GCC Code Coverage Report
Directory: ../../../ffmpeg/ Exec Total Coverage
File: src/libavcodec/cavsdsp.c Lines: 181 181 100.0 %
Date: 2019-11-18 18:00:01 Branches: 156 156 100.0 %

Line Branch Exec Source
1
/*
2
 * Chinese AVS video (AVS1-P2, JiZhun profile) decoder.
3
 *
4
 * DSP functions
5
 *
6
 * Copyright (c) 2006  Stefan Gehrer <stefan.gehrer@gmx.de>
7
 *
8
 * This file is part of FFmpeg.
9
 *
10
 * FFmpeg is free software; you can redistribute it and/or
11
 * modify it under the terms of the GNU Lesser General Public
12
 * License as published by the Free Software Foundation; either
13
 * version 2.1 of the License, or (at your option) any later version.
14
 *
15
 * FFmpeg is distributed in the hope that it will be useful,
16
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
18
 * Lesser General Public License for more details.
19
 *
20
 * You should have received a copy of the GNU Lesser General Public
21
 * License along with FFmpeg; if not, write to the Free Software
22
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23
 */
24
25
#include <stdio.h>
26
27
#include "idctdsp.h"
28
#include "mathops.h"
29
#include "cavsdsp.h"
30
#include "libavutil/common.h"
31
32
/*****************************************************************************
33
 *
34
 * in-loop deblocking filter
35
 *
36
 ****************************************************************************/
37
38
#define P2 p0_p[-3*stride]
39
#define P1 p0_p[-2*stride]
40
#define P0 p0_p[-1*stride]
41
#define Q0 p0_p[ 0*stride]
42
#define Q1 p0_p[ 1*stride]
43
#define Q2 p0_p[ 2*stride]
44
45
1732320
static inline void loop_filter_l2(uint8_t *p0_p, ptrdiff_t stride, int alpha, int beta)
46
{
47
1732320
    int p0 = P0;
48
1732320
    int q0 = Q0;
49
50

1732320
    if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
51
1547636
        int s = p0 + q0 + 2;
52
1547636
        alpha = (alpha>>2) + 2;
53

1547636
        if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {
54
1473212
            P0 = (P1 + p0 + s) >> 2;
55
1473212
            P1 = (2*P1 + s) >> 2;
56
        } else
57
74424
            P0 = (2*P1 + s) >> 2;
58

1547636
        if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {
59
1472174
            Q0 = (Q1 + q0 + s) >> 2;
60
1472174
            Q1 = (2*Q1 + s) >> 2;
61
        } else
62
75462
            Q0 = (2*Q1 + s) >> 2;
63
    }
64
1732320
}
65
66
2746000
static inline void loop_filter_l1(uint8_t *p0_p, ptrdiff_t stride, int alpha, int beta, int tc)
67
{
68
2746000
    int p0 = P0;
69
2746000
    int q0 = Q0;
70
71

2746000
    if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
72
2131300
        int delta = av_clip(((q0-p0)*3+P1-Q1+4)>>3,-tc, tc);
73
2131300
        P0 = av_clip_uint8(p0+delta);
74
2131300
        Q0 = av_clip_uint8(q0-delta);
75
2131300
        if(abs(P2-p0)<beta) {
76
2053935
            delta = av_clip(((P0-P1)*3+P2-Q0+4)>>3, -tc, tc);
77
2053935
            P1 = av_clip_uint8(P1+delta);
78
        }
79
2131300
        if(abs(Q2-q0)<beta) {
80
2050257
            delta = av_clip(((Q1-Q0)*3+P0-Q2+4)>>3, -tc, tc);
81
2050257
            Q1 = av_clip_uint8(Q1-delta);
82
        }
83
    }
84
2746000
}
85
86
921312
static inline void loop_filter_c2(uint8_t *p0_p, ptrdiff_t stride, int alpha, int beta)
87
{
88
921312
    int p0 = P0;
89
921312
    int q0 = Q0;
90
91

921312
    if(abs(p0-q0)<alpha && abs(P1-p0)<beta && abs(Q1-q0)<beta) {
92
855849
        int s = p0 + q0 + 2;
93
855849
        alpha = (alpha>>2) + 2;
94

855849
        if(abs(P2-p0) < beta && abs(p0-q0) < alpha) {
95
820111
            P0 = (P1 + p0 + s) >> 2;
96
        } else
97
35738
            P0 = (2*P1 + s) >> 2;
98

