configure: fix sh_quote function
[libav.git] / libswscale / rgb2rgb.c
CommitLineData
fcfbc150 1/*
8a322796
DB
2 * software RGB to RGB converter
3 * pluralize by software PAL8 to RGB converter
4 * software YUV to YUV converter
5 * software YUV to RGB converter
6 * Written by Nick Kurshev.
7 * palette & YUV & runtime CPU stuff by Michael (michaelni@gmx.at)
4fadc2b4 8 *
d026b45e
DB
9 * This file is part of FFmpeg.
10 *
819ee683
DB
11 * FFmpeg is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public
13 * License as published by the Free Software Foundation; either
14 * version 2.1 of the License, or (at your option) any later version.
4fadc2b4 15 *
d026b45e 16 * FFmpeg is distributed in the hope that it will be useful,
4fadc2b4 17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
819ee683
DB
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
4fadc2b4 20 *
819ee683
DB
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with FFmpeg; if not, write to the Free Software
b19bcbaa 23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
a3aece93 24 */
b234ae81 25#include <inttypes.h>
b2d374c9 26#include "config.h"
83da2c6f
DB
27#include "libavutil/x86_cpu.h"
28#include "libavutil/bswap.h"
b234ae81 29#include "rgb2rgb.h"
700490a4 30#include "swscale.h"
94c4def2 31#include "swscale_internal.h"
a3aece93 32
8a322796 33#define FAST_BGR2YV12 // use 7-bit instead of 15-bit coefficients
21316f3c 34
6107059c
MN
35void (*rgb24tobgr32)(const uint8_t *src, uint8_t *dst, long src_size);
36void (*rgb24tobgr16)(const uint8_t *src, uint8_t *dst, long src_size);
37void (*rgb24tobgr15)(const uint8_t *src, uint8_t *dst, long src_size);
38void (*rgb32tobgr24)(const uint8_t *src, uint8_t *dst, long src_size);
30c48a0a
BF
39void (*rgb32to16)(const uint8_t *src, uint8_t *dst, long src_size);
40void (*rgb32to15)(const uint8_t *src, uint8_t *dst, long src_size);
41void (*rgb15to16)(const uint8_t *src, uint8_t *dst, long src_size);
6107059c 42void (*rgb15tobgr24)(const uint8_t *src, uint8_t *dst, long src_size);
30c48a0a
BF
43void (*rgb15to32)(const uint8_t *src, uint8_t *dst, long src_size);
44void (*rgb16to15)(const uint8_t *src, uint8_t *dst, long src_size);
6107059c 45void (*rgb16tobgr24)(const uint8_t *src, uint8_t *dst, long src_size);
30c48a0a 46void (*rgb16to32)(const uint8_t *src, uint8_t *dst, long src_size);
7f526efd 47void (*rgb24tobgr24)(const uint8_t *src, uint8_t *dst, long src_size);
6107059c
MN
48void (*rgb24to16)(const uint8_t *src, uint8_t *dst, long src_size);
49void (*rgb24to15)(const uint8_t *src, uint8_t *dst, long src_size);
7f526efd 50void (*rgb32tobgr32)(const uint8_t *src, uint8_t *dst, long src_size);
7f526efd
RD
51void (*rgb32tobgr16)(const uint8_t *src, uint8_t *dst, long src_size);
52void (*rgb32tobgr15)(const uint8_t *src, uint8_t *dst, long src_size);
700490a4
MN
53
54void (*yv12toyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
6e42e6c4
DB
55 long width, long height,
56 long lumStride, long chromStride, long dstStride);
caeaabe7 57void (*yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
6e42e6c4
DB
58 long width, long height,
59 long lumStride, long chromStride, long dstStride);
700490a4 60void (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
6e42e6c4
DB
61 long width, long height,
62 long lumStride, long chromStride, long dstStride);
a6100f39
BC
63void (*yuv422ptouyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
64 long width, long height,
65 long lumStride, long chromStride, long dstStride);
700490a4 66void (*yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
6e42e6c4
DB
67 long width, long height,
68 long lumStride, long chromStride, long srcStride);
700490a4 69void (*rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
6e42e6c4
DB
70 long width, long height,
71 long lumStride, long chromStride, long srcStride);
7f526efd 72void (*planar2x)(const uint8_t *src, uint8_t *dst, long width, long height,
6e42e6c4 73 long srcStride, long dstStride);
a959e247 74void (*interleaveBytes)(const uint8_t *src1, const uint8_t *src2, uint8_t *dst,
6e42e6c4
DB
75 long width, long height, long src1Stride,
76 long src2Stride, long dstStride);
700490a4 77void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
6e42e6c4
DB
78 uint8_t *dst1, uint8_t *dst2,
