1 /*
2  * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include <inttypes.h>
22 #include "config.h"
23 #include "libswscale/swscale.h"
24 #include "libswscale/swscale_internal.h"
25 #include "libavutil/attributes.h"
26 #include "libavutil/avassert.h"
27 #include "libavutil/intreadwrite.h"
28 #include "libavutil/x86/cpu.h"
29 #include "libavutil/cpu.h"
30 #include "libavutil/mem_internal.h"
31 #include "libavutil/pixdesc.h"
32 
33 const DECLARE_ALIGNED(8, uint64_t, ff_dither4)[2] = {
34     0x0103010301030103LL,
35     0x0200020002000200LL,};
36 
37 const DECLARE_ALIGNED(8, uint64_t, ff_dither8)[2] = {
38     0x0602060206020602LL,
39     0x0004000400040004LL,};
40 
41 #if HAVE_INLINE_ASM
42 
43 #define DITHER1XBPP
44 
45 DECLARE_ASM_CONST(8, uint64_t, bF8)=       0xF8F8F8F8F8F8F8F8LL;
46 DECLARE_ASM_CONST(8, uint64_t, bFC)=       0xFCFCFCFCFCFCFCFCLL;
47 
48 DECLARE_ASM_ALIGNED(8, const uint64_t, ff_M24A)         = 0x00FF0000FF0000FFLL;
49 DECLARE_ASM_ALIGNED(8, const uint64_t, ff_M24B)         = 0xFF0000FF0000FF00LL;
50 DECLARE_ASM_ALIGNED(8, const uint64_t, ff_M24C)         = 0x0000FF0000FF0000LL;
51 
52 DECLARE_ASM_ALIGNED(8, const uint64_t, ff_bgr2YOffset)  = 0x1010101010101010ULL;
53 DECLARE_ASM_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;
54 DECLARE_ASM_ALIGNED(8, const uint64_t, ff_w1111)        = 0x0001000100010001ULL;
55 
56 
57 // MMXEXT versions
58 #if HAVE_MMXEXT_INLINE
59 #undef RENAME
60 #undef COMPILE_TEMPLATE_MMXEXT
61 #define COMPILE_TEMPLATE_MMXEXT 1
62 #define RENAME(a) a ## _mmxext
63 #include "swscale_template.c"
64 #endif
65 
ff_updateMMXDitherTables(SwsContext *c, int dstY)66 void ff_updateMMXDitherTables(SwsContext *c, int dstY)
67 {
68     const int dstH= c->dstH;
69     const int flags= c->flags;
70 
71     SwsPlane *lumPlane = &c->slice[c->numSlice-2].plane[0];
72     SwsPlane *chrUPlane = &c->slice[c->numSlice-2].plane[1];
73     SwsPlane *alpPlane = &c->slice[c->numSlice-2].plane[3];
74 
75     int hasAlpha = c->needAlpha;
76     int32_t *vLumFilterPos= c->vLumFilterPos;
77     int32_t *vChrFilterPos= c->vChrFilterPos;
78     int16_t *vLumFilter= c->vLumFilter;
79     int16_t *vChrFilter= c->vChrFilter;
80     int32_t *lumMmxFilter= c->lumMmxFilter;
81     int32_t *chrMmxFilter= c->chrMmxFilter;
82     int32_t av_unused *alpMmxFilter= c->alpMmxFilter;
83     const int vLumFilterSize= c->vLumFilterSize;
84     const int vChrFilterSize= c->vChrFilterSize;
85     const int chrDstY= dstY>>c->chrDstVSubSample;
86     const int firstLumSrcY= vLumFilterPos[dstY]; //First line needed as input
87     const int firstChrSrcY= vChrFilterPos[chrDstY]; //First line needed as input
88 
89     c->blueDither= ff_dither8[dstY&1];
90     if (c->dstFormat == AV_PIX_FMT_RGB555 || c->dstFormat == AV_PIX_FMT_BGR555)
91         c->greenDither= ff_dither8[dstY&1];
92     else
93         c->greenDither= ff_dither4[dstY&1];
94     c->redDither= ff_dither8[(dstY+1)&1];
95     if (dstY < dstH - 2) {
96         const int16_t **lumSrcPtr  = (const int16_t **)(void*) lumPlane->line + firstLumSrcY - lumPlane->sliceY;
97         const int16_t **chrUSrcPtr = (const int16_t **)(void*) chrUPlane->line + firstChrSrcY - chrUPlane->sliceY;
98         const int16_t **alpSrcPtr  = (CONFIG_SWSCALE_ALPHA && hasAlpha) ? (const int16_t **)(void*) alpPlane->line + firstLumSrcY - alpPlane->sliceY : NULL;
99 
100         int i;
101         if (firstLumSrcY < 0 || firstLumSrcY + vLumFilterSize > c->srcH) {
102             const int16_t **tmpY = (const int16_t **) lumPlane->tmp;
103 
104             int neg = -firstLumSrcY, i, end = FFMIN(c->srcH - firstLumSrcY, vLumFilterSize);
105             for (i = 0; i < neg;            i++)
106                 tmpY[i] = lumSrcPtr[neg];
107             for (     ; i < end;            i++)
108                 tmpY[i] = lumSrcPtr[i];
109             for (     ; i < vLumFilterSize; i++)
110                 tmpY[i] = tmpY[i-1];
111             lumSrcPtr = tmpY;
112 
113             if (alpSrcPtr) {
114                 const int16_t **tmpA = (const int16_t **) alpPlane->tmp;
115                 for (i = 0; i < neg;            i++)
116                     tmpA[i] = alpSrcPtr[neg];
117                 for (     ; i < end;            i++)
118                     tmpA[i] = alpSrcPtr[i];
119                 for (     ; i < vLumFilterSize; i++)
120                     tmpA[i] = tmpA[i - 1];
121                 alpSrcPtr = tmpA;
122             }
123         }
124         if (firstChrSrcY < 0 || firstChrSrcY + vChrFilterSize > c->chrSrcH) {
