
+
+    if (segments <= 1)
+        return convertSegment(0, src->height);
+
+    QSemaphore semaphore;
+    int y = 0;
+    for (int i = 0; i < segments; ++i) {
+        int yn = (src->height - y) / (segments - i);
+        QThreadPool::globalInstance()->start([&, y, yn]() {
+            convertSegment(y, y + yn);
+            semaphore.release(1);
+        });
+        y += yn;
+    }
+    semaphore.acquire(segments);
+#else
+    convertSegment(0, src->height);
+#endif
 }
 
 void convert_generic_to_rgb64(QImageData *dest, const QImageData *src, Qt::ImageConversionFlags)
 {
     Q_ASSERT(dest->format > QImage::Format_Indexed8);
     Q_ASSERT(src->format > QImage::Format_Indexed8);
-    QRgba64 buf[BufferSize];
-    QRgba64 *buffer = buf;
     const QPixelLayout *srcLayout = &qPixelLayouts[src->format];
     const QPixelLayout *destLayout = &qPixelLayouts[dest->format];
-    const uchar *srcData = src->data;
-    uchar *destData = dest->data;
 
     const FetchAndConvertPixelsFunc64 fetch = srcLayout->fetchToRGBA64PM;
     const ConvertAndStorePixelsFunc64 store = qStoreFromRGBA64PM[dest->format];
 
-    for (int y = 0; y < src->height; ++y) {
-        int x = 0;
-        while (x < src->width) {
-            int l = src->width - x;
-            if (destLayout->bpp == QPixelLayout::BPP64)
-                buffer = reinterpret_cast<QRgba64 *>(destData) + x;
-            else
-                l = qMin(l, BufferSize);
-            const QRgba64 *ptr = fetch(buffer, srcData, x, l, nullptr, nullptr);
-            store(destData, ptr, x, l, nullptr, nullptr);
-            x += l;
+    auto convertSegment = [=](int yStart, int yEnd) {
+        QRgba64 buf[BufferSize];
+        QRgba64 *buffer = buf;
+        const uchar *srcData = src->data + yStart * src->bytes_per_line;
+        uchar *destData = dest->data + yStart * dest->bytes_per_line;
+        for (int y = yStart; y < yEnd; ++y) {
+            int x = 0;
+            while (x < src->width) {
+                int l = src->width - x;
+                if (destLayout->bpp == QPixelLayout::BPP64)
+                    buffer = reinterpret_cast<QRgba64 *>(destData) + x;
+                else
+                    l = qMin(l, BufferSize);
+                const QRgba64 *ptr = fetch(buffer, srcData, x, l, nullptr, nullptr);
+                store(destData, ptr, x, l, nullptr, nullptr);
+                x += l;
+            }
+            srcData += src->bytes_per_line;
+            destData += dest->bytes_per_line;
         }
-        srcData += src->bytes_per_line;
-        destData += dest->bytes_per_line;
-    }
+    };
+#if QT_CONFIG(thread)
+    int segments = src->nbytes / (1<<16);
+    segments = std::min(segments, src->height);
+
+    if (segments <= 1)
+        return convertSegment(0, src->height);
+
+    QSemaphore semaphore;
+    int y = 0;
+    for (int i = 0; i < segments; ++i) {
+        int yn = (src->height - y) / (segments - i);
+        QThreadPool::globalInstance()->start([&, y, yn]() {
+            convertSegment(y, y + yn);
+            semaphore.release(1);
+        });
+        y += yn;
+    }
+    semaphore.acquire(segments);
+#else
+    convertSegment(0, src->height);
+#endif
 }
 
 bool convert_generic_inplace(QImageData *data, QImage::Format dst_format, Qt::ImageConversionFlags flags)
@@ -270,11 +322,6 @@ bool convert_generic_inplace(QImageData *data, QImage::Format dst_format, Qt::Im
             && qt_highColorPrecision(dst_format, !srcLayout->hasAlphaChannel))
         return false;
 
-    uint buf[BufferSize];
-    uint *buffer = buf;
-    uchar *srcData = data->data;
-    uchar *destData = data->data;
-
     QImageData::ImageSizeParameters params = { data->bytes_per_line, data->nbytes };
     if (data->depth != destDepth) {
         params = QImageData::calculateImageParameters(data->width, data->height, destDepth);
@@ -313,28 +360,52 @@ bool convert_generic_inplace(QImageData *data, QImage::Format dst_format, Qt::Im
         else
             store = destLayout->storeFromRGB32;
     }
-    QDitherInfo dither;
-    QDitherInfo *ditherPtr = nullptr;
-    if ((flags & Qt::PreferDither) && (flags & Qt::Dither_Mask) != Qt::ThresholdDither)
-        ditherPtr = &dither;
-
-    for (int y = 0; y < data->height; ++y) {
-        dither.y = y;
-        int x = 0;
-        while (x < data->width) {
-            dither.x = x;
-            int l = data->width - x;
-            if (srcLayout->bpp == QPixelLayout::BPP32)
-                buffer = reinterpret_cast<uint *>(srcData) + x;
-            else
-                l = qMin(l, BufferSize);
-            const uint *ptr = fetch(buffer, srcData, x, l, nullptr, ditherPtr);
-            store(destData, ptr, x, l, nullptr, ditherPtr);
-            x += l;
+
+    auto convertSegment = [=](int yStart, int yEnd) {
+        uint buf[BufferSize];
+        uint *buffer = buf;
+        uchar *srcData = data->data + data->bytes_per_line * yStart;
+        uchar *destData = srcData;
+        QDitherInfo dither;
+        QDitherInfo *ditherPtr = nullptr;
+        if ((flags & Qt::PreferDither) && (flags & Qt::Dither_Mask) != Qt::ThresholdDither)
+            ditherPtr = &dither;
+        for (int y = yStart; y < yEnd; ++y) {
+            dither.y = y;
+            int x = 0;
+            while (x < data->width) {
+                dither.x = x;
+                int l = data->width - x;
+                if (srcLayout->bpp == QPixelLayout::BPP32)
+                    buffer = reinterpret_cast<uint *>(srcData) + x;
+                else
+                    l = qMin(l, BufferSize);
+                const uint *ptr = fetch(buffer, srcData, x, l, nullptr, ditherPtr);
+                store(destData, ptr, x, l, nullptr, ditherPtr);
+                x += l;
+            }
+            srcData += data->bytes_per_line;
+            destData += params.bytesPerLine;
         }
-        srcData += data->bytes_per_line;
-        destData += params.bytesPerLine;
-    }
+    };
+#if QT_CONFIG(thread)
+    int segments = data->nbytes / (1<<16);
+    segments = std::min(segments, data->height);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (data->height - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                convertSegment(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+    } else
+#endif
+        convertSegment(0, data->height);
     if (params.totalSize != data->nbytes) {
         Q_ASSERT(params.totalSize < data->nbytes);
         void *newData = realloc(data->data, params.totalSize);
diff --git a/src/gui/painting/qimagescale.cpp b/src/gui/painting/qimagescale.cpp
index 2e2f65b483..ecb0230e71 100644
--- a/src/gui/painting/qimagescale.cpp
+++ b/src/gui/painting/qimagescale.cpp
@@ -43,6 +43,11 @@
 #include "qcolor.h"
 #include "qrgba64_p.h"
 
+#if QT_CONFIG(thread)
+#include "qsemaphore.h"
+#include "qthreadpool.h"
+#endif
+
 QT_BEGIN_NAMESPACE
 
 /*
@@ -239,6 +244,8 @@ static QImageScaleInfo* QImageScale::qimageCalcScaleInfo(const QImage &img,
     isi = new QImageScaleInfo;
     if (!isi)
         return nullptr;
+    isi->sh = sh;
+    isi->sw = sw;
 
     isi->xup_yup = (qAbs(dw) >= sw) + ((qAbs(dh) >= sh) << 1);
 
@@ -303,33 +310,54 @@ static void qt_qimageScaleAARGBA_up_xy(QImageScaleInfo *isi, unsigned int *dest,
     int *yapoints = isi->yapoints;
 
     /* go through every scanline in the output buffer */
-    for (int y = 0; y < dh; y++) {
-        /* calculate the source line we'll scan from */
-        const unsigned int *sptr = ypoints[y];
-        unsigned int *dptr = dest + (y * dow);
-        const int yap = yapoints[y];
-        if (yap > 0) {
-            for (int x = 0; x < dw; x++) {
-                const unsigned int *pix = sptr + xpoints[x];
-                const int xap = xapoints[x];
-                if (xap > 0)
-                    *dptr = interpolate_4_pixels(pix, pix + sow, xap, yap);
-                else
-                    *dptr = INTERPOLATE_PIXEL_256(pix[0], 256 - yap, pix[sow], yap);
-                dptr++;
-            }
-        } else {
-            for (int x = 0; x < dw; x++) {
-                const unsigned int *pix = sptr + xpoints[x];
-                const int xap = xapoints[x];
-                if (xap > 0)
-                    *dptr = INTERPOLATE_PIXEL_256(pix[0], 256 - xap, pix[1], xap);
-                else
-                    *dptr = pix[0];
-                dptr++;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            /* calculate the source line we'll scan from */
+            const unsigned int *sptr = ypoints[y];
+            unsigned int *dptr = dest + (y * dow);
+            const int yap = yapoints[y];
+            if (yap > 0) {
+                for (int x = 0; x < dw; x++) {
+                    const unsigned int *pix = sptr + xpoints[x];
+                    const int xap = xapoints[x];
+                    if (xap > 0)
+                        *dptr = interpolate_4_pixels(pix, pix + sow, xap, yap);
+                    else
+                        *dptr = INTERPOLATE_PIXEL_256(pix[0], 256 - yap, pix[sow], yap);
+                    dptr++;
+                }
+            } else {
+                for (int x = 0; x < dw; x++) {
+                    const unsigned int *pix = sptr + xpoints[x];
+                    const int xap = xapoints[x];
+                    if (xap > 0)
+                        *dptr = INTERPOLATE_PIXEL_256(pix[0], 256 - xap, pix[1], xap);
+                    else
+                        *dptr = pix[0];
+                    dptr++;
+                }
             }
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 /* scale by area sampling - with alpha */
@@ -411,33 +439,54 @@ static void qt_qimageScaleAARGBA_up_x_down_y(QImageScaleInfo *isi, unsigned int
     int *yapoints = isi->yapoints;
 
     /* go through every scanline in the output buffer */
-    for (int y = 0; y < dh; y++) {
-        int Cy = yapoints[y] >> 16;
-        int yap = yapoints[y] & 0xffff;
-
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            int r, g, b, a;
-            qt_qimageScaleAARGBA_helper(sptr, yap, Cy, sow, r, g, b, a);
-
-            int xap = xapoints[x];
-            if (xap > 0) {
-                int rr, gg, bb, aa;
-                qt_qimageScaleAARGBA_helper(sptr + 1, yap, Cy, sow, rr, gg, bb, aa);
-
-                r = r * (256 - xap);
-                g = g * (256 - xap);
-                b = b * (256 - xap);
-                a = a * (256 - xap);
-                r = (r + (rr * xap)) >> 8;
-                g = (g + (gg * xap)) >> 8;
-                b = (b + (bb * xap)) >> 8;
-                a = (a + (aa * xap)) >> 8;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            int Cy = yapoints[y] >> 16;
+            int yap = yapoints[y] & 0xffff;
+
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                int r, g, b, a;
+                qt_qimageScaleAARGBA_helper(sptr, yap, Cy, sow, r, g, b, a);
+
+                int xap = xapoints[x];
+                if (xap > 0) {
+                    int rr, gg, bb, aa;
+                    qt_qimageScaleAARGBA_helper(sptr + 1, yap, Cy, sow, rr, gg, bb, aa);
+
+                    r = r * (256 - xap);
+                    g = g * (256 - xap);
+                    b = b * (256 - xap);
+                    a = a * (256 - xap);
+                    r = (r + (rr * xap)) >> 8;
+                    g = (g + (gg * xap)) >> 8;
+                    b = (b + (bb * xap)) >> 8;
+                    a = (a + (aa * xap)) >> 8;
+                }
+                *dptr++ = qRgba(r >> 14, g >> 14, b >> 14, a >> 14);
             }
-            *dptr++ = qRgba(r >> 14, g >> 14, b >> 14, a >> 14);
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 static void qt_qimageScaleAARGBA_down_x_up_y(QImageScaleInfo *isi, unsigned int *dest,
@@ -449,34 +498,55 @@ static void qt_qimageScaleAARGBA_down_x_up_y(QImageScaleInfo *isi, unsigned int
     int *yapoints = isi->yapoints;
 
     /* go through every scanline in the output buffer */
-    for (int y = 0; y < dh; y++) {
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            int Cx = xapoints[x] >> 16;
-            int xap = xapoints[x] & 0xffff;
-
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            int r, g, b, a;
-            qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, r, g, b, a);
-
-            int yap = yapoints[y];
-            if (yap > 0) {
-                int rr, gg, bb, aa;
-                qt_qimageScaleAARGBA_helper(sptr + sow, xap, Cx, 1, rr, gg, bb, aa);
-
-                r = r * (256 - yap);
-                g = g * (256 - yap);
-                b = b * (256 - yap);
-                a = a * (256 - yap);
-                r = (r + (rr * yap)) >> 8;
-                g = (g + (gg * yap)) >> 8;
-                b = (b + (bb * yap)) >> 8;
-                a = (a + (aa * yap)) >> 8;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                int Cx = xapoints[x] >> 16;
+                int xap = xapoints[x] & 0xffff;
+
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                int r, g, b, a;
+                qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, r, g, b, a);
+
+                int yap = yapoints[y];
+                if (yap > 0) {
+                    int rr, gg, bb, aa;
+                    qt_qimageScaleAARGBA_helper(sptr + sow, xap, Cx, 1, rr, gg, bb, aa);
+
+                    r = r * (256 - yap);
+                    g = g * (256 - yap);
+                    b = b * (256 - yap);
+                    a = a * (256 - yap);
+                    r = (r + (rr * yap)) >> 8;
+                    g = (g + (gg * yap)) >> 8;
+                    b = (b + (bb * yap)) >> 8;
+                    a = (a + (aa * yap)) >> 8;
+                }
+                *dptr = qRgba(r >> 14, g >> 14, b >> 14, a >> 14);
+                dptr++;
             }
-            *dptr = qRgba(r >> 14, g >> 14, b >> 14, a >> 14);
-            dptr++;
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 static void qt_qimageScaleAARGBA_down_xy(QImageScaleInfo *isi, unsigned int *dest,
@@ -487,45 +557,66 @@ static void qt_qimageScaleAARGBA_down_xy(QImageScaleInfo *isi, unsigned int *des
     int *xapoints = isi->xapoints;
     int *yapoints = isi->yapoints;
 
