FFmpeg coverage

Directory: ../../../ffmpeg/
File: src/libavcodec/vp8dsp.c
Date: 2026-08-19 01:31:13
Exec Total Coverage
Lines: 258 258 100.0%
Functions: 83 83 100.0%
Branches: 216 216 100.0%
Line Branch Exec Source
1 /*
2 * Copyright (C) 2010 David Conrad
3 * Copyright (C) 2010 Ronald S. Bultje
4 * Copyright (C) 2014 Peter Ross
5 *
6 * This file is part of FFmpeg.
7 *
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
12 *
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 */
22
23 /**
24 * @file
25 * VP8 compatible video decoder
26 */
27
28 #include "config_components.h"
29
30 #include "libavutil/common.h"
31 #include "libavutil/intreadwrite.h"
32
33 #include "mathops.h"
34 #include "vp8dsp.h"
35
36 #define MK_IDCT_DC_ADD4_C(name) \
37 static void name ## _idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16], \
38 ptrdiff_t stride) \
39 { \
40 name ## _idct_dc_add_c(dst + stride * 0 + 0, block[0], stride); \
41 name ## _idct_dc_add_c(dst + stride * 0 + 4, block[1], stride); \
42 name ## _idct_dc_add_c(dst + stride * 4 + 0, block[2], stride); \
43 name ## _idct_dc_add_c(dst + stride * 4 + 4, block[3], stride); \
44 } \
45 \
46 static void name ## _idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16], \
47 ptrdiff_t stride) \
48 { \
49 name ## _idct_dc_add_c(dst + 0, block[0], stride); \
50 name ## _idct_dc_add_c(dst + 4, block[1], stride); \
51 name ## _idct_dc_add_c(dst + 8, block[2], stride); \
52 name ## _idct_dc_add_c(dst + 12, block[3], stride); \
53 }
54
55 #if CONFIG_VP7_DECODER
56 3 static void vp7_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16])
57 {
58 int i;
59 unsigned a1, b1, c1, d1;
60 int16_t tmp[16];
61
62
2/2
✓ Branch 0 taken 12 times.
✓ Branch 1 taken 3 times.
 15 for (i = 0; i < 4; i++) {
63 12 a1 = (dc[i * 4 + 0] + dc[i * 4 + 2]) * 23170;
64 12 b1 = (dc[i * 4 + 0] - dc[i * 4 + 2]) * 23170;
65 12 c1 = dc[i * 4 + 1] * 12540 - dc[i * 4 + 3] * 30274;
66 12 d1 = dc[i * 4 + 1] * 30274 + dc[i * 4 + 3] * 12540;
67 12 tmp[i * 4 + 0] = (int)(a1 + d1) >> 14;
68 12 tmp[i * 4 + 3] = (int)(a1 - d1) >> 14;
69 12 tmp[i * 4 + 1] = (int)(b1 + c1) >> 14;
70 12 tmp[i * 4 + 2] = (int)(b1 - c1) >> 14;
71 }
72
73
2/2
✓ Branch 0 taken 12 times.
✓ Branch 1 taken 3 times.
 15 for (i = 0; i < 4; i++) {
74 12 a1 = (tmp[i + 0] + tmp[i + 8]) * 23170;
75 12 b1 = (tmp[i + 0] - tmp[i + 8]) * 23170;
76 12 c1 = tmp[i + 4] * 12540 - tmp[i + 12] * 30274;
77 12 d1 = tmp[i + 4] * 30274 + tmp[i + 12] * 12540;
78 12 AV_ZERO64(dc + i * 4);
79 12 block[0][i][0] = (int)(a1 + d1 + 0x20000) >> 18;
80 12 block[3][i][0] = (int)(a1 - d1 + 0x20000) >> 18;
81 12 block[1][i][0] = (int)(b1 + c1 + 0x20000) >> 18;
82 12 block[2][i][0] = (int)(b1 - c1 + 0x20000) >> 18;
83 }
84 3 }
85
86 30 static void vp7_luma_dc_wht_dc_c(int16_t block[4][4][16], int16_t dc[16])
87 {
88 30 int i, val = (23170 * (23170 * dc[0] >> 14) + 0x20000) >> 18;
89 30 dc[0] = 0;
90
91
2/2
✓ Branch 0 taken 120 times.
✓ Branch 1 taken 30 times.
 150 for (i = 0; i < 4; i++) {
92 120 block[i][0][0] = val;
93 120 block[i][1][0] = val;
94 120 block[i][2][0] = val;
95 120 block[i][3][0] = val;
96 }
97 30 }
98
99 661 static void vp7_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
100 {
101 int i;
102 unsigned a1, b1, c1, d1;
103 int16_t tmp[16];
104
105
2/2
✓ Branch 0 taken 2644 times.
✓ Branch 1 taken 661 times.
 3305 for (i = 0; i < 4; i++) {
106 2644 a1 = (block[i * 4 + 0] + block[i * 4 + 2]) * 23170;
107 2644 b1 = (block[i * 4 + 0] - block[i * 4 + 2]) * 23170;
108 2644 c1 = block[i * 4 + 1] * 12540 - block[i * 4 + 3] * 30274;
109 2644 d1 = block[i * 4 + 1] * 30274 + block[i * 4 + 3] * 12540;
110 2644 AV_ZERO64(block + i * 4);
111 2644 tmp[i * 4 + 0] = (int)(a1 + d1) >> 14;
112 2644 tmp[i * 4 + 3] = (int)(a1 - d1) >> 14;
113 2644 tmp[i * 4 + 1] = (int)(b1 + c1) >> 14;
114 2644 tmp[i * 4 + 2] = (int)(b1 - c1) >> 14;
115 }
116
117
2/2
✓ Branch 0 taken 2644 times.
✓ Branch 1 taken 661 times.
