1 /****************************************************************************
2 **
3 ** Copyright (C) 2016 The Qt Company Ltd.
4 ** Contact: https://www.qt.io/licensing/
5 **
6 ** This file is part of the QtGui module of the Qt Toolkit.
7 **
8 ** $QT_BEGIN_LICENSE:LGPL$
9 ** Commercial License Usage
10 ** Licensees holding valid commercial Qt licenses may use this file in
11 ** accordance with the commercial license agreement provided with the
12 ** Software or, alternatively, in accordance with the terms contained in
13 ** a written agreement between you and The Qt Company. For licensing terms
14 ** and conditions see https://www.qt.io/terms-conditions. For further
15 ** information use the contact form at https://www.qt.io/contact-us.
16 **
17 ** GNU Lesser General Public License Usage
18 ** Alternatively, this file may be used under the terms of the GNU Lesser
19 ** General Public License version 3 as published by the Free Software
20 ** Foundation and appearing in the file LICENSE.LGPL3 included in the
21 ** packaging of this file. Please review the following information to
22 ** ensure the GNU Lesser General Public License version 3 requirements
23 ** will be met: https://www.gnu.org/licenses/lgpl-3.0.html.
24 **
25 ** GNU General Public License Usage
26 ** Alternatively, this file may be used under the terms of the GNU
27 ** General Public License version 2.0 or (at your option) the GNU General
28 ** Public license version 3 or any later version approved by the KDE Free
29 ** Qt Foundation. The licenses are as published by the Free Software
30 ** Foundation and appearing in the file LICENSE.GPL2 and LICENSE.GPL3
31 ** included in the packaging of this file. Please review the following
32 ** information to ensure the GNU General Public License requirements will
33 ** be met: https://www.gnu.org/licenses/gpl-2.0.html and
34 ** https://www.gnu.org/licenses/gpl-3.0.html.
35 **
36 ** $QT_END_LICENSE$
37 **
38 ****************************************************************************/
39
40 #include "private/qmemrotate_p.h"
41
42 QT_BEGIN_NAMESPACE
43
44 static const int tileSize = 32;
45
46 template <class T>
47 static
qt_memrotate90_tiled(const T * src,int w,int h,int sstride,T * dest,int dstride)48 inline void qt_memrotate90_tiled(const T *src, int w, int h, int sstride, T *dest, int dstride)
49 {
50 sstride /= sizeof(T);
51 dstride /= sizeof(T);
52
53 const int pack = sizeof(quint32) / sizeof(T);
54 const int unaligned =
55 qMin(uint((quintptr(dest) & (sizeof(quint32)-1)) / sizeof(T)), uint(h));
56 const int restX = w % tileSize;
57 const int restY = (h - unaligned) % tileSize;
58 const int unoptimizedY = restY % pack;
59 const int numTilesX = w / tileSize + (restX > 0);
60 const int numTilesY = (h - unaligned) / tileSize + (restY >= pack);
61
62 for (int tx = 0; tx < numTilesX; ++tx) {
63 const int startx = w - tx * tileSize - 1;
64 const int stopx = qMax(startx - tileSize, 0);
65
66 if (unaligned) {
67 for (int x = startx; x >= stopx; --x) {
68 T *d = dest + (w - x - 1) * dstride;
69 for (int y = 0; y < unaligned; ++y) {
70 *d++ = src[y * sstride + x];
71 }
72 }
73 }
74
75 for (int ty = 0; ty < numTilesY; ++ty) {
76 const int starty = ty * tileSize + unaligned;
77 const int stopy = qMin(starty + tileSize, h - unoptimizedY);
78
79 for (int x = startx; x >= stopx; --x) {
