FFmpeg coverage

GCC Code Coverage Report

Directory:	../../../ffmpeg/		Exec	Total	Coverage
File:	src/libavcodec/xvididct.c	Lines:	198	205	96.6 %
Date:	2020-08-14 10:39:37	Branches:	33	38	86.8 %


/*
 * Xvid MPEG-4 IDCT
 *
 * Copyright (C) 2006-2011 Xvid Solutions GmbH
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

/**
 * @file
 * Walken IDCT
 * Alternative IDCT implementation for decoding compatibility.
 *
 * @author Skal
 * @note This C version is not the original IDCT, but a modified one that
 *       yields the same error profile as the MMX/MMXEXT/SSE2 versions.
 */

#include "config.h"
#include "libavutil/attributes.h"
#include "avcodec.h"
#include "idctdsp.h"
#include "xvididct.h"

#define ROW_SHIFT 11
#define COL_SHIFT  6

// #define FIX(x)   (int)((x) * (1 << ROW_SHIFT))
#define RND0 65536 // 1 << (COL_SHIFT + ROW_SHIFT - 1);
#define RND1 3597  // FIX (1.75683487303);
#define RND2 2260  // FIX (1.10355339059);
#define RND3 1203  // FIX (0.587788325588);
#define RND4 0
#define RND5 120   // FIX (0.058658283817);
#define RND6 512   // FIX (0.25);
#define RND7 512   // FIX (0.25);

static const int TAB04[] = { 22725, 21407, 19266, 16384, 12873,  8867, 4520 };
static const int TAB17[] = { 31521, 29692, 26722, 22725, 17855, 12299, 6270 };
static const int TAB26[] = { 29692, 27969, 25172, 21407, 16819, 11585, 5906 };
static const int TAB35[] = { 26722, 25172, 22654, 19266, 15137, 10426, 5315 };

static int idct_row(short *in, const int *const tab, int rnd)
{
    const int c1 = tab[0];
    const int c2 = tab[1];
    const int c3 = tab[2];
    const int c4 = tab[3];
    const int c5 = tab[4];
    const int c6 = tab[5];
    const int c7 = tab[6];

    const int right = in[5] | in[6] | in[7];
    const int left  = in[1] | in[2] | in[3];
    if (!(right | in[4])) {
        const int k = c4 * in[0] + rnd;
        if (left) {
            const int a0 = k + c2 * in[2];
            const int a1 = k + c6 * in[2];
            const int a2 = k - c6 * in[2];
            const int a3 = k - c2 * in[2];

            const int b0 = c1 * in[1] + c3 * in[3];
            const int b1 = c3 * in[1] - c7 * in[3];
            const int b2 = c5 * in[1] - c1 * in[3];
            const int b3 = c7 * in[1] - c5 * in[3];

            in[0] = (a0 + b0) >> ROW_SHIFT;
            in[1] = (a1 + b1) >> ROW_SHIFT;
            in[2] = (a2 + b2) >> ROW_SHIFT;
            in[3] = (a3 + b3) >> ROW_SHIFT;
            in[4] = (a3 - b3) >> ROW_SHIFT;
            in[5] = (a2 - b2) >> ROW_SHIFT;
            in[6] = (a1 - b1) >> ROW_SHIFT;
            in[7] = (a0 - b0) >> ROW_SHIFT;
        } else {
            const int a0 = k >> ROW_SHIFT;
            if (a0) {
                in[0] =
                in[1] =
                in[2] =
                in[3] =
                in[4] =
                in[5] =
                in[6] =
                in[7] = a0;
            } else
                return 0;
        }
    } else if (!(left | right)) {
        const int a0 = (rnd + c4 * (in[0] + in[4])) >> ROW_SHIFT;
        const int a1 = (rnd + c4 * (in[0] - in[4])) >> ROW_SHIFT;