125             const int16_t **tmpU = (const int16_t **) chrUPlane->tmp;
126             int neg = -firstChrSrcY, i, end = FFMIN(c->chrSrcH - firstChrSrcY, vChrFilterSize);
127             for (i = 0; i < neg;            i++) {
128                 tmpU[i] = chrUSrcPtr[neg];
129             }
130             for (     ; i < end;            i++) {
131                 tmpU[i] = chrUSrcPtr[i];
132             }
133             for (     ; i < vChrFilterSize; i++) {
134                 tmpU[i] = tmpU[i - 1];
135             }
136             chrUSrcPtr = tmpU;
137         }
138 
139         if (flags & SWS_ACCURATE_RND) {
140             int s= APCK_SIZE / 8;
141             for (i=0; i<vLumFilterSize; i+=2) {
142                 *(const void**)&lumMmxFilter[s*i              ]= lumSrcPtr[i  ];
143                 *(const void**)&lumMmxFilter[s*i+APCK_PTR2/4  ]= lumSrcPtr[i+(vLumFilterSize>1)];
144                 lumMmxFilter[s*i+APCK_COEF/4  ]=
145                 lumMmxFilter[s*i+APCK_COEF/4+1]= vLumFilter[dstY*vLumFilterSize + i    ]
146                     + (vLumFilterSize>1 ? vLumFilter[dstY*vLumFilterSize + i + 1] * (1 << 16) : 0);
147                 if (CONFIG_SWSCALE_ALPHA && hasAlpha) {
148                     *(const void**)&alpMmxFilter[s*i              ]= alpSrcPtr[i  ];
149                     *(const void**)&alpMmxFilter[s*i+APCK_PTR2/4  ]= alpSrcPtr[i+(vLumFilterSize>1)];
150                     alpMmxFilter[s*i+APCK_COEF/4  ]=
151                     alpMmxFilter[s*i+APCK_COEF/4+1]= lumMmxFilter[s*i+APCK_COEF/4  ];
152                 }
153             }
154             for (i=0; i<vChrFilterSize; i+=2) {
155                 *(const void**)&chrMmxFilter[s*i              ]= chrUSrcPtr[i  ];
156                 *(const void**)&chrMmxFilter[s*i+APCK_PTR2/4  ]= chrUSrcPtr[i+(vChrFilterSize>1)];
157                 chrMmxFilter[s*i+APCK_COEF/4  ]=
158                 chrMmxFilter[s*i+APCK_COEF/4+1]= vChrFilter[chrDstY*vChrFilterSize + i    ]
159                     + (vChrFilterSize>1 ? vChrFilter[chrDstY*vChrFilterSize + i + 1] * (1 << 16) : 0);
160             }
161         } else {
162             for (i=0; i<vLumFilterSize; i++) {
163                 *(const void**)&lumMmxFilter[4*i+0]= lumSrcPtr[i];
164                 lumMmxFilter[4*i+2]=
165                 lumMmxFilter[4*i+3]=
166                 ((uint16_t)vLumFilter[dstY*vLumFilterSize + i])*0x10001U;
167                 if (CONFIG_SWSCALE_ALPHA && hasAlpha) {
168                     *(const void**)&alpMmxFilter[4*i+0]= alpSrcPtr[i];
169                     alpMmxFilter[4*i+2]=
170                     alpMmxFilter[4*i+3]= lumMmxFilter[4*i+2];
171                 }
172             }
173             for (i=0; i<vChrFilterSize; i++) {
174                 *(const void**)&chrMmxFilter[4*i+0]= chrUSrcPtr[i];
175                 chrMmxFilter[4*i+2]=
176                 chrMmxFilter[4*i+3]=
177                 ((uint16_t)vChrFilter[chrDstY*vChrFilterSize + i])*0x10001U;
178             }
179         }
180     }
181 }
182 #endif /* HAVE_INLINE_ASM */
183 
184 #define YUV2YUVX_FUNC_MMX(opt, step)  \
185 void ff_yuv2yuvX_ ##opt(const int16_t *filter, int filterSize, int srcOffset, \
186                            uint8_t *dest, int dstW,  \
187                            const uint8_t *dither, int offset); \
188 static void yuv2yuvX_ ##opt(const int16_t *filter, int filterSize, \
189                            const int16_t **src, uint8_t *dest, int dstW, \
190                            const uint8_t *dither, int offset) \
191 { \
192     if(dstW > 0) \
193         ff_yuv2yuvX_ ##opt(filter, filterSize - 1, 0, dest - offset, dstW + offset, dither, offset); \
194     return; \
195 }
196 
197 #define YUV2YUVX_FUNC(opt, step)  \
198 void ff_yuv2yuvX_ ##opt(const int16_t *filter, int filterSize, int srcOffset, \
199                            uint8_t *dest, int dstW,  \
200                            const uint8_t *dither, int offset); \
201 static void yuv2yuvX_ ##opt(const int16_t *filter, int filterSize, \
202                            const int16_t **src, uint8_t *dest, int dstW, \
203                            const uint8_t *dither, int offset) \
204 { \
205     int remainder = (dstW % step); \
206     int pixelsProcessed = dstW - remainder; \
207     if(((uintptr_t)dest) & 15){ \
208         yuv2yuvX_mmx(filter, filterSize, src, dest, dstW, dither, offset); \
209         return; \
210     } \
211     if(pixelsProcessed > 0) \
212         ff_yuv2yuvX_ ##opt(filter, filterSize - 1, 0, dest - offset, pixelsProcessed + offset, dither, offset); \