 3305 for (i = 0; i < 4; i++) {
118 2644 a1 = (tmp[i + 0] + tmp[i + 8]) * 23170;
119 2644 b1 = (tmp[i + 0] - tmp[i + 8]) * 23170;
120 2644 c1 = tmp[i + 4] * 12540 - tmp[i + 12] * 30274;
121 2644 d1 = tmp[i + 4] * 30274 + tmp[i + 12] * 12540;
122 2644 dst[0 * stride + i] = av_clip_uint8(dst[0 * stride + i] +
123 2644 ((int)(a1 + d1 + 0x20000) >> 18));
124 2644 dst[3 * stride + i] = av_clip_uint8(dst[3 * stride + i] +
125 2644 ((int)(a1 - d1 + 0x20000) >> 18));
126 2644 dst[1 * stride + i] = av_clip_uint8(dst[1 * stride + i] +
127 2644 ((int)(b1 + c1 + 0x20000) >> 18));
128 2644 dst[2 * stride + i] = av_clip_uint8(dst[2 * stride + i] +
129 2644 ((int)(b1 - c1 + 0x20000) >> 18));
130 }
131 661 }
132
133 554 static void vp7_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
134 {
135 554 int i, dc = (23170 * (23170 * block[0] >> 14) + 0x20000) >> 18;
136 554 block[0] = 0;
137
138
2/2
✓ Branch 0 taken 2216 times.
✓ Branch 1 taken 554 times.
 2770 for (i = 0; i < 4; i++) {
139 2216 dst[0] = av_clip_uint8(dst[0] + dc);
140 2216 dst[1] = av_clip_uint8(dst[1] + dc);
141 2216 dst[2] = av_clip_uint8(dst[2] + dc);
142 2216 dst[3] = av_clip_uint8(dst[3] + dc);
143 2216 dst += stride;
144 }
145 554 }
146
147 262 MK_IDCT_DC_ADD4_C(vp7)
148 #endif /* CONFIG_VP7_DECODER */
149
150 // TODO: Maybe add dequant
151 #if CONFIG_VP8_DECODER
152 24581 static void vp8_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16])
153 {
154 int i, t0, t1, t2, t3;
155
156
2/2
✓ Branch 0 taken 98324 times.
✓ Branch 1 taken 24581 times.
 122905 for (i = 0; i < 4; i++) {
157 98324 t0 = dc[0 * 4 + i] + dc[3 * 4 + i];
158 98324 t1 = dc[1 * 4 + i] + dc[2 * 4 + i];
159 98324 t2 = dc[1 * 4 + i] - dc[2 * 4 + i];
160 98324 t3 = dc[0 * 4 + i] - dc[3 * 4 + i];
161
162 98324 dc[0 * 4 + i] = t0 + t1;
163 98324 dc[1 * 4 + i] = t3 + t2;
164 98324 dc[2 * 4 + i] = t0 - t1;
165 98324 dc[3 * 4 + i] = t3 - t2;
166 }
167
168
2/2
✓ Branch 0 taken 98324 times.
✓ Branch 1 taken 24581 times.
 122905 for (i = 0; i < 4; i++) {
169 98324 t0 = dc[i * 4 + 0] + dc[i * 4 + 3] + 3; // rounding
170 98324 t1 = dc[i * 4 + 1] + dc[i * 4 + 2];
171 98324 t2 = dc[i * 4 + 1] - dc[i * 4 + 2];
172 98324 t3 = dc[i * 4 + 0] - dc[i * 4 + 3] + 3; // rounding
173 98324 AV_ZERO64(dc + i * 4);
174
175 98324 block[i][0][0] = (t0 + t1) >> 3;
176 98324 block[i][1][0] = (t3 + t2) >> 3;
177 98324 block[i][2][0] = (t0 - t1) >> 3;
178 98324 block[i][3][0] = (t3 - t2) >> 3;
179 }
180 24581 }
181
182 10719 static void vp8_luma_dc_wht_dc_c(int16_t block[4][4][16], int16_t dc[16])
183 {
184 10719 int i, val = (dc[0] + 3) >> 3;
185 10719 dc[0] = 0;
186
187
2/2
✓ Branch 0 taken 42876 times.
✓ Branch 1 taken 10719 times.
 53595 for (i = 0; i < 4; i++) {
188 42876 block[i][0][0] = val;
189 42876 block[i][1][0] = val;
190 42876 block[i][2][0] = val;
191 42876 block[i][3][0] = val;
192 }
193 10719 }
194
195 #define MUL_20091(a) ((((a) * 20091) >> 16) + (a))
196 #define MUL_35468(a) (((a) * 35468) >> 16)
197
198 258620 static void vp8_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
199 {
200 int i, t0, t1, t2, t3;
201 int16_t tmp[16];
202
203
2/2
✓ Branch 0 taken 1034480 times.
✓ Branch 1 taken 258620 times.
 1293100 for (i = 0; i < 4; i++) {
204 1034480 t0 = block[0 * 4 + i] + block[2 * 4 + i];
205 1034480 t1 = block[0 * 4 + i] - block[2 * 4 + i];
206 1034480 t2 = MUL_35468(block[1 * 4 + i]) - MUL_20091(block[3 * 4 + i]);
207 1034480 t3 = MUL_20091(block[1 * 4 + i]) + MUL_35468(block[3 * 4 + i]);
208 1034480 block[0 * 4 + i] = 0;
209 1034480 block[1 * 4 + i] = 0;
210 1034480 block[2 * 4 + i] = 0;
211 1034480 block[3 * 4 + i] = 0;
212
213 1034480 tmp[i * 4 + 0] = t0 + t3;
214 1034480 tmp[i * 4 + 1] = t1 + t2;
215 1034480 tmp[i * 4 + 2] = t1 - t2;
216 1034480 tmp[i * 4 + 3] = t0 - t3;
217 }
218
219
2/2
✓ Branch 0 taken 1034480 times.
✓ Branch 1 taken 258620 times.