-    for (int y = 0; y < dh; y++) {
-        int Cy = (yapoints[y]) >> 16;
-        int yap = (yapoints[y]) & 0xffff;
-
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            int Cx = xapoints[x] >> 16;
-            int xap = xapoints[x] & 0xffff;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            int Cy = (yapoints[y]) >> 16;
+            int yap = (yapoints[y]) & 0xffff;
 
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            int rx, gx, bx, ax;
-            qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                int Cx = xapoints[x] >> 16;
+                int xap = xapoints[x] & 0xffff;
 
-            int r = ((rx>>4) * yap);
-            int g = ((gx>>4) * yap);
-            int b = ((bx>>4) * yap);
-            int a = ((ax>>4) * yap);
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                int rx, gx, bx, ax;
+                qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
 
-            int j;
-            for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+                int r = ((rx>>4) * yap);
+                int g = ((gx>>4) * yap);
+                int b = ((bx>>4) * yap);
+                int a = ((ax>>4) * yap);
+
+                int j;
+                for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+                    sptr += sow;
+                    qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
+                    r += ((rx>>4) * Cy);
+                    g += ((gx>>4) * Cy);
+                    b += ((bx>>4) * Cy);
+                    a += ((ax>>4) * Cy);
+                }
                 sptr += sow;
                 qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
-                r += ((rx>>4) * Cy);
-                g += ((gx>>4) * Cy);
-                b += ((bx>>4) * Cy);
-                a += ((ax>>4) * Cy);
-            }
-            sptr += sow;
-            qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
 
-            r += ((rx>>4) * j);
-            g += ((gx>>4) * j);
-            b += ((bx>>4) * j);
-            a += ((ax>>4) * j);
+                r += ((rx>>4) * j);
+                g += ((gx>>4) * j);
+                b += ((bx>>4) * j);
+                a += ((ax>>4) * j);
 
-            *dptr = qRgba(r >> 24, g >> 24, b >> 24, a >> 24);
-            dptr++;
+                *dptr = qRgba(r >> 24, g >> 24, b >> 24, a >> 24);
+                dptr++;
+            }
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 #if QT_CONFIG(raster_64bit)
@@ -546,32 +637,53 @@ static void qt_qimageScaleRgba64_up_xy(QImageScaleInfo *isi, QRgba64 *dest,
     int *xapoints = isi->xapoints;
     int *yapoints = isi->yapoints;
 
-    for (int y = 0; y < dh; y++) {
-        const QRgba64 *sptr = ypoints[y];
-        QRgba64 *dptr = dest + (y * dow);
-        const int yap = yapoints[y];
-        if (yap > 0) {
-            for (int x = 0; x < dw; x++) {
-                const QRgba64 *pix = sptr + xpoints[x];
-                const int xap = xapoints[x];
-                if (xap > 0)
-                    *dptr = interpolate_4_pixels_rgb64(pix, pix + sow, xap * 256, yap * 256);
-                else
-                    *dptr = interpolate256(pix[0], 256 - yap, pix[sow], yap);
-                dptr++;
-            }
-        } else {
-            for (int x = 0; x < dw; x++) {
-                const QRgba64 *pix = sptr + xpoints[x];
-                const int xap = xapoints[x];
-                if (xap > 0)
-                    *dptr = interpolate256(pix[0], 256 - xap, pix[1], xap);
-                else
-                    *dptr = pix[0];
-                dptr++;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            const QRgba64 *sptr = ypoints[y];
+            QRgba64 *dptr = dest + (y * dow);
+            const int yap = yapoints[y];
+            if (yap > 0) {
+                for (int x = 0; x < dw; x++) {
+                    const QRgba64 *pix = sptr + xpoints[x];
+                    const int xap = xapoints[x];
+                    if (xap > 0)
+                        *dptr = interpolate_4_pixels_rgb64(pix, pix + sow, xap * 256, yap * 256);
+                    else
+                        *dptr = interpolate256(pix[0], 256 - yap, pix[sow], yap);
+                    dptr++;
+                }
+            } else {
+                for (int x = 0; x < dw; x++) {
+                    const QRgba64 *pix = sptr + xpoints[x];
+                    const int xap = xapoints[x];
+                    if (xap > 0)
+                        *dptr = interpolate256(pix[0], 256 - xap, pix[1], xap);
+                    else
+                        *dptr = pix[0];
+                    dptr++;
+                }
             }
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 void qt_qimageScaleRgba64(QImageScaleInfo *isi, QRgba64 *dest,
@@ -616,33 +728,54 @@ static void qt_qimageScaleRgba64_up_x_down_y(QImageScaleInfo *isi, QRgba64 *dest
     int *xapoints = isi->xapoints;
     int *yapoints = isi->yapoints;
 
-    for (int y = 0; y < dh; y++) {
-        int Cy = (yapoints[y]) >> 16;
-        int yap = (yapoints[y]) & 0xffff;
-
-        QRgba64 *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            const QRgba64 *sptr = ypoints[y] + xpoints[x];
-            qint64 r, g, b, a;
-            qt_qimageScaleRgba64_helper(sptr, yap, Cy, sow, r, g, b, a);
-
-            int xap = xapoints[x];
-            if (xap > 0) {
-                qint64 rr, gg, bb, aa;
-                qt_qimageScaleRgba64_helper(sptr + 1, yap, Cy, sow, rr, gg, bb, aa);
-
-                r = r * (256 - xap);
-                g = g * (256 - xap);
-                b = b * (256 - xap);
-                a = a * (256 - xap);
-                r = (r + (rr * xap)) >> 8;
-                g = (g + (gg * xap)) >> 8;
-                b = (b + (bb * xap)) >> 8;
-                a = (a + (aa * xap)) >> 8;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            int Cy = (yapoints[y]) >> 16;
+            int yap = (yapoints[y]) & 0xffff;
+
+            QRgba64 *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                const QRgba64 *sptr = ypoints[y] + xpoints[x];
+                qint64 r, g, b, a;
+                qt_qimageScaleRgba64_helper(sptr, yap, Cy, sow, r, g, b, a);
+
+                int xap = xapoints[x];
+                if (xap > 0) {
+                    qint64 rr, gg, bb, aa;
+                    qt_qimageScaleRgba64_helper(sptr + 1, yap, Cy, sow, rr, gg, bb, aa);
+
+                    r = r * (256 - xap);
+                    g = g * (256 - xap);
+                    b = b * (256 - xap);
+                    a = a * (256 - xap);
+                    r = (r + (rr * xap)) >> 8;
+                    g = (g + (gg * xap)) >> 8;
+                    b = (b + (bb * xap)) >> 8;
+                    a = (a + (aa * xap)) >> 8;
+                }
+                *dptr++ = qRgba64(r >> 14, g >> 14, b >> 14, a >> 14);
             }
-            *dptr++ = qRgba64(r >> 14, g >> 14, b >> 14, a >> 14);
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 static void qt_qimageScaleRgba64_down_x_up_y(QImageScaleInfo *isi, QRgba64 *dest,
@@ -653,34 +786,55 @@ static void qt_qimageScaleRgba64_down_x_up_y(QImageScaleInfo *isi, QRgba64 *dest
     int *xapoints = isi->xapoints;
     int *yapoints = isi->yapoints;
 
-    for (int y = 0; y < dh; y++) {
-        QRgba64 *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            int Cx = xapoints[x] >> 16;
-            int xap = xapoints[x] & 0xffff;
-
-            const QRgba64 *sptr = ypoints[y] + xpoints[x];
-            qint64 r, g, b, a;
-            qt_qimageScaleRgba64_helper(sptr, xap, Cx, 1, r, g, b, a);
-
-            int yap = yapoints[y];
-            if (yap > 0) {
-                qint64 rr, gg, bb, aa;
-                qt_qimageScaleRgba64_helper(sptr + sow, xap, Cx, 1, rr, gg, bb, aa);
-
-                r = r * (256 - yap);
-                g = g * (256 - yap);
-                b = b * (256 - yap);
-                a = a * (256 - yap);
-                r = (r + (rr * yap)) >> 8;
-                g = (g + (gg * yap)) >> 8;
-                b = (b + (bb * yap)) >> 8;
-                a = (a + (aa * yap)) >> 8;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            QRgba64 *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                int Cx = xapoints[x] >> 16;
+                int xap = xapoints[x] & 0xffff;
+
+                const QRgba64 *sptr = ypoints[y] + xpoints[x];
+                qint64 r, g, b, a;
+                qt_qimageScaleRgba64_helper(sptr, xap, Cx, 1, r, g, b, a);
+
+                int yap = yapoints[y];
+                if (yap > 0) {
+                    qint64 rr, gg, bb, aa;
+                    qt_qimageScaleRgba64_helper(sptr + sow, xap, Cx, 1, rr, gg, bb, aa);
+
+                    r = r * (256 - yap);
+                    g = g * (256 - yap);
+                    b = b * (256 - yap);
+                    a = a * (256 - yap);
+                    r = (r + (rr * yap)) >> 8;
+                    g = (g + (gg * yap)) >> 8;
+                    b = (b + (bb * yap)) >> 8;
+                    a = (a + (aa * yap)) >> 8;
+                }
+                *dptr = qRgba64(r >> 14, g >> 14, b >> 14, a >> 14);
+                dptr++;
             }
-            *dptr = qRgba64(r >> 14, g >> 14, b >> 14, a >> 14);
-            dptr++;
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 static void qt_qimageScaleRgba64_down_xy(QImageScaleInfo *isi, QRgba64 *dest,
@@ -691,43 +845,64 @@ static void qt_qimageScaleRgba64_down_xy(QImageScaleInfo *isi, QRgba64 *dest,
     int *xapoints = isi->xapoints;
     int *yapoints = isi->yapoints;
 
-    for (int y = 0; y < dh; y++) {
-        int Cy = (yapoints[y]) >> 16;
-        int yap = (yapoints[y]) & 0xffff;
-
-        QRgba64 *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            int Cx = xapoints[x] >> 16;
-            int xap = xapoints[x] & 0xffff;
-
-            const QRgba64 *sptr = ypoints[y] + xpoints[x];
-            qint64 rx, gx, bx, ax;
-            qt_qimageScaleRgba64_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
-
-            qint64 r = rx * yap;
-            qint64 g = gx * yap;
-            qint64 b = bx * yap;
-            qint64 a = ax * yap;
-            int j;
-            for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            int Cy = (yapoints[y]) >> 16;
+            int yap = (yapoints[y]) & 0xffff;
+
+            QRgba64 *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                int Cx = xapoints[x] >> 16;
+                int xap = xapoints[x] & 0xffff;
+
+                const QRgba64 *sptr = ypoints[y] + xpoints[x];
+                qint64 rx, gx, bx, ax;
+                qt_qimageScaleRgba64_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
+
+                qint64 r = rx * yap;
+                qint64 g = gx * yap;
+                qint64 b = bx * yap;
+                qint64 a = ax * yap;
+                int j;
+                for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+                    sptr += sow;
+                    qt_qimageScaleRgba64_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
+                    r += rx * Cy;
+                    g += gx * Cy;
+                    b += bx * Cy;
+                    a += ax * Cy;
+                }
                 sptr += sow;
                 qt_qimageScaleRgba64_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
-                r += rx * Cy;
-                g += gx * Cy;
-                b += bx * Cy;
-                a += ax * Cy;
+                r += rx * j;
+                g += gx * j;
+                b += bx * j;
+                a += ax * j;
+
+                *dptr = qRgba64(r >> 28, g >> 28, b >> 28, a >> 28);
+                dptr++;
             }
-            sptr += sow;
-            qt_qimageScaleRgba64_helper(sptr, xap, Cx, 1, rx, gx, bx, ax);
-            r += rx * j;
-            g += gx * j;
-            b += bx * j;
-            a += ax * j;
-
-            *dptr = qRgba64(r >> 28, g >> 28, b >> 28, a >> 28);
-            dptr++;
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 #endif
 
@@ -817,31 +992,52 @@ static void qt_qimageScaleAARGB_up_x_down_y(QImageScaleInfo *isi, unsigned int *
     int *yapoints = isi->yapoints;
 
     /* go through every scanline in the output buffer */
-    for (int y = 0; y < dh; y++) {
-        int Cy = yapoints[y] >> 16;
-        int yap = yapoints[y] & 0xffff;
-
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            int r, g, b;
-            qt_qimageScaleAARGB_helper(sptr, yap, Cy, sow, r, g, b);
-
-            int xap = xapoints[x];
-            if (xap > 0) {
-                int rr, bb, gg;
-                qt_qimageScaleAARGB_helper(sptr + 1, yap, Cy, sow, rr, gg, bb);
-
-                r = r * (256 - xap);
-                g = g * (256 - xap);
-                b = b * (256 - xap);
-                r = (r + (rr * xap)) >> 8;
-                g = (g + (gg * xap)) >> 8;
-                b = (b + (bb * xap)) >> 8;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            int Cy = yapoints[y] >> 16;
+            int yap = yapoints[y] & 0xffff;
+
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                int r, g, b;
+                qt_qimageScaleAARGB_helper(sptr, yap, Cy, sow, r, g, b);
+
+                int xap = xapoints[x];
+                if (xap > 0) {
+                    int rr, bb, gg;
+                    qt_qimageScaleAARGB_helper(sptr + 1, yap, Cy, sow, rr, gg, bb);
+
+                    r = r * (256 - xap);
+                    g = g * (256 - xap);
+                    b = b * (256 - xap);
+                    r = (r + (rr * xap)) >> 8;
+                    g = (g + (gg * xap)) >> 8;
+                    b = (b + (bb * xap)) >> 8;
+                }
+                *dptr++ = qRgb(r >> 14, g >> 14, b >> 14);
             }
-            *dptr++ = qRgb(r >> 14, g >> 14, b >> 14);
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 static void qt_qimageScaleAARGB_down_x_up_y(QImageScaleInfo *isi, unsigned int *dest,
@@ -853,31 +1049,52 @@ static void qt_qimageScaleAARGB_down_x_up_y(QImageScaleInfo *isi, unsigned int *
     int *yapoints = isi->yapoints;
 