        in[0] = a0;
        in[3] = a0;
        in[4] = a0;
        in[7] = a0;
        in[1] = a1;
        in[2] = a1;
        in[5] = a1;
        in[6] = a1;
    } else {
        const int k  = c4 * in[0] + rnd;
        const unsigned int a0 = k + c2 * in[2] + c4 * in[4] + c6 * in[6];
        const unsigned int a1 = k + c6 * in[2] - c4 * in[4] - c2 * in[6];
        const unsigned int a2 = k - c6 * in[2] - c4 * in[4] + c2 * in[6];
        const unsigned int a3 = k - c2 * in[2] + c4 * in[4] - c6 * in[6];

        const unsigned int b0 = c1 * in[1] + c3 * in[3] + c5 * in[5] + c7 * in[7];
        const unsigned int b1 = c3 * in[1] - c7 * in[3] - c1 * in[5] - c5 * in[7];
        const unsigned int b2 = c5 * in[1] - c1 * in[3] + c7 * in[5] + c3 * in[7];
        const unsigned int b3 = c7 * in[1] - c5 * in[3] + c3 * in[5] - c1 * in[7];

        in[0] = (int)(a0 + b0) >> ROW_SHIFT;
        in[1] = (int)(a1 + b1) >> ROW_SHIFT;
        in[2] = (int)(a2 + b2) >> ROW_SHIFT;
        in[3] = (int)(a3 + b3) >> ROW_SHIFT;
        in[4] = (int)(a3 - b3) >> ROW_SHIFT;
        in[5] = (int)(a2 - b2) >> ROW_SHIFT;
        in[6] = (int)(a1 - b1) >> ROW_SHIFT;
        in[7] = (int)(a0 - b0) >> ROW_SHIFT;
    }
    return 1;
}

#define TAN1  0x32EC
#define TAN2  0x6A0A
#define TAN3  0xAB0E
#define SQRT2 0x5A82

#define MULT(c, x, n)  ((unsigned)((int)((c) * (unsigned)(x)) >> (n)))
// 12b version => #define MULT(c,x, n)  ((((c) >> 3) * (x)) >> ((n) - 3))
// 12b zero-testing version:

#define BUTTERFLY(a, b, tmp)     \
    (tmp) = (a) + (b);           \
    (b)   = (a) - (b);           \
    (a)   = (tmp)

#define LOAD_BUTTERFLY(m1, m2, a, b, tmp, s)   \
    (m1) = (s)[(a)] + (s)[(b)];                \
    (m2) = (s)[(a)] - (s)[(b)]

static void idct_col_8(short *const in)
{
    int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, spill;

    // odd

    mm4 = (int) in[7 * 8];
    mm5 = (int) in[5 * 8];
    mm6 = (int) in[3 * 8];
    mm7 = (int) in[1 * 8];

    mm0 = MULT(TAN1, mm4, 16) + mm7;
    mm1 = MULT(TAN1, mm7, 16) - mm4;
    mm2 = MULT(TAN3, mm5, 16) + mm6;
    mm3 = MULT(TAN3, mm6, 16) - mm5;

    mm7 = mm0 + mm2;
    mm4 = mm1 - mm3;
    mm0 = mm0 - mm2;
    mm1 = mm1 + mm3;
    mm6 = mm0 + mm1;
    mm5 = mm0 - mm1;
    mm5 = 2 * MULT(SQRT2, mm5, 16); // 2*sqrt2
    mm6 = 2 * MULT(SQRT2, mm6, 16); // Watch out: precision loss but done to match
                                    // the pmulhw used in MMX/MMXEXT/SSE2 versions

    // even

    mm1 = (int) in[2 * 8];
    mm2 = (int) in[6 * 8];
    mm3 = MULT(TAN2, mm2, 16) + mm1;
    mm2 = MULT(TAN2, mm1, 16) - mm2;

    LOAD_BUTTERFLY(mm0, mm1, 0 * 8, 4 * 8, spill, in);