 1293100 for (i = 0; i < 4; i++) {
220 1034480 t0 = tmp[0 * 4 + i] + tmp[2 * 4 + i];
221 1034480 t1 = tmp[0 * 4 + i] - tmp[2 * 4 + i];
222 1034480 t2 = MUL_35468(tmp[1 * 4 + i]) - MUL_20091(tmp[3 * 4 + i]);
223 1034480 t3 = MUL_20091(tmp[1 * 4 + i]) + MUL_35468(tmp[3 * 4 + i]);
224
225 1034480 dst[0] = av_clip_uint8(dst[0] + ((t0 + t3 + 4) >> 3));
226 1034480 dst[1] = av_clip_uint8(dst[1] + ((t1 + t2 + 4) >> 3));
227 1034480 dst[2] = av_clip_uint8(dst[2] + ((t1 - t2 + 4) >> 3));
228 1034480 dst[3] = av_clip_uint8(dst[3] + ((t0 - t3 + 4) >> 3));
229 1034480 dst += stride;
230 }
231 258620 }
232
233 769871 static void vp8_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
234 {
235 769871 int i, dc = (block[0] + 4) >> 3;
236 769871 block[0] = 0;
237
238
2/2
✓ Branch 0 taken 3079484 times.
✓ Branch 1 taken 769871 times.
 3849355 for (i = 0; i < 4; i++) {
239 3079484 dst[0] = av_clip_uint8(dst[0] + dc);
240 3079484 dst[1] = av_clip_uint8(dst[1] + dc);
241 3079484 dst[2] = av_clip_uint8(dst[2] + dc);
242 3079484 dst[3] = av_clip_uint8(dst[3] + dc);
243 3079484 dst += stride;
244 }
245 769871 }
246
247 328258 MK_IDCT_DC_ADD4_C(vp8)
248 #endif /* CONFIG_VP8_DECODER */
249
250 // because I like only having two parameters to pass functions...
251 #define LOAD_PIXELS \
252 av_unused int p3 = p[-4 * stride]; \
253 av_unused int p2 = p[-3 * stride]; \
254 av_unused int p1 = p[-2 * stride]; \
255 av_unused int p0 = p[-1 * stride]; \
256 av_unused int q0 = p[ 0 * stride]; \
257 av_unused int q1 = p[ 1 * stride]; \
258 av_unused int q2 = p[ 2 * stride]; \
259 av_unused int q3 = p[ 3 * stride];
260
261 #define clip_int8(n) (cm[(n) + 0x80] - 0x80)
262
263 10499788 static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride,
264 int is4tap, int is_vp7)
265 {
266 10499788 LOAD_PIXELS
267 int a, f1, f2;
268 10499788 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;
269
270 10499788 a = 3 * (q0 - p0);
271
272
2/2
✓ Branch 0 taken 5584564 times.
✓ Branch 1 taken 4915224 times.
 10499788 if (is4tap)
273 5584564 a += clip_int8(p1 - q1);
274
275 10499788 a = clip_int8(a);
276
277 // We deviate from the spec here with c(a+3) >> 3
278 // since that's what libvpx does.
279 10499788 f1 = FFMIN(a + 4, 127) >> 3;
280
281
2/2
✓ Branch 0 taken 807129 times.
✓ Branch 1 taken 9692659 times.
 10499788 if (is_vp7)
282 807129 f2 = f1 - ((a & 7) == 4);
283 else
284 9692659 f2 = FFMIN(a + 3, 127) >> 3;
285
286 // Despite what the spec says, we do need to clamp here to
287 // be bitexact with libvpx.
288 10499788 p[-1 * stride] = cm[p0 + f2];
289 10499788 p[ 0 * stride] = cm[q0 - f1];
290
291 // only used for _inner on blocks without high edge variance
292
2/2
✓ Branch 0 taken 4915224 times.
✓ Branch 1 taken 5584564 times.
 10499788 if (!is4tap) {
293 4915224 a = (f1 + 1) >> 1;
294 4915224 p[-2 * stride] = cm[p1 + a];
295 4915224 p[ 1 * stride] = cm[q1 - a];
296 }
297 10499788 }
298
299 807129 static av_always_inline void vp7_filter_common(uint8_t *p, ptrdiff_t stride,
300 int is4tap)
301 {
302 807129 filter_common(p, stride, is4tap, IS_VP7);
303 807129 }
304
305 9692659 static av_always_inline void vp8_filter_common(uint8_t *p, ptrdiff_t stride,
306 int is4tap)
307 {
308 9692659 filter_common(p, stride, is4tap, IS_VP8);
309 9692659 }
310
311 1320928 static av_always_inline int vp7_simple_limit(uint8_t *p, ptrdiff_t stride,
312 int flim)
313 {
314 1320928 LOAD_PIXELS
315 1320928 return FFABS(p0 - q0) <= flim;
316 }
317
318 34066176 static av_always_inline int vp8_simple_limit(uint8_t *p, ptrdiff_t stride,
319 int flim)
320 {
321 34066176 LOAD_PIXELS
322 34066176 return 2 * FFABS(p0 - q0) + (FFABS(p1 - q1) >> 1) <= flim;
323 }
324
325 /**
326 * E - limit at the macroblock edge
327 * I - limit for interior difference
328 */
329 #define NORMAL_LIMIT(vpn) \
330 static av_always_inline int vp ## vpn ## _normal_limit(uint8_t *p, \
331 ptrdiff_t stride, \
332 int E, int I) \
333 { \
334 LOAD_PIXELS \
335 return vp ## vpn ## _simple_limit(p, stride, E) && \
336 FFABS(p3 - p2) <= I && FFABS(p2 - p1) <= I && \
337 FFABS(p1 - p0) <= I && FFABS(q3 - q2) <= I && \
338 FFABS(q2 - q1) <= I && FFABS(q1 - q0) <= I; \
339 }
340
341
14/14
✓ Branch 1 taken 1242772 times.
✓ Branch 2 taken 78060 times.
✓ Branch 3 taken 1209677 times.
✓ Branch 4 taken 33095 times.
✓ Branch 5 taken 1193091 times.
✓ Branch 6 taken 16586 times.
✓ Branch 7 taken 1188830 times.
✓ Branch 8 taken 4261 times.