     /* go through every scanline in the output buffer */
-    for (int y = 0; y < dh; y++) {
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            int Cx = xapoints[x] >> 16;
-            int xap = xapoints[x] & 0xffff;
-
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            int r, g, b;
-            qt_qimageScaleAARGB_helper(sptr, xap, Cx, 1, r, g, b);
-
-            int yap = yapoints[y];
-            if (yap > 0) {
-                int rr, bb, gg;
-                qt_qimageScaleAARGB_helper(sptr + sow, xap, Cx, 1, rr, gg, bb);
-
-                r = r * (256 - yap);
-                g = g * (256 - yap);
-                b = b * (256 - yap);
-                r = (r + (rr * yap)) >> 8;
-                g = (g + (gg * yap)) >> 8;
-                b = (b + (bb * yap)) >> 8;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                int Cx = xapoints[x] >> 16;
+                int xap = xapoints[x] & 0xffff;
+
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                int r, g, b;
+                qt_qimageScaleAARGB_helper(sptr, xap, Cx, 1, r, g, b);
+
+                int yap = yapoints[y];
+                if (yap > 0) {
+                    int rr, bb, gg;
+                    qt_qimageScaleAARGB_helper(sptr + sow, xap, Cx, 1, rr, gg, bb);
+
+                    r = r * (256 - yap);
+                    g = g * (256 - yap);
+                    b = b * (256 - yap);
+                    r = (r + (rr * yap)) >> 8;
+                    g = (g + (gg * yap)) >> 8;
+                    b = (b + (bb * yap)) >> 8;
+                }
+                *dptr++ = qRgb(r >> 14, g >> 14, b >> 14);
             }
-            *dptr++ = qRgb(r >> 14, g >> 14, b >> 14);
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 static void qt_qimageScaleAARGB_down_xy(QImageScaleInfo *isi, unsigned int *dest,
@@ -888,43 +1105,64 @@ static void qt_qimageScaleAARGB_down_xy(QImageScaleInfo *isi, unsigned int *dest
     int *xapoints = isi->xapoints;
     int *yapoints = isi->yapoints;
 
-    for (int y = 0; y < dh; y++) {
-        int Cy = yapoints[y] >> 16;
-        int yap = yapoints[y] & 0xffff;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            int Cy = yapoints[y] >> 16;
+            int yap = yapoints[y] & 0xffff;
 
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            int Cx = xapoints[x] >> 16;
-            int xap = xapoints[x] & 0xffff;
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                int Cx = xapoints[x] >> 16;
+                int xap = xapoints[x] & 0xffff;
+
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                int rx, gx, bx;
+                qt_qimageScaleAARGB_helper(sptr, xap, Cx, 1, rx, gx, bx);
 
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            int rx, gx, bx;
-            qt_qimageScaleAARGB_helper(sptr, xap, Cx, 1, rx, gx, bx);
+                int r = (rx >> 4) * yap;
+                int g = (gx >> 4) * yap;
+                int b = (bx >> 4) * yap;
 
-            int r = (rx >> 4) * yap;
-            int g = (gx >> 4) * yap;
-            int b = (bx >> 4) * yap;
+                int j;
+                for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+                    sptr += sow;
+                    qt_qimageScaleAARGB_helper(sptr, xap, Cx, 1, rx, gx, bx);
 
-            int j;
-            for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+                    r += (rx >> 4) * Cy;
+                    g += (gx >> 4) * Cy;
+                    b += (bx >> 4) * Cy;
+                }
                 sptr += sow;
                 qt_qimageScaleAARGB_helper(sptr, xap, Cx, 1, rx, gx, bx);
 
-                r += (rx >> 4) * Cy;
-                g += (gx >> 4) * Cy;
-                b += (bx >> 4) * Cy;
-            }
-            sptr += sow;
-            qt_qimageScaleAARGB_helper(sptr, xap, Cx, 1, rx, gx, bx);
-
-            r += (rx >> 4) * j;
-            g += (gx >> 4) * j;
-            b += (bx >> 4) * j;
+                r += (rx >> 4) * j;
+                g += (gx >> 4) * j;
+                b += (bx >> 4) * j;
 
-            *dptr = qRgb(r >> 24, g >> 24, b >> 24);
-            dptr++;
+                *dptr = qRgb(r >> 24, g >> 24, b >> 24);
+                dptr++;
+            }
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 QImage qSmoothScaleImage(const QImage &src, int dw, int dh)
diff --git a/src/gui/painting/qimagescale_neon.cpp b/src/gui/painting/qimagescale_neon.cpp
index 4ae113b002..416155e139 100644
--- a/src/gui/painting/qimagescale_neon.cpp
+++ b/src/gui/painting/qimagescale_neon.cpp
@@ -41,6 +41,11 @@
 #include "qimage.h"
 #include <private/qsimd_p.h>
 
+#if QT_CONFIG(thread)
+#include "qsemaphore.h"
+#include "qthreadpool.h"
+#endif
+
 #if defined(__ARM_NEON__)
 
 QT_BEGIN_NAMESPACE
@@ -76,33 +81,54 @@ void qt_qimageScaleAARGBA_up_x_down_y_neon(QImageScaleInfo *isi, unsigned int *d
     int *yapoints = isi->yapoints;
 
     /* go through every scanline in the output buffer */
-    for (int y = 0; y < dh; y++) {
-        int Cy = yapoints[y] >> 16;
-        int yap = yapoints[y] & 0xffff;
-
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            uint32x4_t vx = qt_qimageScaleAARGBA_helper(sptr, yap, Cy, sow);
-
-            int xap = xapoints[x];
-            if (xap > 0) {
-                uint32x4_t vr = qt_qimageScaleAARGBA_helper(sptr + 1, yap, Cy, sow);
-
-                vx = vmulq_n_u32(vx, 256 - xap);
-                vr = vmulq_n_u32(vr, xap);
-                vx = vaddq_u32(vx, vr);
-                vx = vshrq_n_u32(vx, 8);
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            int Cy = yapoints[y] >> 16;
+            int yap = yapoints[y] & 0xffff;
+
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                uint32x4_t vx = qt_qimageScaleAARGBA_helper(sptr, yap, Cy, sow);
+
+                int xap = xapoints[x];
+                if (xap > 0) {
+                    uint32x4_t vr = qt_qimageScaleAARGBA_helper(sptr + 1, yap, Cy, sow);
+
+                    vx = vmulq_n_u32(vx, 256 - xap);
+                    vr = vmulq_n_u32(vr, xap);
+                    vx = vaddq_u32(vx, vr);
+                    vx = vshrq_n_u32(vx, 8);
+                }
+                vx = vshrq_n_u32(vx, 14);
+                const uint16x4_t vx16 = vmovn_u32(vx);
+                const uint8x8_t vx8 = vmovn_u16(vcombine_u16(vx16, vx16));
+                *dptr = vget_lane_u32(vreinterpret_u32_u8(vx8), 0);
+                if (RGB)
+                    *dptr |= 0xff000000;
+                dptr++;
             }
-            vx = vshrq_n_u32(vx, 14);
-            const uint16x4_t vx16 = vmovn_u32(vx);
-            const uint8x8_t vx8 = vmovn_u16(vcombine_u16(vx16, vx16));
-            *dptr = vget_lane_u32(vreinterpret_u32_u8(vx8), 0);
-            if (RGB)
-                *dptr |= 0xff000000;
-            dptr++;
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 template<bool RGB>
@@ -115,33 +141,54 @@ void qt_qimageScaleAARGBA_down_x_up_y_neon(QImageScaleInfo *isi, unsigned int *d
     int *yapoints = isi->yapoints;
 
     /* go through every scanline in the output buffer */
-    for (int y = 0; y < dh; y++) {
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            int Cx = xapoints[x] >> 16;
-            int xap = xapoints[x] & 0xffff;
-
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            uint32x4_t vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1);
-
-            int yap = yapoints[y];
-            if (yap > 0) {
-                uint32x4_t vr = qt_qimageScaleAARGBA_helper(sptr + sow, xap, Cx, 1);
-
-                vx = vmulq_n_u32(vx, 256 - yap);
-                vr = vmulq_n_u32(vr, yap);
-                vx = vaddq_u32(vx, vr);
-                vx = vshrq_n_u32(vx, 8);
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                int Cx = xapoints[x] >> 16;
+                int xap = xapoints[x] & 0xffff;
+
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                uint32x4_t vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1);
+
+                int yap = yapoints[y];
+                if (yap > 0) {
+                    uint32x4_t vr = qt_qimageScaleAARGBA_helper(sptr + sow, xap, Cx, 1);
+
+                    vx = vmulq_n_u32(vx, 256 - yap);
+                    vr = vmulq_n_u32(vr, yap);
+                    vx = vaddq_u32(vx, vr);
+                    vx = vshrq_n_u32(vx, 8);
+                }
+                vx = vshrq_n_u32(vx, 14);
+                const uint16x4_t vx16 = vmovn_u32(vx);
+                const uint8x8_t vx8 = vmovn_u16(vcombine_u16(vx16, vx16));
+                *dptr = vget_lane_u32(vreinterpret_u32_u8(vx8), 0);
+                if (RGB)
+                    *dptr |= 0xff000000;
+                dptr++;
             }
-            vx = vshrq_n_u32(vx, 14);
-            const uint16x4_t vx16 = vmovn_u32(vx);
-            const uint8x8_t vx8 = vmovn_u16(vcombine_u16(vx16, vx16));
-            *dptr = vget_lane_u32(vreinterpret_u32_u8(vx8), 0);
-            if (RGB)
-                *dptr |= 0xff000000;
-            dptr++;
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 template<bool RGB>
@@ -153,43 +200,64 @@ void qt_qimageScaleAARGBA_down_xy_neon(QImageScaleInfo *isi, unsigned int *dest,
     int *xapoints = isi->xapoints;
     int *yapoints = isi->yapoints;
 
-    for (int y = 0; y < dh; y++) {
-        int Cy = yapoints[y] >> 16;
-        int yap = yapoints[y] & 0xffff;
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            int Cy = yapoints[y] >> 16;
+            int yap = yapoints[y] & 0xffff;
 
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            const int Cx = xapoints[x] >> 16;
-            const int xap = xapoints[x] & 0xffff;
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                const int Cx = xapoints[x] >> 16;
+                const int xap = xapoints[x] & 0xffff;
 
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            uint32x4_t vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1);
-            vx = vshrq_n_u32(vx, 4);
-            uint32x4_t vr = vmulq_n_u32(vx, yap);
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                uint32x4_t vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1);
+                vx = vshrq_n_u32(vx, 4);
+                uint32x4_t vr = vmulq_n_u32(vx, yap);
 
-            int j;
-            for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+                int j;
+                for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+                    sptr += sow;
+                    vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1);
+                    vx = vshrq_n_u32(vx, 4);
+                    vx = vmulq_n_u32(vx, Cy);
+                    vr = vaddq_u32(vr, vx);
+                }
                 sptr += sow;
                 vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1);
                 vx = vshrq_n_u32(vx, 4);
-                vx = vmulq_n_u32(vx, Cy);
+                vx = vmulq_n_u32(vx, j);
                 vr = vaddq_u32(vr, vx);
+
+                vx = vshrq_n_u32(vr, 24);
+                const uint16x4_t vx16 = vmovn_u32(vx);
+                const uint8x8_t vx8 = vmovn_u16(vcombine_u16(vx16, vx16));
+                *dptr = vget_lane_u32(vreinterpret_u32_u8(vx8), 0);
+                if (RGB)
+                    *dptr |= 0xff000000;
+                dptr++;
             }
-            sptr += sow;
-            vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1);
-            vx = vshrq_n_u32(vx, 4);
-            vx = vmulq_n_u32(vx, j);
-            vr = vaddq_u32(vr, vx);
-
-            vx = vshrq_n_u32(vr, 24);
-            const uint16x4_t vx16 = vmovn_u32(vx);
-            const uint8x8_t vx8 = vmovn_u16(vcombine_u16(vx16, vx16));
-            *dptr = vget_lane_u32(vreinterpret_u32_u8(vx8), 0);
-            if (RGB)
-                *dptr |= 0xff000000;
-            dptr++;
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 template void qt_qimageScaleAARGBA_up_x_down_y_neon<false>(QImageScaleInfo *isi, unsigned int *dest,
diff --git a/src/gui/painting/qimagescale_p.h b/src/gui/painting/qimagescale_p.h
index 244d681718..a9a4c0f858 100644
--- a/src/gui/painting/qimagescale_p.h
+++ b/src/gui/painting/qimagescale_p.h
@@ -66,6 +66,8 @@ namespace QImageScale {
         int *xapoints{nullptr};
         int *yapoints{nullptr};
         int xup_yup{0};
+        int sh = 0;
+        int sw = 0;
     };
 }
 
diff --git a/src/gui/painting/qimagescale_sse4.cpp b/src/gui/painting/qimagescale_sse4.cpp
index 5861a2e2ff..902ae61ed2 100644
--- a/src/gui/painting/qimagescale_sse4.cpp
+++ b/src/gui/painting/qimagescale_sse4.cpp
@@ -42,6 +42,11 @@
 #include <private/qdrawhelper_x86_p.h>
 #include <private/qsimd_p.h>
 
+#if QT_CONFIG(thread)
+#include "qsemaphore.h"
+#include "qthreadpool.h"
+#endif
+
 #if defined(QT_COMPILER_SUPPORTS_SSE4_1)
 
 QT_BEGIN_NAMESPACE
@@ -70,44 +75,65 @@ void qt_qimageScaleAARGBA_up_x_down_y_sse4(QImageScaleInfo *isi, unsigned int *d
                                            int dw, int dh, int dow, int sow)
 {
     const unsigned int **ypoints = isi->ypoints;
-    int *xpoints = isi->xpoints;
-    int *xapoints = isi->xapoints;
-    int *yapoints = isi->yapoints;
+    const int *xpoints = isi->xpoints;
+    const int *xapoints = isi->xapoints;
+    const int *yapoints = isi->yapoints;
 
     const __m128i v256 = _mm_set1_epi32(256);
 
     /* go through every scanline in the output buffer */
-    for (int y = 0; y < dh; y++) {
-        int Cy = yapoints[y] >> 16;
-        int yap = yapoints[y] & 0xffff;
-        const __m128i vCy = _mm_set1_epi32(Cy);
-        const __m128i vyap = _mm_set1_epi32(yap);
-
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            __m128i vx = qt_qimageScaleAARGBA_helper(sptr, yap, Cy, sow, vyap, vCy);
-
-            int xap = xapoints[x];
-            if (xap > 0) {
-                const __m128i vxap = _mm_set1_epi32(xap);
-                const __m128i vinvxap = _mm_sub_epi32(v256, vxap);
-                __m128i vr = qt_qimageScaleAARGBA_helper(sptr + 1, yap, Cy, sow, vyap, vCy);
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            const int Cy = yapoints[y] >> 16;
+            const int yap = yapoints[y] & 0xffff;
+            const __m128i vCy = _mm_set1_epi32(Cy);
+            const __m128i vyap = _mm_set1_epi32(yap);
 