    BUTTERFLY(mm0, mm3, spill);
    BUTTERFLY(mm0, mm7, spill);
    in[8 * 0] = (int16_t) (mm0 >> COL_SHIFT);
    in[8 * 7] = (int16_t) (mm7 >> COL_SHIFT);
    BUTTERFLY(mm3, mm4, mm0);
    in[8 * 3] = (int16_t) (mm3 >> COL_SHIFT);
    in[8 * 4] = (int16_t) (mm4 >> COL_SHIFT);

    BUTTERFLY(mm1, mm2, mm0);
    BUTTERFLY(mm1, mm6, mm0);
    in[8 * 1] = (int16_t) (mm1 >> COL_SHIFT);
    in[8 * 6] = (int16_t) (mm6 >> COL_SHIFT);
    BUTTERFLY(mm2, mm5, mm0);
    in[8 * 2] = (int16_t) (mm2 >> COL_SHIFT);
    in[8 * 5] = (int16_t) (mm5 >> COL_SHIFT);
}

static void idct_col_4(short *const in)
{
    int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, spill;

    // odd

    mm0 = (int) in[1 * 8];
    mm2 = (int) in[3 * 8];

    mm1 = MULT(TAN1, mm0, 16);
    mm3 = MULT(TAN3, mm2, 16);

    mm7 = mm0 + mm2;
    mm4 = mm1 - mm3;
    mm0 = mm0 - mm2;
    mm1 = mm1 + mm3;
    mm6 = mm0 + mm1;
    mm5 = mm0 - mm1;
    mm6 = 2 * MULT(SQRT2, mm6, 16); // 2*sqrt2
    mm5 = 2 * MULT(SQRT2, mm5, 16);

    // even

    mm0 = mm1 = (int) in[0 * 8];
    mm3 = (int) in[2 * 8];
    mm2 = MULT(TAN2, mm3, 16);

    BUTTERFLY(mm0, mm3, spill);
    BUTTERFLY(mm0, mm7, spill);
    in[8 * 0] = (int16_t) (mm0 >> COL_SHIFT);
    in[8 * 7] = (int16_t) (mm7 >> COL_SHIFT);
    BUTTERFLY(mm3, mm4, mm0);
    in[8 * 3] = (int16_t) (mm3 >> COL_SHIFT);
    in[8 * 4] = (int16_t) (mm4 >> COL_SHIFT);

    BUTTERFLY(mm1, mm2, mm0);
    BUTTERFLY(mm1, mm6, mm0);
    in[8 * 1] = (int16_t) (mm1 >> COL_SHIFT);
    in[8 * 6] = (int16_t) (mm6 >> COL_SHIFT);
    BUTTERFLY(mm2, mm5, mm0);
    in[8 * 2] = (int16_t) (mm2 >> COL_SHIFT);
    in[8 * 5] = (int16_t) (mm5 >> COL_SHIFT);
}

static void idct_col_3(short *const in)
{
    int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, spill;

    // odd

    mm7 = (int) in[1 * 8];
    mm4 = MULT(TAN1, mm7, 16);

    mm6 = mm7 + mm4;
    mm5 = mm7 - mm4;
    mm6 = 2 * MULT(SQRT2, mm6, 16); // 2*sqrt2
    mm5 = 2 * MULT(SQRT2, mm5, 16);

    // even

    mm0 = mm1 = (int) in[0 * 8];
    mm3 = (int) in[2 * 8];
    mm2 = MULT(TAN2, mm3, 16);

    BUTTERFLY(mm0, mm3, spill);
    BUTTERFLY(mm0, mm7, spill);
    in[8 * 0] = (int16_t) (mm0 >> COL_SHIFT);
    in[8 * 7] = (int16_t) (mm7 >> COL_SHIFT);
    BUTTERFLY(mm3, mm4, mm0);
    in[8 * 3] = (int16_t) (mm3 >> COL_SHIFT);
    in[8 * 4] = (int16_t) (mm4 >> COL_SHIFT);

