GCC Code Coverage Report
Directory: ../../../ffmpeg/ Exec Total Coverage
File: src/libavcodec/x86/lossless_videodsp_init.c Lines: 17 17 100.0 %
Date: 2020-08-14 10:39:37 Branches: 10 12 83.3 %

Line Branch Exec Source
1
/*
2
 * Lossless video DSP utils
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 "config.h"
22
#include "libavutil/x86/asm.h"
23
#include "../lossless_videodsp.h"
24
#include "libavutil/x86/cpu.h"
25
26
void ff_add_bytes_mmx(uint8_t *dst, uint8_t *src, ptrdiff_t w);
27
void ff_add_bytes_sse2(uint8_t *dst, uint8_t *src, ptrdiff_t w);
28
void ff_add_bytes_avx2(uint8_t *dst, uint8_t *src, ptrdiff_t w);
29
30
void ff_add_median_pred_mmxext(uint8_t *dst, const uint8_t *top,
31
                               const uint8_t *diff, ptrdiff_t w,
32
                               int *left, int *left_top);
33
void ff_add_median_pred_sse2(uint8_t *dst, const uint8_t *top,
34
                             const uint8_t *diff, ptrdiff_t w,
35
                             int *left, int *left_top);
36
37
int  ff_add_left_pred_ssse3(uint8_t *dst, const uint8_t *src,
38
                            ptrdiff_t w, int left);
39
int  ff_add_left_pred_unaligned_ssse3(uint8_t *dst, const uint8_t *src,
40
                                      ptrdiff_t w, int left);
41
int  ff_add_left_pred_unaligned_avx2(uint8_t *dst, const uint8_t *src,
42
                                     ptrdiff_t w, int left);
43
44
int ff_add_left_pred_int16_ssse3(uint16_t *dst, const uint16_t *src, unsigned mask, ptrdiff_t w, unsigned acc);
45
int ff_add_left_pred_int16_unaligned_ssse3(uint16_t *dst, const uint16_t *src, unsigned mask, ptrdiff_t w, unsigned acc);
46
47
void ff_add_gradient_pred_ssse3(uint8_t *src, const ptrdiff_t stride, const ptrdiff_t width);
48
void ff_add_gradient_pred_avx2(uint8_t *src, const ptrdiff_t stride, const ptrdiff_t width);
49
50
#if HAVE_INLINE_ASM && HAVE_7REGS && ARCH_X86_32
51
static void add_median_pred_cmov(uint8_t *dst, const uint8_t *top,
52
                                 const uint8_t *diff, ptrdiff_t w,
53
                                 int *left, int *left_top)
54
{
55
    x86_reg w2 = -w;
56
    x86_reg x;
57
    int l  = *left     & 0xff;
58
    int tl = *left_top & 0xff;
59
    int t;
60
    __asm__ volatile (
61
        "mov          %7, %3            \n"
62
        "1:                             \n"
63
        "movzbl (%3, %4), %2            \n"
64
        "mov          %2, %k3           \n"
65
        "sub         %b1, %b3           \n"
66
        "add         %b0, %b3           \n"
67
        "mov          %2, %1            \n"
68
        "cmp          %0, %2            \n"
69
        "cmovg        %0, %2            \n"
70
        "cmovg        %1, %0            \n"
71
        "cmp         %k3, %0            \n"
72
        "cmovg       %k3, %0            \n"
73
        "mov          %7, %3            \n"
74
        "cmp          %2, %0            \n"
75
        "cmovl        %2, %0            \n"
76
        "add    (%6, %4), %b0           \n"
77
        "mov         %b0, (%5, %4)      \n"
78
        "inc          %4                \n"
79
        "jl           1b                \n"
80
        : "+&q"(l), "+&q"(tl), "=&r"(t), "=&q"(x), "+&r"(w2)
81
        : "r"(dst + w), "r"(diff + w), "rm"(top + w)
82
    );
83
    *left     = l;
84
    *left_top = tl;
85
}
86
#endif
87
88
180
void ff_llviddsp_init_x86(LLVidDSPContext *c)
89
{
90
180
    int cpu_flags = av_get_cpu_flags();
91
92
#if HAVE_INLINE_ASM && HAVE_7REGS && ARCH_X86_32
93
    if (cpu_flags & AV_CPU_FLAG_CMOV)
94
        c->add_median_pred = add_median_pred_cmov;
95
#endif
96
97
    if (ARCH_X86_32 && EXTERNAL_MMX(cpu_flags)) {
98
        c->add_bytes = ff_add_bytes_mmx;
99
    }
100
101
    if (ARCH_X86_32 && EXTERNAL_MMXEXT(cpu_flags)) {
102
        /* slower than cmov version on AMD */
103
        if (!(cpu_flags & AV_CPU_FLAG_3DNOW))
104
            c->add_median_pred = ff_add_median_pred_mmxext;
105
    }
106
107
180
    if (EXTERNAL_SSE2(cpu_flags)) {
108
30
        c->add_bytes       = ff_add_bytes_sse2;
109
30
        c->add_median_pred = ff_add_median_pred_sse2;
110
    }
111
112
180
    if (EXTERNAL_SSSE3(cpu_flags)) {
113
28
        c->add_left_pred = ff_add_left_pred_ssse3;
114
28
        c->add_left_pred_int16 = ff_add_left_pred_int16_ssse3;
115
28
        c->add_gradient_pred   = ff_add_gradient_pred_ssse3;
116
    }
117
118

180
    if (EXTERNAL_SSSE3_FAST(cpu_flags)) {
119
28
        c->add_left_pred = ff_add_left_pred_unaligned_ssse3;
120
28
        c->add_left_pred_int16 = ff_add_left_pred_int16_unaligned_ssse3;
121
    }
122
123

180
    if (EXTERNAL_AVX2_FAST(cpu_flags)) {
124
22
        c->add_bytes       = ff_add_bytes_avx2;
125
22
        c->add_left_pred   = ff_add_left_pred_unaligned_avx2;
126
22
        c->add_gradient_pred = ff_add_gradient_pred_avx2;
127
    }
128
180
}