-                vx = _mm_mullo_epi32(vx, vinvxap);
-                vr = _mm_mullo_epi32(vr, vxap);
-                vx = _mm_add_epi32(vx, vr);
-                vx = _mm_srli_epi32(vx, 8);
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                __m128i vx = qt_qimageScaleAARGBA_helper(sptr, yap, Cy, sow, vyap, vCy);
+
+                const int xap = xapoints[x];
+                if (xap > 0) {
+                    const __m128i vxap = _mm_set1_epi32(xap);
+                    const __m128i vinvxap = _mm_sub_epi32(v256, vxap);
+                    __m128i vr = qt_qimageScaleAARGBA_helper(sptr + 1, yap, Cy, sow, vyap, vCy);
+
+                    vx = _mm_mullo_epi32(vx, vinvxap);
+                    vr = _mm_mullo_epi32(vr, vxap);
+                    vx = _mm_add_epi32(vx, vr);
+                    vx = _mm_srli_epi32(vx, 8);
+                }
+                vx = _mm_srli_epi32(vx, 14);
+                vx = _mm_packus_epi32(vx, vx);
+                vx = _mm_packus_epi16(vx, vx);
+                *dptr = _mm_cvtsi128_si32(vx);
+                if (RGB)
+                    *dptr |= 0xff000000;
+                dptr++;
             }
-            vx = _mm_srli_epi32(vx, 14);
-            vx = _mm_packus_epi32(vx, _mm_setzero_si128());
-            vx = _mm_packus_epi16(vx, _mm_setzero_si128());
-            *dptr = _mm_cvtsi128_si32(vx);
-            if (RGB)
-                *dptr |= 0xff000000;
-            dptr++;
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 template<bool RGB>
@@ -122,37 +148,58 @@ void qt_qimageScaleAARGBA_down_x_up_y_sse4(QImageScaleInfo *isi, unsigned int *d
     const __m128i v256 = _mm_set1_epi32(256);
 
     /* go through every scanline in the output buffer */
-    for (int y = 0; y < dh; y++) {
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            int Cx = xapoints[x] >> 16;
-            int xap = xapoints[x] & 0xffff;
-            const __m128i vCx = _mm_set1_epi32(Cx);
-            const __m128i vxap = _mm_set1_epi32(xap);
-
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            __m128i vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx);
-
-            int yap = yapoints[y];
-            if (yap > 0) {
-                const __m128i vyap = _mm_set1_epi32(yap);
-                const __m128i vinvyap = _mm_sub_epi32(v256, vyap);
-                __m128i vr = qt_qimageScaleAARGBA_helper(sptr + sow, xap, Cx, 1, vxap, vCx);
-
-                vx = _mm_mullo_epi32(vx, vinvyap);
-                vr = _mm_mullo_epi32(vr, vyap);
-                vx = _mm_add_epi32(vx, vr);
-                vx = _mm_srli_epi32(vx, 8);
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                int Cx = xapoints[x] >> 16;
+                int xap = xapoints[x] & 0xffff;
+                const __m128i vCx = _mm_set1_epi32(Cx);
+                const __m128i vxap = _mm_set1_epi32(xap);
+
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                __m128i vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx);
+
+                int yap = yapoints[y];
+                if (yap > 0) {
+                    const __m128i vyap = _mm_set1_epi32(yap);
+                    const __m128i vinvyap = _mm_sub_epi32(v256, vyap);
+                    __m128i vr = qt_qimageScaleAARGBA_helper(sptr + sow, xap, Cx, 1, vxap, vCx);
+
+                    vx = _mm_mullo_epi32(vx, vinvyap);
+                    vr = _mm_mullo_epi32(vr, vyap);
+                    vx = _mm_add_epi32(vx, vr);
+                    vx = _mm_srli_epi32(vx, 8);
+                }
+                vx = _mm_srli_epi32(vx, 14);
+                vx = _mm_packus_epi32(vx, vx);
+                vx = _mm_packus_epi16(vx, vx);
+                *dptr = _mm_cvtsi128_si32(vx);
+                if (RGB)
+                    *dptr |= 0xff000000;
+                dptr++;
             }
-            vx = _mm_srli_epi32(vx, 14);
-            vx = _mm_packus_epi32(vx, _mm_setzero_si128());
-            vx = _mm_packus_epi16(vx, _mm_setzero_si128());
-            *dptr = _mm_cvtsi128_si32(vx);
-            if (RGB)
-                *dptr |= 0xff000000;
-            dptr++;
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 template<bool RGB>
@@ -164,42 +211,63 @@ void qt_qimageScaleAARGBA_down_xy_sse4(QImageScaleInfo *isi, unsigned int *dest,
     int *xapoints = isi->xapoints;
     int *yapoints = isi->yapoints;
 
-    for (int y = 0; y < dh; y++) {
-        int Cy = yapoints[y] >> 16;
-        int yap = yapoints[y] & 0xffff;
-        const __m128i vCy = _mm_set1_epi32(Cy);
-        const __m128i vyap = _mm_set1_epi32(yap);
-
-        unsigned int *dptr = dest + (y * dow);
-        for (int x = 0; x < dw; x++) {
-            const int Cx = xapoints[x] >> 16;
-            const int xap = xapoints[x] & 0xffff;
-            const __m128i vCx = _mm_set1_epi32(Cx);
-            const __m128i vxap = _mm_set1_epi32(xap);
-
-            const unsigned int *sptr = ypoints[y] + xpoints[x];
-            __m128i vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx);
-            __m128i vr = _mm_mullo_epi32(_mm_srli_epi32(vx, 4), vyap);
-
-            int j;
-            for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+    auto scaleSection = [&] (int yStart, int yEnd) {
+        for (int y = yStart; y < yEnd; ++y) {
+            int Cy = yapoints[y] >> 16;
+            int yap = yapoints[y] & 0xffff;
+            const __m128i vCy = _mm_set1_epi32(Cy);
+            const __m128i vyap = _mm_set1_epi32(yap);
+
+            unsigned int *dptr = dest + (y * dow);
+            for (int x = 0; x < dw; x++) {
+                const int Cx = xapoints[x] >> 16;
+                const int xap = xapoints[x] & 0xffff;
+                const __m128i vCx = _mm_set1_epi32(Cx);
+                const __m128i vxap = _mm_set1_epi32(xap);
+
+                const unsigned int *sptr = ypoints[y] + xpoints[x];
+                __m128i vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx);
+                __m128i vr = _mm_mullo_epi32(_mm_srli_epi32(vx, 4), vyap);
+
+                int j;
+                for (j = (1 << 14) - yap; j > Cy; j -= Cy) {
+                    sptr += sow;
+                    vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx);
+                    vr = _mm_add_epi32(vr, _mm_mullo_epi32(_mm_srli_epi32(vx, 4), vCy));
+                }
                 sptr += sow;
                 vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx);
-                vr = _mm_add_epi32(vr, _mm_mullo_epi32(_mm_srli_epi32(vx, 4), vCy));
+                vr = _mm_add_epi32(vr, _mm_mullo_epi32(_mm_srli_epi32(vx, 4), _mm_set1_epi32(j)));
+
+                vr = _mm_srli_epi32(vr, 24);
+                vr = _mm_packus_epi32(vr, _mm_setzero_si128());
+                vr = _mm_packus_epi16(vr, _mm_setzero_si128());
+                *dptr = _mm_cvtsi128_si32(vr);
+                if (RGB)
+                    *dptr |= 0xff000000;
+                dptr++;
             }
-            sptr += sow;
-            vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx);
-            vr = _mm_add_epi32(vr, _mm_mullo_epi32(_mm_srli_epi32(vx, 4), _mm_set1_epi32(j)));
-
-            vr = _mm_srli_epi32(vr, 24);
-            vr = _mm_packus_epi32(vr, _mm_setzero_si128());
-            vr = _mm_packus_epi16(vr, _mm_setzero_si128());
-            *dptr = _mm_cvtsi128_si32(vr);
-            if (RGB)
-                *dptr |= 0xff000000;
-            dptr++;
         }
+    };
+#if QT_CONFIG(thread)
+    int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16);
+    segments = std::min(segments, dh);
+    if (segments > 1) {
+        QSemaphore semaphore;
+        int y = 0;
+        for (int i = 0; i < segments; ++i) {
+            int yn = (dh - y) / (segments - i);
+            QThreadPool::globalInstance()->start([&, y, yn]() {
+                scaleSection(y, y + yn);
+                semaphore.release(1);
+            });
+            y += yn;
+        }
+        semaphore.acquire(segments);
+        return;
     }
+#endif
+    scaleSection(0, dh);
 }
 
 template void qt_qimageScaleAARGBA_up_x_down_y_sse4<false>(QImageScaleInfo *isi, unsigned int *dest,
diff --git a/src/gui/painting/qpainterpath.cpp b/src/gui/painting/qpainterpath.cpp
index 11623c78f0..e3cdc218b2 100644
--- a/src/gui/painting/qpainterpath.cpp
+++ b/src/gui/painting/qpainterpath.cpp
@@ -1603,25 +1603,6 @@ QPainterPath QPainterPath::toReversed() const
 }
 
 /*!
-    \overload
-
-    Converts the path into a list of polygons without any transformation,
-    and returns the list.
-
-    This function creates one polygon for each subpath regardless of
-    intersecting subpaths (i.e. overlapping bounding rectangles). To
-    make sure that such overlapping subpaths are filled correctly, use
-    the toFillPolygons() function instead.
-
-    \sa toFillPolygons(), toFillPolygon(), {QPainterPath#QPainterPath
-    Conversion}{QPainterPath Conversion}
-*/
-QList<QPolygonF> QPainterPath::toSubpathPolygons() const
-{
-    return toSubpathPolygons(QTransform());
-}
-
-/*!
     Converts the path into a list of polygons using the QTransform
     \a matrix, and returns the list.
 
@@ -1679,35 +1660,6 @@ QList<QPolygonF> QPainterPath::toSubpathPolygons(const QTransform &matrix) const
 }
 
 /*!
-    \overload
-
-    Converts the path into a list of polygons without any transformation,
-    and returns the list.
-
-    The function differs from the toFillPolygon() function in that it
-    creates several polygons. It is provided because it is usually
-    faster to draw several small polygons than to draw one large
-    polygon, even though the total number of points drawn is the same.
-
-    The toFillPolygons() function differs from the toSubpathPolygons()
-    function in that it create only polygon for subpaths that have
-    overlapping bounding rectangles.
-
-    Like the toFillPolygon() function, this function uses a rewinding
-    technique to make sure that overlapping subpaths can be filled
-    using the correct fill rule. Note that rewinding inserts addition
-    lines in the polygons so the outline of the fill polygon does not
-    match the outline of the path.
-
-    \sa toSubpathPolygons(), toFillPolygon(),
-    {QPainterPath#QPainterPath Conversion}{QPainterPath Conversion}
-*/
-QList<QPolygonF> QPainterPath::toFillPolygons() const
-{
-    return toFillPolygons(QTransform());
-}
-
-/*!
     Converts the path into a list of polygons using the
     QTransform \a matrix, and returns the list.
 
@@ -2903,28 +2855,6 @@ void QPainterPathStroker::setDashOffset(qreal offset)
 }
 
 /*!
-  \overload
-
-  Converts the path into a polygon without any transformation,
-  and returns the polygon.
-
-  The polygon is created by first converting all subpaths to
-  polygons, then using a rewinding technique to make sure that
-  overlapping subpaths can be filled using the correct fill rule.
-
-  Note that rewinding inserts addition lines in the polygon so
-  the outline of the fill polygon does not match the outline of
-  the path.
-
-  \sa toSubpathPolygons(), toFillPolygons(),
-  {QPainterPath#QPainterPath Conversion}{QPainterPath Conversion}
-*/
-QPolygonF QPainterPath::toFillPolygon() const
-{
-    return toFillPolygon(QTransform());
-}
-
-/*!
   Converts the path into a polygon using the QTransform
   \a matrix, and returns the polygon.
 
diff --git a/src/gui/painting/qpainterpath.h b/src/gui/painting/qpainterpath.h
index 078b665222..aa43c74bee 100644
--- a/src/gui/painting/qpainterpath.h
+++ b/src/gui/painting/qpainterpath.h
@@ -41,6 +41,7 @@
 #define QPAINTERPATH_H
 
 #include <QtGui/qtguiglobal.h>
+#include <QtGui/qtransform.h>
 #include <QtCore/qglobal.h>
 #include <QtCore/qrect.h>
 #include <QtCore/qline.h>
@@ -176,12 +177,9 @@ public:
 
     Q_REQUIRED_RESULT QPainterPath toReversed() const;
 
-    QList<QPolygonF> toSubpathPolygons() const;
-    QList<QPolygonF> toSubpathPolygons(const QTransform &matrix) const;
-    QList<QPolygonF> toFillPolygons() const;
-    QList<QPolygonF> toFillPolygons(const QTransform &matrix) const;
-    QPolygonF toFillPolygon() const;
-    QPolygonF toFillPolygon(const QTransform &matrix) const;
+    QList<QPolygonF> toSubpathPolygons(const QTransform &matrix = QTransform()) const;
+    QList<QPolygonF> toFillPolygons(const QTransform &matrix = QTransform()) const;
+    QPolygonF toFillPolygon(const QTransform &matrix = QTransform()) const;
 
     int elementCount() const;
     QPainterPath::Element elementAt(int i) const;
@@ -356,6 +354,8 @@ inline void QPainterPath::translate(const QPointF &offset)
 inline QPainterPath QPainterPath::translated(const QPointF &offset) const
 { return translated(offset.x(), offset.y()); }
 
+inline QPainterPath operator *(const QPainterPath &p, const QTransform &m)
+{ return m.map(p); }
 
 #ifndef QT_NO_DEBUG_STREAM
 Q_GUI_EXPORT QDebug operator<<(QDebug, const QPainterPath &);
diff --git a/src/gui/painting/qrasterizer.cpp b/src/gui/painting/qrasterizer.cpp
index cd31d75f83..f3c193d799 100644
--- a/src/gui/painting/qrasterizer.cpp
+++ b/src/gui/painting/qrasterizer.cpp
@@ -46,6 +46,8 @@
 #include <private/qdatabuffer_p.h>
 #include <private/qdrawhelper_p.h>
 
+#include <QtGui/qpainterpath.h>
+
 #include <algorithm>
 
 QT_BEGIN_NAMESPACE
diff --git a/src/gui/painting/qtextureglyphcache.cpp b/src/gui/painting/qtextureglyphcache.cpp
index e906397520..3728060caf 100644
--- a/src/gui/painting/qtextureglyphcache.cpp
+++ b/src/gui/painting/qtextureglyphcache.cpp
@@ -43,6 +43,8 @@
 #include "private/qfontengine_p.h"
 #include "private/qnumeric_p.h"
 
+#include <QtGui/qpainterpath.h>
+
 QT_BEGIN_NAMESPACE
 
 // #define CACHE_DEBUG
diff --git a/src/gui/painting/qtransform.h b/src/gui/painting/qtransform.h
index 351f2b2e4e..d235f15029 100644
--- a/src/gui/painting/qtransform.h
+++ b/src/gui/painting/qtransform.h
@@ -40,7 +40,6 @@
 #define QTRANSFORM_H
 