213     if(remainder > 0){ \
214       ff_yuv2yuvX_mmx(filter, filterSize - 1, pixelsProcessed, dest - offset, pixelsProcessed + remainder + offset, dither, offset); \
215     } \
216     return; \
217 }
218 
219 #if HAVE_MMX_EXTERNAL
220 YUV2YUVX_FUNC_MMX(mmx, 16)
221 #endif
222 #if HAVE_MMXEXT_EXTERNAL
223 YUV2YUVX_FUNC_MMX(mmxext, 16)
224 #endif
225 #if HAVE_SSE3_EXTERNAL
226 YUV2YUVX_FUNC(sse3, 32)
227 #endif
228 #if HAVE_AVX2_EXTERNAL
229 YUV2YUVX_FUNC(avx2, 64)
230 #endif
231 
232 #define SCALE_FUNC(filter_n, from_bpc, to_bpc, opt) \
233 void ff_hscale ## from_bpc ## to ## to_bpc ## _ ## filter_n ## _ ## opt( \
234                                                 SwsContext *c, int16_t *data, \
235                                                 int dstW, const uint8_t *src, \
236                                                 const int16_t *filter, \
237                                                 const int32_t *filterPos, int filterSize)
238 
239 #define SCALE_FUNCS(filter_n, opt) \
240     SCALE_FUNC(filter_n,  8, 15, opt); \
241     SCALE_FUNC(filter_n,  9, 15, opt); \
242     SCALE_FUNC(filter_n, 10, 15, opt); \
243     SCALE_FUNC(filter_n, 12, 15, opt); \
244     SCALE_FUNC(filter_n, 14, 15, opt); \
245     SCALE_FUNC(filter_n, 16, 15, opt); \
246     SCALE_FUNC(filter_n,  8, 19, opt); \
247     SCALE_FUNC(filter_n,  9, 19, opt); \
248     SCALE_FUNC(filter_n, 10, 19, opt); \
249     SCALE_FUNC(filter_n, 12, 19, opt); \
250     SCALE_FUNC(filter_n, 14, 19, opt); \
251     SCALE_FUNC(filter_n, 16, 19, opt)
252 
253 #define SCALE_FUNCS_MMX(opt) \
254     SCALE_FUNCS(4, opt); \
255     SCALE_FUNCS(8, opt); \
256     SCALE_FUNCS(X, opt)
257 
258 #define SCALE_FUNCS_SSE(opt) \
259     SCALE_FUNCS(4, opt); \
260     SCALE_FUNCS(8, opt); \
261     SCALE_FUNCS(X4, opt); \
262     SCALE_FUNCS(X8, opt)
263 
264 SCALE_FUNCS_SSE(sse2);
265 SCALE_FUNCS_SSE(ssse3);
266 SCALE_FUNCS_SSE(sse4);
267 
268 SCALE_FUNC(4, 8, 15, avx2);
269 SCALE_FUNC(X4, 8, 15, avx2);
270 
271 #define VSCALEX_FUNC(size, opt) \
272 void ff_yuv2planeX_ ## size ## _ ## opt(const int16_t *filter, int filterSize, \
273                                         const int16_t **src, uint8_t *dest, int dstW, \
274                                         const uint8_t *dither, int offset)
275 #define VSCALEX_FUNCS(opt) \
276     VSCALEX_FUNC(8,  opt); \
277     VSCALEX_FUNC(9,  opt); \
278     VSCALEX_FUNC(10, opt)
279 
280 VSCALEX_FUNC(8, mmxext);
281 VSCALEX_FUNCS(sse2);
282 VSCALEX_FUNCS(sse4);
283 VSCALEX_FUNC(16, sse4);
284 VSCALEX_FUNCS(avx);
285 
286 #define VSCALE_FUNC(size, opt) \
287 void ff_yuv2plane1_ ## size ## _ ## opt(const int16_t *src, uint8_t *dst, int dstW, \
288                                         const uint8_t *dither, int offset)
289 #define VSCALE_FUNCS(opt1, opt2) \
290     VSCALE_FUNC(8,  opt1); \
291     VSCALE_FUNC(9,  opt2); \
292     VSCALE_FUNC(10, opt2); \
293     VSCALE_FUNC(16, opt1)
294 
295 VSCALE_FUNCS(sse2, sse2);
296 VSCALE_FUNC(16, sse4);
297 VSCALE_FUNCS(avx, avx);
298 
299 #define INPUT_Y_FUNC(fmt, opt) \
300 void ff_ ## fmt ## ToY_  ## opt(uint8_t *dst, const uint8_t *src, \
301                                 const uint8_t *unused1, const uint8_t *unused2, \
302                                 int w, uint32_t *unused)
303 #define INPUT_UV_FUNC(fmt, opt) \
304 void ff_ ## fmt ## ToUV_ ## opt(uint8_t *dstU, uint8_t *dstV, \
305                                 const uint8_t *unused0, \
306                                 const uint8_t *src1, \
307                                 const uint8_t *src2, \
308                                 int w, uint32_t *unused)
309 #define INPUT_FUNC(fmt, opt) \
310     INPUT_Y_FUNC(fmt, opt); \
311     INPUT_UV_FUNC(fmt, opt)
312 #define INPUT_FUNCS(opt) \
313     INPUT_FUNC(uyvy, opt); \
314     INPUT_FUNC(yuyv, opt); \
315     INPUT_UV_FUNC(nv12, opt); \
316     INPUT_UV_FUNC(nv21, opt); \
317     INPUT_FUNC(rgba, opt); \
318     INPUT_FUNC(bgra, opt); \
319     INPUT_FUNC(argb, opt); \
320     INPUT_FUNC(abgr, opt); \
321     INPUT_FUNC(rgb24, opt); \
322     INPUT_FUNC(bgr24, opt)
323 
324 INPUT_FUNCS(sse2);
325 INPUT_FUNCS(ssse3);
326 INPUT_FUNCS(avx);
327 
328 #if ARCH_X86_64
329 #define YUV2NV_DECL(fmt, opt) \
330 void ff_yuv2 ## fmt ## cX_ ## opt(enum AVPixelFormat format, const uint8_t *dither, \
331                                   const int16_t *filter, int filterSize, \
332                                   const int16_t **u, const int16_t **v, \
333                                   uint8_t *dst, int dstWidth)
334 