855849
        if(abs(Q2-q0) < beta && abs(q0-p0) < alpha) {
99
820095
            Q0 = (Q1 + q0 + s) >> 2;
100
        } else
101
35754
            Q0 = (2*Q1 + s) >> 2;
102
    }
103
921312
}
104
105
1980296
static inline void loop_filter_c1(uint8_t *p0_p, ptrdiff_t stride, int alpha, int beta,
106
                                  int tc)
107
{
108

1980296
    if(abs(P0-Q0)<alpha && abs(P1-P0)<beta && abs(Q1-Q0)<beta) {
109
1684429
        int delta = av_clip(((Q0-P0)*3+P1-Q1+4)>>3, -tc, tc);
110
1684429
        P0 = av_clip_uint8(P0+delta);
111
1684429
        Q0 = av_clip_uint8(Q0-delta);
112
    }
113
1980296
}
114
115
#undef P0
116
#undef P1
117
#undef P2
118
#undef Q0
119
#undef Q1
120
#undef Q2
121
122
262099
static void cavs_filter_lv_c(uint8_t *d, ptrdiff_t stride, int alpha, int beta, int tc,
123
                             int bs1, int bs2)
124
{
125
    int i;
126
262099
    if(bs1==2)
127
915756
        for(i=0;i<16;i++)
128
861888
            loop_filter_l2(d + i*stride,1,alpha,beta);
129
    else {
130
208231
        if(bs1)
131
775233
            for(i=0;i<8;i++)
132
689096
                loop_filter_l1(d + i*stride,1,alpha,beta,tc);
133
208231
        if (bs2)
134
756180
            for(i=8;i<16;i++)
135
672160
                loop_filter_l1(d + i*stride,1,alpha,beta,tc);
136
    }
137
262099
}
138
139
260436
static void cavs_filter_lh_c(uint8_t *d, ptrdiff_t stride, int alpha, int beta, int tc,
140
                             int bs1, int bs2)
141
{
142
    int i;
143
260436
    if(bs1==2)
144
924834
        for(i=0;i<16;i++)
145
870432
            loop_filter_l2(d + i,stride,alpha,beta);
146
    else {
147
206034
        if(bs1)
148
773244
            for(i=0;i<8;i++)
149
687328
                loop_filter_l1(d + i,stride,alpha,beta,tc);
150
206034
        if (bs2)
151
784593
            for(i=8;i<16;i++)
152
697416
                loop_filter_l1(d + i,stride,alpha,beta,tc);
153
    }
154
260436
}
155
156
255482
static void cavs_filter_cv_c(uint8_t *d, ptrdiff_t stride, int alpha, int beta, int tc,
157
                             int bs1, int bs2)
158
{
159
    int i;
160
255482
    if(bs1==2)
161
513432
        for(i=0;i<8;i++)
162
456384
            loop_filter_c2(d + i*stride,1,alpha,beta);
163
    else {
164
198434
        if(bs1)
165
626070
            for(i=0;i<4;i++)
166
500856
                loop_filter_c1(d + i*stride,1,alpha,beta,tc);
167
198434
        if (bs2)
168
610140
            for(i=4;i<8;i++)
169
488112
                loop_filter_c1(d + i*stride,1,alpha,beta,tc);
170
    }
171
255482
}
172
173
252156
static void cavs_filter_ch_c(uint8_t *d, ptrdiff_t stride, int alpha, int beta, int tc,
174
                             int bs1, int bs2)
175
{
176
    int i;
177
252156
    if(bs1==2)
178
523044
        for(i=0;i<8;i++)
179
464928
            loop_filter_c2(d + i,stride,alpha,beta);
180
    else {
181
194040
        if(bs1)
182
612820
            for(i=0;i<4;i++)
183
490256
                loop_filter_c1(d + i,stride,alpha,beta,tc);
184
194040
        if (bs2)
185
626340
            for(i=4;i<8;i++)
186
501072
                loop_filter_c1(d + i,stride,alpha,beta,tc);
187
    }
188
252156
}
189
190
/*****************************************************************************
191
 *
192
 * inverse transform
193
 *
194
 ****************************************************************************/
195
196
351600
static void cavs_idct8_add_c(uint8_t *dst, int16_t *block, ptrdiff_t stride)
197
{
198
    int i;
199
351600
    int16_t (*src)[8] = (int16_t(*)[8])block;
200
201
351600
    src[0][0] += 8;
202
203
3164400
    for( i = 0; i < 8; i++ ) {
204
2812800
        const int a0 =  3*src[i][1] - (src[i][7]<<1);
205
2812800
        const int a1 =  3*src[i][3] + (src[i][5]<<1);
206
2812800
        const int a2 =  (src[i][3]<<1) - 3*src[i][5];
207
2812800
        const int a3 =  (src[i][1]<<1) + 3*src[i][7];
208
209
2812800
        const int b4 = ((a0 + a1 + a3)<<1) + a1;
210
2812800
        const int b5 = ((a0 - a1 + a2)<<1) + a0;
211
2812800
        const int b6 = ((a3 - a2 - a1)<<1) + a3;
212
2812800
        const int b7 = ((a0 - a2 - a3)<<1) - a2;