✓ Branch 9 taken 1168415 times.
✓ Branch 10 taken 20415 times.
✓ Branch 11 taken 1165610 times.
✓ Branch 12 taken 2805 times.
✓ Branch 13 taken 1164716 times.
✓ Branch 14 taken 894 times.
 1320832 NORMAL_LIMIT(7)
342
14/14
✓ Branch 1 taken 32252652 times.
✓ Branch 2 taken 1262036 times.
✓ Branch 3 taken 31586901 times.
✓ Branch 4 taken 665751 times.
✓ Branch 5 taken 31292422 times.
✓ Branch 6 taken 294479 times.
✓ Branch 7 taken 31179789 times.
✓ Branch 8 taken 112633 times.
✓ Branch 9 taken 30739050 times.
✓ Branch 10 taken 440739 times.
✓ Branch 11 taken 30536792 times.
✓ Branch 12 taken 202258 times.
✓ Branch 13 taken 30462751 times.
✓ Branch 14 taken 74041 times.
 33514688 NORMAL_LIMIT(8)
343
344 // high edge variance
345 31627467 static av_always_inline int hev(uint8_t *p, ptrdiff_t stride, int thresh)
346 {
347 31627467 LOAD_PIXELS
348
4/4
✓ Branch 0 taken 27954305 times.
✓ Branch 1 taken 3673162 times.
✓ Branch 2 taken 1420245 times.
✓ Branch 3 taken 26534060 times.
 31627467 return FFABS(p1 - p0) > thresh || FFABS(q1 - q0) > thresh;
349 }
350
351 21618836 static av_always_inline void filter_mbedge(uint8_t *p, ptrdiff_t stride)
352 {
353 int a0, a1, a2, w;
354 21618836 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;
355
356 21618836 LOAD_PIXELS
357
358 21618836 w = clip_int8(p1 - q1);
359 21618836 w = clip_int8(w + 3 * (q0 - p0));
360
361 21618836 a0 = (27 * w + 63) >> 7;
362 21618836 a1 = (18 * w + 63) >> 7;
363 21618836 a2 = (9 * w + 63) >> 7;
364
365 21618836 p[-3 * stride] = cm[p2 + a2];
366 21618836 p[-2 * stride] = cm[p1 + a1];
367 21618836 p[-1 * stride] = cm[p0 + a0];
368 21618836 p[ 0 * stride] = cm[q0 - a0];
369 21618836 p[ 1 * stride] = cm[q1 - a1];
370 21618836 p[ 2 * stride] = cm[q2 - a2];
371 21618836 }
372
373 #define LOOP_FILTER(vpn, dir, size, stridea, strideb, maybe_inline) \
374 static maybe_inline \
375 void vpn ## _ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, \
376 ptrdiff_t stride, \
377 int flim_E, int flim_I, \
378 int hev_thresh) \
379 { \
380 int i; \
381 for (i = 0; i < size; i++) \
382 if (vpn ## _normal_limit(dst + i * stridea, strideb, \
383 flim_E, flim_I)) { \
384 if (hev(dst + i * stridea, strideb, hev_thresh)) \
385 vpn ## _filter_common(dst + i * stridea, strideb, 1); \
386 else \
387 filter_mbedge(dst + i * stridea, strideb); \
388 } \
389 } \
390 \
391 static maybe_inline \
392 void vpn ## _ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst, \
393 ptrdiff_t stride, \
394 int flim_E, \
395 int flim_I, \
396 int hev_thresh) \
397 { \
398 int i; \
399 for (i = 0; i < size; i++) \
400 if (vpn ## _normal_limit(dst + i * stridea, strideb, \
401 flim_E, flim_I)) { \
402 int hv = hev(dst + i * stridea, strideb, hev_thresh); \
403 if (hv) \
404 vpn ## _filter_common(dst + i * stridea, strideb, 1); \
405 else \
406 vpn ## _filter_common(dst + i * stridea, strideb, 0); \
407 } \
408 }
409
410 #define UV_LOOP_FILTER(vpn, dir, stridea, strideb) \
411 LOOP_FILTER(vpn, dir, 8, stridea, strideb, av_always_inline) \
412 static void vpn ## _ ## dir ## _loop_filter8uv_c(uint8_t *dstU, \
413 uint8_t *dstV, \
414 ptrdiff_t stride, int fE, \
415 int fI, int hev_thresh) \
416 { \
417 vpn ## _ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh); \
418 vpn ## _ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh); \
419 } \
420 \
421 static void vpn ## _ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU, \
422 uint8_t *dstV, \
423 ptrdiff_t stride, \
424 int fE, int fI, \
425 int hev_thresh) \
426 { \
427 vpn ## _ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI, \
428 hev_thresh); \
429 vpn ## _ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI, \
430 hev_thresh); \
431 }
432
433 #define LOOP_FILTER_SIMPLE(vpn) \
434 static void vpn ## _v_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, \
435 int flim) \
436 { \
437 int i; \
438 for (i = 0; i < 16; i++) \
439 if (vpn ## _simple_limit(dst + i, stride, flim)) \
440 vpn ## _filter_common(dst + i, stride, 1); \
441 } \
442 \
443 static void vpn ## _h_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, \
444 int flim) \
445 { \
446 int i; \
447 for (i = 0; i < 16; i++) \
448 if (vpn ## _simple_limit(dst + i * stride, 1, flim)) \
449 vpn ## _filter_common(dst + i * stride, 1, 1); \
450 }
451
452 #define LOOP_FILTERS(vpn) \
453 LOOP_FILTER(vpn, v, 16, 1, stride, ) \
454 LOOP_FILTER(vpn, h, 16, stride, 1, ) \
455 UV_LOOP_FILTER(vpn, v, 1, stride) \
456 UV_LOOP_FILTER(vpn, h, stride, 1) \
457 LOOP_FILTER_SIMPLE(vpn) \
458
459 static const uint8_t subpel_filters[7][6] = {
460 { 0, 6, 123, 12, 1, 0 },
461 { 2, 11, 108, 36, 8, 1 },
462 { 0, 9, 93, 50, 6, 0 },
463 { 3, 16, 77, 77, 16, 3 },
464 { 0, 6, 50, 93, 9, 0 },
465 { 1, 8, 36, 108, 11, 2 },
466 { 0, 1, 12, 123, 6, 0 },
467 };
468
469 #define PUT_PIXELS(WIDTH) \
470 static void put_vp8_pixels ## WIDTH ## _c(uint8_t *dst, ptrdiff_t dststride, \
471 const uint8_t *src, ptrdiff_t srcstride, \
472 int h, int x, int y) \
473 { \
474 int i; \
475 for (i = 0; i < h; i++, dst += dststride, src += srcstride) \
476 memcpy(dst, src, WIDTH); \
477 }
478
479
2/2
✓ Branch 0 taken 5391088 times.