335 YUV2NV_DECL(nv12, avx2);
336 YUV2NV_DECL(nv21, avx2);
337 
338 #define YUV2GBRP_FN_DECL(fmt, opt)                                                      \
339 void ff_yuv2##fmt##_full_X_ ##opt(SwsContext *c, const int16_t *lumFilter,           \
340                                  const int16_t **lumSrcx, int lumFilterSize,         \
341                                  const int16_t *chrFilter, const int16_t **chrUSrcx, \
342                                  const int16_t **chrVSrcx, int chrFilterSize,        \
343                                  const int16_t **alpSrcx, uint8_t **dest,            \
344                                  int dstW, int y)
345 
346 #define YUV2GBRP_DECL(opt)     \
347 YUV2GBRP_FN_DECL(gbrp,       opt); \
348 YUV2GBRP_FN_DECL(gbrap,      opt); \
349 YUV2GBRP_FN_DECL(gbrp9le,    opt); \
350 YUV2GBRP_FN_DECL(gbrp10le,   opt); \
351 YUV2GBRP_FN_DECL(gbrap10le,  opt); \
352 YUV2GBRP_FN_DECL(gbrp12le,   opt); \
353 YUV2GBRP_FN_DECL(gbrap12le,  opt); \
354 YUV2GBRP_FN_DECL(gbrp14le,   opt); \
355 YUV2GBRP_FN_DECL(gbrp16le,   opt); \
356 YUV2GBRP_FN_DECL(gbrap16le,  opt); \
357 YUV2GBRP_FN_DECL(gbrpf32le,  opt); \
358 YUV2GBRP_FN_DECL(gbrapf32le, opt); \
359 YUV2GBRP_FN_DECL(gbrp9be,    opt); \
360 YUV2GBRP_FN_DECL(gbrp10be,   opt); \
361 YUV2GBRP_FN_DECL(gbrap10be,  opt); \
362 YUV2GBRP_FN_DECL(gbrp12be,   opt); \
363 YUV2GBRP_FN_DECL(gbrap12be,  opt); \
364 YUV2GBRP_FN_DECL(gbrp14be,   opt); \
365 YUV2GBRP_FN_DECL(gbrp16be,   opt); \
366 YUV2GBRP_FN_DECL(gbrap16be,  opt); \
367 YUV2GBRP_FN_DECL(gbrpf32be,  opt); \
368 YUV2GBRP_FN_DECL(gbrapf32be, opt)
369 
370 YUV2GBRP_DECL(sse2);
371 YUV2GBRP_DECL(sse4);
372 YUV2GBRP_DECL(avx2);
373 
374 #define INPUT_PLANAR_RGB_Y_FN_DECL(fmt, opt)                               \
375 void ff_planar_##fmt##_to_y_##opt(uint8_t *dst,                            \
376                            const uint8_t *src[4], int w, int32_t *rgb2yuv)
377 
378 #define INPUT_PLANAR_RGB_UV_FN_DECL(fmt, opt)                              \
379 void ff_planar_##fmt##_to_uv_##opt(uint8_t *dstU, uint8_t *dstV,           \
380                            const uint8_t *src[4], int w, int32_t *rgb2yuv)
381 
382 #define INPUT_PLANAR_RGB_A_FN_DECL(fmt, opt)                               \
383 void ff_planar_##fmt##_to_a_##opt(uint8_t *dst,                            \
384                            const uint8_t *src[4], int w, int32_t *rgb2yuv)
385 
386 
387 #define INPUT_PLANAR_RGBXX_A_DECL(fmt, opt) \
388 INPUT_PLANAR_RGB_A_FN_DECL(fmt##le,  opt);  \
389 INPUT_PLANAR_RGB_A_FN_DECL(fmt##be,  opt)
390 
391 #define INPUT_PLANAR_RGBXX_Y_DECL(fmt, opt) \
392 INPUT_PLANAR_RGB_Y_FN_DECL(fmt##le,  opt);  \
393 INPUT_PLANAR_RGB_Y_FN_DECL(fmt##be,  opt)
394 
395 #define INPUT_PLANAR_RGBXX_UV_DECL(fmt, opt) \
396 INPUT_PLANAR_RGB_UV_FN_DECL(fmt##le,  opt);  \
397 INPUT_PLANAR_RGB_UV_FN_DECL(fmt##be,  opt)
398 
399 #define INPUT_PLANAR_RGBXX_YUVA_DECL(fmt, opt) \
400 INPUT_PLANAR_RGBXX_Y_DECL(fmt,  opt);          \
401 INPUT_PLANAR_RGBXX_UV_DECL(fmt, opt);          \
402 INPUT_PLANAR_RGBXX_A_DECL(fmt,  opt)
403 
404 #define INPUT_PLANAR_RGBXX_YUV_DECL(fmt, opt) \
405 INPUT_PLANAR_RGBXX_Y_DECL(fmt,  opt);         \
406 INPUT_PLANAR_RGBXX_UV_DECL(fmt, opt)
407 
408 #define INPUT_PLANAR_RGBXX_UVA_DECL(fmt, opt) \
409 INPUT_PLANAR_RGBXX_UV_DECL(fmt, opt);         \
410 INPUT_PLANAR_RGBXX_A_DECL(fmt,  opt)
411 
412 #define INPUT_PLANAR_RGB_A_ALL_DECL(opt) \
413 INPUT_PLANAR_RGB_A_FN_DECL(rgb,   opt);  \
414 INPUT_PLANAR_RGBXX_A_DECL(rgb10,  opt);  \
415 INPUT_PLANAR_RGBXX_A_DECL(rgb12,  opt);  \
416 INPUT_PLANAR_RGBXX_A_DECL(rgb16,  opt);  \
417 INPUT_PLANAR_RGBXX_A_DECL(rgbf32, opt)
418 
419 #define INPUT_PLANAR_RGB_Y_ALL_DECL(opt) \
420 INPUT_PLANAR_RGB_Y_FN_DECL(rgb,   opt);  \
421 INPUT_PLANAR_RGBXX_Y_DECL(rgb9,   opt);  \
422 INPUT_PLANAR_RGBXX_Y_DECL(rgb10,  opt);  \
423 INPUT_PLANAR_RGBXX_Y_DECL(rgb12,  opt);  \
424 INPUT_PLANAR_RGBXX_Y_DECL(rgb14,  opt);  \
425 INPUT_PLANAR_RGBXX_Y_DECL(rgb16,  opt);  \
426 INPUT_PLANAR_RGBXX_Y_DECL(rgbf32, opt)
427 
428 #define INPUT_PLANAR_RGB_UV_ALL_DECL(opt) \
429 INPUT_PLANAR_RGB_UV_FN_DECL(rgb,    opt); \
430 INPUT_PLANAR_RGBXX_UV_DECL(rgb9,   opt);  \
431 INPUT_PLANAR_RGBXX_UV_DECL(rgb10,  opt);  \
432 INPUT_PLANAR_RGBXX_UV_DECL(rgb12,  opt);  \
433 INPUT_PLANAR_RGBXX_UV_DECL(rgb14,  opt);  \
434 INPUT_PLANAR_RGBXX_UV_DECL(rgb16,  opt);  \
435 INPUT_PLANAR_RGBXX_UV_DECL(rgbf32, opt)
436 
437 INPUT_PLANAR_RGBXX_Y_DECL(rgbf32, sse2);
438 INPUT_PLANAR_RGB_UV_ALL_DECL(sse2);
439 INPUT_PLANAR_RGB_A_ALL_DECL(sse2);
440 