213
214
2812800
        const int a7 = (src[i][2]<<2) - 10*src[i][6];
215
2812800
        const int a6 = (src[i][6]<<2) + 10*src[i][2];
216
2812800
        const int a5 = ((src[i][0] - src[i][4]) << 3) + 4;
217
2812800
        const int a4 = ((src[i][0] + src[i][4]) << 3) + 4;
218
219
2812800
        const int b0 = a4 + a6;
220
2812800
        const int b1 = a5 + a7;
221
2812800
        const int b2 = a5 - a7;
222
2812800
        const int b3 = a4 - a6;
223
224
2812800
        src[i][0] = (b0 + b4) >> 3;
225
2812800
        src[i][1] = (b1 + b5) >> 3;
226
2812800
        src[i][2] = (b2 + b6) >> 3;
227
2812800
        src[i][3] = (b3 + b7) >> 3;
228
2812800
        src[i][4] = (b3 - b7) >> 3;
229
2812800
        src[i][5] = (b2 - b6) >> 3;
230
2812800
        src[i][6] = (b1 - b5) >> 3;
231
2812800
        src[i][7] = (b0 - b4) >> 3;
232
    }
233
3164400
    for( i = 0; i < 8; i++ ) {
234
2812800
        const int a0 =  3*src[1][i] - (src[7][i]<<1);
235
2812800
        const int a1 =  3*src[3][i] + (src[5][i]<<1);
236
2812800
        const int a2 =  (src[3][i]<<1) - 3*src[5][i];
237
2812800
        const int a3 =  (src[1][i]<<1) + 3*src[7][i];
238
239
2812800
        const int b4 = ((a0 + a1 + a3)<<1) + a1;
240
2812800
        const int b5 = ((a0 - a1 + a2)<<1) + a0;
241
2812800
        const int b6 = ((a3 - a2 - a1)<<1) + a3;
242
2812800
        const int b7 = ((a0 - a2 - a3)<<1) - a2;
243
244
2812800
        const int a7 = (src[2][i]<<2) - 10*src[6][i];
245
2812800
        const int a6 = (src[6][i]<<2) + 10*src[2][i];
246
2812800
        const int a5 = (src[0][i] - src[4][i]) << 3;
247
2812800
        const int a4 = (src[0][i] + src[4][i]) << 3;
248
249
2812800
        const int b0 = a4 + a6;
250
2812800
        const int b1 = a5 + a7;
251
2812800
        const int b2 = a5 - a7;
252
2812800
        const int b3 = a4 - a6;
253
254
2812800
        dst[i + 0*stride] = av_clip_uint8( dst[i + 0*stride] + ((b0 + b4) >> 7));
255
2812800
        dst[i + 1*stride] = av_clip_uint8( dst[i + 1*stride] + ((b1 + b5) >> 7));
256
2812800
        dst[i + 2*stride] = av_clip_uint8( dst[i + 2*stride] + ((b2 + b6) >> 7));
257
2812800
        dst[i + 3*stride] = av_clip_uint8( dst[i + 3*stride] + ((b3 + b7) >> 7));
258
2812800
        dst[i + 4*stride] = av_clip_uint8( dst[i + 4*stride] + ((b3 - b7) >> 7));
259
2812800
        dst[i + 5*stride] = av_clip_uint8( dst[i + 5*stride] + ((b2 - b6) >> 7));
260
2812800
        dst[i + 6*stride] = av_clip_uint8( dst[i + 6*stride] + ((b1 - b5) >> 7));
261
2812800
        dst[i + 7*stride] = av_clip_uint8( dst[i + 7*stride] + ((b0 - b4) >> 7));
262
    }
263
351600
}
264
265
/*****************************************************************************
266
 *
267
 * motion compensation
268
 *
269
 ****************************************************************************/
270
271
#define CAVS_SUBPIX(OPNAME, OP, NAME, A, B, C, D, E, F) \
272
static void OPNAME ## cavs_filt8_h_ ## NAME(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
273
{                                                                       \
274
    const int h=8;\
275
    const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
276
    int i;\
277
    for(i=0; i<h; i++)\
278
    {\
279
        OP(dst[0], A*src[-2] + B*src[-1] + C*src[0] + D*src[1] + E*src[2] + F*src[3]);\
280
        OP(dst[1], A*src[-1] + B*src[ 0] + C*src[1] + D*src[2] + E*src[3] + F*src[4]);\
281
        OP(dst[2], A*src[ 0] + B*src[ 1] + C*src[2] + D*src[3] + E*src[4] + F*src[5]);\
282
        OP(dst[3], A*src[ 1] + B*src[ 2] + C*src[3] + D*src[4] + E*src[5] + F*src[6]);\
283
        OP(dst[4], A*src[ 2] + B*src[ 3] + C*src[4] + D*src[5] + E*src[6] + F*src[7]);\
284
        OP(dst[5], A*src[ 3] + B*src[ 4] + C*src[5] + D*src[6] + E*src[7] + F*src[8]);\
285
        OP(dst[6], A*src[ 4] + B*src[ 5] + C*src[6] + D*src[7] + E*src[8] + F*src[9]);\
286
        OP(dst[7], A*src[ 5] + B*src[ 6] + C*src[7] + D*src[8] + E*src[9] + F*src[10]);\
287
        dst+=dstStride;\
288
        src+=srcStride;\
289
    }\
290
}\
291
\
292
static void OPNAME ## cavs_filt8_v_  ## NAME(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