    BUTTERFLY(mm1, mm2, mm0);
    BUTTERFLY(mm1, mm6, mm0);
    in[8 * 1] = (int16_t) (mm1 >> COL_SHIFT);
    in[8 * 6] = (int16_t) (mm6 >> COL_SHIFT);
    BUTTERFLY(mm2, mm5, mm0);
    in[8 * 2] = (int16_t) (mm2 >> COL_SHIFT);
    in[8 * 5] = (int16_t) (mm5 >> COL_SHIFT);
}

void ff_xvid_idct(int16_t *const in)
{
    int i, rows = 0x07;

    idct_row(in + 0 * 8, TAB04, RND0);
    idct_row(in + 1 * 8, TAB17, RND1);
    idct_row(in + 2 * 8, TAB26, RND2);
    if (idct_row(in + 3 * 8, TAB35, RND3))
        rows |= 0x08;
    if (idct_row(in + 4 * 8, TAB04, RND4))
        rows |= 0x10;
    if (idct_row(in + 5 * 8, TAB35, RND5))
        rows |= 0x20;
    if (idct_row(in + 6 * 8, TAB26, RND6))
        rows |= 0x40;
    if (idct_row(in + 7 * 8, TAB17, RND7))
        rows |= 0x80;

    if (rows & 0xF0) {
        for (i = 0; i < 8; i++)
            idct_col_8(in + i);
    } else if (rows & 0x08) {
        for (i = 0; i < 8; i++)
            idct_col_4(in + i);
    } else {
        for (i = 0; i < 8; i++)
            idct_col_3(in + i);
    }
}

static void xvid_idct_put(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
{
    ff_xvid_idct(block);
    ff_put_pixels_clamped_c(block, dest, line_size);
}

static void xvid_idct_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
{
    ff_xvid_idct(block);
    ff_add_pixels_clamped_c(block, dest, line_size);
}

av_cold void ff_xvid_idct_init(IDCTDSPContext *c, AVCodecContext *avctx)
{
    const unsigned high_bit_depth = avctx->bits_per_raw_sample > 8;

    if (high_bit_depth || avctx->lowres ||
        !(avctx->idct_algo == FF_IDCT_AUTO ||
          avctx->idct_algo == FF_IDCT_XVID))
        return;

    if (avctx->idct_algo == FF_IDCT_XVID) {
        c->idct_put  = xvid_idct_put;
        c->idct_add  = xvid_idct_add;
        c->idct      = ff_xvid_idct;
        c->perm_type = FF_IDCT_PERM_NONE;
    }

    if (ARCH_X86)
        ff_xvid_idct_init_x86(c, avctx, high_bit_depth);
    if (ARCH_MIPS)
        ff_xvid_idct_init_mips(c, avctx, high_bit_depth);

    ff_init_scantable_permutation(c->idct_permutation, c->perm_type);
}


Generated by: GCOVR (Version 4.2)