441 INPUT_PLANAR_RGB_Y_ALL_DECL(sse4);
442 INPUT_PLANAR_RGB_UV_ALL_DECL(sse4);
443 INPUT_PLANAR_RGBXX_A_DECL(rgbf32, sse4);
444 
445 INPUT_PLANAR_RGB_Y_ALL_DECL(avx2);
446 INPUT_PLANAR_RGB_UV_ALL_DECL(avx2);
447 INPUT_PLANAR_RGB_A_ALL_DECL(avx2);
448 #endif
449 
ff_sws_init_swscale_x86(SwsContext *c)450 av_cold void ff_sws_init_swscale_x86(SwsContext *c)
451 {
452     int cpu_flags = av_get_cpu_flags();
453 
454 #if HAVE_MMXEXT_INLINE
455     if (INLINE_MMXEXT(cpu_flags))
456         sws_init_swscale_mmxext(c);
457 #endif
458     if(c->use_mmx_vfilter && !(c->flags & SWS_ACCURATE_RND)) {
459 #if HAVE_MMXEXT_EXTERNAL
460         if (EXTERNAL_MMXEXT(cpu_flags))
461             c->yuv2planeX = yuv2yuvX_mmxext;
462 #endif
463 #if HAVE_SSE3_EXTERNAL
464         if (EXTERNAL_SSE3(cpu_flags))
465             c->yuv2planeX = yuv2yuvX_sse3;
466 #endif
467 #if HAVE_AVX2_EXTERNAL
468         if (EXTERNAL_AVX2_FAST(cpu_flags))
469             c->yuv2planeX = yuv2yuvX_avx2;
470 #endif
471     }
472 #if ARCH_X86_32 && !HAVE_ALIGNED_STACK
473     // The better yuv2planeX_8 functions need aligned stack on x86-32,
474     // so we use MMXEXT in this case if they are not available.
475     if (EXTERNAL_MMXEXT(cpu_flags)) {
476         if (c->dstBpc == 8 && !c->use_mmx_vfilter)
477             c->yuv2planeX = ff_yuv2planeX_8_mmxext;
478     }
479 #endif /* ARCH_X86_32 && !HAVE_ALIGNED_STACK */
480 
481 #define ASSIGN_SCALE_FUNC2(hscalefn, filtersize, opt1, opt2) do { \
482     if (c->srcBpc == 8) { \
483         hscalefn = c->dstBpc <= 14 ? ff_hscale8to15_ ## filtersize ## _ ## opt2 : \
484                                      ff_hscale8to19_ ## filtersize ## _ ## opt1; \
485     } else if (c->srcBpc == 9) { \
486         hscalefn = c->dstBpc <= 14 ? ff_hscale9to15_ ## filtersize ## _ ## opt2 : \
487                                      ff_hscale9to19_ ## filtersize ## _ ## opt1; \
488     } else if (c->srcBpc == 10) { \
489         hscalefn = c->dstBpc <= 14 ? ff_hscale10to15_ ## filtersize ## _ ## opt2 : \
490                                      ff_hscale10to19_ ## filtersize ## _ ## opt1; \
491     } else if (c->srcBpc == 12) { \
492         hscalefn = c->dstBpc <= 14 ? ff_hscale12to15_ ## filtersize ## _ ## opt2 : \
493                                      ff_hscale12to19_ ## filtersize ## _ ## opt1; \
494     } else if (c->srcBpc == 14 || ((c->srcFormat==AV_PIX_FMT_PAL8||isAnyRGB(c->srcFormat)) && av_pix_fmt_desc_get(c->srcFormat)->comp[0].depth<16)) { \
495         hscalefn = c->dstBpc <= 14 ? ff_hscale14to15_ ## filtersize ## _ ## opt2 : \
496                                      ff_hscale14to19_ ## filtersize ## _ ## opt1; \
497     } else { /* c->srcBpc == 16 */ \
498         av_assert0(c->srcBpc == 16);\
499         hscalefn = c->dstBpc <= 14 ? ff_hscale16to15_ ## filtersize ## _ ## opt2 : \
500                                      ff_hscale16to19_ ## filtersize ## _ ## opt1; \
501     } \
502 } while (0)
503 #define ASSIGN_VSCALEX_FUNC(vscalefn, opt, do_16_case, condition_8bit) \
504 switch(c->dstBpc){ \
505     case 16:                          do_16_case;                          break; \
506     case 10: if (!isBE(c->dstFormat) && !isSemiPlanarYUV(c->dstFormat)) vscalefn = ff_yuv2planeX_10_ ## opt; break; \
507     case 9:  if (!isBE(c->dstFormat)) vscalefn = ff_yuv2planeX_9_  ## opt; break; \
508     case 8: if ((condition_8bit) && !c->use_mmx_vfilter) vscalefn = ff_yuv2planeX_8_  ## opt; break; \
509     }
510 #define ASSIGN_VSCALE_FUNC(vscalefn, opt) \
511     switch(c->dstBpc){ \
512     case 16: if (!isBE(c->dstFormat)) vscalefn = ff_yuv2plane1_16_ ## opt; break; \
513     case 10: if (!isBE(c->dstFormat) && !isSemiPlanarYUV(c->dstFormat)) vscalefn = ff_yuv2plane1_10_ ## opt; break; \
514     case 9:  if (!isBE(c->dstFormat)) vscalefn = ff_yuv2plane1_9_  ## opt;  break; \
515     case 8:                           vscalefn = ff_yuv2plane1_8_  ## opt;  break; \
516     default: av_assert0(c->dstBpc>8); \
517     }
518 #define case_rgb(x, X, opt) \
519         case AV_PIX_FMT_ ## X: \
520             c->lumToYV12 = ff_ ## x ## ToY_ ## opt; \
521             if (!c->chrSrcHSubSample) \
522                 c->chrToYV12 = ff_ ## x ## ToUV_ ## opt; \
523             break
524 #define ASSIGN_SSE_SCALE_FUNC(hscalefn, filtersize, opt1, opt2) \
525     switch (filtersize) { \
526     case 4:  ASSIGN_SCALE_FUNC2(hscalefn, 4, opt1, opt2); break; \
527     case 8:  ASSIGN_SCALE_FUNC2(hscalefn, 8, opt1, opt2); break; \
528     default: if (filtersize & 4) ASSIGN_SCALE_FUNC2(hscalefn, X4, opt1, opt2); \