293
{                                                                       \
294
    const int w=8;\
295
    const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
296
    int i;\
297
    for(i=0; i<w; i++)\
298
    {\
299
        const int srcB= src[-2*srcStride];\
300
        const int srcA= src[-1*srcStride];\
301
        const int src0= src[0 *srcStride];\
302
        const int src1= src[1 *srcStride];\
303
        const int src2= src[2 *srcStride];\
304
        const int src3= src[3 *srcStride];\
305
        const int src4= src[4 *srcStride];\
306
        const int src5= src[5 *srcStride];\
307
        const int src6= src[6 *srcStride];\
308
        const int src7= src[7 *srcStride];\
309
        const int src8= src[8 *srcStride];\
310
        const int src9= src[9 *srcStride];\
311
        const int src10= src[10 *srcStride];\
312
        OP(dst[0*dstStride], A*srcB + B*srcA + C*src0 + D*src1 + E*src2 + F*src3);\
313
        OP(dst[1*dstStride], A*srcA + B*src0 + C*src1 + D*src2 + E*src3 + F*src4);\
314
        OP(dst[2*dstStride], A*src0 + B*src1 + C*src2 + D*src3 + E*src4 + F*src5);\
315
        OP(dst[3*dstStride], A*src1 + B*src2 + C*src3 + D*src4 + E*src5 + F*src6);\
316
        OP(dst[4*dstStride], A*src2 + B*src3 + C*src4 + D*src5 + E*src6 + F*src7);\
317
        OP(dst[5*dstStride], A*src3 + B*src4 + C*src5 + D*src6 + E*src7 + F*src8);\
318
        OP(dst[6*dstStride], A*src4 + B*src5 + C*src6 + D*src7 + E*src8 + F*src9);\
319
        OP(dst[7*dstStride], A*src5 + B*src6 + C*src7 + D*src8 + E*src9 + F*src10);\
320
        dst++;\
321
        src++;\
322
    }\
323
}\
324
\
325
static void OPNAME ## cavs_filt16_v_ ## NAME(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
326
{                                                                       \
327
    OPNAME ## cavs_filt8_v_ ## NAME(dst  , src  , dstStride, srcStride);\
328
    OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\
329
    src += 8*srcStride;\
330
    dst += 8*dstStride;\
331
    OPNAME ## cavs_filt8_v_ ## NAME(dst  , src  , dstStride, srcStride);\
332
    OPNAME ## cavs_filt8_v_ ## NAME(dst+8, src+8, dstStride, srcStride);\
333
}\
334
\
335
static void OPNAME ## cavs_filt16_h_ ## NAME(uint8_t *dst, const uint8_t *src, ptrdiff_t dstStride, ptrdiff_t srcStride)\
336
{                                                                       \
337
    OPNAME ## cavs_filt8_h_ ## NAME(dst  , src  , dstStride, srcStride);\
338
    OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\
339
    src += 8*srcStride;\
340
    dst += 8*dstStride;\
341
    OPNAME ## cavs_filt8_h_ ## NAME(dst  , src  , dstStride, srcStride);\
342
    OPNAME ## cavs_filt8_h_ ## NAME(dst+8, src+8, dstStride, srcStride);\
343
}\
344
345
#define CAVS_SUBPIX_HV(OPNAME, OP, NAME, AH, BH, CH, DH, EH, FH, AV, BV, CV, DV, EV, FV, FULL) \
346
static void OPNAME ## cavs_filt8_hv_ ## NAME(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t srcStride)\
347
{                                                                       \
348
    int16_t temp[8*(8+5)];\
349
    int16_t *tmp = temp;\
350
    const int h=8;\
351
    const int w=8;\
352
    const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
353
    int i;\
354
    src1 -= 2*srcStride;\
355
    for(i=0; i<h+5; i++)\
356
    {\
357
        tmp[0]= AH*src1[-2] + BH*src1[-1] + CH*src1[0] + DH*src1[1] + EH*src1[2] + FH*src1[3];\
358
        tmp[1]= AH*src1[-1] + BH*src1[ 0] + CH*src1[1] + DH*src1[2] + EH*src1[3] + FH*src1[4];\
359
        tmp[2]= AH*src1[ 0] + BH*src1[ 1] + CH*src1[2] + DH*src1[3] + EH*src1[4] + FH*src1[5];\
360
        tmp[3]= AH*src1[ 1] + BH*src1[ 2] + CH*src1[3] + DH*src1[4] + EH*src1[5] + FH*src1[6];\
361
        tmp[4]= AH*src1[ 2] + BH*src1[ 3] + CH*src1[4] + DH*src1[5] + EH*src1[6] + FH*src1[7];\
362
        tmp[5]= AH*src1[ 3] + BH*src1[ 4] + CH*src1[5] + DH*src1[6] + EH*src1[7] + FH*src1[8];\
363
        tmp[6]= AH*src1[ 4] + BH*src1[ 5] + CH*src1[6] + DH*src1[7] + EH*src1[8] + FH*src1[9];\
364
        tmp[7]= AH*src1[ 5] + BH*src1[ 6] + CH*src1[7] + DH*src1[8] + EH*src1[9] + FH*src1[10];\
365