Line	Branch	Exec	Source
1			/*
2			* Xvid MPEG-4 IDCT
3			*
4			* Copyright (C) 2006-2011 Xvid Solutions GmbH
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			* Walken IDCT
26			* Alternative IDCT implementation for decoding compatibility.
27			*
28			* @author Skal
29			* @note This C version is not the original IDCT, but a modified one that
30			* yields the same error profile as the MMX/MMXEXT/SSE2 versions.
31			*/
32
33			#include "config.h"
34			#include "libavutil/attributes.h"
35			#include "avcodec.h"
36			#include "idctdsp.h"
37			#include "xvididct.h"
38
39			#define ROW_SHIFT 11
40			#define COL_SHIFT 6
41
42			// #define FIX(x) (int)((x) * (1 << ROW_SHIFT))
43			#define RND0 65536 // 1 << (COL_SHIFT + ROW_SHIFT - 1);
44			#define RND1 3597 // FIX (1.75683487303);
45			#define RND2 2260 // FIX (1.10355339059);
46			#define RND3 1203 // FIX (0.587788325588);
47			#define RND4 0
48			#define RND5 120 // FIX (0.058658283817);
49			#define RND6 512 // FIX (0.25);
50			#define RND7 512 // FIX (0.25);
51
52			static const int TAB04[] = { 22725, 21407, 19266, 16384, 12873, 8867, 4520 };
53			static const int TAB17[] = { 31521, 29692, 26722, 22725, 17855, 12299, 6270 };
54			static const int TAB26[] = { 29692, 27969, 25172, 21407, 16819, 11585, 5906 };
55			static const int TAB35[] = { 26722, 25172, 22654, 19266, 15137, 10426, 5315 };
56
57		480000	static int idct_row(short in, const int const tab, int rnd)
58			{
59		480000	const int c1 = tab[0];
60		480000	const int c2 = tab[1];
61		480000	const int c3 = tab[2];
62		480000	const int c4 = tab[3];
63		480000	const int c5 = tab[4];
64		480000	const int c6 = tab[5];
65		480000	const int c7 = tab[6];
66
67		480000	const int right = in[5] \| in[6] \| in[7];
68		480000	const int left = in[1] \| in[2] \| in[3];
69	✓✓	480000	if (!(right \| in[4])) {
70		264353	const int k = c4 * in[0] + rnd;
71	✓✓	264353	if (left) {
72		24730	const int a0 = k + c2 * in[2];
73		24730	const int a1 = k + c6 * in[2];
74		24730	const int a2 = k - c6 * in[2];
75		24730	const int a3 = k - c2 * in[2];
76
77		24730	const int b0 = c1 * in[1] + c3 * in[3];
78		24730	const int b1 = c3 * in[1] - c7 * in[3];
79		24730	const int b2 = c5 * in[1] - c1 * in[3];
80		24730	const int b3 = c7 * in[1] - c5 * in[3];
81
82		24730	in[0] = (a0 + b0) >> ROW_SHIFT;
83		24730	in[1] = (a1 + b1) >> ROW_SHIFT;
84		24730	in[2] = (a2 + b2) >> ROW_SHIFT;
85		24730	in[3] = (a3 + b3) >> ROW_SHIFT;
86		24730	in[4] = (a3 - b3) >> ROW_SHIFT;
87		24730	in[5] = (a2 - b2) >> ROW_SHIFT;
88		24730	in[6] = (a1 - b1) >> ROW_SHIFT;
89		24730	in[7] = (a0 - b0) >> ROW_SHIFT;
90			} else {
91		239623	const int a0 = k >> ROW_SHIFT;
92	✓✓	239623	if (a0) {
93		97939	in[0] =
94		97939	in[1] =
95		97939	in[2] =
96		97939	in[3] =
97		97939	in[4] =
98		97939	in[5] =
99		97939	in[6] =
100		97939	in[7] = a0;