 #include <QtGui/qtguiglobal.h>
-#include <QtGui/qpainterpath.h>
 #include <QtGui/qpolygon.h>
 #include <QtGui/qregion.h>
 #include <QtGui/qwindowdefs.h>
@@ -50,8 +49,8 @@
 
 QT_BEGIN_NAMESPACE
 
-
 class QVariant;
+class QPainterPath;
 
 class Q_GUI_EXPORT QTransform
 {
@@ -406,8 +405,6 @@ inline QPolygonF operator *(const QPolygonF &a, const QTransform &m)
 { return m.map(a); }
 inline QRegion operator *(const QRegion &r, const QTransform &m)
 { return m.map(r); }
-inline QPainterPath operator *(const QPainterPath &p, const QTransform &m)
-{ return m.map(p); }
 
 inline QTransform operator *(const QTransform &a, qreal n)
 { QTransform t(a); t *= n; return t; }
diff --git a/src/gui/rhi/qrhi.cpp b/src/gui/rhi/qrhi.cpp
index c805e23ad0..83c1e8eaa2 100644
--- a/src/gui/rhi/qrhi.cpp
+++ b/src/gui/rhi/qrhi.cpp
@@ -2885,16 +2885,57 @@ QRhiShaderResourceBinding QRhiShaderResourceBinding::uniformBufferWithDynamicOff
     \return a shader resource binding for the given binding number, pipeline
     stages, texture, and sampler specified by \a binding, \a stage, \a tex,
     \a sampler.
+
+    \note This function is equivalent to calling sampledTextures() with a
+    \c count of 1.
+
+    \sa sampledTextures()
  */
 QRhiShaderResourceBinding QRhiShaderResourceBinding::sampledTexture(
         int binding, StageFlags stage, QRhiTexture *tex, QRhiSampler *sampler)
 {
+    const TextureAndSampler texSampler = { tex, sampler };
+    return sampledTextures(binding, stage, 1, &texSampler);
+}
+
+/*!
+    \return a shader resource binding for the given binding number, pipeline
+    stages, and the array of texture-sampler pairs specified by \a binding, \a
+    stage, \a count, and \a texSamplers.
+
+    \note \a count must be at least 1, and not larger than 16.
+
+    \note When \a count is 1, this function is equivalent to sampledTexture().
+
+    This function is relevant when arrays of combined image samplers are
+    involved. For example, in GLSL \c{layout(binding = 5) uniform sampler2D
+    shadowMaps[8];} declares an array of combined image samplers. The
+    application is then expected provide a QRhiShaderResourceBinding for
+    binding point 5, set up by calling this function with \a count set to 8 and
+    a valid texture and sampler for each element of the array.
+
+    \warning All elements of the array must be specified. With the above
+    example, the only valid, portable approach is calling this function with a
+    \a count of 8. Additionally, all QRhiTexture and QRhiSampler instances must
+    be valid, meaning nullptr is not an accepted value. This is due to some of
+    the underlying APIs, such as, Vulkan, that require a valid image and
+    sampler object for each element in descriptor arrays. Applications are
+    advised to provide "dummy" samplers and textures if some array elements are
+    not relevant (due to not being accessed in the shader).
+
+    \sa sampledTexture()
+ */
+QRhiShaderResourceBinding QRhiShaderResourceBinding::sampledTextures(
+        int binding, StageFlags stage, int count, const TextureAndSampler *texSamplers)
+{
+    Q_ASSERT(count >= 1 && count <= Data::MAX_TEX_SAMPLER_ARRAY_SIZE);
     QRhiShaderResourceBinding b;
     b.d.binding = binding;
     b.d.stage = stage;
     b.d.type = SampledTexture;
-    b.d.u.stex.tex = tex;
-    b.d.u.stex.sampler = sampler;
+    b.d.u.stex.count = count;
+    for (int i = 0; i < count; ++i)
+        b.d.u.stex.texSamplers[i] = texSamplers[i];
     return b;
 }
 
@@ -3084,10 +3125,14 @@ bool operator==(const QRhiShaderResourceBinding &a, const QRhiShaderResourceBind
         }
         break;
     case QRhiShaderResourceBinding::SampledTexture:
-        if (da->u.stex.tex != db->u.stex.tex
-                || da->u.stex.sampler != db->u.stex.sampler)
-        {
+        if (da->u.stex.count != db->u.stex.count)
             return false;
+        for (int i = 0; i < da->u.stex.count; ++i) {
+            if (da->u.stex.texSamplers[i].tex != db->u.stex.texSamplers[i].tex
+                    || da->u.stex.texSamplers[i].sampler != db->u.stex.texSamplers[i].sampler)
+            {
+                return false;
+            }
         }
         break;
     case QRhiShaderResourceBinding::ImageLoad:
@@ -3162,10 +3207,13 @@ QDebug operator<<(QDebug dbg, const QRhiShaderResourceBinding &b)
                       << ')';
         break;
     case QRhiShaderResourceBinding::SampledTexture:
-        dbg.nospace() << " SampledTexture("
-                      << "texture=" << d->u.stex.tex
-                      << " sampler=" << d->u.stex.sampler
-                      << ')';
+        dbg.nospace() << " SampledTextures("
+                      << "count=" << d->u.stex.count;
+        for (int i = 0; i < d->u.stex.count; ++i) {
+            dbg.nospace() << " texture=" << d->u.stex.texSamplers[i].tex
+                          << " sampler=" << d->u.stex.texSamplers[i].sampler;
+        }
+        dbg.nospace() << ')';
         break;
     case QRhiShaderResourceBinding::ImageLoad:
         dbg.nospace() << " ImageLoad("
diff --git a/src/gui/rhi/qrhi_p.h b/src/gui/rhi/qrhi_p.h
index 17c911a5ff..9d906d7bbd 100644
--- a/src/gui/rhi/qrhi_p.h
+++ b/src/gui/rhi/qrhi_p.h
@@ -348,6 +348,12 @@ public:
 
     static QRhiShaderResourceBinding sampledTexture(int binding, StageFlags stage, QRhiTexture *tex, QRhiSampler *sampler);
 
+    struct TextureAndSampler {
+        QRhiTexture *tex;
+        QRhiSampler *sampler;
+    };
+    static QRhiShaderResourceBinding sampledTextures(int binding, StageFlags stage, int count, const TextureAndSampler *texSamplers);
+
     static QRhiShaderResourceBinding imageLoad(int binding, StageFlags stage, QRhiTexture *tex, int level);
     static QRhiShaderResourceBinding imageStore(int binding, StageFlags stage, QRhiTexture *tex, int level);
     static QRhiShaderResourceBinding imageLoadStore(int binding, StageFlags stage, QRhiTexture *tex, int level);
@@ -370,9 +376,10 @@ public:
             int maybeSize;
             bool hasDynamicOffset;
         };
+        static const int MAX_TEX_SAMPLER_ARRAY_SIZE = 16;
         struct SampledTextureData {
-            QRhiTexture *tex;
-            QRhiSampler *sampler;
+            int count;
+            TextureAndSampler texSamplers[MAX_TEX_SAMPLER_ARRAY_SIZE];
         };
         struct StorageImageData {
             QRhiTexture *tex;
diff --git a/src/gui/rhi/qrhid3d11.cpp b/src/gui/rhi/qrhid3d11.cpp
index 7b583e6fd2..ed202958f3 100644
--- a/src/gui/rhi/qrhid3d11.cpp
+++ b/src/gui/rhi/qrhid3d11.cpp
@@ -113,6 +113,10 @@ QT_BEGIN_NAMESPACE
 #define DXGI_ADAPTER_FLAG_SOFTWARE 2
 #endif
 
+#ifndef D3D11_1_UAV_SLOT_COUNT
+#define D3D11_1_UAV_SLOT_COUNT 64
+#endif
+
 QRhiD3D11::QRhiD3D11(QRhiD3D11InitParams *params, QRhiD3D11NativeHandles *importDevice)
     : ofr(this),
       deviceCurse(this)
@@ -627,18 +631,25 @@ void QRhiD3D11::setShaderResources(QRhiCommandBuffer *cb, QRhiShaderResourceBind
             break;
         case QRhiShaderResourceBinding::SampledTexture:
         {
-            QD3D11Texture *texD = QRHI_RES(QD3D11Texture, b->u.stex.tex);
-            QD3D11Sampler *samplerD = QRHI_RES(QD3D11Sampler, b->u.stex.sampler);
-            if (texD->generation != bd.stex.texGeneration
-                    || texD->m_id != bd.stex.texId
-                    || samplerD->generation != bd.stex.samplerGeneration
-                    || samplerD->m_id != bd.stex.samplerId)
-            {
+            const QRhiShaderResourceBinding::Data::SampledTextureData *data = &b->u.stex;
+            if (bd.stex.count != data->count) {
+                bd.stex.count = data->count;
                 srbUpdate = true;
-                bd.stex.texId = texD->m_id;
-                bd.stex.texGeneration = texD->generation;
-                bd.stex.samplerId = samplerD->m_id;
-                bd.stex.samplerGeneration = samplerD->generation;
+            }
+            for (int elem = 0; elem < data->count; ++elem) {
+                QD3D11Texture *texD = QRHI_RES(QD3D11Texture, data->texSamplers[elem].tex);
+                QD3D11Sampler *samplerD = QRHI_RES(QD3D11Sampler, data->texSamplers[elem].sampler);
+                if (texD->generation != bd.stex.d[elem].texGeneration
+                        || texD->m_id != bd.stex.d[elem].texId
+                        || samplerD->generation != bd.stex.d[elem].samplerGeneration
+                        || samplerD->m_id != bd.stex.d[elem].samplerId)
+                {
+                    srbUpdate = true;
+                    bd.stex.d[elem].texId = texD->m_id;
+                    bd.stex.d[elem].texGeneration = texD->generation;
+                    bd.stex.d[elem].samplerId = samplerD->m_id;
+                    bd.stex.d[elem].samplerGeneration = samplerD->generation;
+                }
             }
         }
             break;
@@ -1894,31 +1905,38 @@ void QRhiD3D11::updateShaderResourceBindings(QD3D11ShaderResourceBindings *srbD,
             break;
         case QRhiShaderResourceBinding::SampledTexture:
         {
-            QD3D11Texture *texD = QRHI_RES(QD3D11Texture, b->u.stex.tex);
-            QD3D11Sampler *samplerD = QRHI_RES(QD3D11Sampler, b->u.stex.sampler);
-            bd.stex.texId = texD->m_id;
-            bd.stex.texGeneration = texD->generation;
-            bd.stex.samplerId = samplerD->m_id;
-            bd.stex.samplerGeneration = samplerD->generation;
-            if (b->stage.testFlag(QRhiShaderResourceBinding::VertexStage)) {
-                QPair<int, int> nativeBinding = mapBinding(b->binding, RBM_VERTEX, nativeResourceBindingMaps);
-                if (nativeBinding.first >= 0 && nativeBinding.second >= 0) {
-                    res[RBM_VERTEX].textures.append({ nativeBinding.first, texD->srv });
-                    res[RBM_VERTEX].samplers.append({ nativeBinding.second, samplerD->samplerState });
+            const QRhiShaderResourceBinding::Data::SampledTextureData *data = &b->u.stex;
+            bd.stex.count = data->count;
+            const QPair<int, int> nativeBindingVert = mapBinding(b->binding, RBM_VERTEX, nativeResourceBindingMaps);
+            const QPair<int, int> nativeBindingFrag = mapBinding(b->binding, RBM_FRAGMENT, nativeResourceBindingMaps);
+            const QPair<int, int> nativeBindingComp = mapBinding(b->binding, RBM_COMPUTE, nativeResourceBindingMaps);
+            // if SPIR-V binding b is mapped to tN and sN in HLSL, and it
+            // is an array, then it will use tN, tN+1, tN+2, ..., and sN,
+            // sN+1, sN+2, ...
+            for (int elem = 0; elem < data->count; ++elem) {
+                QD3D11Texture *texD = QRHI_RES(QD3D11Texture, data->texSamplers[elem].tex);
+                QD3D11Sampler *samplerD = QRHI_RES(QD3D11Sampler, data->texSamplers[elem].sampler);
+                bd.stex.d[elem].texId = texD->m_id;
+                bd.stex.d[elem].texGeneration = texD->generation;
+                bd.stex.d[elem].samplerId = samplerD->m_id;
+                bd.stex.d[elem].samplerGeneration = samplerD->generation;
+                if (b->stage.testFlag(QRhiShaderResourceBinding::VertexStage)) {
+                    if (nativeBindingVert.first >= 0 && nativeBindingVert.second >= 0) {
+                        res[RBM_VERTEX].textures.append({ nativeBindingVert.first + elem, texD->srv });
+                        res[RBM_VERTEX].samplers.append({ nativeBindingVert.second + elem, samplerD->samplerState });
+                    }
                 }
-            }
-            if (b->stage.testFlag(QRhiShaderResourceBinding::FragmentStage)) {
-                QPair<int, int> nativeBinding = mapBinding(b->binding, RBM_FRAGMENT, nativeResourceBindingMaps);
-                if (nativeBinding.first >= 0 && nativeBinding.second >= 0) {
-                    res[RBM_FRAGMENT].textures.append({ nativeBinding.first, texD->srv });
-                    res[RBM_FRAGMENT].samplers.append({ nativeBinding.second, samplerD->samplerState });
+                if (b->stage.testFlag(QRhiShaderResourceBinding::FragmentStage)) {
+                    if (nativeBindingFrag.first >= 0 && nativeBindingFrag.second >= 0) {
+                        res[RBM_FRAGMENT].textures.append({ nativeBindingFrag.first + elem, texD->srv });
+                        res[RBM_FRAGMENT].samplers.append({ nativeBindingFrag.second + elem, samplerD->samplerState });
+                    }
                 }
-            }
-            if (b->stage.testFlag(QRhiShaderResourceBinding::ComputeStage)) {
-                QPair<int, int> nativeBinding = mapBinding(b->binding, RBM_COMPUTE, nativeResourceBindingMaps);
-                if (nativeBinding.first >= 0 && nativeBinding.second >= 0) {
-                    res[RBM_COMPUTE].textures.append({ nativeBinding.first, texD->srv });
-                    res[RBM_COMPUTE].samplers.append({ nativeBinding.second, samplerD->samplerState });
+                if (b->stage.testFlag(QRhiShaderResourceBinding::ComputeStage)) {
+                    if (nativeBindingComp.first >= 0 && nativeBindingComp.second >= 0) {
+                        res[RBM_COMPUTE].textures.append({ nativeBindingComp.first + elem, texD->srv });
+                        res[RBM_COMPUTE].samplers.append({ nativeBindingComp.second + elem, samplerD->samplerState });
+                    }
                 }
             }
         }
@@ -2077,102 +2095,156 @@ static void applyDynamicOffsets(QVarLengthArray<UINT, 4> *offsets,
     }
 }
 