        tmp+=8;\
366
        src1+=srcStride;\
367
    }\
368
    if(FULL) {\
369
      tmp = temp+8*2;                           \
370
      for(i=0; i<w; i++)                        \
371
        {                                       \
372
          const int tmpB= tmp[-2*8];    \
373
          const int tmpA= tmp[-1*8];    \
374
          const int tmp0= tmp[0 *8];    \
375
          const int tmp1= tmp[1 *8];    \
376
          const int tmp2= tmp[2 *8];    \
377
          const int tmp3= tmp[3 *8];    \
378
          const int tmp4= tmp[4 *8];    \
379
          const int tmp5= tmp[5 *8];    \
380
          const int tmp6= tmp[6 *8];    \
381
          const int tmp7= tmp[7 *8];    \
382
          const int tmp8= tmp[8 *8];    \
383
          const int tmp9= tmp[9 *8];    \
384
          const int tmp10=tmp[10*8];                            \
385
          OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3 + 64*src2[0*srcStride]); \
386
          OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4 + 64*src2[1*srcStride]); \
387
          OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5 + 64*src2[2*srcStride]); \
388
          OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6 + 64*src2[3*srcStride]); \
389
          OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7 + 64*src2[4*srcStride]); \
390
          OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8 + 64*src2[5*srcStride]); \
391
          OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9 + 64*src2[6*srcStride]); \
392
          OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10 + 64*src2[7*srcStride]); \
393
          dst++;                                                        \
394
          tmp++;                                                        \
395
          src2++;                                                       \
396
        }                                                               \
397
    } else {\
398
      tmp = temp+8*2;                           \
399
      for(i=0; i<w; i++)                        \
400
        {                                       \
401
          const int tmpB= tmp[-2*8];    \
402
          const int tmpA= tmp[-1*8];    \
403
          const int tmp0= tmp[0 *8];    \
404
          const int tmp1= tmp[1 *8];    \
405
          const int tmp2= tmp[2 *8];    \
406
          const int tmp3= tmp[3 *8];    \
407
          const int tmp4= tmp[4 *8];    \
408
          const int tmp5= tmp[5 *8];    \
409
          const int tmp6= tmp[6 *8];    \
410
          const int tmp7= tmp[7 *8];    \
411
          const int tmp8= tmp[8 *8];    \
412
          const int tmp9= tmp[9 *8];    \
413
          const int tmp10=tmp[10*8];                            \
414
          OP(dst[0*dstStride], AV*tmpB + BV*tmpA + CV*tmp0 + DV*tmp1 + EV*tmp2 + FV*tmp3); \
415
          OP(dst[1*dstStride], AV*tmpA + BV*tmp0 + CV*tmp1 + DV*tmp2 + EV*tmp3 + FV*tmp4); \
416
          OP(dst[2*dstStride], AV*tmp0 + BV*tmp1 + CV*tmp2 + DV*tmp3 + EV*tmp4 + FV*tmp5); \
417
          OP(dst[3*dstStride], AV*tmp1 + BV*tmp2 + CV*tmp3 + DV*tmp4 + EV*tmp5 + FV*tmp6); \
418
          OP(dst[4*dstStride], AV*tmp2 + BV*tmp3 + CV*tmp4 + DV*tmp5 + EV*tmp6 + FV*tmp7); \
419
          OP(dst[5*dstStride], AV*tmp3 + BV*tmp4 + CV*tmp5 + DV*tmp6 + EV*tmp7 + FV*tmp8); \
420
          OP(dst[6*dstStride], AV*tmp4 + BV*tmp5 + CV*tmp6 + DV*tmp7 + EV*tmp8 + FV*tmp9); \
421
          OP(dst[7*dstStride], AV*tmp5 + BV*tmp6 + CV*tmp7 + DV*tmp8 + EV*tmp9 + FV*tmp10); \
422
          dst++;                                                        \
423
          tmp++;                                                        \
424
        }                                                               \
425
    }\
426
}\
427
\
428
static void OPNAME ## cavs_filt16_hv_ ## NAME(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, ptrdiff_t dstStride, ptrdiff_t srcStride)\
429
{                                                                       \
430
    OPNAME ## cavs_filt8_hv_ ## NAME(dst  , src1,   src2  , dstStride, srcStride); \
431
    OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, src2+8, dstStride, srcStride); \
432
    src1 += 8*srcStride;\
433
    src2 += 8*srcStride;\
434
    dst += 8*dstStride;\
435
    OPNAME ## cavs_filt8_hv_ ## NAME(dst  , src1,   src2  , dstStride, srcStride); \
436
    OPNAME ## cavs_filt8_hv_ ## NAME(dst+8, src1+8, src2+8, dstStride, srcStride); \
437
}\
438
439
#define CAVS_MC(OPNAME, SIZE) \
440
static void OPNAME ## cavs_qpel ## SIZE ## _mc10_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
441
{\
442
    OPNAME ## cavs_filt ## SIZE ## _h_qpel_l(dst, src, stride, stride);\
443
}\
444
\
445
static void OPNAME ## cavs_qpel ## SIZE ## _mc20_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
446
{\
447
    OPNAME ## cavs_filt ## SIZE ## _h_hpel(dst, src, stride, stride);\
448
}\
449
\
450
static void OPNAME ## cavs_qpel ## SIZE ## _mc30_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
451
{\
452
    OPNAME ## cavs_filt ## SIZE ## _h_qpel_r(dst, src, stride, stride);\
453
}\
454
\
455
static void OPNAME ## cavs_qpel ## SIZE ## _mc01_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
456
{\
457
    OPNAME ## cavs_filt ## SIZE ## _v_qpel_l(dst, src, stride, stride);\
458
}\
459
\
460
static void OPNAME ## cavs_qpel ## SIZE ## _mc02_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
461
{\
462
    OPNAME ## cavs_filt ## SIZE ## _v_hpel(dst, src, stride, stride);\
463
}\
464
\
465
static void OPNAME ## cavs_qpel ## SIZE ## _mc03_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
466
{\
467
    OPNAME ## cavs_filt ## SIZE ## _v_qpel_r(dst, src, stride, stride);\
468
}\
469
\
470
static void OPNAME ## cavs_qpel ## SIZE ## _mc22_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
471
{\
472
  OPNAME ## cavs_filt ## SIZE ## _hv_jj(dst, src, NULL, stride, stride); \
473
}\
474
\
475
static void OPNAME ## cavs_qpel ## SIZE ## _mc11_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
476
{\
477
  OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src, stride, stride); \
478
}\
479
\
480
static void OPNAME ## cavs_qpel ## SIZE ## _mc13_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
481
{\
482
  OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride, stride, stride); \
483
}\
484
\
485
static void OPNAME ## cavs_qpel ## SIZE ## _mc31_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
486
{\
487
  OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+1, stride, stride); \
488
}\
489
\
490
static void OPNAME ## cavs_qpel ## SIZE ## _mc33_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
491
{\
492
  OPNAME ## cavs_filt ## SIZE ## _hv_egpr(dst, src, src+stride+1,stride, stride); \
493
}\
494
\
495
static void OPNAME ## cavs_qpel ## SIZE ## _mc21_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
496
{\
497
  OPNAME ## cavs_filt ## SIZE ## _hv_ff(dst, src, src+stride+1,stride, stride); \
498
}\
499
\
500
static void OPNAME ## cavs_qpel ## SIZE ## _mc12_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
501
{\
502
  OPNAME ## cavs_filt ## SIZE ## _hv_ii(dst, src, src+stride+1,stride, stride); \
503
}\
504
\
505
static void OPNAME ## cavs_qpel ## SIZE ## _mc32_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
506
{\
507
  OPNAME ## cavs_filt ## SIZE ## _hv_kk(dst, src, src+stride+1,stride, stride); \
508
}\
509
\
510
static void OPNAME ## cavs_qpel ## SIZE ## _mc23_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
511
{\
512
  OPNAME ## cavs_filt ## SIZE ## _hv_qq(dst, src, src+stride+1,stride, stride); \
513
}\
514
515
#define op_put1(a, b)  a = cm[((b)+4)>>3]
516
#define op_put2(a, b)  a = cm[((b)+64)>>7]
517
#define op_put3(a, b)  a = cm[((b)+32)>>6]
518
#define op_put4(a, b)  a = cm[((b)+512)>>10]
519
#define op_avg1(a, b)  a = ((a)+cm[((b)+4)>>3]   +1)>>1
520
#define op_avg2(a, b)  a = ((a)+cm[((b)+64)>>7]  +1)>>1
521
#define op_avg3(a, b)  a = ((a)+cm[((b)+32)>>6]  +1)>>1
522
#define op_avg4(a, b)  a = ((a)+cm[((b)+512)>>10]+1)>>1
523
148095
CAVS_SUBPIX(put_   , op_put1, hpel,    0, -1,  5,  5, -1,  0)
524
200007
CAVS_SUBPIX(put_   , op_put2, qpel_l, -1, -2, 96, 42, -7,  0)
525
173142
CAVS_SUBPIX(put_   , op_put2, qpel_r,  0, -7, 42, 96, -2, -1)
526