80 quint32 *d = reinterpret_cast<quint32*>(dest + (w - x - 1) * dstride + starty);
81 for (int y = starty; y < stopy; y += pack) {
82 quint32 c = src[y * sstride + x];
83 for (int i = 1; i < pack; ++i) {
84 const int shift = (sizeof(T) * 8 * i);
85 const T color = src[(y + i) * sstride + x];
86 c |= color << shift;
87 }
88 *d++ = c;
89 }
90 }
91 }
92
93 if (unoptimizedY) {
94 const int starty = h - unoptimizedY;
95 for (int x = startx; x >= stopx; --x) {
96 T *d = dest + (w - x - 1) * dstride + starty;
97 for (int y = starty; y < h; ++y) {
98 *d++ = src[y * sstride + x];
99 }
100 }
101 }
102 }
103 }
104
105 template <class T>
106 static
qt_memrotate90_tiled_unpacked(const T * src,int w,int h,int sstride,T * dest,int dstride)107 inline void qt_memrotate90_tiled_unpacked(const T *src, int w, int h, int sstride, T *dest,
108 int dstride)
109 {
110 const int numTilesX = (w + tileSize - 1) / tileSize;
111 const int numTilesY = (h + tileSize - 1) / tileSize;
112
113 for (int tx = 0; tx < numTilesX; ++tx) {
114 const int startx = w - tx * tileSize - 1;
115 const int stopx = qMax(startx - tileSize, 0);
116
117 for (int ty = 0; ty < numTilesY; ++ty) {
118 const int starty = ty * tileSize;
119 const int stopy = qMin(starty + tileSize, h);
120
121 for (int x = startx; x >= stopx; --x) {
122 T *d = (T *)((char*)dest + (w - x - 1) * dstride) + starty;
123 const char *s = (const char*)(src + x) + starty * sstride;
124 for (int y = starty; y < stopy; ++y) {
125 *d++ = *(const T *)(s);
126 s += sstride;
127 }
128 }
129 }
130 }
131 }
132
133 template <class T>
134 static
qt_memrotate270_tiled(const T * src,int w,int h,int sstride,T * dest,int dstride)135 inline void qt_memrotate270_tiled(const T *src, int w, int h, int sstride, T *dest, int dstride)
136 {
137 sstride /= sizeof(T);
138 dstride /= sizeof(T);
139
140 const int pack = sizeof(quint32) / sizeof(T);
141 const int unaligned =
142 qMin(uint((quintptr(dest) & (sizeof(quint32)-1)) / sizeof(T)), uint(h));
143 const int restX = w % tileSize;
144 const int restY = (h - unaligned) % tileSize;
145 const int unoptimizedY = restY % pack;
146 const int numTilesX = w / tileSize + (restX > 0);
147 const int numTilesY = (h - unaligned) / tileSize + (restY >= pack);
148
149 for (int tx = 0; tx < numTilesX; ++tx) {
150 const int startx = tx * tileSize;
151 const int stopx = qMin(startx + tileSize, w);
152
153 if (unaligned) {
154 for (int x = startx; x < stopx; ++x) {
155 T *d = dest + x * dstride;
156 for (int y = h - 1; y >= h - unaligned; --y) {
157 *d++ = src[y * sstride + x];
158 }
159 }
160 }
161
162 for (int ty = 0; ty < numTilesY; ++ty) {
163 const int starty = h - 1 - unaligned - ty * tileSize;
164 const int stopy = qMax(starty - tileSize, unoptimizedY);
165
166 for (int x = startx; x < stopx; ++x) {
167 quint32 *d = reinterpret_cast<quint32*>(dest + x * dstride
168 + h - 1 - starty);
169 for (int y = starty; y >= stopy; y -= pack) {
170 quint32 c = src[y * sstride + x];
171 for (int i = 1; i < pack; ++i) {
172 const int shift = (sizeof(T) * 8 * i);
173 const T color = src[(y - i) * sstride + x];
174 c |= color << shift;
175 }
176 *d++ = c;
177 }
178 }
179 }
180 if (unoptimizedY) {
181 const int starty = unoptimizedY - 1;
182 for (int x = startx; x < stopx; ++x) {
183 T *d = dest + x * dstride + h - 1 - starty;
184 for (int y = starty; y >= 0; --y) {
185 *d++ = src[y * sstride + x];
186 }