101			} else
102		141684	return 0;
103			}
104	✓✓	215647	} else if (!(left \| right)) {
105		7429	const int a0 = (rnd + c4 * (in[0] + in[4])) >> ROW_SHIFT;
106		7429	const int a1 = (rnd + c4 * (in[0] - in[4])) >> ROW_SHIFT;
107
108		7429	in[0] = a0;
109		7429	in[3] = a0;
110		7429	in[4] = a0;
111		7429	in[7] = a0;
112		7429	in[1] = a1;
113		7429	in[2] = a1;
114		7429	in[5] = a1;
115		7429	in[6] = a1;
116			} else {
117		208218	const int k = c4 * in[0] + rnd;
118		208218	const unsigned int a0 = k + c2 * in[2] + c4 * in[4] + c6 * in[6];
119		208218	const unsigned int a1 = k + c6 * in[2] - c4 * in[4] - c2 * in[6];
120		208218	const unsigned int a2 = k - c6 * in[2] - c4 * in[4] + c2 * in[6];
121		208218	const unsigned int a3 = k - c2 * in[2] + c4 * in[4] - c6 * in[6];
122
123		208218	const unsigned int b0 = c1 * in[1] + c3 * in[3] + c5 * in[5] + c7 * in[7];
124		208218	const unsigned int b1 = c3 * in[1] - c7 * in[3] - c1 * in[5] - c5 * in[7];
125		208218	const unsigned int b2 = c5 * in[1] - c1 * in[3] + c7 * in[5] + c3 * in[7];
126		208218	const unsigned int b3 = c7 * in[1] - c5 * in[3] + c3 * in[5] - c1 * in[7];
127
128		208218	in[0] = (int)(a0 + b0) >> ROW_SHIFT;
129		208218	in[1] = (int)(a1 + b1) >> ROW_SHIFT;
130		208218	in[2] = (int)(a2 + b2) >> ROW_SHIFT;
131		208218	in[3] = (int)(a3 + b3) >> ROW_SHIFT;
132		208218	in[4] = (int)(a3 - b3) >> ROW_SHIFT;
133		208218	in[5] = (int)(a2 - b2) >> ROW_SHIFT;
134		208218	in[6] = (int)(a1 - b1) >> ROW_SHIFT;
135		208218	in[7] = (int)(a0 - b0) >> ROW_SHIFT;
136			}
137		338316	return 1;
138			}
139
140			#define TAN1 0x32EC
141			#define TAN2 0x6A0A
142			#define TAN3 0xAB0E
143			#define SQRT2 0x5A82
144
145			#define MULT(c, x, n) ((unsigned)((int)((c) * (unsigned)(x)) >> (n)))
146			// 12b version => #define MULT(c,x, n) ((((c) >> 3) * (x)) >> ((n) - 3))
147			// 12b zero-testing version:
148
149			#define BUTTERFLY(a, b, tmp) \
150			(tmp) = (a) + (b); \
151			(b) = (a) - (b); \
152			(a) = (tmp)
153
154			#define LOAD_BUTTERFLY(m1, m2, a, b, tmp, s) \
155			(m1) = (s)[(a)] + (s)[(b)]; \
156			(m2) = (s)[(a)] - (s)[(b)]
157
158		384504	static void idct_col_8(short *const in)
159			{
160			int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, spill;
161
162			// odd
163
164		384504	mm4 = (int) in[7 * 8];
165		384504	mm5 = (int) in[5 * 8];
166		384504	mm6 = (int) in[3 * 8];
167		384504	mm7 = (int) in[1 * 8];
168
169		384504	mm0 = MULT(TAN1, mm4, 16) + mm7;
170		384504	mm1 = MULT(TAN1, mm7, 16) - mm4;
171		384504	mm2 = MULT(TAN3, mm5, 16) + mm6;
172		384504	mm3 = MULT(TAN3, mm6, 16) - mm5;
173
174		384504	mm7 = mm0 + mm2;
175		384504	mm4 = mm1 - mm3;
176		384504	mm0 = mm0 - mm2;
177		384504	mm1 = mm1 + mm3;
178		384504	mm6 = mm0 + mm1;
179		384504	mm5 = mm0 - mm1;
180		384504	mm5 = 2 * MULT(SQRT2, mm5, 16); // 2*sqrt2