79 long width, long height,
80 long srcStride1, long srcStride2,
81 long dstStride1, long dstStride2);
700490a4 82void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
6e42e6c4
DB
83 uint8_t *dst,
84 long width, long height,
85 long srcStride1, long srcStride2,
86 long srcStride3, long dstStride);
0411072e 87void (*uyvytoyuv420)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
9b734d44
RP
88 long width, long height,
89 long lumStride, long chromStride, long srcStride);
0411072e 90void (*uyvytoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
9b734d44
RP
91 long width, long height,
92 long lumStride, long chromStride, long srcStride);
0411072e 93void (*yuyvtoyuv420)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
9b734d44
RP
94 long width, long height,
95 long lumStride, long chromStride, long srcStride);
0411072e 96void (*yuyvtoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst, const uint8_t *src,
9b734d44
RP
97 long width, long height,
98 long lumStride, long chromStride, long srcStride);
0411072e 99
700490a4 100
819ee683 101#if ARCH_X86
d334c7c2
RD
102DECLARE_ASM_CONST(8, uint64_t, mmx_null) = 0x0000000000000000ULL;
103DECLARE_ASM_CONST(8, uint64_t, mmx_one) = 0xFFFFFFFFFFFFFFFFULL;
104DECLARE_ASM_CONST(8, uint64_t, mask32b) = 0x000000FF000000FFULL;
105DECLARE_ASM_CONST(8, uint64_t, mask32g) = 0x0000FF000000FF00ULL;
106DECLARE_ASM_CONST(8, uint64_t, mask32r) = 0x00FF000000FF0000ULL;
f8a138be 107DECLARE_ASM_CONST(8, uint64_t, mask32a) = 0xFF000000FF000000ULL;
d334c7c2
RD
108DECLARE_ASM_CONST(8, uint64_t, mask32) = 0x00FFFFFF00FFFFFFULL;
109DECLARE_ASM_CONST(8, uint64_t, mask3216br) = 0x00F800F800F800F8ULL;
110DECLARE_ASM_CONST(8, uint64_t, mask3216g) = 0x0000FC000000FC00ULL;
111DECLARE_ASM_CONST(8, uint64_t, mask3215g) = 0x0000F8000000F800ULL;
112DECLARE_ASM_CONST(8, uint64_t, mul3216) = 0x2000000420000004ULL;
113DECLARE_ASM_CONST(8, uint64_t, mul3215) = 0x2000000820000008ULL;
114DECLARE_ASM_CONST(8, uint64_t, mask24b) = 0x00FF0000FF0000FFULL;
115DECLARE_ASM_CONST(8, uint64_t, mask24g) = 0xFF0000FF0000FF00ULL;
116DECLARE_ASM_CONST(8, uint64_t, mask24r) = 0x0000FF0000FF0000ULL;
117DECLARE_ASM_CONST(8, uint64_t, mask24l) = 0x0000000000FFFFFFULL;
118DECLARE_ASM_CONST(8, uint64_t, mask24h) = 0x0000FFFFFF000000ULL;
119DECLARE_ASM_CONST(8, uint64_t, mask24hh) = 0xffff000000000000ULL;
120DECLARE_ASM_CONST(8, uint64_t, mask24hhh) = 0xffffffff00000000ULL;
121DECLARE_ASM_CONST(8, uint64_t, mask24hhhh) = 0xffffffffffff0000ULL;
122DECLARE_ASM_CONST(8, uint64_t, mask15b) = 0x001F001F001F001FULL; /* 00000000 00011111 xxB */
123DECLARE_ASM_CONST(8, uint64_t, mask15rg) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000 RGx */
124DECLARE_ASM_CONST(8, uint64_t, mask15s) = 0xFFE0FFE0FFE0FFE0ULL;
125DECLARE_ASM_CONST(8, uint64_t, mask15g) = 0x03E003E003E003E0ULL;
126DECLARE_ASM_CONST(8, uint64_t, mask15r) = 0x7C007C007C007C00ULL;
0d9f3d85 127#define mask16b mask15b
d334c7c2
RD
128DECLARE_ASM_CONST(8, uint64_t, mask16g) = 0x07E007E007E007E0ULL;
129DECLARE_ASM_CONST(8, uint64_t, mask16r) = 0xF800F800F800F800ULL;
130DECLARE_ASM_CONST(8, uint64_t, red_16mask) = 0x0000f8000000f800ULL;
131DECLARE_ASM_CONST(8, uint64_t, green_16mask) = 0x000007e0000007e0ULL;
132DECLARE_ASM_CONST(8, uint64_t, blue_16mask) = 0x0000001f0000001fULL;
133DECLARE_ASM_CONST(8, uint64_t, red_15mask) = 0x00007c0000007c00ULL;
134DECLARE_ASM_CONST(8, uint64_t, green_15mask) = 0x000003e0000003e0ULL;
135DECLARE_ASM_CONST(8, uint64_t, blue_15mask) = 0x0000001f0000001fULL;
b63f641e 136#endif /* ARCH_X86 */
79811694 137
1de97d84
MN
138#define RGB2YUV_SHIFT 8
139#define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
140#define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
141#define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
142#define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
143#define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
144#define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
145#define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
146#define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
147#define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
148
8a322796
DB
149//Note: We have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW + MMX2 one.
