1 files changed, 75 insertions, 48 deletions
diff --git a/src/gui/painting/qdrawhelper_sse4.cpp b/src/gui/painting/qdrawhelper_sse4.cpp
index dc9755e414..a7b4e6ba76 100644
--- a/src/gui/painting/qdrawhelper_sse4.cpp
+++ b/src/gui/painting/qdrawhelper_sse4.cpp
@@ -1,41 +1,5 @@
-/****************************************************************************
-**
-** Copyright (C) 2016 The Qt Company Ltd.
-** Contact: https://www.qt.io/licensing/
-**
-** This file is part of the QtGui module of the Qt Toolkit.
-**
-** $QT_BEGIN_LICENSE:LGPL$
-** Commercial License Usage
-** Licensees holding valid commercial Qt licenses may use this file in
-** accordance with the commercial license agreement provided with the
-** Software or, alternatively, in accordance with the terms contained in
-** a written agreement between you and The Qt Company. For licensing terms
-** and conditions see https://www.qt.io/terms-conditions. For further
-** information use the contact form at https://www.qt.io/contact-us.
-**
-** GNU Lesser General Public License Usage
-** Alternatively, this file may be used under the terms of the GNU Lesser
-** General Public License version 3 as published by the Free Software
-** Foundation and appearing in the file LICENSE.LGPL3 included in the
-** packaging of this file. Please review the following information to
-** ensure the GNU Lesser General Public License version 3 requirements
-** will be met: https://www.gnu.org/licenses/lgpl-3.0.html.
-**
-** GNU General Public License Usage
-** Alternatively, this file may be used under the terms of the GNU
-** General Public License version 2.0 or (at your option) the GNU General
-** Public license version 3 or any later version approved by the KDE Free
-** Qt Foundation. The licenses are as published by the Free Software
-** Foundation and appearing in the file LICENSE.GPL2 and LICENSE.GPL3
-** included in the packaging of this file. Please review the following
-** information to ensure the GNU General Public License requirements will
-** be met: https://www.gnu.org/licenses/gpl-2.0.html and
-** https://www.gnu.org/licenses/gpl-3.0.html.
-**
-** $QT_END_LICENSE$
-**
-****************************************************************************/
+// Copyright (C) 2016 The Qt Company Ltd.
+// SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only
 
 #include <private/qdrawhelper_p.h>
 #include <private/qdrawingprimitive_sse2_p.h>
@@ -46,7 +10,7 @@
 
 QT_BEGIN_NAMESPACE
 
-#ifndef __AVX2__
+#ifndef __haswell__
 template<bool RGBA>
 static void convertARGBToARGB32PM_sse4(uint *buffer, const uint *src, int count)
 {
@@ -142,7 +106,7 @@ static void convertARGBToRGBA64PM_sse4(QRgba64 *buffer, const uint *src, int cou
         buffer[i] = QRgba64::fromArgb32(s).premultiplied();
     }
 }
-#endif // __AVX2__
+#endif // __haswell__
 
 static inline __m128 Q_DECL_VECTORCALL reciprocal_mul_ps(__m128 a, float mul)
 {
@@ -411,7 +375,7 @@ static inline void convertRGBA64FromRGBA64PM_sse4(QRgba64 *buffer, const QRgba64
     }
 }
 
-#ifndef __AVX2__
+#ifndef __haswell__
 void QT_FASTCALL convertARGB32ToARGB32PM_sse4(uint *buffer, int count, const QList<QRgb> *)
 {
     convertARGBToARGB32PM_sse4<false>(buffer, buffer, count);
@@ -463,7 +427,7 @@ const QRgba64 *QT_FASTCALL fetchRGBA8888ToRGBA64PM_sse4(QRgba64 *buffer, const u
     convertARGBToRGBA64PM_sse4<true>(buffer, reinterpret_cast<const uint *>(src) + index, count);
     return buffer;
 }
-#endif // __AVX2__
+#endif // __haswell__
 
 void QT_FASTCALL storeRGB32FromARGB32PM_sse4(uchar *dest, const uint *src, int index, int count,
                                              const QList<QRgb> *, QDitherInfo *)
@@ -502,6 +466,13 @@ void QT_FASTCALL storeA2RGB30PMFromARGB32PM_sse4(uchar *dest, const uint *src, i
         d[i] = qConvertArgb32ToA2rgb30_sse4<PixelOrder>(src[i]);
 }
 