+static inline uint clampedResourceCount(uint startSlot, int countSlots, uint maxSlots, const char *resType)
+{
+    if (startSlot + countSlots > maxSlots) {
+        qWarning("Not enough D3D11 %s slots to bind %d resources starting at slot %d, max slots is %d",
+                 resType, countSlots, startSlot, maxSlots);
+        countSlots = maxSlots > startSlot ? maxSlots - startSlot : 0;
+    }
+    return countSlots;
+}
+
 void QRhiD3D11::bindShaderResources(QD3D11ShaderResourceBindings *srbD,
                                     const uint *dynOfsPairs, int dynOfsPairCount,
                                     bool offsetOnlyChange)
 {
     if (!offsetOnlyChange) {
-        for (const auto &batch : srbD->vssamplers.batches)
-            context->VSSetSamplers(batch.startBinding, UINT(batch.resources.count()), batch.resources.constData());
+        for (const auto &batch : srbD->vssamplers.batches) {
+            const uint count = clampedResourceCount(batch.startBinding, batch.resources.count(),
+                                                    D3D11_COMMONSHADER_SAMPLER_SLOT_COUNT, "VS sampler");
+            if (count)
+                context->VSSetSamplers(batch.startBinding, count, batch.resources.constData());
+        }
 
         for (const auto &batch : srbD->vsshaderresources.batches) {
-            context->VSSetShaderResources(batch.startBinding, UINT(batch.resources.count()), batch.resources.constData());
-            contextState.vsHighestActiveSrvBinding = qMax<int>(contextState.vsHighestActiveSrvBinding,
-                                                               int(batch.startBinding) + batch.resources.count() - 1);
+            const uint count = clampedResourceCount(batch.startBinding, batch.resources.count(),
+                                                    D3D11_COMMONSHADER_INPUT_RESOURCE_SLOT_COUNT, "VS SRV");
+            if (count) {
+                context->VSSetShaderResources(batch.startBinding, count, batch.resources.constData());
+                contextState.vsHighestActiveSrvBinding = qMax(contextState.vsHighestActiveSrvBinding,
+                                                            int(batch.startBinding + count) - 1);
+            }
         }
 
-        for (const auto &batch : srbD->fssamplers.batches)
-            context->PSSetSamplers(batch.startBinding, UINT(batch.resources.count()), batch.resources.constData());
+        for (const auto &batch : srbD->fssamplers.batches) {
+            const uint count = clampedResourceCount(batch.startBinding, batch.resources.count(),
+                                                    D3D11_COMMONSHADER_SAMPLER_SLOT_COUNT, "PS sampler");
+            if (count)
+                context->PSSetSamplers(batch.startBinding, count, batch.resources.constData());
+        }
 
         for (const auto &batch : srbD->fsshaderresources.batches) {
-            context->PSSetShaderResources(batch.startBinding, UINT(batch.resources.count()), batch.resources.constData());
-            contextState.fsHighestActiveSrvBinding = qMax<int>(contextState.fsHighestActiveSrvBinding,
-                                                               int(batch.startBinding) + batch.resources.count() - 1);
+            const uint count = clampedResourceCount(batch.startBinding, batch.resources.count(),
+                                                    D3D11_COMMONSHADER_INPUT_RESOURCE_SLOT_COUNT, "PS SRV");
+            if (count) {
+                context->PSSetShaderResources(batch.startBinding, count, batch.resources.constData());
+                contextState.fsHighestActiveSrvBinding = qMax(contextState.fsHighestActiveSrvBinding,
+                                                            int(batch.startBinding + count) - 1);
+            }
         }
 
-        for (const auto &batch : srbD->cssamplers.batches)
-            context->CSSetSamplers(batch.startBinding, UINT(batch.resources.count()), batch.resources.constData());
+        for (const auto &batch : srbD->cssamplers.batches) {
+            const uint count = clampedResourceCount(batch.startBinding, batch.resources.count(),
+                                                    D3D11_COMMONSHADER_SAMPLER_SLOT_COUNT, "CS sampler");
+            if (count)
+                context->CSSetSamplers(batch.startBinding, count, batch.resources.constData());
+        }
 
         for (const auto &batch : srbD->csshaderresources.batches) {
-            context->CSSetShaderResources(batch.startBinding, UINT(batch.resources.count()), batch.resources.constData());
-            contextState.csHighestActiveSrvBinding = qMax<int>(contextState.csHighestActiveSrvBinding,
-                                                               int(batch.startBinding) + batch.resources.count() - 1);
+            const uint count = clampedResourceCount(batch.startBinding, batch.resources.count(),
+                                                    D3D11_COMMONSHADER_INPUT_RESOURCE_SLOT_COUNT, "CS SRV");
+            if (count) {
+                context->CSSetShaderResources(batch.startBinding, count, batch.resources.constData());
+                contextState.csHighestActiveSrvBinding = qMax(contextState.csHighestActiveSrvBinding,
+                                                              int(batch.startBinding + count) - 1);
+            }
         }
     }
 
     for (int i = 0, ie = srbD->vsubufs.batches.count(); i != ie; ++i) {
-        if (!dynOfsPairCount) {
-            context->VSSetConstantBuffers1(srbD->vsubufs.batches[i].startBinding,
-                                           UINT(srbD->vsubufs.batches[i].resources.count()),
-                                           srbD->vsubufs.batches[i].resources.constData(),
-                                           srbD->vsubufoffsets.batches[i].resources.constData(),
-                                           srbD->vsubufsizes.batches[i].resources.constData());
-        } else {
-            QVarLengthArray<UINT, 4> offsets;
-            applyDynamicOffsets(&offsets, i, &srbD->vsubufs, &srbD->vsubufoffsets, dynOfsPairs, dynOfsPairCount);
-            context->VSSetConstantBuffers1(srbD->vsubufs.batches[i].startBinding,
-                                           UINT(srbD->vsubufs.batches[i].resources.count()),
-                                           srbD->vsubufs.batches[i].resources.constData(),
-                                           offsets.constData(),
-                                           srbD->vsubufsizes.batches[i].resources.constData());
+        const uint count = clampedResourceCount(srbD->vsubufs.batches[i].startBinding,
+                                                srbD->vsubufs.batches[i].resources.count(),
+                                                D3D11_COMMONSHADER_CONSTANT_BUFFER_API_SLOT_COUNT,
+                                                "VS cbuf");
+        if (count) {
+            if (!dynOfsPairCount) {
+                context->VSSetConstantBuffers1(srbD->vsubufs.batches[i].startBinding,
+                                               count,
+                                               srbD->vsubufs.batches[i].resources.constData(),
+                                               srbD->vsubufoffsets.batches[i].resources.constData(),
+                                               srbD->vsubufsizes.batches[i].resources.constData());
+            } else {
+                QVarLengthArray<UINT, 4> offsets;
+                applyDynamicOffsets(&offsets, i, &srbD->vsubufs, &srbD->vsubufoffsets, dynOfsPairs, dynOfsPairCount);
+                context->VSSetConstantBuffers1(srbD->vsubufs.batches[i].startBinding,
+                                               count,
+                                               srbD->vsubufs.batches[i].resources.constData(),
+                                               offsets.constData(),
+                                               srbD->vsubufsizes.batches[i].resources.constData());
+            }
         }
     }
 
     for (int i = 0, ie = srbD->fsubufs.batches.count(); i != ie; ++i) {
-        if (!dynOfsPairCount) {
-            context->PSSetConstantBuffers1(srbD->fsubufs.batches[i].startBinding,
-                                           UINT(srbD->fsubufs.batches[i].resources.count()),
-                                           srbD->fsubufs.batches[i].resources.constData(),
-                                           srbD->fsubufoffsets.batches[i].resources.constData(),
-                                           srbD->fsubufsizes.batches[i].resources.constData());
-        } else {
-            QVarLengthArray<UINT, 4> offsets;
-            applyDynamicOffsets(&offsets, i, &srbD->fsubufs, &srbD->fsubufoffsets, dynOfsPairs, dynOfsPairCount);
-            context->PSSetConstantBuffers1(srbD->fsubufs.batches[i].startBinding,
-                                           UINT(srbD->fsubufs.batches[i].resources.count()),
-                                           srbD->fsubufs.batches[i].resources.constData(),
-                                           offsets.constData(),
-                                           srbD->fsubufsizes.batches[i].resources.constData());
+        const uint count = clampedResourceCount(srbD->fsubufs.batches[i].startBinding,
+                                                srbD->fsubufs.batches[i].resources.count(),
+                                                D3D11_COMMONSHADER_CONSTANT_BUFFER_API_SLOT_COUNT,
+                                                "PS cbuf");
+        if (count) {
+            if (!dynOfsPairCount) {
+                context->PSSetConstantBuffers1(srbD->fsubufs.batches[i].startBinding,
+                                               count,
+                                               srbD->fsubufs.batches[i].resources.constData(),
+                                               srbD->fsubufoffsets.batches[i].resources.constData(),
+                                               srbD->fsubufsizes.batches[i].resources.constData());
+            } else {
+                QVarLengthArray<UINT, 4> offsets;
+                applyDynamicOffsets(&offsets, i, &srbD->fsubufs, &srbD->fsubufoffsets, dynOfsPairs, dynOfsPairCount);
+                context->PSSetConstantBuffers1(srbD->fsubufs.batches[i].startBinding,
+                                               count,
+                                               srbD->fsubufs.batches[i].resources.constData(),
+                                               offsets.constData(),
+                                               srbD->fsubufsizes.batches[i].resources.constData());
+            }
         }
     }
 
     for (int i = 0, ie = srbD->csubufs.batches.count(); i != ie; ++i) {
-        if (!dynOfsPairCount) {
-            context->CSSetConstantBuffers1(srbD->csubufs.batches[i].startBinding,
-                                           UINT(srbD->csubufs.batches[i].resources.count()),
-                                           srbD->csubufs.batches[i].resources.constData(),
-                                           srbD->csubufoffsets.batches[i].resources.constData(),
-                                           srbD->csubufsizes.batches[i].resources.constData());
-        } else {
-            QVarLengthArray<UINT, 4> offsets;
-            applyDynamicOffsets(&offsets, i, &srbD->csubufs, &srbD->csubufoffsets, dynOfsPairs, dynOfsPairCount);
-            context->CSSetConstantBuffers1(srbD->csubufs.batches[i].startBinding,
-                                           UINT(srbD->csubufs.batches[i].resources.count()),
-                                           srbD->csubufs.batches[i].resources.constData(),
-                                           offsets.constData(),
-                                           srbD->csubufsizes.batches[i].resources.constData());
+        const uint count = clampedResourceCount(srbD->csubufs.batches[i].startBinding,
+                                                srbD->csubufs.batches[i].resources.count(),
+                                                D3D11_COMMONSHADER_CONSTANT_BUFFER_API_SLOT_COUNT,
+                                                "CS cbuf");
+        if (count) {
+            if (!dynOfsPairCount) {
+                context->CSSetConstantBuffers1(srbD->csubufs.batches[i].startBinding,
+                                               count,
+                                               srbD->csubufs.batches[i].resources.constData(),
+                                               srbD->csubufoffsets.batches[i].resources.constData(),
+                                               srbD->csubufsizes.batches[i].resources.constData());
+            } else {
+                QVarLengthArray<UINT, 4> offsets;
+                applyDynamicOffsets(&offsets, i, &srbD->csubufs, &srbD->csubufoffsets, dynOfsPairs, dynOfsPairCount);
+                context->CSSetConstantBuffers1(srbD->csubufs.batches[i].startBinding,
+                                               count,
+                                               srbD->csubufs.batches[i].resources.constData(),
+                                               offsets.constData(),
+                                               srbD->csubufsizes.batches[i].resources.constData());
+            }
         }
     }
 
-    for (int i = 0, ie = srbD->csUAVs.batches.count(); i != ie; ++i) {
-        const uint startBinding = srbD->csUAVs.batches[i].startBinding;
-        const uint count = uint(srbD->csUAVs.batches[i].resources.count());
-        context->CSSetUnorderedAccessViews(startBinding,
-                                           count,
-                                           srbD->csUAVs.batches[i].resources.constData(),
-                                           nullptr);
-        contextState.csHighestActiveUavBinding = qMax<int>(contextState.csHighestActiveUavBinding,
-                                                           int(startBinding + count - 1));
+    for (const auto &batch : srbD->csUAVs.batches) {
+        const uint count = clampedResourceCount(batch.startBinding, batch.resources.count(),
+                                                D3D11_1_UAV_SLOT_COUNT, "CS UAV");
+        if (count) {
+            context->CSSetUnorderedAccessViews(batch.startBinding,
+                                               count,
+                                               batch.resources.constData(),
+                                               nullptr);
+            contextState.csHighestActiveUavBinding = qMax(contextState.csHighestActiveUavBinding,
+                                                          int(batch.startBinding + count) - 1);
+        }
     }
 }
 
@@ -3529,11 +3601,15 @@ static pD3DCompile resolveD3DCompile()
 
 static QByteArray compileHlslShaderSource(const QShader &shader, QShader::Variant shaderVariant, QString *error, QShaderKey *usedShaderKey)
 {
-    QShaderCode dxbc = shader.shader({ QShader::DxbcShader, 50, shaderVariant });
-    if (!dxbc.shader().isEmpty())
+    QShaderKey key = { QShader::DxbcShader, 50, shaderVariant };
+    QShaderCode dxbc = shader.shader(key);
+    if (!dxbc.shader().isEmpty()) {
+        if (usedShaderKey)
+            *usedShaderKey = key;
         return dxbc.shader();
+    }
 