✓ Branch 1 taken 339529 times.
 5730617 PUT_PIXELS(16)
480
2/2
✓ Branch 0 taken 5415384 times.
✓ Branch 1 taken 679380 times.
 6094764 PUT_PIXELS(8)
481
2/2
✓ Branch 0 taken 327080 times.
✓ Branch 1 taken 78419 times.
 405499 PUT_PIXELS(4)
482
483 #define FILTER_6TAP(src, F, stride) \
484 cm[(F[2] * src[x + 0 * stride] - F[1] * src[x - 1 * stride] + \
485 F[0] * src[x - 2 * stride] + F[3] * src[x + 1 * stride] - \
486 F[4] * src[x + 2 * stride] + F[5] * src[x + 3 * stride] + 64) >> 7]
487
488 #define FILTER_4TAP(src, F, stride) \
489 cm[(F[2] * src[x + 0 * stride] - F[1] * src[x - 1 * stride] + \
490 F[3] * src[x + 1 * stride] - F[4] * src[x + 2 * stride] + 64) >> 7]
491
492 #define VP8_EPEL_H(SIZE, TAPS) \
493 static void put_vp8_epel ## SIZE ## _h ## TAPS ## _c(uint8_t *dst, \
494 ptrdiff_t dststride, \
495 const uint8_t *src, \
496 ptrdiff_t srcstride, \
497 int h, int mx, int my) \
498 { \
499 const uint8_t *filter = subpel_filters[mx - 1]; \
500 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \
501 int x, y; \
502 for (y = 0; y < h; y++) { \
503 for (x = 0; x < SIZE; x++) \
504 dst[x] = FILTER_ ## TAPS ## TAP(src, filter, 1); \
505 dst += dststride; \
506 src += srcstride; \
507 } \
508 }
509
510 #define VP8_EPEL_V(SIZE, TAPS) \
511 static void put_vp8_epel ## SIZE ## _v ## TAPS ## _c(uint8_t *dst, \
512 ptrdiff_t dststride, \
513 const uint8_t *src, \
514 ptrdiff_t srcstride, \
515 int h, int mx, int my) \
516 { \
517 const uint8_t *filter = subpel_filters[my - 1]; \
518 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \
519 int x, y; \
520 for (y = 0; y < h; y++) { \
521 for (x = 0; x < SIZE; x++) \
522 dst[x] = FILTER_ ## TAPS ## TAP(src, filter, srcstride); \
523 dst += dststride; \
524 src += srcstride; \
525 } \
526 }
527
528 #define VP8_EPEL_HV(SIZE, HTAPS, VTAPS) \
529 static void \
530 put_vp8_epel ## SIZE ## _h ## HTAPS ## v ## VTAPS ## _c(uint8_t *dst, \
531 ptrdiff_t dststride, \
532 const uint8_t *src, \
533 ptrdiff_t srcstride, \
534 int h, int mx, \
535 int my) \
536 { \
537 const uint8_t *filter = subpel_filters[mx - 1]; \
538 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \
539 int x, y; \
540 uint8_t tmp_array[(2 * SIZE + VTAPS - 1) * SIZE]; \
541 uint8_t *tmp = tmp_array; \
542 src -= (2 - (VTAPS == 4)) * srcstride; \
543 \
544 for (y = 0; y < h + VTAPS - 1; y++) { \
545 for (x = 0; x < SIZE; x++) \
546 tmp[x] = FILTER_ ## HTAPS ## TAP(src, filter, 1); \
547 tmp += SIZE; \
548 src += srcstride; \
549 } \
550 tmp = tmp_array + (2 - (VTAPS == 4)) * SIZE; \
551 filter = subpel_filters[my - 1]; \
552 \
553 for (y = 0; y < h; y++) { \
554 for (x = 0; x < SIZE; x++) \
555 dst[x] = FILTER_ ## VTAPS ## TAP(tmp, filter, SIZE); \
556 dst += dststride; \
557 tmp += SIZE; \
558 } \
559 }
560
561
4/4
✓ Branch 0 taken 812160 times.
✓ Branch 1 taken 101520 times.
✓ Branch 2 taken 101520 times.
✓ Branch 3 taken 13324 times.
 927004 VP8_EPEL_H(8, 4)
562
4/4
✓ Branch 0 taken 158080 times.
✓ Branch 1 taken 39520 times.
✓ Branch 2 taken 39520 times.
✓ Branch 3 taken 8656 times.
 206256 VP8_EPEL_H(4, 4)
563
4/4
✓ Branch 0 taken 2627584 times.
✓ Branch 1 taken 164224 times.
✓ Branch 2 taken 164224 times.
✓ Branch 3 taken 10861 times.
 2802669 VP8_EPEL_H(16, 6)
564
4/4
✓ Branch 0 taken 845888 times.
✓ Branch 1 taken 105736 times.
✓ Branch 2 taken 105736 times.
✓ Branch 3 taken 12799 times.
 964423 VP8_EPEL_H(8, 6)
565
4/4
✓ Branch 0 taken 366880 times.
✓ Branch 1 taken 91720 times.
✓ Branch 2 taken 91720 times.
✓ Branch 3 taken 21990 times.