181		384504	mm6 = 2 * MULT(SQRT2, mm6, 16); // Watch out: precision loss but done to match
182			// the pmulhw used in MMX/MMXEXT/SSE2 versions
183
184			// even
185
186		384504	mm1 = (int) in[2 * 8];
187		384504	mm2 = (int) in[6 * 8];
188		384504	mm3 = MULT(TAN2, mm2, 16) + mm1;
189		384504	mm2 = MULT(TAN2, mm1, 16) - mm2;
190
191		384504	LOAD_BUTTERFLY(mm0, mm1, 0 * 8, 4 * 8, spill, in);
192
193		384504	BUTTERFLY(mm0, mm3, spill);
194		384504	BUTTERFLY(mm0, mm7, spill);
195		384504	in[8 * 0] = (int16_t) (mm0 >> COL_SHIFT);
196		384504	in[8 * 7] = (int16_t) (mm7 >> COL_SHIFT);
197		384504	BUTTERFLY(mm3, mm4, mm0);
198		384504	in[8 * 3] = (int16_t) (mm3 >> COL_SHIFT);
199		384504	in[8 * 4] = (int16_t) (mm4 >> COL_SHIFT);
200
201		384504	BUTTERFLY(mm1, mm2, mm0);
202		384504	BUTTERFLY(mm1, mm6, mm0);
203		384504	in[8 * 1] = (int16_t) (mm1 >> COL_SHIFT);
204		384504	in[8 * 6] = (int16_t) (mm6 >> COL_SHIFT);
205		384504	BUTTERFLY(mm2, mm5, mm0);
206		384504	in[8 * 2] = (int16_t) (mm2 >> COL_SHIFT);
207		384504	in[8 * 5] = (int16_t) (mm5 >> COL_SHIFT);
208		384504	}
209
210		5896	static void idct_col_4(short *const in)
211			{
212			int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, spill;
213
214			// odd
215
216		5896	mm0 = (int) in[1 * 8];
217		5896	mm2 = (int) in[3 * 8];
218
219		5896	mm1 = MULT(TAN1, mm0, 16);
220		5896	mm3 = MULT(TAN3, mm2, 16);
221
222		5896	mm7 = mm0 + mm2;
223		5896	mm4 = mm1 - mm3;
224		5896	mm0 = mm0 - mm2;
225		5896	mm1 = mm1 + mm3;
226		5896	mm6 = mm0 + mm1;
227		5896	mm5 = mm0 - mm1;
228		5896	mm6 = 2 * MULT(SQRT2, mm6, 16); // 2*sqrt2
229		5896	mm5 = 2 * MULT(SQRT2, mm5, 16);
230
231			// even
232
233		5896	mm0 = mm1 = (int) in[0 * 8];
234		5896	mm3 = (int) in[2 * 8];
235		5896	mm2 = MULT(TAN2, mm3, 16);
236
237		5896	BUTTERFLY(mm0, mm3, spill);
238		5896	BUTTERFLY(mm0, mm7, spill);
239		5896	in[8 * 0] = (int16_t) (mm0 >> COL_SHIFT);
240		5896	in[8 * 7] = (int16_t) (mm7 >> COL_SHIFT);
241		5896	BUTTERFLY(mm3, mm4, mm0);
242		5896	in[8 * 3] = (int16_t) (mm3 >> COL_SHIFT);
243		5896	in[8 * 4] = (int16_t) (mm4 >> COL_SHIFT);
244
245		5896	BUTTERFLY(mm1, mm2, mm0);
246		5896	BUTTERFLY(mm1, mm6, mm0);
247		5896	in[8 * 1] = (int16_t) (mm1 >> COL_SHIFT);
248		5896	in[8 * 6] = (int16_t) (mm6 >> COL_SHIFT);
249		5896	BUTTERFLY(mm2, mm5, mm0);
250		5896	in[8 * 2] = (int16_t) (mm2 >> COL_SHIFT);
251		5896	in[8 * 5] = (int16_t) (mm5 >> COL_SHIFT);
252		5896	}
253
254		89600	static void idct_col_3(short *const in)
255			{
256			int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, spill;
257
258			// odd
259
260		89600	mm7 = (int) in[1 * 8];
261		89600	mm4 = MULT(TAN1, mm7, 16);
262
263		89600	mm6 = mm7 + mm4;
264		89600	mm5 = mm7 - mm4;
265		89600	mm6 = 2 * MULT(SQRT2, mm6, 16); // 2*sqrt2
266		89600	mm5 = 2 * MULT(SQRT2, mm5, 16);
267
268			// even
269
270		89600	mm0 = mm1 = (int) in[0 * 8];