-    const QShaderKey key = { QShader::HlslShader, 50, shaderVariant };
+    key = { QShader::HlslShader, 50, shaderVariant };
     QShaderCode hlslSource = shader.shader(key);
     if (hlslSource.shader().isEmpty()) {
         qWarning() << "No HLSL (shader model 5.0) code found in baked shader" << shader;
diff --git a/src/gui/rhi/qrhid3d11_p_p.h b/src/gui/rhi/qrhid3d11_p_p.h
index f749b612b5..33412b8011 100644
--- a/src/gui/rhi/qrhid3d11_p_p.h
+++ b/src/gui/rhi/qrhid3d11_p_p.h
@@ -210,10 +210,13 @@ struct QD3D11ShaderResourceBindings : public QRhiShaderResourceBindings
         uint generation;
     };
     struct BoundSampledTextureData {
-        quint64 texId;
-        uint texGeneration;
-        quint64 samplerId;
-        uint samplerGeneration;
+        int count;
+        struct {
+            quint64 texId;
+            uint texGeneration;
+            quint64 samplerId;
+            uint samplerGeneration;
+        } d[QRhiShaderResourceBinding::Data::MAX_TEX_SAMPLER_ARRAY_SIZE];
     };
     struct BoundStorageImageData {
         quint64 id;
diff --git a/src/gui/rhi/qrhigles2.cpp b/src/gui/rhi/qrhigles2.cpp
index 4a442bc582..ea9bce08e4 100644
--- a/src/gui/rhi/qrhigles2.cpp
+++ b/src/gui/rhi/qrhigles2.cpp
@@ -917,10 +917,12 @@ void QRhiGles2::setShaderResources(QRhiCommandBuffer *cb, QRhiShaderResourceBind
                 hasDynamicOffsetInSrb = true;
             break;
         case QRhiShaderResourceBinding::SampledTexture:
-            trackedRegisterTexture(&passResTracker,
-                                   QRHI_RES(QGles2Texture, b->u.stex.tex),
-                                   QRhiPassResourceTracker::TexSample,
-                                   QRhiPassResourceTracker::toPassTrackerTextureStage(b->stage));
+            for (int elem = 0; elem < b->u.stex.count; ++elem) {
+                trackedRegisterTexture(&passResTracker,
+                                       QRHI_RES(QGles2Texture, b->u.stex.texSamplers[elem].tex),
+                                       QRhiPassResourceTracker::TexSample,
+                                       QRhiPassResourceTracker::toPassTrackerTextureStage(b->stage));
+            }
             break;
         case QRhiShaderResourceBinding::ImageLoad:
         case QRhiShaderResourceBinding::ImageStore:
@@ -2572,36 +2574,37 @@ void QRhiGles2::bindShaderResources(QRhiGraphicsPipeline *maybeGraphicsPs, QRhiC
             break;
         case QRhiShaderResourceBinding::SampledTexture:
         {
-            QGles2Texture *texD = QRHI_RES(QGles2Texture, b->u.stex.tex);
-            QGles2Sampler *samplerD = QRHI_RES(QGles2Sampler, b->u.stex.sampler);
             QVector<QGles2SamplerDescription> &samplers(maybeGraphicsPs ? QRHI_RES(QGles2GraphicsPipeline, maybeGraphicsPs)->samplers
                                                                         : QRHI_RES(QGles2ComputePipeline, maybeComputePs)->samplers);
-
-            for (QGles2SamplerDescription &sampler : samplers) {
-                if (sampler.binding == b->binding) {
-                    f->glActiveTexture(GL_TEXTURE0 + uint(texUnit));
-                    f->glBindTexture(texD->target, texD->texture);
-
-                    if (texD->samplerState != samplerD->d) {
-                        f->glTexParameteri(texD->target, GL_TEXTURE_MIN_FILTER, GLint(samplerD->d.glminfilter));
-                        f->glTexParameteri(texD->target, GL_TEXTURE_MAG_FILTER, GLint(samplerD->d.glmagfilter));
-                        f->glTexParameteri(texD->target, GL_TEXTURE_WRAP_S, GLint(samplerD->d.glwraps));
-                        f->glTexParameteri(texD->target, GL_TEXTURE_WRAP_T, GLint(samplerD->d.glwrapt));
-                        // 3D textures not supported by GLES 2.0 or by us atm...
-                        //f->glTexParameteri(texD->target, GL_TEXTURE_WRAP_R, samplerD->d.glwrapr);
-                        if (caps.textureCompareMode) {
-                            if (samplerD->d.gltexcomparefunc != GL_NEVER) {
-                                f->glTexParameteri(texD->target, GL_TEXTURE_COMPARE_MODE, GL_COMPARE_REF_TO_TEXTURE);
-                                f->glTexParameteri(texD->target, GL_TEXTURE_COMPARE_FUNC, GLint(samplerD->d.gltexcomparefunc));
-                            } else {
-                                f->glTexParameteri(texD->target, GL_TEXTURE_COMPARE_MODE, GL_NONE);
+            for (int elem = 0; elem < b->u.stex.count; ++elem) {
+                QGles2Texture *texD = QRHI_RES(QGles2Texture, b->u.stex.texSamplers[elem].tex);
+                QGles2Sampler *samplerD = QRHI_RES(QGles2Sampler, b->u.stex.texSamplers[elem].sampler);
+                for (QGles2SamplerDescription &sampler : samplers) {
+                    if (sampler.binding == b->binding) {
+                        f->glActiveTexture(GL_TEXTURE0 + uint(texUnit));
+                        f->glBindTexture(texD->target, texD->texture);
+
+                        if (texD->samplerState != samplerD->d) {
+                            f->glTexParameteri(texD->target, GL_TEXTURE_MIN_FILTER, GLint(samplerD->d.glminfilter));
+                            f->glTexParameteri(texD->target, GL_TEXTURE_MAG_FILTER, GLint(samplerD->d.glmagfilter));
+                            f->glTexParameteri(texD->target, GL_TEXTURE_WRAP_S, GLint(samplerD->d.glwraps));
+                            f->glTexParameteri(texD->target, GL_TEXTURE_WRAP_T, GLint(samplerD->d.glwrapt));
+                            // 3D textures not supported by GLES 2.0 or by us atm...
+                            //f->glTexParameteri(texD->target, GL_TEXTURE_WRAP_R, samplerD->d.glwrapr);
+                            if (caps.textureCompareMode) {
+                                if (samplerD->d.gltexcomparefunc != GL_NEVER) {
+                                    f->glTexParameteri(texD->target, GL_TEXTURE_COMPARE_MODE, GL_COMPARE_REF_TO_TEXTURE);
+                                    f->glTexParameteri(texD->target, GL_TEXTURE_COMPARE_FUNC, GLint(samplerD->d.gltexcomparefunc));
+                                } else {
+                                    f->glTexParameteri(texD->target, GL_TEXTURE_COMPARE_MODE, GL_NONE);
+                                }
                             }
+                            texD->samplerState = samplerD->d;
                         }
-                        texD->samplerState = samplerD->d;
-                    }
 
-                    f->glUniform1i(sampler.glslLocation, texUnit);
-                    ++texUnit;
+                        f->glUniform1i(sampler.glslLocation + elem, texUnit);
+                        ++texUnit;
+                    }
                 }
             }
         }
diff --git a/src/gui/rhi/qrhimetal.mm b/src/gui/rhi/qrhimetal.mm
index 314c58b0b7..0806c8a052 100644
--- a/src/gui/rhi/qrhimetal.mm
+++ b/src/gui/rhi/qrhimetal.mm
@@ -748,30 +748,33 @@ void QRhiMetal::enqueueShaderResourceBindings(QMetalShaderResourceBindings *srbD
             break;
         case QRhiShaderResourceBinding::SampledTexture:
         {
-            QMetalTexture *texD = QRHI_RES(QMetalTexture, b->u.stex.tex);
-            QMetalSampler *samplerD = QRHI_RES(QMetalSampler, b->u.stex.sampler);
-            if (b->stage.testFlag(QRhiShaderResourceBinding::VertexStage)) {
-                const int nativeBindingTexture = mapBinding(b->binding, VERTEX, nativeResourceBindingMaps, BindingType::Texture);
-                const int nativeBindingSampler = mapBinding(b->binding, VERTEX, nativeResourceBindingMaps, BindingType::Sampler);
-                if (nativeBindingTexture >= 0 && nativeBindingSampler >= 0) {
-                    res[VERTEX].textures.append({ nativeBindingTexture, texD->d->tex });
-                    res[VERTEX].samplers.append({ nativeBindingSampler, samplerD->d->samplerState });
+            const QRhiShaderResourceBinding::Data::SampledTextureData *data = &b->u.stex;
+            for (int elem = 0; elem < data->count; ++elem) {
+                QMetalTexture *texD = QRHI_RES(QMetalTexture, b->u.stex.texSamplers[elem].tex);
+                QMetalSampler *samplerD = QRHI_RES(QMetalSampler, b->u.stex.texSamplers[elem].sampler);
+                if (b->stage.testFlag(QRhiShaderResourceBinding::VertexStage)) {
+                    const int nativeBindingTexture = mapBinding(b->binding, VERTEX, nativeResourceBindingMaps, BindingType::Texture);
+                    const int nativeBindingSampler = mapBinding(b->binding, VERTEX, nativeResourceBindingMaps, BindingType::Sampler);
+                    if (nativeBindingTexture >= 0 && nativeBindingSampler >= 0) {
+                        res[VERTEX].textures.append({ nativeBindingTexture + elem, texD->d->tex });
+                        res[VERTEX].samplers.append({ nativeBindingSampler + elem, samplerD->d->samplerState });
+                    }
                 }
-            }
-            if (b->stage.testFlag(QRhiShaderResourceBinding::FragmentStage)) {
-                const int nativeBindingTexture = mapBinding(b->binding, FRAGMENT, nativeResourceBindingMaps, BindingType::Texture);
-                const int nativeBindingSampler = mapBinding(b->binding, FRAGMENT, nativeResourceBindingMaps, BindingType::Sampler);
-                if (nativeBindingTexture >= 0 && nativeBindingSampler >= 0) {
-                    res[FRAGMENT].textures.append({ nativeBindingTexture, texD->d->tex });
-                    res[FRAGMENT].samplers.append({ nativeBindingSampler, samplerD->d->samplerState });
+                if (b->stage.testFlag(QRhiShaderResourceBinding::FragmentStage)) {
+                    const int nativeBindingTexture = mapBinding(b->binding, FRAGMENT, nativeResourceBindingMaps, BindingType::Texture);
+                    const int nativeBindingSampler = mapBinding(b->binding, FRAGMENT, nativeResourceBindingMaps, BindingType::Sampler);
+                    if (nativeBindingTexture >= 0 && nativeBindingSampler >= 0) {
+                        res[FRAGMENT].textures.append({ nativeBindingTexture + elem, texD->d->tex });
+                        res[FRAGMENT].samplers.append({ nativeBindingSampler + elem, samplerD->d->samplerState });
+                    }
                 }
-            }
-            if (b->stage.testFlag(QRhiShaderResourceBinding::ComputeStage)) {
-                const int nativeBindingTexture = mapBinding(b->binding, COMPUTE, nativeResourceBindingMaps, BindingType::Texture);
-                const int nativeBindingSampler = mapBinding(b->binding, COMPUTE, nativeResourceBindingMaps, BindingType::Sampler);
-                if (nativeBindingTexture >= 0 && nativeBindingSampler >= 0) {
-                    res[COMPUTE].textures.append({ nativeBindingTexture, texD->d->tex });
-                    res[COMPUTE].samplers.append({ nativeBindingSampler, samplerD->d->samplerState });
+                if (b->stage.testFlag(QRhiShaderResourceBinding::ComputeStage)) {
+                    const int nativeBindingTexture = mapBinding(b->binding, COMPUTE, nativeResourceBindingMaps, BindingType::Texture);
+                    const int nativeBindingSampler = mapBinding(b->binding, COMPUTE, nativeResourceBindingMaps, BindingType::Sampler);
+                    if (nativeBindingTexture >= 0 && nativeBindingSampler >= 0) {
+                        res[COMPUTE].textures.append({ nativeBindingTexture + elem, texD->d->tex });
+                        res[COMPUTE].samplers.append({ nativeBindingSampler + elem, samplerD->d->samplerState });
+                    }
                 }
             }
         }
@@ -1020,21 +1023,28 @@ void QRhiMetal::setShaderResources(QRhiCommandBuffer *cb, QRhiShaderResourceBind
             break;
         case QRhiShaderResourceBinding::SampledTexture:
         {
-            QMetalTexture *texD = QRHI_RES(QMetalTexture, b->u.stex.tex);
-            QMetalSampler *samplerD = QRHI_RES(QMetalSampler, b->u.stex.sampler);
-            if (texD->generation != bd.stex.texGeneration
-                    || texD->m_id != bd.stex.texId
-                    || samplerD->generation != bd.stex.samplerGeneration
-                    || samplerD->m_id != bd.stex.samplerId)
-            {
+            const QRhiShaderResourceBinding::Data::SampledTextureData *data = &b->u.stex;
+            if (bd.stex.count != data->count) {
+                bd.stex.count = data->count;
                 resNeedsRebind = true;
-                bd.stex.texId = texD->m_id;
-                bd.stex.texGeneration = texD->generation;
-                bd.stex.samplerId = samplerD->m_id;
-                bd.stex.samplerGeneration = samplerD->generation;
             }
-            texD->lastActiveFrameSlot = currentFrameSlot;
-            samplerD->lastActiveFrameSlot = currentFrameSlot;
+            for (int elem = 0; elem < data->count; ++elem) {
+                QMetalTexture *texD = QRHI_RES(QMetalTexture, data->texSamplers[elem].tex);
+                QMetalSampler *samplerD = QRHI_RES(QMetalSampler, data->texSamplers[elem].sampler);
+                if (texD->generation != bd.stex.d[elem].texGeneration
+                        || texD->m_id != bd.stex.d[elem].texId
+                        || samplerD->generation != bd.stex.d[elem].samplerGeneration
+                        || samplerD->m_id != bd.stex.d[elem].samplerId)
+                {
+                    resNeedsRebind = true;
+                    bd.stex.d[elem].texId = texD->m_id;
+                    bd.stex.d[elem].texGeneration = texD->generation;
+                    bd.stex.d[elem].samplerId = samplerD->m_id;
+                    bd.stex.d[elem].samplerGeneration = samplerD->generation;
+                }
+                texD->lastActiveFrameSlot = currentFrameSlot;
+                samplerD->lastActiveFrameSlot = currentFrameSlot;
+            }
         }
             break;
         case QRhiShaderResourceBinding::ImageLoad:
@@ -2981,12 +2991,16 @@ bool QMetalShaderResourceBindings::build()
             break;
         case QRhiShaderResourceBinding::SampledTexture:
         {
-            QMetalTexture *texD = QRHI_RES(QMetalTexture, b->u.stex.tex);
-            QMetalSampler *samplerD = QRHI_RES(QMetalSampler, b->u.stex.sampler);
-            bd.stex.texId = texD->m_id;
-            bd.stex.texGeneration = texD->generation;
-            bd.stex.samplerId = samplerD->m_id;
-            bd.stex.samplerGeneration = samplerD->generation;
+            const QRhiShaderResourceBinding::Data::SampledTextureData *data = &b->u.stex;
+            bd.stex.count = data->count;
+            for (int elem = 0; elem < data->count; ++elem) {
+                QMetalTexture *texD = QRHI_RES(QMetalTexture, data->texSamplers[elem].tex);
+                QMetalSampler *samplerD = QRHI_RES(QMetalSampler, data->texSamplers[elem].sampler);
+                bd.stex.d[elem].texId = texD->m_id;
+                bd.stex.d[elem].texGeneration = texD->generation;
+                bd.stex.d[elem].samplerId = samplerD->m_id;
+                bd.stex.d[elem].samplerGeneration = samplerD->generation;
+            }
         }
             break;
         case QRhiShaderResourceBinding::ImageLoad:
@@ -3241,8 +3255,12 @@ static inline MTLCullMode toMetalCullMode(QRhiGraphicsPipeline::CullMode c)
 id<MTLLibrary> QRhiMetalData::createMetalLib(const QShader &shader, QShader::Variant shaderVariant,
                                              QString *error, QByteArray *entryPoint, QShaderKey *activeKey)
 {
-    QShaderKey key = { QShader::MetalLibShader, 12, shaderVariant };
+    QShaderKey key = { QShader::MetalLibShader, 20, shaderVariant };
     QShaderCode mtllib = shader.shader(key);
+    if (mtllib.shader().isEmpty()) {
+        key.setSourceVersion(12);
+        mtllib = shader.shader(key);
+    }
     if (!mtllib.shader().isEmpty()) {
         dispatch_data_t data = dispatch_data_create(mtllib.shader().constData(),
                                                     size_t(mtllib.shader().size()),
@@ -3261,16 +3279,20 @@ id<MTLLibrary> QRhiMetalData::createMetalLib(const QShader &shader, QShader::Var
         }
     }
 