 480590 VP8_EPEL_H(4, 6)
566
4/4
✓ Branch 0 taken 646016 times.
✓ Branch 1 taken 80752 times.
✓ Branch 2 taken 80752 times.
✓ Branch 3 taken 10644 times.
 737412 VP8_EPEL_V(8, 4)
567
4/4
✓ Branch 0 taken 144192 times.
✓ Branch 1 taken 36048 times.
✓ Branch 2 taken 36048 times.
✓ Branch 3 taken 7922 times.
 188162 VP8_EPEL_V(4, 4)
568
4/4
✓ Branch 0 taken 2145024 times.
✓ Branch 1 taken 134064 times.
✓ Branch 2 taken 134064 times.
✓ Branch 3 taken 8896 times.
 2287984 VP8_EPEL_V(16, 6)
569
4/4
✓ Branch 0 taken 664576 times.
✓ Branch 1 taken 83072 times.
✓ Branch 2 taken 83072 times.
✓ Branch 3 taken 9792 times.
 757440 VP8_EPEL_V(8, 6)
570
4/4
✓ Branch 0 taken 330896 times.
✓ Branch 1 taken 82724 times.
✓ Branch 2 taken 82724 times.
✓ Branch 3 taken 19993 times.
 433613 VP8_EPEL_V(4, 6)
571
572
8/8
✓ Branch 0 taken 1280928 times.
✓ Branch 1 taken 160116 times.
✓ Branch 2 taken 160116 times.
✓ Branch 3 taken 15284 times.
✓ Branch 4 taken 914112 times.
✓ Branch 5 taken 114264 times.
✓ Branch 6 taken 114264 times.
✓ Branch 7 taken 15284 times.
 2484704 VP8_EPEL_HV(8, 4, 4)
573
8/8
✓ Branch 0 taken 417656 times.
✓ Branch 1 taken 104414 times.
✓ Branch 2 taken 104414 times.
✓ Branch 3 taken 13914 times.
✓ Branch 4 taken 250688 times.
✓ Branch 5 taken 62672 times.
✓ Branch 6 taken 62672 times.
✓ Branch 7 taken 13914 times.
 849344 VP8_EPEL_HV(4, 4, 4)
574
8/8
✓ Branch 0 taken 778656 times.
✓ Branch 1 taken 97332 times.
✓ Branch 2 taken 97332 times.
✓ Branch 3 taken 7836 times.
✓ Branch 4 taken 465216 times.
✓ Branch 5 taken 58152 times.
✓ Branch 6 taken 58152 times.
✓ Branch 7 taken 7836 times.
 1407192 VP8_EPEL_HV(8, 4, 6)
575
8/8
✓ Branch 0 taken 307584 times.
✓ Branch 1 taken 76896 times.
✓ Branch 2 taken 76896 times.
✓ Branch 3 taken 8088 times.
✓ Branch 4 taken 145824 times.
✓ Branch 5 taken 36456 times.
✓ Branch 6 taken 36456 times.
✓ Branch 7 taken 8088 times.
 574848 VP8_EPEL_HV(4, 4, 6)
576
8/8
✓ Branch 0 taken 830352 times.
✓ Branch 1 taken 103794 times.
✓ Branch 2 taken 103794 times.
✓ Branch 3 taken 9934 times.
✓ Branch 4 taken 591936 times.
✓ Branch 5 taken 73992 times.
✓ Branch 6 taken 73992 times.
✓ Branch 7 taken 9934 times.
 1610008 VP8_EPEL_HV(8, 6, 4)
577
8/8
✓ Branch 0 taken 276152 times.
✓ Branch 1 taken 69038 times.
✓ Branch 2 taken 69038 times.
✓ Branch 3 taken 9114 times.
✓ Branch 4 taken 166784 times.
✓ Branch 5 taken 41696 times.
✓ Branch 6 taken 41696 times.
✓ Branch 7 taken 9114 times.
 562784 VP8_EPEL_HV(4, 6, 4)
578
8/8
✓ Branch 0 taken 5355552 times.
✓ Branch 1 taken 334722 times.
✓ Branch 2 taken 334722 times.
✓ Branch 3 taken 16794 times.
✓ Branch 4 taken 4012032 times.
✓ Branch 5 taken 250752 times.
✓ Branch 6 taken 250752 times.
✓ Branch 7 taken 16794 times.
 9969852 VP8_EPEL_HV(16, 6, 6)
579
8/8
✓ Branch 0 taken 1489760 times.
✓ Branch 1 taken 186220 times.
✓ Branch 2 taken 186220 times.
✓ Branch 3 taken 13476 times.
✓ Branch 4 taken 950720 times.
✓ Branch 5 taken 118840 times.
✓ Branch 6 taken 118840 times.
✓ Branch 7 taken 13476 times.
 2759016 VP8_EPEL_HV(8, 6, 6)
580
8/8
✓ Branch 0 taken 1004540 times.
✓ Branch 1 taken 251135 times.
✓ Branch 2 taken 251135 times.
✓ Branch 3 taken 27527 times.
✓ Branch 4 taken 454000 times.
✓ Branch 5 taken 113500 times.
✓ Branch 6 taken 113500 times.
✓ Branch 7 taken 27527 times.
