1/*
2 * Copyright (c) 2010 Mans Rullgard <mans@mansr.com>
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 <stdint.h>
22
23#include "libavutil/attributes.h"
24#include "libavutil/arm/cpu.h"
25#include "libavcodec/h264dsp.h"
26
27int ff_startcode_find_candidate_armv6(const uint8_t *buf, int size);
28
29void ff_h264_v_loop_filter_luma_neon(uint8_t *pix, int stride, int alpha,
30                                     int beta, int8_t *tc0);
31void ff_h264_h_loop_filter_luma_neon(uint8_t *pix, int stride, int alpha,
32                                     int beta, int8_t *tc0);
33void ff_h264_v_loop_filter_chroma_neon(uint8_t *pix, int stride, int alpha,
34                                       int beta, int8_t *tc0);
35void ff_h264_h_loop_filter_chroma_neon(uint8_t *pix, int stride, int alpha,
36                                       int beta, int8_t *tc0);
37
38void ff_weight_h264_pixels_16_neon(uint8_t *dst, int stride, int height,
39                                   int log2_den, int weight, int offset);
40void ff_weight_h264_pixels_8_neon(uint8_t *dst, int stride, int height,
41                                  int log2_den, int weight, int offset);
42void ff_weight_h264_pixels_4_neon(uint8_t *dst, int stride, int height,
43                                  int log2_den, int weight, int offset);
44
45void ff_biweight_h264_pixels_16_neon(uint8_t *dst, uint8_t *src, int stride,
46                                     int height, int log2_den, int weightd,
47                                     int weights, int offset);
48void ff_biweight_h264_pixels_8_neon(uint8_t *dst, uint8_t *src, int stride,
49                                    int height, int log2_den, int weightd,
50                                    int weights, int offset);
51void ff_biweight_h264_pixels_4_neon(uint8_t *dst, uint8_t *src, int stride,
52                                    int height, int log2_den, int weightd,
53                                    int weights, int offset);
54
55void ff_h264_idct_add_neon(uint8_t *dst, int16_t *block, int stride);
56void ff_h264_idct_dc_add_neon(uint8_t *dst, int16_t *block, int stride);
57void ff_h264_idct_add16_neon(uint8_t *dst, const int *block_offset,
58                             int16_t *block, int stride,
59                             const uint8_t nnzc[6*8]);
60void ff_h264_idct_add16intra_neon(uint8_t *dst, const int *block_offset,
61                                  int16_t *block, int stride,
62                                  const uint8_t nnzc[6*8]);
63void ff_h264_idct_add8_neon(uint8_t **dest, const int *block_offset,
64                            int16_t *block, int stride,
65                            const uint8_t nnzc[6*8]);
66
67void ff_h264_idct8_add_neon(uint8_t *dst, int16_t *block, int stride);
68void ff_h264_idct8_dc_add_neon(uint8_t *dst, int16_t *block, int stride);
69void ff_h264_idct8_add4_neon(uint8_t *dst, const int *block_offset,
70                             int16_t *block, int stride,
71                             const uint8_t nnzc[6*8]);
72
73static av_cold void h264dsp_init_neon(H264DSPContext *c, const int bit_depth,
74                                      const int chroma_format_idc)
75{
76#if HAVE_NEON
77    if (bit_depth == 8) {
78        c->h264_v_loop_filter_luma   = ff_h264_v_loop_filter_luma_neon;
79        c->h264_h_loop_filter_luma   = ff_h264_h_loop_filter_luma_neon;
80        if(chroma_format_idc == 1){
81        c->h264_v_loop_filter_chroma = ff_h264_v_loop_filter_chroma_neon;
82        c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma_neon;
83        }
84
85        c->weight_h264_pixels_tab[0] = ff_weight_h264_pixels_16_neon;
86        c->weight_h264_pixels_tab[1] = ff_weight_h264_pixels_8_neon;
87        c->weight_h264_pixels_tab[2] = ff_weight_h264_pixels_4_neon;
88
89        c->biweight_h264_pixels_tab[0] = ff_biweight_h264_pixels_16_neon;
90        c->biweight_h264_pixels_tab[1] = ff_biweight_h264_pixels_8_neon;
91        c->biweight_h264_pixels_tab[2] = ff_biweight_h264_pixels_4_neon;
92
93        c->h264_idct_add        = ff_h264_idct_add_neon;
94        c->h264_idct_dc_add     = ff_h264_idct_dc_add_neon;
95        c->h264_idct_add16      = ff_h264_idct_add16_neon;
96        c->h264_idct_add16intra = ff_h264_idct_add16intra_neon;
97        if (chroma_format_idc <= 1)
98            c->h264_idct_add8   = ff_h264_idct_add8_neon;
99        c->h264_idct8_add       = ff_h264_idct8_add_neon;
100        c->h264_idct8_dc_add    = ff_h264_idct8_dc_add_neon;
101        c->h264_idct8_add4      = ff_h264_idct8_add4_neon;
102    }
103#endif // HAVE_NEON
104}
105
106av_cold void ff_h264dsp_init_arm(H264DSPContext *c, const int bit_depth,
107                                 const int chroma_format_idc)
108{
109    int cpu_flags = av_get_cpu_flags();
110
111    if (have_armv6(cpu_flags) && !(have_vfpv3(cpu_flags) || have_neon(cpu_flags))) {
112        // This function uses the 'setend' instruction which is deprecated
113        // on ARMv8. This instruction is serializing on some ARMv7 cores as
114        // well. Therefore, only use the function on ARMv6.
115        c->h264_find_start_code_candidate = ff_startcode_find_candidate_armv6;
116    }
117    if (have_neon(cpu_flags))
118        h264dsp_init_neon(c, bit_depth, chroma_format_idc);
119}
120