58740
CAVS_SUBPIX_HV(put_, op_put3, jj,      0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 0)
527

80828
CAVS_SUBPIX_HV(put_, op_put4, ff,      0, -1,  5,  5, -1,  0, -1, -2, 96, 42, -7, 0, 0)
528

77814
CAVS_SUBPIX_HV(put_, op_put4, ii,     -1, -2, 96, 42, -7,  0,  0, -1,  5,  5, -1, 0, 0)
529

89606
CAVS_SUBPIX_HV(put_, op_put4, kk,      0, -7, 42, 96, -2, -1,  0, -1,  5,  5, -1, 0, 0)
530

79618
CAVS_SUBPIX_HV(put_, op_put4, qq,      0, -1,  5,  5, -1,  0,  0, -7, 42, 96, -2,-1, 0)
531

485562
CAVS_SUBPIX_HV(put_, op_put2, egpr,    0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 1)
532
95643
CAVS_SUBPIX(avg_   , op_avg1, hpel,    0, -1,  5,  5, -1,  0)
533
42804
CAVS_SUBPIX(avg_   , op_avg2, qpel_l, -1, -2, 96, 42, -7,  0)
534
47313
CAVS_SUBPIX(avg_   , op_avg2, qpel_r,  0, -7, 42, 96, -2, -1)
535