 1850702 VP8_EPEL_HV(4, 6, 6)
581
582 #define VP8_BILINEAR(SIZE) \
583 static void put_vp8_bilinear ## SIZE ## _h_c(uint8_t *dst, ptrdiff_t dstride, \
584 const uint8_t *src, ptrdiff_t sstride, \
585 int h, int mx, int my) \
586 { \
587 int a = 8 - mx, b = mx; \
588 int x, y; \
589 for (y = 0; y < h; y++) { \
590 for (x = 0; x < SIZE; x++) \
591 dst[x] = (a * src[x] + b * src[x + 1] + 4) >> 3; \
592 dst += dstride; \
593 src += sstride; \
594 } \
595 } \
596 \
597 static void put_vp8_bilinear ## SIZE ## _v_c(uint8_t *dst, ptrdiff_t dstride, \
598 const uint8_t *src, ptrdiff_t sstride, \
599 int h, int mx, int my) \
600 { \
601 int c = 8 - my, d = my; \
602 int x, y; \
603 for (y = 0; y < h; y++) { \
604 for (x = 0; x < SIZE; x++) \
605 dst[x] = (c * src[x] + d * src[x + sstride] + 4) >> 3; \
606 dst += dstride; \
607 src += sstride; \
608 } \
609 } \
610 \
611 static void put_vp8_bilinear ## SIZE ## _hv_c(uint8_t *dst, \
612 ptrdiff_t dstride, \
613 const uint8_t *src, \
614 ptrdiff_t sstride, \
615 int h, int mx, int my) \
616 { \
617 int a = 8 - mx, b = mx; \
618 int c = 8 - my, d = my; \
619 int x, y; \
620 uint8_t tmp_array[(2 * SIZE + 1) * SIZE]; \
621 uint8_t *tmp = tmp_array; \
622 for (y = 0; y < h + 1; y++) { \
623 for (x = 0; x < SIZE; x++) \
624 tmp[x] = (a * src[x] + b * src[x + 1] + 4) >> 3; \
625 tmp += SIZE; \
626 src += sstride; \
627 } \
628 tmp = tmp_array; \
629 for (y = 0; y < h; y++) { \
630 for (x = 0; x < SIZE; x++) \
631 dst[x] = (c * tmp[x] + d * tmp[x + SIZE] + 4) >> 3; \
632 dst += dstride; \
633 tmp += SIZE; \
634 } \
635 }
636
637
8/8
✓ Branch 0 taken 445088 times.
✓ Branch 1 taken 27818 times.
✓ Branch 2 taken 27818 times.
✓ Branch 3 taken 1778 times.
✓ Branch 4 taken 185472 times.
✓ Branch 5 taken 11592 times.
✓ Branch 6 taken 11592 times.
✓ Branch 7 taken 762 times.
 1343496 VP8_BILINEAR(16)
638
8/8
✓ Branch 0 taken 283872 times.
✓ Branch 1 taken 35484 times.
✓ Branch 2 taken 35484 times.
✓ Branch 3 taken 4185 times.
✓ Branch 4 taken 125120 times.
✓ Branch 5 taken 15640 times.
✓ Branch 6 taken 15640 times.
✓ Branch 7 taken 1964 times.
 928602 VP8_BILINEAR(8)
639
8/8
✓ Branch 0 taken 96124 times.
✓ Branch 1 taken 24031 times.
✓ Branch 2 taken 24031 times.
✓ Branch 3 taken 5076 times.
✓ Branch 4 taken 39888 times.
✓ Branch 5 taken 9972 times.
✓ Branch 6 taken 9972 times.
✓ Branch 7 taken 2303 times.
 350182 VP8_BILINEAR(4)
640
641 #define VP78_MC_FUNC(IDX, SIZE) \
642 dsp->put_vp8_epel_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
643 dsp->put_vp8_epel_pixels_tab[IDX][0][1] = put_vp8_epel ## SIZE ## _h4_c; \
644 dsp->put_vp8_epel_pixels_tab[IDX][0][2] = put_vp8_epel ## SIZE ## _h6_c; \
645 dsp->put_vp8_epel_pixels_tab[IDX][1][0] = put_vp8_epel ## SIZE ## _v4_c; \
646 dsp->put_vp8_epel_pixels_tab[IDX][1][1] = put_vp8_epel ## SIZE ## _h4v4_c; \
647 dsp->put_vp8_epel_pixels_tab[IDX][1][2] = put_vp8_epel ## SIZE ## _h6v4_c; \
648 dsp->put_vp8_epel_pixels_tab[IDX][2][0] = put_vp8_epel ## SIZE ## _v6_c; \
649 dsp->put_vp8_epel_pixels_tab[IDX][2][1] = put_vp8_epel ## SIZE ## _h4v6_c; \
650 dsp->put_vp8_epel_pixels_tab[IDX][2][2] = put_vp8_epel ## SIZE ## _h6v6_c
651
652 #define VP78_BILINEAR_MC_FUNC(IDX, SIZE) \
653 dsp->put_vp8_bilinear_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
654 dsp->put_vp8_bilinear_pixels_tab[IDX][0][1] = put_vp8_bilinear ## SIZE ## _h_c; \
655 dsp->put_vp8_bilinear_pixels_tab[IDX][0][2] = put_vp8_bilinear ## SIZE ## _h_c; \
656 dsp->put_vp8_bilinear_pixels_tab[IDX][1][0] = put_vp8_bilinear ## SIZE ## _v_c; \
657 dsp->put_vp8_bilinear_pixels_tab[IDX][1][1] = put_vp8_bilinear ## SIZE ## _hv_c; \
658 dsp->put_vp8_bilinear_pixels_tab[IDX][1][2] = put_vp8_bilinear ## SIZE ## _hv_c; \
659 dsp->put_vp8_bilinear_pixels_tab[IDX][2][0] = put_vp8_bilinear ## SIZE ## _v_c; \
660 dsp->put_vp8_bilinear_pixels_tab[IDX][2][1] = put_vp8_bilinear ## SIZE ## _hv_c; \
661 dsp->put_vp8_bilinear_pixels_tab[IDX][2][2] = put_vp8_bilinear ## SIZE ## _hv_c
662
663 79 av_cold void ff_vp78dsp_init(VP8DSPContext *dsp)
664 {
665 79 dsp->put_vp8_epel_pixels_tab[0][0][0] = put_vp8_pixels16_c;
666 79 dsp->put_vp8_epel_pixels_tab[0][0][2] = put_vp8_epel16_h6_c;
667 79 dsp->put_vp8_epel_pixels_tab[0][2][0] = put_vp8_epel16_v6_c;
668 79 dsp->put_vp8_epel_pixels_tab[0][2][2] = put_vp8_epel16_h6v6_c;
669
670 79 VP78_MC_FUNC(1, 8);
671 79 VP78_MC_FUNC(2, 4);
672
673 79 VP78_BILINEAR_MC_FUNC(0, 16);
674 79 VP78_BILINEAR_MC_FUNC(1, 8);
675 79 VP78_BILINEAR_MC_FUNC(2, 4);
676
677 #if ARCH_AARCH64
678 ff_vp78dsp_init_aarch64(dsp);
679 #elif ARCH_ARM
680 ff_vp78dsp_init_arm(dsp);
681 #elif ARCH_PPC
682 ff_vp78dsp_init_ppc(dsp);
683 #elif ARCH_RISCV
684 ff_vp78dsp_init_riscv(dsp);
685 #elif ARCH_X86 && HAVE_X86ASM
686 79 ff_vp78dsp_init_x86(dsp);
687 #endif
688 79 }
689
690 #if CONFIG_VP7_DECODER
691
6/6
✓ Branch 1 taken 1164772 times.