-    key = { QShader::MslShader, 12, shaderVariant };
+    key = { QShader::MslShader, 20, shaderVariant };
     QShaderCode mslSource = shader.shader(key);
     if (mslSource.shader().isEmpty()) {
-        qWarning() << "No MSL 1.2 code found in baked shader" << shader;
+        key.setSourceVersion(12);
+        mslSource = shader.shader(key);
+    }
+    if (mslSource.shader().isEmpty()) {
+        qWarning() << "No MSL 2.0 or 1.2 code found in baked shader" << shader;
         return nil;
     }
 
     NSString *src = [NSString stringWithUTF8String: mslSource.shader().constData()];
     MTLCompileOptions *opts = [[MTLCompileOptions alloc] init];
-    opts.languageVersion = MTLLanguageVersion1_2;
+    opts.languageVersion = key.sourceVersion() == 20 ? MTLLanguageVersion2_0 : MTLLanguageVersion1_2;
     NSError *err = nil;
     id<MTLLibrary> lib = [dev newLibraryWithSource: src options: opts error: &err];
     [opts release];
diff --git a/src/gui/rhi/qrhimetal_p_p.h b/src/gui/rhi/qrhimetal_p_p.h
index a5af5611a6..cb4b777d88 100644
--- a/src/gui/rhi/qrhimetal_p_p.h
+++ b/src/gui/rhi/qrhimetal_p_p.h
@@ -197,10 +197,13 @@ struct QMetalShaderResourceBindings : public QRhiShaderResourceBindings
         uint generation;
     };
     struct BoundSampledTextureData {
-        quint64 texId;
-        uint texGeneration;
-        quint64 samplerId;
-        uint samplerGeneration;
+        int count;
+        struct {
+            quint64 texId;
+            uint texGeneration;
+            quint64 samplerId;
+            uint samplerGeneration;
+        } d[QRhiShaderResourceBinding::Data::MAX_TEX_SAMPLER_ARRAY_SIZE];
     };
     struct BoundStorageImageData {
         quint64 id;
diff --git a/src/gui/rhi/qrhivulkan.cpp b/src/gui/rhi/qrhivulkan.cpp
index a92c3e14e9..26c153afff 100644
--- a/src/gui/rhi/qrhivulkan.cpp
+++ b/src/gui/rhi/qrhivulkan.cpp
@@ -345,6 +345,15 @@ static bool qvk_debug_filter(VkDebugReportFlagsEXT flags, VkDebugReportObjectTyp
         return true;
     }
 
+    // In certain cases allocateDescriptorSet() will attempt to allocate from a
+    // pool that does not have enough descriptors of a certain type. This makes
+    // the validation layer shout. However, this is not an error since we will
+    // then move on to another pool. If there is a real error, a qWarning
+    // message is shown by allocateDescriptorSet(), so the validation warning
+    // does not have any value and is just noise.
+    if (strstr(pMessage, "VUID-VkDescriptorSetAllocateInfo-descriptorPool-00307"))
+        return true;
+
     return false;
 }
 
@@ -2487,7 +2496,8 @@ void QRhiVulkan::updateShaderResourceBindings(QRhiShaderResourceBindings *srb, i
     QVkShaderResourceBindings *srbD = QRHI_RES(QVkShaderResourceBindings, srb);
 
     QVarLengthArray<VkDescriptorBufferInfo, 8> bufferInfos;
-    QVarLengthArray<VkDescriptorImageInfo, 8> imageInfos;
+    using ArrayOfImageDesc = QVarLengthArray<VkDescriptorImageInfo, 8>;
+    QVarLengthArray<ArrayOfImageDesc, 8> imageInfos;
     QVarLengthArray<VkWriteDescriptorSet, 12> writeInfos;
     QVarLengthArray<QPair<int, int>, 12> infoIndices;
 
@@ -2530,17 +2540,22 @@ void QRhiVulkan::updateShaderResourceBindings(QRhiShaderResourceBindings *srb, i
                 break;
             case QRhiShaderResourceBinding::SampledTexture:
             {
-                QVkTexture *texD = QRHI_RES(QVkTexture, b->u.stex.tex);
-                QVkSampler *samplerD = QRHI_RES(QVkSampler, b->u.stex.sampler);
+                const QRhiShaderResourceBinding::Data::SampledTextureData *data = &b->u.stex;
+                writeInfo.descriptorCount = data->count; // arrays of combined image samplers are supported
                 writeInfo.descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER;
-                bd.stex.texId = texD->m_id;
-                bd.stex.texGeneration = texD->generation;
-                bd.stex.samplerId = samplerD->m_id;
-                bd.stex.samplerGeneration = samplerD->generation;
-                VkDescriptorImageInfo imageInfo;
-                imageInfo.sampler = samplerD->sampler;
-                imageInfo.imageView = texD->imageView;
-                imageInfo.imageLayout = VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL;
+                ArrayOfImageDesc imageInfo(data->count);
+                for (int elem = 0; elem < data->count; ++elem) {
+                    QVkTexture *texD = QRHI_RES(QVkTexture, data->texSamplers[elem].tex);
+                    QVkSampler *samplerD = QRHI_RES(QVkSampler, data->texSamplers[elem].sampler);
+                    bd.stex.d[elem].texId = texD->m_id;
+                    bd.stex.d[elem].texGeneration = texD->generation;
+                    bd.stex.d[elem].samplerId = samplerD->m_id;
+                    bd.stex.d[elem].samplerGeneration = samplerD->generation;
+                    imageInfo[elem].sampler = samplerD->sampler;
+                    imageInfo[elem].imageView = texD->imageView;
+                    imageInfo[elem].imageLayout = VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL;
+                }
+                bd.stex.count = data->count;
                 imageInfoIndex = imageInfos.count();
                 imageInfos.append(imageInfo);
             }
@@ -2555,10 +2570,10 @@ void QRhiVulkan::updateShaderResourceBindings(QRhiShaderResourceBindings *srb, i
                     writeInfo.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE;
                     bd.simage.id = texD->m_id;
                     bd.simage.generation = texD->generation;
-                    VkDescriptorImageInfo imageInfo;
-                    imageInfo.sampler = VK_NULL_HANDLE;
-                    imageInfo.imageView = view;
-                    imageInfo.imageLayout = VK_IMAGE_LAYOUT_GENERAL;
+                    ArrayOfImageDesc imageInfo(1);
+                    imageInfo[0].sampler = VK_NULL_HANDLE;
+                    imageInfo[0].imageView = view;
+                    imageInfo[0].imageLayout = VK_IMAGE_LAYOUT_GENERAL;
                     imageInfoIndex = imageInfos.count();
                     imageInfos.append(imageInfo);
                 }
@@ -2596,7 +2611,7 @@ void QRhiVulkan::updateShaderResourceBindings(QRhiShaderResourceBindings *srb, i
         if (bufferInfoIndex >= 0)
             writeInfos[i].pBufferInfo = &bufferInfos[bufferInfoIndex];
         else if (imageInfoIndex >= 0)
-            writeInfos[i].pImageInfo = &imageInfos[imageInfoIndex];
+            writeInfos[i].pImageInfo = imageInfos[imageInfoIndex].constData();
     }
 
     df->vkUpdateDescriptorSets(dev, uint32_t(writeInfos.count()), writeInfos.constData(), 0, nullptr);
@@ -4210,24 +4225,30 @@ void QRhiVulkan::setShaderResources(QRhiCommandBuffer *cb, QRhiShaderResourceBin
             break;
         case QRhiShaderResourceBinding::SampledTexture:
         {
-            QVkTexture *texD = QRHI_RES(QVkTexture, b->u.stex.tex);
-            QVkSampler *samplerD = QRHI_RES(QVkSampler, b->u.stex.sampler);
-            texD->lastActiveFrameSlot = currentFrameSlot;
-            samplerD->lastActiveFrameSlot = currentFrameSlot;
-            trackedRegisterTexture(&passResTracker, texD,
-                                   QRhiPassResourceTracker::TexSample,
-                                   QRhiPassResourceTracker::toPassTrackerTextureStage(b->stage));
-
-            if (texD->generation != bd.stex.texGeneration
-                    || texD->m_id != bd.stex.texId
-                    || samplerD->generation != bd.stex.samplerGeneration
-                    || samplerD->m_id != bd.stex.samplerId)
-            {
+            const QRhiShaderResourceBinding::Data::SampledTextureData *data = &b->u.stex;
+            if (bd.stex.count != data->count) {
+                bd.stex.count = data->count;
                 rewriteDescSet = true;
-                bd.stex.texId = texD->m_id;
-                bd.stex.texGeneration = texD->generation;
-                bd.stex.samplerId = samplerD->m_id;
-                bd.stex.samplerGeneration = samplerD->generation;
+            }
+            for (int elem = 0; elem < data->count; ++elem) {
+                QVkTexture *texD = QRHI_RES(QVkTexture, data->texSamplers[elem].tex);
+                QVkSampler *samplerD = QRHI_RES(QVkSampler, data->texSamplers[elem].sampler);
+                texD->lastActiveFrameSlot = currentFrameSlot;
+                samplerD->lastActiveFrameSlot = currentFrameSlot;
+                trackedRegisterTexture(&passResTracker, texD,
+                                       QRhiPassResourceTracker::TexSample,
+                                       QRhiPassResourceTracker::toPassTrackerTextureStage(b->stage));
+                if (texD->generation != bd.stex.d[elem].texGeneration
+                        || texD->m_id != bd.stex.d[elem].texId
+                        || samplerD->generation != bd.stex.d[elem].samplerGeneration
+                        || samplerD->m_id != bd.stex.d[elem].samplerId)
+                {
+                    rewriteDescSet = true;
+                    bd.stex.d[elem].texId = texD->m_id;
+                    bd.stex.d[elem].texGeneration = texD->generation;
+                    bd.stex.d[elem].samplerId = samplerD->m_id;
+                    bd.stex.d[elem].samplerGeneration = samplerD->generation;
+                }
             }
         }
             break;
@@ -6065,7 +6086,10 @@ bool QVkShaderResourceBindings::build()
         memset(&vkbinding, 0, sizeof(vkbinding));
         vkbinding.binding = uint32_t(b->binding);
         vkbinding.descriptorType = toVkDescriptorType(b);
-        vkbinding.descriptorCount = 1; // no array support yet
+        if (b->type == QRhiShaderResourceBinding::SampledTexture)
+            vkbinding.descriptorCount = b->u.stex.count;
+        else
+            vkbinding.descriptorCount = 1;
         vkbinding.stageFlags = toVkShaderStageFlags(b->stage);
         vkbindings.append(vkbinding);
     }
diff --git a/src/gui/rhi/qrhivulkan_p_p.h b/src/gui/rhi/qrhivulkan_p_p.h
index fd65417e75..62516e268d 100644
--- a/src/gui/rhi/qrhivulkan_p_p.h
+++ b/src/gui/rhi/qrhivulkan_p_p.h
@@ -254,10 +254,13 @@ struct QVkShaderResourceBindings : public QRhiShaderResourceBindings
         uint generation;
     };
     struct BoundSampledTextureData {
-        quint64 texId;
-        uint texGeneration;
-        quint64 samplerId;
-        uint samplerGeneration;
+        int count;
+        struct {
+            quint64 texId;
+            uint texGeneration;
+            quint64 samplerId;
+            uint samplerGeneration;
+        } d[QRhiShaderResourceBinding::Data::MAX_TEX_SAMPLER_ARRAY_SIZE];
     };
     struct BoundStorageImageData {
         quint64 id;
diff --git a/src/gui/text/qrawfont.cpp b/src/gui/text/qrawfont.cpp
index 884525bd76..b4b60cbaaf 100644
--- a/src/gui/text/qrawfont.cpp
+++ b/src/gui/text/qrawfont.cpp
@@ -51,6 +51,7 @@
 
 #include <QtCore/qendian.h>
 #include <QtCore/qfile.h>
+#include <QtGui/qpainterpath.h>
 
 QT_BEGIN_NAMESPACE
 
diff --git a/src/gui/vulkan/qvulkanwindow.cpp b/src/gui/vulkan/qvulkanwindow.cpp
index ee49cf0999..cb89b0b1e6 100644
--- a/src/gui/vulkan/qvulkanwindow.cpp
+++ b/src/gui/vulkan/qvulkanwindow.cpp
@@ -1602,7 +1602,7 @@ bool QVulkanWindow::event(QEvent *e)
 
     \since 5.15
  */
-void QVulkanWindow::setQueueCreateInfoModifier(QueueCreateInfoModifier modifier)
+void QVulkanWindow::setQueueCreateInfoModifier(const QueueCreateInfoModifier &modifier)
 {
     Q_D(QVulkanWindow);
     d->queueCreateInfoModifier = modifier;
diff --git a/src/gui/vulkan/qvulkanwindow.h b/src/gui/vulkan/qvulkanwindow.h
index 511b9501bd..a2d1c9995c 100644
--- a/src/gui/vulkan/qvulkanwindow.h
+++ b/src/gui/vulkan/qvulkanwindow.h
@@ -106,7 +106,7 @@ public:
     typedef std::function<void(const VkQueueFamilyProperties *,
                                uint32_t,
                                QVector<VkDeviceQueueCreateInfo> &)> QueueCreateInfoModifier;
-    void setQueueCreateInfoModifier(QueueCreateInfoModifier modifier);
+    void setQueueCreateInfoModifier(const QueueCreateInfoModifier &modifier);
 
     bool isValid() const;