187 }
188 }
189 }
190 }
191
192 template <class T>
193 static
qt_memrotate270_tiled_unpacked(const T * src,int w,int h,int sstride,T * dest,int dstride)194 inline void qt_memrotate270_tiled_unpacked(const T *src, int w, int h, int sstride, T *dest,
195 int dstride)
196 {
197 const int numTilesX = (w + tileSize - 1) / tileSize;
198 const int numTilesY = (h + tileSize - 1) / tileSize;
199
200 for (int tx = 0; tx < numTilesX; ++tx) {
201 const int startx = tx * tileSize;
202 const int stopx = qMin(startx + tileSize, w);
203
204 for (int ty = 0; ty < numTilesY; ++ty) {
205 const int starty = h - 1 - ty * tileSize;
206 const int stopy = qMax(starty - tileSize, 0);
207
208 for (int x = startx; x < stopx; ++x) {
209 T *d = (T*)((char*)dest + x * dstride) + h - 1 - starty;
210 const char *s = (const char*)(src + x) + starty * sstride;
211 for (int y = starty; y >= stopy; --y) {
212 *d++ = *(const T*)s;
213 s -= sstride;
214 }
215 }
216 }
217 }
218 }
219
220
221 template <class T>
222 static
qt_memrotate90_template(const T * src,int srcWidth,int srcHeight,int srcStride,T * dest,int dstStride)223 inline void qt_memrotate90_template(const T *src, int srcWidth, int srcHeight, int srcStride,
224 T *dest, int dstStride)
225 {
226 #if Q_BYTE_ORDER == Q_LITTLE_ENDIAN
227 // packed algorithm assumes little endian and that sizeof(quint32)/sizeof(T) is an integer
228 if (sizeof(quint32) % sizeof(T) == 0)
229 qt_memrotate90_tiled<T>(src, srcWidth, srcHeight, srcStride, dest, dstStride);
230 else
231 #endif
232 qt_memrotate90_tiled_unpacked<T>(src, srcWidth, srcHeight, srcStride, dest, dstStride);
233 }
234
235 template <>
qt_memrotate90_template(const quint32 * src,int w,int h,int sstride,quint32 * dest,int dstride)236 inline void qt_memrotate90_template<quint32>(const quint32 *src, int w, int h, int sstride, quint32 *dest, int dstride)
237 {
238 // packed algorithm doesn't have any benefit for quint32
239 qt_memrotate90_tiled_unpacked(src, w, h, sstride, dest, dstride);
240 }
241
242 template <>
qt_memrotate90_template(const quint64 * src,int w,int h,int sstride,quint64 * dest,int dstride)243 inline void qt_memrotate90_template<quint64>(const quint64 *src, int w, int h, int sstride, quint64 *dest, int dstride)
244 {
245 qt_memrotate90_tiled_unpacked(src, w, h, sstride, dest, dstride);
246 }
247
248 template <class T>
249 static
qt_memrotate180_template(const T * src,int w,int h,int sstride,T * dest,int dstride)250 inline void qt_memrotate180_template(const T *src, int w, int h, int sstride, T *dest, int dstride)
251 {
252 const char *s = (const char*)(src) + (h - 1) * sstride;
253 for (int dy = 0; dy < h; ++dy) {
254 T *d = reinterpret_cast<T*>((char *)(dest) + dy * dstride);
255 src = reinterpret_cast<const T*>(s);
256 for (int dx = 0; dx < w; ++dx) {
257 d[dx] = src[w - 1 - dx];
258 }
259 s -= sstride;
260 }
261 }
262
263 template <class T>
264 static
qt_memrotate270_template(const T * src,int srcWidth,int srcHeight,int srcStride,T * dest,int dstStride)265 inline void qt_memrotate270_template(const T *src, int srcWidth, int srcHeight, int srcStride,
266 T *dest, int dstStride)
267 {
268 #if Q_BYTE_ORDER == Q_LITTLE_ENDIAN
269 // packed algorithm assumes little endian and that sizeof(quint32)/sizeof(T) is an integer
270 if (sizeof(quint32) % sizeof(T) == 0)
271 qt_memrotate270_tiled<T>(src, srcWidth, srcHeight, srcStride, dest, dstStride);