✓ Branch 2 taken 156156 times.
✓ Branch 4 taken 20885 times.
✓ Branch 5 taken 1143933 times.
✓ Branch 8 taken 1320832 times.
✓ Branch 9 taken 109748 times.
 2915756 LOOP_FILTERS(vp7)
692
693 17 av_cold void ff_vp7dsp_init(VP8DSPContext *dsp)
694 {
695 17 dsp->vp8_luma_dc_wht = vp7_luma_dc_wht_c;
696 17 dsp->vp8_luma_dc_wht_dc = vp7_luma_dc_wht_dc_c;
697 17 dsp->vp8_idct_add = vp7_idct_add_c;
698 17 dsp->vp8_idct_dc_add = vp7_idct_dc_add_c;
699 17 dsp->vp8_idct_dc_add4y = vp7_idct_dc_add4y_c;
700 17 dsp->vp8_idct_dc_add4uv = vp7_idct_dc_add4uv_c;
701
702 17 dsp->vp8_v_loop_filter16y = vp7_v_loop_filter16_c;
703 17 dsp->vp8_h_loop_filter16y = vp7_h_loop_filter16_c;
704 17 dsp->vp8_v_loop_filter8uv = vp7_v_loop_filter8uv_c;
705 17 dsp->vp8_h_loop_filter8uv = vp7_h_loop_filter8uv_c;
706
707 17 dsp->vp8_v_loop_filter16y_inner = vp7_v_loop_filter16_inner_c;
708 17 dsp->vp8_h_loop_filter16y_inner = vp7_h_loop_filter16_inner_c;
709 17 dsp->vp8_v_loop_filter8uv_inner = vp7_v_loop_filter8uv_inner_c;
710 17 dsp->vp8_h_loop_filter8uv_inner = vp7_h_loop_filter8uv_inner_c;
711
712 17 dsp->vp8_v_loop_filter_simple = vp7_v_loop_filter_simple_c;
713 17 dsp->vp8_h_loop_filter_simple = vp7_h_loop_filter_simple_c;
714
715 #if ARCH_RISCV
716 ff_vp7dsp_init_riscv(dsp);
717 #endif
718 17 }
719 #endif /* CONFIG_VP7_DECODER */
720
721 #if CONFIG_VP8_DECODER
722
6/6
✓ Branch 1 taken 30953852 times.
✓ Branch 2 taken 3112324 times.
✓ Branch 4 taken 5624106 times.
✓ Branch 5 taken 25424601 times.
✓ Branch 8 taken 33514688 times.
✓ Branch 9 taken 3010770 times.
 76055032 LOOP_FILTERS(vp8)
723
724 76 av_cold void ff_vp8dsp_init(VP8DSPContext *dsp)
725 {
726 76 dsp->vp8_luma_dc_wht = vp8_luma_dc_wht_c;
727 76 dsp->vp8_luma_dc_wht_dc = vp8_luma_dc_wht_dc_c;
728 76 dsp->vp8_idct_add = vp8_idct_add_c;
729 76 dsp->vp8_idct_dc_add = vp8_idct_dc_add_c;
730 76 dsp->vp8_idct_dc_add4y = vp8_idct_dc_add4y_c;
731 76 dsp->vp8_idct_dc_add4uv = vp8_idct_dc_add4uv_c;
732
733 76 dsp->vp8_v_loop_filter16y = vp8_v_loop_filter16_c;
734 76 dsp->vp8_h_loop_filter16y = vp8_h_loop_filter16_c;
735 76 dsp->vp8_v_loop_filter8uv = vp8_v_loop_filter8uv_c;
736 76 dsp->vp8_h_loop_filter8uv = vp8_h_loop_filter8uv_c;
737
738 76 dsp->vp8_v_loop_filter16y_inner = vp8_v_loop_filter16_inner_c;
739 76 dsp->vp8_h_loop_filter16y_inner = vp8_h_loop_filter16_inner_c;
740 76 dsp->vp8_v_loop_filter8uv_inner = vp8_v_loop_filter8uv_inner_c;
741 76 dsp->vp8_h_loop_filter8uv_inner = vp8_h_loop_filter8uv_inner_c;
742
743 76 dsp->vp8_v_loop_filter_simple = vp8_v_loop_filter_simple_c;
744 76 dsp->vp8_h_loop_filter_simple = vp8_h_loop_filter_simple_c;
745
746 #if ARCH_AARCH64
747 ff_vp8dsp_init_aarch64(dsp);
748 #elif ARCH_ARM
749 ff_vp8dsp_init_arm(dsp);
750 #elif ARCH_RISCV
751 ff_vp8dsp_init_riscv(dsp);
752 #elif ARCH_X86 && HAVE_X86ASM
753 76 ff_vp8dsp_init_x86(dsp);
754 #elif ARCH_MIPS
755 ff_vp8dsp_init_mips(dsp);
756 #elif ARCH_LOONGARCH
757 ff_vp8dsp_init_loongarch(dsp);
758 #endif
759 76 }
760 #endif /* CONFIG_VP8_DECODER */
761