529              else                ASSIGN_SCALE_FUNC2(hscalefn, X8, opt1, opt2); \
530              break; \
531     }
532     if (EXTERNAL_SSE2(cpu_flags)) {
533         ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, sse2, sse2);
534         ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, sse2, sse2);
535         ASSIGN_VSCALEX_FUNC(c->yuv2planeX, sse2, ,
536                             HAVE_ALIGNED_STACK || ARCH_X86_64);
537         ASSIGN_VSCALE_FUNC(c->yuv2plane1, sse2);
538 
539         switch (c->srcFormat) {
540         case AV_PIX_FMT_YA8:
541             c->lumToYV12 = ff_yuyvToY_sse2;
542             if (c->needAlpha)
543                 c->alpToYV12 = ff_uyvyToY_sse2;
544             break;
545         case AV_PIX_FMT_YUYV422:
546             c->lumToYV12 = ff_yuyvToY_sse2;
547             c->chrToYV12 = ff_yuyvToUV_sse2;
548             break;
549         case AV_PIX_FMT_UYVY422:
550             c->lumToYV12 = ff_uyvyToY_sse2;
551             c->chrToYV12 = ff_uyvyToUV_sse2;
552             break;
553         case AV_PIX_FMT_NV12:
554             c->chrToYV12 = ff_nv12ToUV_sse2;
555             break;
556         case AV_PIX_FMT_NV21:
557             c->chrToYV12 = ff_nv21ToUV_sse2;
558             break;
559         case_rgb(rgb24, RGB24, sse2);
560         case_rgb(bgr24, BGR24, sse2);
561         case_rgb(bgra,  BGRA,  sse2);
562         case_rgb(rgba,  RGBA,  sse2);
563         case_rgb(abgr,  ABGR,  sse2);
564         case_rgb(argb,  ARGB,  sse2);
565         default:
566             break;
567         }
568     }
569     if (EXTERNAL_SSSE3(cpu_flags)) {
570         ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, ssse3, ssse3);
571         ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, ssse3, ssse3);
572         switch (c->srcFormat) {
573         case_rgb(rgb24, RGB24, ssse3);
574         case_rgb(bgr24, BGR24, ssse3);
575         default:
576             break;
577         }
578     }
579     if (EXTERNAL_SSE4(cpu_flags)) {
580         /* Xto15 don't need special sse4 functions */
581         ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, sse4, ssse3);
582         ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, sse4, ssse3);
583         ASSIGN_VSCALEX_FUNC(c->yuv2planeX, sse4,
584                             if (!isBE(c->dstFormat)) c->yuv2planeX = ff_yuv2planeX_16_sse4,
585                             HAVE_ALIGNED_STACK || ARCH_X86_64);
586         if (c->dstBpc == 16 && !isBE(c->dstFormat))
587             c->yuv2plane1 = ff_yuv2plane1_16_sse4;
588     }
589 
590     if (EXTERNAL_AVX(cpu_flags)) {
591         ASSIGN_VSCALEX_FUNC(c->yuv2planeX, avx, ,
592                             HAVE_ALIGNED_STACK || ARCH_X86_64);
593         ASSIGN_VSCALE_FUNC(c->yuv2plane1, avx);
594 
595         switch (c->srcFormat) {
596         case AV_PIX_FMT_YUYV422:
597             c->chrToYV12 = ff_yuyvToUV_avx;
598             break;
599         case AV_PIX_FMT_UYVY422:
600             c->chrToYV12 = ff_uyvyToUV_avx;
601             break;
602         case AV_PIX_FMT_NV12:
603             c->chrToYV12 = ff_nv12ToUV_avx;
604             break;
605         case AV_PIX_FMT_NV21:
606             c->chrToYV12 = ff_nv21ToUV_avx;
607             break;
608         case_rgb(rgb24, RGB24, avx);
609         case_rgb(bgr24, BGR24, avx);
610         case_rgb(bgra,  BGRA,  avx);
611         case_rgb(rgba,  RGBA,  avx);
612         case_rgb(abgr,  ABGR,  avx);
613         case_rgb(argb,  ARGB,  avx);
614         default:
615             break;
616         }
617     }
618 
619 #if ARCH_X86_64
620 #define ASSIGN_AVX2_SCALE_FUNC(hscalefn, filtersize) \
621     switch (filtersize) { \
622     case 4:  hscalefn = ff_hscale8to15_4_avx2; break; \
623     default:  hscalefn = ff_hscale8to15_X4_avx2; break; \
624              break; \
625     }
626 
627     if (EXTERNAL_AVX2_FAST(cpu_flags) && !(cpu_flags & AV_CPU_FLAG_SLOW_GATHER)) {
628         if ((c->srcBpc == 8) && (c->dstBpc <= 14)) {
629             if (c->chrDstW % 16 == 0)
630                 ASSIGN_AVX2_SCALE_FUNC(c->hcScale, c->hChrFilterSize);
631             if (c->dstW % 16 == 0)
632                 ASSIGN_AVX2_SCALE_FUNC(c->hyScale, c->hLumFilterSize);
633         }
634     }
635 
636     if (EXTERNAL_AVX2_FAST(cpu_flags)) {
637         switch (c->dstFormat) {
638         case AV_PIX_FMT_NV12:
639         case AV_PIX_FMT_NV24:
640             c->yuv2nv12cX = ff_yuv2nv12cX_avx2;
641             break;
642         case AV_PIX_FMT_NV21:
643         case AV_PIX_FMT_NV42:
644             c->yuv2nv12cX = ff_yuv2nv21cX_avx2;
645             break;
646         default:
647             break;
648         }
649     }
650 
651 
652 #define INPUT_PLANER_RGB_A_FUNC_CASE(fmt, name, opt)                  \
653         case fmt:                                                     \
654             c->readAlpPlanar = ff_planar_##name##_to_a_##opt;
655 
656 #define INPUT_PLANER_RGBA_YUV_FUNC_CASE(rgb_fmt, rgba_fmt, name, opt) \
657         case rgba_fmt:                                                \
658         case rgb_fmt:                                                 \
659             c->readLumPlanar = ff_planar_##name##_to_y_##opt;         \
660             c->readChrPlanar = ff_planar_##name##_to_uv_##opt;        \
661             break;
662 
663 #define INPUT_PLANER_RGB_YUV_FUNC_CASE(fmt, name, opt)                \
664         case fmt:                                                     \
665             c->readLumPlanar = ff_planar_##name##_to_y_##opt;         \
666             c->readChrPlanar = ff_planar_##name##_to_uv_##opt;        \
667             break;
668 
669 #define INPUT_PLANER_RGB_UV_FUNC_CASE(fmt, name, opt)                 \
670         case fmt:                                                     \
671             c->readChrPlanar = ff_planar_##name##_to_uv_##opt;        \
672             break;
673 
674 #define INPUT_PLANER_RGBAXX_YUVA_FUNC_CASE(rgb_fmt, rgba_fmt, name, opt) \
675         INPUT_PLANER_RGB_A_FUNC_CASE(rgba_fmt##LE,  name##le, opt)       \
676         INPUT_PLANER_RGB_YUV_FUNC_CASE(rgb_fmt##LE, name##le, opt)       \
677         INPUT_PLANER_RGB_A_FUNC_CASE(rgba_fmt##BE,  name##be, opt)       \
678         INPUT_PLANER_RGB_YUV_FUNC_CASE(rgb_fmt##BE, name##be, opt)
679 
680 #define INPUT_PLANER_RGBAXX_UVA_FUNC_CASE(rgb_fmt, rgba_fmt, name, opt) \
681         INPUT_PLANER_RGB_A_FUNC_CASE(rgba_fmt##LE, name##le, opt)       \
682         INPUT_PLANER_RGB_UV_FUNC_CASE(rgb_fmt##LE, name##le, opt)       \
683         INPUT_PLANER_RGB_A_FUNC_CASE(rgba_fmt##BE, name##be, opt)       \
684         INPUT_PLANER_RGB_UV_FUNC_CASE(rgb_fmt##BE, name##be, opt)
685 
686 #define INPUT_PLANER_RGBAXX_YUV_FUNC_CASE(rgb_fmt, rgba_fmt, name, opt)           \
687         INPUT_PLANER_RGBA_YUV_FUNC_CASE(rgb_fmt##LE, rgba_fmt##LE, name##le, opt) \
688         INPUT_PLANER_RGBA_YUV_FUNC_CASE(rgb_fmt##BE, rgba_fmt##BE, name##be, opt)
689 
690 #define INPUT_PLANER_RGBXX_YUV_FUNC_CASE(rgb_fmt, name, opt)       \
691         INPUT_PLANER_RGB_YUV_FUNC_CASE(rgb_fmt##LE, name##le, opt) \
692         INPUT_PLANER_RGB_YUV_FUNC_CASE(rgb_fmt##BE, name##be, opt)
693 
694 #define INPUT_PLANER_RGBXX_UV_FUNC_CASE(rgb_fmt, name, opt)        \
695         INPUT_PLANER_RGB_UV_FUNC_CASE(rgb_fmt##LE, name##le, opt)  \
696         INPUT_PLANER_RGB_UV_FUNC_CASE(rgb_fmt##BE, name##be, opt)
697 
698 #define INPUT_PLANER_RGB_YUVA_ALL_CASES(opt)                                                     \
699         INPUT_PLANER_RGB_A_FUNC_CASE(      AV_PIX_FMT_GBRAP,                           rgb, opt) \
700         INPUT_PLANER_RGB_YUV_FUNC_CASE(    AV_PIX_FMT_GBRP,                            rgb, opt) \
701         INPUT_PLANER_RGBXX_YUV_FUNC_CASE(  AV_PIX_FMT_GBRP9,                          rgb9, opt) \
702         INPUT_PLANER_RGBAXX_YUVA_FUNC_CASE(AV_PIX_FMT_GBRP10,  AV_PIX_FMT_GBRAP10,   rgb10, opt) \
703         INPUT_PLANER_RGBAXX_YUVA_FUNC_CASE(AV_PIX_FMT_GBRP12,  AV_PIX_FMT_GBRAP12,   rgb12, opt) \
704         INPUT_PLANER_RGBXX_YUV_FUNC_CASE(  AV_PIX_FMT_GBRP14,                        rgb14, opt) \
705         INPUT_PLANER_RGBAXX_YUVA_FUNC_CASE(AV_PIX_FMT_GBRP16,  AV_PIX_FMT_GBRAP16,   rgb16, opt) \
706         INPUT_PLANER_RGBAXX_YUVA_FUNC_CASE(AV_PIX_FMT_GBRPF32, AV_PIX_FMT_GBRAPF32, rgbf32, opt)
707 
708 
709     if (EXTERNAL_SSE2(cpu_flags)) {
710         switch (c->srcFormat) {
711         INPUT_PLANER_RGB_A_FUNC_CASE(      AV_PIX_FMT_GBRAP,                           rgb, sse2);
712         INPUT_PLANER_RGB_UV_FUNC_CASE(     AV_PIX_FMT_GBRP,                            rgb, sse2);
713         INPUT_PLANER_RGBXX_UV_FUNC_CASE(   AV_PIX_FMT_GBRP9,                          rgb9, sse2);
714         INPUT_PLANER_RGBAXX_UVA_FUNC_CASE( AV_PIX_FMT_GBRP10,  AV_PIX_FMT_GBRAP10,   rgb10, sse2);
715         INPUT_PLANER_RGBAXX_UVA_FUNC_CASE( AV_PIX_FMT_GBRP12,  AV_PIX_FMT_GBRAP12,   rgb12, sse2);
716         INPUT_PLANER_RGBXX_UV_FUNC_CASE(   AV_PIX_FMT_GBRP14,                        rgb14, sse2);
717         INPUT_PLANER_RGBAXX_UVA_FUNC_CASE( AV_PIX_FMT_GBRP16,  AV_PIX_FMT_GBRAP16,   rgb16, sse2);