272 else
273 #endif
274 qt_memrotate270_tiled_unpacked<T>(src, srcWidth, srcHeight, srcStride, dest, dstStride);
275 }
276
277 template <>
qt_memrotate270_template(const quint32 * src,int w,int h,int sstride,quint32 * dest,int dstride)278 inline void qt_memrotate270_template<quint32>(const quint32 *src, int w, int h, int sstride, quint32 *dest, int dstride)
279 {
280 // packed algorithm doesn't have any benefit for quint32
281 qt_memrotate270_tiled_unpacked(src, w, h, sstride, dest, dstride);
282 }
283
284 template <>
qt_memrotate270_template(const quint64 * src,int w,int h,int sstride,quint64 * dest,int dstride)285 inline void qt_memrotate270_template<quint64>(const quint64 *src, int w, int h, int sstride, quint64 *dest, int dstride)
286 {
287 qt_memrotate270_tiled_unpacked(src, w, h, sstride, dest, dstride);
288 }
289
290 #define QT_IMPL_MEMROTATE(type) \
291 Q_GUI_EXPORT void qt_memrotate90(const type *src, int w, int h, int sstride, \
292 type *dest, int dstride) \
293 { \
294 qt_memrotate90_template(src, w, h, sstride, dest, dstride); \
295 } \
296 Q_GUI_EXPORT void qt_memrotate180(const type *src, int w, int h, int sstride, \
297 type *dest, int dstride) \
298 { \
299 qt_memrotate180_template(src, w, h, sstride, dest, dstride); \
300 } \
301 Q_GUI_EXPORT void qt_memrotate270(const type *src, int w, int h, int sstride, \
302 type *dest, int dstride) \
303 { \
304 qt_memrotate270_template(src, w, h, sstride, dest, dstride); \
305 }
306
307 #define QT_IMPL_SIMPLE_MEMROTATE(type) \
308 Q_GUI_EXPORT void qt_memrotate90(const type *src, int w, int h, int sstride, \
309 type *dest, int dstride) \
310 { \
311 qt_memrotate90_tiled_unpacked(src, w, h, sstride, dest, dstride); \
312 } \
313 Q_GUI_EXPORT void qt_memrotate180(const type *src, int w, int h, int sstride, \
314 type *dest, int dstride) \
315 { \
316 qt_memrotate180_template(src, w, h, sstride, dest, dstride); \
317 } \
318 Q_GUI_EXPORT void qt_memrotate270(const type *src, int w, int h, int sstride, \
319 type *dest, int dstride) \
320 { \
321 qt_memrotate270_tiled_unpacked(src, w, h, sstride, dest, dstride); \
322 }
323
324 QT_IMPL_MEMROTATE(quint64)
QT_IMPL_MEMROTATE(quint32)325 QT_IMPL_MEMROTATE(quint32)
326 QT_IMPL_MEMROTATE(quint16)
327 QT_IMPL_MEMROTATE(quint24)
328 QT_IMPL_MEMROTATE(quint8)
329
330 void qt_memrotate90_8(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
331 {
332 qt_memrotate90(srcPixels, w, h, sbpl, destPixels, dbpl);
333 }
334
qt_memrotate180_8(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)335 void qt_memrotate180_8(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
336 {
337 qt_memrotate180(srcPixels, w, h, sbpl, destPixels, dbpl);
338 }
339
qt_memrotate270_8(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)340 void qt_memrotate270_8(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
341 {
342 qt_memrotate270(srcPixels, w, h, sbpl, destPixels, dbpl);
343 }
344
qt_memrotate90_16(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)345 void qt_memrotate90_16(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
346 {
347 qt_memrotate90((const ushort *)srcPixels, w, h, sbpl, (ushort *)destPixels, dbpl);
348 }
349
qt_memrotate180_16(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)350 void qt_memrotate180_16(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
351 {