271		89600	mm3 = (int) in[2 * 8];
272		89600	mm2 = MULT(TAN2, mm3, 16);
273
274		89600	BUTTERFLY(mm0, mm3, spill);
275		89600	BUTTERFLY(mm0, mm7, spill);
276		89600	in[8 * 0] = (int16_t) (mm0 >> COL_SHIFT);
277		89600	in[8 * 7] = (int16_t) (mm7 >> COL_SHIFT);
278		89600	BUTTERFLY(mm3, mm4, mm0);
279		89600	in[8 * 3] = (int16_t) (mm3 >> COL_SHIFT);
280		89600	in[8 * 4] = (int16_t) (mm4 >> COL_SHIFT);
281
282		89600	BUTTERFLY(mm1, mm2, mm0);
283		89600	BUTTERFLY(mm1, mm6, mm0);
284		89600	in[8 * 1] = (int16_t) (mm1 >> COL_SHIFT);
285		89600	in[8 * 6] = (int16_t) (mm6 >> COL_SHIFT);
286		89600	BUTTERFLY(mm2, mm5, mm0);
287		89600	in[8 * 2] = (int16_t) (mm2 >> COL_SHIFT);
288		89600	in[8 * 5] = (int16_t) (mm5 >> COL_SHIFT);
289		89600	}
290
291		60000	void ff_xvid_idct(int16_t *const in)
292			{
293		60000	int i, rows = 0x07;
294
295		60000	idct_row(in + 0 * 8, TAB04, RND0);
296		60000	idct_row(in + 1 * 8, TAB17, RND1);
297		60000	idct_row(in + 2 * 8, TAB26, RND2);
298	✓✓	60000	if (idct_row(in + 3 * 8, TAB35, RND3))
299		29558	rows \|= 0x08;
300	✓✓	60000	if (idct_row(in + 4 * 8, TAB04, RND4))
301		29661	rows \|= 0x10;
302	✓✓	60000	if (idct_row(in + 5 * 8, TAB35, RND5))
303		29720	rows \|= 0x20;
304	✓✓	60000	if (idct_row(in + 6 * 8, TAB26, RND6))
305		29731	rows \|= 0x40;
306	✓✓	60000	if (idct_row(in + 7 * 8, TAB17, RND7))
307		39653	rows \|= 0x80;
308
309	✓✓	60000	if (rows & 0xF0) {
310	✓✓	432567	for (i = 0; i < 8; i++)
311		384504	idct_col_8(in + i);
312	✓✓	11937	} else if (rows & 0x08) {
313	✓✓	6633	for (i = 0; i < 8; i++)
314		5896	idct_col_4(in + i);
315			} else {
316	✓✓	100800	for (i = 0; i < 8; i++)
317		89600	idct_col_3(in + i);
318			}
319		60000	}
320
321			static void xvid_idct_put(uint8_t dest, ptrdiff_t line_size, int16_t block)
322			{
323			ff_xvid_idct(block);
324			ff_put_pixels_clamped_c(block, dest, line_size);
325			}
326
327			static void xvid_idct_add(uint8_t dest, ptrdiff_t line_size, int16_t block)
328			{
329			ff_xvid_idct(block);
330			ff_add_pixels_clamped_c(block, dest, line_size);
331			}
332
333		4	av_cold void ff_xvid_idct_init(IDCTDSPContext c, AVCodecContext avctx)
334			{
335		4	const unsigned high_bit_depth = avctx->bits_per_raw_sample > 8;
336
337	✓✗✓✗	4	if (high_bit_depth \|\| avctx->lowres \|\|
338	✓✗	4	!(avctx->idct_algo == FF_IDCT_AUTO \|\|
339	✗✓	4	avctx->idct_algo == FF_IDCT_XVID))
340			return;
341
342	✓✗	4	if (avctx->idct_algo == FF_IDCT_XVID) {
343		4	c->idct_put = xvid_idct_put;
344		4	c->idct_add = xvid_idct_add;
345		4	c->idct = ff_xvid_idct;
346		4	c->perm_type = FF_IDCT_PERM_NONE;
347			}
348
349			if (ARCH_X86)
350		4	ff_xvid_idct_init_x86(c, avctx, high_bit_depth);
351			if (ARCH_MIPS)
352			ff_xvid_idct_init_mips(c, avctx, high_bit_depth);
353
354		4	ff_init_scantable_permutation(c->idct_permutation, c->perm_type);
355			}