718         INPUT_PLANER_RGBAXX_YUVA_FUNC_CASE(AV_PIX_FMT_GBRPF32, AV_PIX_FMT_GBRAPF32, rgbf32, sse2);
719         default:
720             break;
721         }
722     }
723 
724     if (EXTERNAL_SSE4(cpu_flags)) {
725         switch (c->srcFormat) {
726         case AV_PIX_FMT_GBRAP:
727         INPUT_PLANER_RGB_YUV_FUNC_CASE(    AV_PIX_FMT_GBRP,                            rgb, sse4);
728         INPUT_PLANER_RGBXX_YUV_FUNC_CASE(  AV_PIX_FMT_GBRP9,                          rgb9, sse4);
729         INPUT_PLANER_RGBAXX_YUV_FUNC_CASE( AV_PIX_FMT_GBRP10,  AV_PIX_FMT_GBRAP10,   rgb10, sse4);
730         INPUT_PLANER_RGBAXX_YUV_FUNC_CASE( AV_PIX_FMT_GBRP12,  AV_PIX_FMT_GBRAP12,   rgb12, sse4);
731         INPUT_PLANER_RGBXX_YUV_FUNC_CASE(  AV_PIX_FMT_GBRP14,                        rgb14, sse4);
732         INPUT_PLANER_RGBAXX_YUV_FUNC_CASE( AV_PIX_FMT_GBRP16,  AV_PIX_FMT_GBRAP16,   rgb16, sse4);
733         INPUT_PLANER_RGBAXX_YUVA_FUNC_CASE(AV_PIX_FMT_GBRPF32, AV_PIX_FMT_GBRAPF32, rgbf32, sse4);
734         default:
735             break;
736         }
737     }
738 
739     if (EXTERNAL_AVX2_FAST(cpu_flags)) {
740         switch (c->srcFormat) {
741         INPUT_PLANER_RGB_YUVA_ALL_CASES(avx2)
742         default:
743             break;
744         }
745     }
746 
747     if(c->flags & SWS_FULL_CHR_H_INT) {
748 
749         /* yuv2gbrp uses the SwsContext for yuv coefficients
750            if struct offsets change the asm needs to be updated too */
751         av_assert0(offsetof(SwsContext, yuv2rgb_y_offset) == 40292);
752 
753 #define YUV2ANYX_FUNC_CASE(fmt, name, opt)              \
754         case fmt:                                       \
755             c->yuv2anyX = ff_yuv2##name##_full_X_##opt; \
756             break;
757 
758 #define YUV2ANYX_GBRAP_CASES(opt)                                  \
759         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP,       gbrp,       opt) \
760         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRAP,      gbrap,      opt) \
761         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP9LE,    gbrp9le,    opt) \
762         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP10LE,   gbrp10le,   opt) \
763         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRAP10LE,  gbrap10le,  opt) \
764         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP12LE,   gbrp12le,   opt) \
765         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRAP12LE,  gbrap12le,  opt) \
766         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP14LE,   gbrp14le,   opt) \
767         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP16LE,   gbrp16le,   opt) \
768         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRAP16LE,  gbrap16le,  opt) \
769         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRPF32LE,  gbrpf32le,  opt) \
770         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRAPF32LE, gbrapf32le, opt) \
771         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP9BE,    gbrp9be,    opt) \
772         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP10BE,   gbrp10be,   opt) \
773         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRAP10BE,  gbrap10be,  opt) \
774         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP12BE,   gbrp12be,   opt) \
775         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRAP12BE,  gbrap12be,  opt) \
776         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP14BE,   gbrp14be,   opt) \
777         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRP16BE,   gbrp16be,   opt) \
778         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRAP16BE,  gbrap16be,  opt) \
779         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRPF32BE,  gbrpf32be,  opt) \
780         YUV2ANYX_FUNC_CASE(AV_PIX_FMT_GBRAPF32BE, gbrapf32be, opt)
781 
782         if (EXTERNAL_SSE2(cpu_flags)) {
783             switch (c->dstFormat) {
784             YUV2ANYX_GBRAP_CASES(sse2)
785             default:
786                 break;
787             }
788         }
789 
790         if (EXTERNAL_SSE4(cpu_flags)) {
791             switch (c->dstFormat) {
792             YUV2ANYX_GBRAP_CASES(sse4)
793             default:
794                 break;
795             }
796         }
797 
798         if (EXTERNAL_AVX2_FAST(cpu_flags)) {
799             switch (c->dstFormat) {
800             YUV2ANYX_GBRAP_CASES(avx2)
801             default:
802                 break;
803             }
804         }
805     }
806 
807 #endif
808 }
809