43494
CAVS_SUBPIX_HV(avg_, op_avg3, jj,      0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 0)
536

14762
CAVS_SUBPIX_HV(avg_, op_avg4, ff,      0, -1,  5,  5, -1,  0, -1, -2, 96, 42, -7, 0, 0)
537

13552
CAVS_SUBPIX_HV(avg_, op_avg4, ii,     -1, -2, 96, 42, -7,  0,  0, -1,  5,  5, -1, 0, 0)
538

15708
CAVS_SUBPIX_HV(avg_, op_avg4, kk,      0, -7, 42, 96, -2, -1,  0, -1,  5,  5, -1, 0, 0)
539

19382
CAVS_SUBPIX_HV(avg_, op_avg4, qq,      0, -1,  5,  5, -1,  0,  0, -7, 42, 96, -2,-1, 0)
540

65868
CAVS_SUBPIX_HV(avg_, op_avg2, egpr,    0, -1,  5,  5, -1,  0,  0, -1,  5,  5, -1, 0, 1)
541
15707
CAVS_MC(put_, 8)
542
1629
CAVS_MC(put_, 16)
543
4444
CAVS_MC(avg_, 8)
544
78
CAVS_MC(avg_, 16)
545
546
#define put_cavs_qpel8_mc00_c  ff_put_pixels8x8_c
547
#define avg_cavs_qpel8_mc00_c  ff_avg_pixels8x8_c
548
#define put_cavs_qpel16_mc00_c ff_put_pixels16x16_c
549
#define avg_cavs_qpel16_mc00_c ff_avg_pixels16x16_c
550
551
4
av_cold void ff_cavsdsp_init(CAVSDSPContext* c, AVCodecContext *avctx) {
552
#define dspfunc(PFX, IDX, NUM) \
553
    c->PFX ## _pixels_tab[IDX][ 0] = PFX ## NUM ## _mc00_c; \
554
    c->PFX ## _pixels_tab[IDX][ 1] = PFX ## NUM ## _mc10_c; \
555
    c->PFX ## _pixels_tab[IDX][ 2] = PFX ## NUM ## _mc20_c; \
556
    c->PFX ## _pixels_tab[IDX][ 3] = PFX ## NUM ## _mc30_c; \
557
    c->PFX ## _pixels_tab[IDX][ 4] = PFX ## NUM ## _mc01_c; \
558
    c->PFX ## _pixels_tab[IDX][ 5] = PFX ## NUM ## _mc11_c; \
559
    c->PFX ## _pixels_tab[IDX][ 6] = PFX ## NUM ## _mc21_c; \
560
    c->PFX ## _pixels_tab[IDX][ 7] = PFX ## NUM ## _mc31_c; \
561
    c->PFX ## _pixels_tab[IDX][ 8] = PFX ## NUM ## _mc02_c; \
562
    c->PFX ## _pixels_tab[IDX][ 9] = PFX ## NUM ## _mc12_c; \
563
    c->PFX ## _pixels_tab[IDX][10] = PFX ## NUM ## _mc22_c; \
564
    c->PFX ## _pixels_tab[IDX][11] = PFX ## NUM ## _mc32_c; \
565
    c->PFX ## _pixels_tab[IDX][12] = PFX ## NUM ## _mc03_c; \
566
    c->PFX ## _pixels_tab[IDX][13] = PFX ## NUM ## _mc13_c; \
567
    c->PFX ## _pixels_tab[IDX][14] = PFX ## NUM ## _mc23_c; \
568
    c->PFX ## _pixels_tab[IDX][15] = PFX ## NUM ## _mc33_c
569
4
    dspfunc(put_cavs_qpel, 0, 16);
570
4
    dspfunc(put_cavs_qpel, 1, 8);
571
4
    dspfunc(avg_cavs_qpel, 0, 16);
572
4
    dspfunc(avg_cavs_qpel, 1, 8);
573
4
    c->cavs_filter_lv = cavs_filter_lv_c;
574
4
    c->cavs_filter_lh = cavs_filter_lh_c;
575
4
    c->cavs_filter_cv = cavs_filter_cv_c;
576
4
    c->cavs_filter_ch = cavs_filter_ch_c;
577
4
    c->cavs_idct8_add = cavs_idct8_add_c;
578
4
    c->idct_perm = FF_IDCT_PERM_NONE;
579
580
    if (ARCH_X86)
581
4
        ff_cavsdsp_init_x86(c, avctx);
582
4
}