150//plain C versions
1de97d84
MN
151#undef HAVE_MMX
152#undef HAVE_MMX2
f4406ec1 153#undef HAVE_AMD3DNOW
52075cde 154#undef HAVE_SSE2
b63f641e
AJ
155#define HAVE_MMX 0
156#define HAVE_MMX2 0
f4406ec1 157#define HAVE_AMD3DNOW 0
b63f641e 158#define HAVE_SSE2 0
1de97d84
MN
159#define RENAME(a) a ## _C
160#include "rgb2rgb_template.c"
161
819ee683 162#if ARCH_X86
1de97d84
MN
163
164//MMX versions
165#undef RENAME
b63f641e
AJ
166#undef HAVE_MMX
167#define HAVE_MMX 1
1de97d84
MN
168#define RENAME(a) a ## _MMX
169#include "rgb2rgb_template.c"
170
171//MMX2 versions
172#undef RENAME
b63f641e
AJ
173#undef HAVE_MMX2
174#define HAVE_MMX2 1
1de97d84
MN
175#define RENAME(a) a ## _MMX2
176#include "rgb2rgb_template.c"
177
178//3DNOW versions
179#undef RENAME
1de97d84 180#undef HAVE_MMX2
f4406ec1 181#undef HAVE_AMD3DNOW
b63f641e 182#define HAVE_MMX2 0
f4406ec1 183#define HAVE_AMD3DNOW 1
abdcd337 184#define RENAME(a) a ## _3DNOW
1de97d84
MN
185#include "rgb2rgb_template.c"
186
6e1c66bc 187#endif //ARCH_X86 || ARCH_X86_64
b238eb2e 188
a3aece93 189/*
8a322796 190 RGB15->RGB16 original by Strepto/Astral
a3aece93 191 ported to gcc & bugfixed : A'rpi
51da31f1 192 MMX2, 3DNOW optimization by Nick Kurshev
8a322796 193 32-bit C version, and and&add trick by Michael Niedermayer
a3aece93 194*/
fcfbc150 195
dd68318c
RP
196void sws_rgb2rgb_init(int flags)
197{
819ee683 198#if HAVE_MMX2 || HAVE_AMD3DNOW || HAVE_MMX
6e42e6c4
DB
199 if (flags & SWS_CPU_CAPS_MMX2)
200 rgb2rgb_init_MMX2();
201 else if (flags & SWS_CPU_CAPS_3DNOW)
202 rgb2rgb_init_3DNOW();
203 else if (flags & SWS_CPU_CAPS_MMX)
204 rgb2rgb_init_MMX();
205 else
f4406ec1 206#endif /* HAVE_MMX2 || HAVE_AMD3DNOW || HAVE_MMX */
6e42e6c4 207 rgb2rgb_init_C();
ac4d0aea 208}
700490a4 209
2b991422 210#if LIBSWSCALE_VERSION_MAJOR < 1
522ce957 211void palette8topacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
fcfbc150 212{
2b991422 213 sws_convertPalette8ToPacked32(src, dst, num_pixels, palette);
6c1baeb0 214}
6c1baeb0 215
522ce957 216void palette8topacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
9ce6584e 217{
2b991422 218 sws_convertPalette8ToPacked24(src, dst, num_pixels, palette);
6c1baeb0
AB
219}
220
fcfbc150 221/**
8a322796 222 * Palette is assumed to contain BGR16, see rgb32to16 to convert the palette.