352 qt_memrotate180((const ushort *)srcPixels, w, h, sbpl, (ushort *)destPixels, dbpl);
353 }
354
qt_memrotate270_16(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)355 void qt_memrotate270_16(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
356 {
357 qt_memrotate270((const ushort *)srcPixels, w, h, sbpl, (ushort *)destPixels, dbpl);
358 }
359
qt_memrotate90_24(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)360 void qt_memrotate90_24(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
361 {
362 qt_memrotate90((const quint24 *)srcPixels, w, h, sbpl, (quint24 *)destPixels, dbpl);
363 }
364
qt_memrotate180_24(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)365 void qt_memrotate180_24(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
366 {
367 qt_memrotate180((const quint24 *)srcPixels, w, h, sbpl, (quint24 *)destPixels, dbpl);
368 }
369
qt_memrotate270_24(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)370 void qt_memrotate270_24(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
371 {
372 qt_memrotate270((const quint24 *)srcPixels, w, h, sbpl, (quint24 *)destPixels, dbpl);
373 }
374
qt_memrotate90_32(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)375 void qt_memrotate90_32(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
376 {
377 qt_memrotate90((const uint *)srcPixels, w, h, sbpl, (uint *)destPixels, dbpl);
378 }
379
qt_memrotate180_32(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)380 void qt_memrotate180_32(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
381 {
382 qt_memrotate180((const uint *)srcPixels, w, h, sbpl, (uint *)destPixels, dbpl);
383 }
384
qt_memrotate270_32(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)385 void qt_memrotate270_32(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
386 {
387 qt_memrotate270((const uint *)srcPixels, w, h, sbpl, (uint *)destPixels, dbpl);
388 }
389
390
qt_memrotate90_64(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)391 void qt_memrotate90_64(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
392 {
393 qt_memrotate90((const quint64 *)srcPixels, w, h, sbpl, (quint64 *)destPixels, dbpl);
394 }
395
qt_memrotate180_64(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)396 void qt_memrotate180_64(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
397 {
398 qt_memrotate180((const quint64 *)srcPixels, w, h, sbpl, (quint64 *)destPixels, dbpl);
399 }
400
qt_memrotate270_64(const uchar * srcPixels,int w,int h,int sbpl,uchar * destPixels,int dbpl)401 void qt_memrotate270_64(const uchar *srcPixels, int w, int h, int sbpl, uchar *destPixels, int dbpl)
402 {
403 qt_memrotate270((const quint64 *)srcPixels, w, h, sbpl, (quint64 *)destPixels, dbpl);
404 }
405
406 MemRotateFunc qMemRotateFunctions[QPixelLayout::BPPCount][3] =
407 // 90, 180, 270
408 {
409 { nullptr, nullptr, nullptr }, // BPPNone,
410 { nullptr, nullptr, nullptr }, // BPP1MSB,
411 { nullptr, nullptr, nullptr }, // BPP1LSB,
412 { qt_memrotate90_8, qt_memrotate180_8, qt_memrotate270_8 }, // BPP8,
413 { qt_memrotate90_16, qt_memrotate180_16, qt_memrotate270_16 }, // BPP16,
414 { qt_memrotate90_24, qt_memrotate180_24, qt_memrotate270_24 }, // BPP24
415 { qt_memrotate90_32, qt_memrotate180_32, qt_memrotate270_32 }, // BPP32
416 { qt_memrotate90_64, qt_memrotate180_64, qt_memrotate270_64 }, // BPP64
417 };
418
419 QT_END_NAMESPACE
420