+template
+void QT_FASTCALL storeA2RGB30PMFromARGB32PM_sse4<PixelOrderBGR>(uchar *dest, const uint *src, int index, int count,
+                                                                const QList<QRgb> *, QDitherInfo *);
+template
+void QT_FASTCALL storeA2RGB30PMFromARGB32PM_sse4<PixelOrderRGB>(uchar *dest, const uint *src, int index, int count,
+                                                                const QList<QRgb> *, QDitherInfo *);
+
 #if QT_CONFIG(raster_64bit)
 void QT_FASTCALL destStore64ARGB32_sse4(QRasterBuffer *rasterBuffer, int x, int y, const QRgba64 *buffer, int length)
 {
@@ -544,12 +515,68 @@ void QT_FASTCALL storeRGBx64FromRGBA64PM_sse4(uchar *dest, const QRgba64 *src, i
     convertRGBA64FromRGBA64PM_sse4<true>(d, src, count);
 }
 
-template
-void QT_FASTCALL storeA2RGB30PMFromARGB32PM_sse4<PixelOrderBGR>(uchar *dest, const uint *src, int index, int count,
-                                                                const QList<QRgb> *, QDitherInfo *);
-template
-void QT_FASTCALL storeA2RGB30PMFromARGB32PM_sse4<PixelOrderRGB>(uchar *dest, const uint *src, int index, int count,
-                                                                const QList<QRgb> *, QDitherInfo *);
+#if QT_CONFIG(raster_fp)
+const QRgbaFloat32 *QT_FASTCALL fetchRGBA32FToRGBA32F_sse4(QRgbaFloat32 *buffer, const uchar *src, int index, int count,
+                                                       const QList<QRgb> *, QDitherInfo *)
+{
+    const QRgbaFloat32 *s = reinterpret_cast<const QRgbaFloat32 *>(src) + index;
+    for (int i = 0; i < count; ++i) {
+        __m128 vsf = _mm_load_ps(reinterpret_cast<const float *>(s + i));
+        __m128 vsa = _mm_shuffle_ps(vsf, vsf, _MM_SHUFFLE(3, 3, 3, 3));
+        vsf = _mm_mul_ps(vsf, vsa);
+        vsf = _mm_insert_ps(vsf, vsa, 0x30);
+        _mm_store_ps(reinterpret_cast<float *>(buffer + i), vsf);
+    }
+    return buffer;
+}
+
+void QT_FASTCALL storeRGBX32FFromRGBA32F_sse4(uchar *dest, const QRgbaFloat32 *src, int index, int count,
+                                              const QList<QRgb> *, QDitherInfo *)
+{
+    QRgbaFloat32 *d = reinterpret_cast<QRgbaFloat32 *>(dest) + index;
+    const __m128 zero = _mm_set_ps(1.0f, 0.0f, 0.0f, 0.0f);
+    for (int i = 0; i < count; ++i) {
+        __m128 vsf = _mm_load_ps(reinterpret_cast<const float *>(src + i));
+        const __m128 vsa = _mm_shuffle_ps(vsf, vsf, _MM_SHUFFLE(3, 3, 3, 3));
+        const float a = _mm_cvtss_f32(vsa);
+        if (a == 1.0f)
+        { }
+        else if (a == 0.0f)
+            vsf = zero;
+        else {
+            __m128 vsr = _mm_rcp_ps(vsa);
+            vsr = _mm_sub_ps(_mm_add_ps(vsr, vsr), _mm_mul_ps(vsr, _mm_mul_ps(vsr, vsa)));
+            vsf = _mm_mul_ps(vsf, vsr);
+            vsf = _mm_insert_ps(vsf, _mm_set_ss(1.0f), 0x30);
+        }
+        _mm_store_ps(reinterpret_cast<float *>(d + i), vsf);
+    }
+}
+
+void QT_FASTCALL storeRGBA32FFromRGBA32F_sse4(uchar *dest, const QRgbaFloat32 *src, int index, int count,
+                                              const QList<QRgb> *, QDitherInfo *)
+{
+    QRgbaFloat32 *d = reinterpret_cast<QRgbaFloat32 *>(dest) + index;
+    const __m128 zero = _mm_set1_ps(0.0f);
+    for (int i = 0; i < count; ++i) {
+        __m128 vsf = _mm_load_ps(reinterpret_cast<const float *>(src + i));
+        const __m128 vsa = _mm_shuffle_ps(vsf, vsf, _MM_SHUFFLE(3, 3, 3, 3));
+        const float a = _mm_cvtss_f32(vsa);
+        if (a == 1.0f)
+        { }
+        else if (a == 0.0f)
+            vsf = zero;
+        else {
+            __m128 vsr = _mm_rcp_ps(vsa);
+            vsr = _mm_sub_ps(_mm_add_ps(vsr, vsr), _mm_mul_ps(vsr, _mm_mul_ps(vsr, vsa)));
+            vsr = _mm_insert_ps(vsr, _mm_set_ss(1.0f), 0x30);
+            vsf = _mm_mul_ps(vsf, vsr);
+        }
+        _mm_store_ps(reinterpret_cast<float *>(d + i), vsf);
+    }
+}
+#endif
+
 
 QT_END_NAMESPACE