fcfbc150 223 */
7f526efd 224void palette8torgb16(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
fcfbc150 225{
6e42e6c4
DB
226 long i;
227 for (i=0; i<num_pixels; i++)
774c386a 228 ((uint16_t *)dst)[i] = ((const uint16_t *)palette)[src[i]];
fcfbc150 229}
7f526efd 230void palette8tobgr16(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
6c1baeb0 231{
6e42e6c4
DB
232 long i;
233 for (i=0; i<num_pixels; i++)
6a01eab7 234 ((uint16_t *)dst)[i] = av_bswap16(((const uint16_t *)palette)[src[i]]);
6c1baeb0 235}
2b991422 236#endif
fcfbc150 237
6107059c 238void rgb32to24(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 239{
6e42e6c4
DB
240 long i;
241 long num_pixels = src_size >> 2;
dd68318c 242 for (i=0; i<num_pixels; i++) {
7d73d1c3
RP
243#if HAVE_BIGENDIAN
244 /* RGB32 (= A,B,G,R) -> BGR24 (= B,G,R) */
245 dst[3*i + 0] = src[4*i + 1];
246 dst[3*i + 1] = src[4*i + 2];
247 dst[3*i + 2] = src[4*i + 3];
248#else
249 dst[3*i + 0] = src[4*i + 2];
250 dst[3*i + 1] = src[4*i + 1];
251 dst[3*i + 2] = src[4*i + 0];
252#endif
6e42e6c4 253 }
ac4d0aea
MN
254}
255
6107059c 256void rgb24to32(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 257{
6e42e6c4 258 long i;
dd68318c 259 for (i=0; 3*i<src_size; i++) {
7d73d1c3
RP
260#if HAVE_BIGENDIAN
261 /* RGB24 (= R,G,B) -> BGR32 (= A,R,G,B) */
262 dst[4*i + 0] = 255;
263 dst[4*i + 1] = src[3*i + 0];
264 dst[4*i + 2] = src[3*i + 1];
265 dst[4*i + 3] = src[3*i + 2];
266#else
267 dst[4*i + 0] = src[3*i + 2];
268 dst[4*i + 1] = src[3*i + 1];
269 dst[4*i + 2] = src[3*i + 0];
270 dst[4*i + 3] = 255;
271#endif
6e42e6c4 272 }
ac4d0aea
MN
273}
274
7f526efd 275void rgb16tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 276{
6e42e6c4 277 const uint16_t *end;
72d6b610 278 uint8_t *d = dst;
774c386a 279 const uint16_t *s = (const uint16_t *)src;
6e42e6c4 280 end = s + src_size/2;
dd68318c 281 while (s < end) {
6e42e6c4
DB
282 register uint16_t bgr;
283 bgr = *s++;
7d73d1c3
RP
284#if HAVE_BIGENDIAN
285 *d++ = 255;
286 *d++ = (bgr&0x1F)<<3;
287 *d++ = (bgr&0x7E0)>>3;
288 *d++ = (bgr&0xF800)>>8;
289#else
290 *d++ = (bgr&0xF800)>>8;
291 *d++ = (bgr&0x7E0)>>3;
292 *d++ = (bgr&0x1F)<<3;
293 *d++ = 255;
294#endif
6e42e6c4 295 }
ac4d0aea
MN
296}
297
6107059c 298void rgb16to24(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 299{
6e42e6c4 300 const uint16_t *end;
72d6b610 301 uint8_t *d = dst;
6e42e6c4
DB
302 const uint16_t *s = (const uint16_t *)src;
303 end = s + src_size/2;
dd68318c 304 while (s < end) {
6e42e6c4
DB
305 register uint16_t bgr;
306 bgr = *s++;
307 *d++ = (bgr&0xF800)>>8;
308 *d++ = (bgr&0x7E0)>>3;
309 *d++ = (bgr&0x1F)<<3;
310 }
ac4d0aea
MN
311}
312
7f526efd 313void rgb16tobgr16(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 314{
6e42e6c4
DB
315 long i;
316 long num_pixels = src_size >> 1;
317
dd68318c 318 for (i=0; i<num_pixels; i++) {
7372e9bb
MN
319 unsigned rgb = ((const uint16_t*)src)[i];
320 ((uint16_t*)dst)[i] = (rgb>>11) | (rgb&0x7E0) | (rgb<<11);
6e42e6c4 321 }
ac4d0aea
MN
322}
323
7f526efd 324void rgb16tobgr15(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 325{
6e42e6c4
DB
326 long i;
327 long num_pixels = src_size >> 1;
328
dd68318c 329 for (i=0; i<num_pixels; i++) {
7372e9bb
MN
330 unsigned rgb = ((const uint16_t*)src)[i];
331 ((uint16_t*)dst)[i] = (rgb>>11) | ((rgb&0x7C0)>>1) | ((rgb&0x1F)<<10);
6e42e6c4 332 }
ac4d0aea
MN
333}
334
7f526efd 335void rgb15tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 336{
6e42e6c4 337 const uint16_t *end;
72d6b610 338 uint8_t *d = dst;
6e42e6c4
DB
339 const uint16_t *s = (const uint16_t *)src;
340 end = s + src_size/2;
dd68318c 341 while (s < end) {
6e42e6c4
DB
342 register uint16_t bgr;
343 bgr = *s++;
7d73d1c3
RP
344#if HAVE_BIGENDIAN
345 *d++ = 255;
346 *d++ = (bgr&0x1F)<<3;
347 *d++ = (bgr&0x3E0)>>2;
348 *d++ = (bgr&0x7C00)>>7;
349#else
350 *d++ = (bgr&0x7C00)>>7;
351 *d++ = (bgr&0x3E0)>>2;
352 *d++ = (bgr&0x1F)<<3;
353 *d++ = 255;
354#endif
6e42e6c4 355 }
ac4d0aea
MN
356}
357
6107059c 358void rgb15to24(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 359{
6e42e6c4 360 const uint16_t *end;
72d6b610 361 uint8_t *d = dst;
774c386a 362 const uint16_t *s = (const uint16_t *)src;
6e42e6c4 363 end = s + src_size/2;
dd68318c 364 while (s < end) {
6e42e6c4
DB
365 register uint16_t bgr;
366 bgr = *s++;
367 *d++ = (bgr&0x7C00)>>7;
368 *d++ = (bgr&0x3E0)>>2;
369 *d++ = (bgr&0x1F)<<3;
370 }
ac4d0aea
MN
371}
372
7f526efd 373void rgb15tobgr16(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 374{
6e42e6c4
DB
375 long i;
376 long num_pixels = src_size >> 1;
377
dd68318c 378 for (i=0; i<num_pixels; i++) {
7372e9bb
MN
379 unsigned rgb = ((const uint16_t*)src)[i];
380 ((uint16_t*)dst)[i] = ((rgb&0x7C00)>>10) | ((rgb&0x3E0)<<1) | (rgb<<11);
6e42e6c4 381 }
ac4d0aea
MN
382}
383
7f526efd 384void rgb15tobgr15(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 385{
6e42e6c4
DB
386 long i;
387 long num_pixels = src_size >> 1;
388
dd68318c 389 for (i=0; i<num_pixels; i++) {
7372e9bb
MN
390 unsigned br;
391 unsigned rgb = ((const uint16_t*)src)[i];
392 br = rgb&0x7c1F;
393 ((uint16_t*)dst)[i] = (br>>10) | (rgb&0x3E0) | (br<<10);
6e42e6c4 394 }
ac4d0aea
MN
395}
396
6107059c 397void bgr8torgb8(const uint8_t *src, uint8_t *dst, long src_size)
ac4d0aea 398{
6e42e6c4
DB
399 long i;
400 long num_pixels = src_size;
dd68318c 401 for (i=0; i<num_pixels; i++) {
6e42e6c4
DB
402 unsigned b,g,r;
403 register uint8_t rgb;
404 rgb = src[i];
405 r = (rgb&0x07);
406 g = (rgb&0x38)>>3;
407 b = (rgb&0xC0)>>6;
408 dst[i] = ((b<<1)&0x07) | ((g&0x07)<<3) | ((r&0x03)<<6);
409 }
ac4d0aea 410}
82e5f86b
SS
411
412#define DEFINE_SHUFFLE_BYTES(a, b, c, d) \
413void shuffle_bytes_##a##b##c##d(const uint8_t *src, uint8_t *dst, long src_size) \
414{ \
415 long i; \
416 \
417 for (i = 0; i < src_size; i+=4) { \
418 dst[i + 0] = src[i + a]; \
419 dst[i + 1] = src[i + b]; \
420 dst[i + 2] = src[i + c]; \
421 dst[i + 3] = src[i + d]; \
422 } \
423}
424
425DEFINE_SHUFFLE_BYTES(0, 3, 2, 1);
426DEFINE_SHUFFLE_BYTES(1, 2, 3, 0);
427DEFINE_SHUFFLE_BYTES(2, 1, 0, 3);
428DEFINE_SHUFFLE_BYTES(3, 0, 1, 2);
429DEFINE_SHUFFLE_BYTES(3, 2, 1, 0);
430