1 /*
2 * This file is part of FFmpeg.
3 *
4 * FFmpeg is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
8 *
9 * FFmpeg is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
13 *
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with FFmpeg; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17 */
18
19 #include <stdint.h>
20
21 #include "libavutil/attributes.h"
22 #include "libavutil/cpu.h"
23 #include "libavutil/aarch64/cpu.h"
24 #include "libavutil/intreadwrite.h"
25 #include "libavcodec/vc1dsp.h"
26
27 #include "config.h"
28
29 void ff_vc1_inv_trans_8x8_neon(int16_t *block);
30 void ff_vc1_inv_trans_8x4_neon(uint8_t *dest, ptrdiff_t stride, int16_t *block);
31 void ff_vc1_inv_trans_4x8_neon(uint8_t *dest, ptrdiff_t stride, int16_t *block);
32 void ff_vc1_inv_trans_4x4_neon(uint8_t *dest, ptrdiff_t stride, int16_t *block);
33
34 void ff_vc1_inv_trans_8x8_dc_neon(uint8_t *dest, ptrdiff_t stride, int16_t *block);
35 void ff_vc1_inv_trans_8x4_dc_neon(uint8_t *dest, ptrdiff_t stride, int16_t *block);
36 void ff_vc1_inv_trans_4x8_dc_neon(uint8_t *dest, ptrdiff_t stride, int16_t *block);
37 void ff_vc1_inv_trans_4x4_dc_neon(uint8_t *dest, ptrdiff_t stride, int16_t *block);
38
39 void ff_vc1_v_loop_filter4_neon(uint8_t *src, ptrdiff_t stride, int pq);
40 void ff_vc1_h_loop_filter4_neon(uint8_t *src, ptrdiff_t stride, int pq);
41 void ff_vc1_v_loop_filter8_neon(uint8_t *src, ptrdiff_t stride, int pq);
42 void ff_vc1_h_loop_filter8_neon(uint8_t *src, ptrdiff_t stride, int pq);
43 void ff_vc1_v_loop_filter16_neon(uint8_t *src, ptrdiff_t stride, int pq);
44 void ff_vc1_h_loop_filter16_neon(uint8_t *src, ptrdiff_t stride, int pq);
45
46 void ff_put_vc1_chroma_mc8_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
47 int h, int x, int y);
48 void ff_avg_vc1_chroma_mc8_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
49 int h, int x, int y);
50 void ff_put_vc1_chroma_mc4_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
51 int h, int x, int y);
52 void ff_avg_vc1_chroma_mc4_neon(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
53 int h, int x, int y);
54
55 int ff_vc1_unescape_buffer_helper_neon(const uint8_t *src, int size, uint8_t *dst);
56
vc1_unescape_buffer_neon(const uint8_t * src,int size,uint8_t * dst)57 static int vc1_unescape_buffer_neon(const uint8_t *src, int size, uint8_t *dst)
58 {
59 /* Dealing with starting and stopping, and removing escape bytes, are
60 * comparatively less time-sensitive, so are more clearly expressed using
61 * a C wrapper around the assembly inner loop. Note that we assume a
62 * little-endian machine that supports unaligned loads. */
63 int dsize = 0;
64 while (size >= 4)
65 {
66 int found = 0;
67 while (!found && (((uintptr_t) dst) & 7) && size >= 4)
68 {
69 found = (AV_RL32(src) &~ 0x03000000) == 0x00030000;
70 if (!found)
71 {
72 *dst++ = *src++;
73 --size;
74 ++dsize;
75 }
76 }
77 if (!found)
78 {
79 int skip = size - ff_vc1_unescape_buffer_helper_neon(src, size, dst);
80 dst += skip;
81 src += skip;
82 size -= skip;
83 dsize += skip;
84 while (!found && size >= 4)
85 {
86 found = (AV_RL32(src) &~ 0x03000000) == 0x00030000;
87 if (!found)
88 {
89 *dst++ = *src++;
90 --size;
91 ++dsize;
92 }
93 }
94 }
95 if (found)
96 {
97 *dst++ = *src++;
98 *dst++ = *src++;
99 ++src;
100 size -= 3;
101 dsize += 2;
102 }
103 }
104 while (size > 0)
105 {
106 *dst++ = *src++;
107 --size;
108 ++dsize;
109 }
110 return dsize;
111 }
112
ff_vc1dsp_init_aarch64(VC1DSPContext * dsp)113 av_cold void ff_vc1dsp_init_aarch64(VC1DSPContext *dsp)
114 {
115 int cpu_flags = av_get_cpu_flags();
116
117 if (have_neon(cpu_flags)) {
118 dsp->vc1_inv_trans_8x8 = ff_vc1_inv_trans_8x8_neon;
119 dsp->vc1_inv_trans_8x4 = ff_vc1_inv_trans_8x4_neon;
120 dsp->vc1_inv_trans_4x8 = ff_vc1_inv_trans_4x8_neon;
121 dsp->vc1_inv_trans_4x4 = ff_vc1_inv_trans_4x4_neon;
122 dsp->vc1_inv_trans_8x8_dc = ff_vc1_inv_trans_8x8_dc_neon;
123 dsp->vc1_inv_trans_8x4_dc = ff_vc1_inv_trans_8x4_dc_neon;
124 dsp->vc1_inv_trans_4x8_dc = ff_vc1_inv_trans_4x8_dc_neon;
125 dsp->vc1_inv_trans_4x4_dc = ff_vc1_inv_trans_4x4_dc_neon;
126
127 dsp->vc1_v_loop_filter4 = ff_vc1_v_loop_filter4_neon;
128 dsp->vc1_h_loop_filter4 = ff_vc1_h_loop_filter4_neon;
129 dsp->vc1_v_loop_filter8 = ff_vc1_v_loop_filter8_neon;
130 dsp->vc1_h_loop_filter8 = ff_vc1_h_loop_filter8_neon;
131 dsp->vc1_v_loop_filter16 = ff_vc1_v_loop_filter16_neon;
132 dsp->vc1_h_loop_filter16 = ff_vc1_h_loop_filter16_neon;
133
134 dsp->put_no_rnd_vc1_chroma_pixels_tab[0] = ff_put_vc1_chroma_mc8_neon;
135 dsp->avg_no_rnd_vc1_chroma_pixels_tab[0] = ff_avg_vc1_chroma_mc8_neon;
136 dsp->put_no_rnd_vc1_chroma_pixels_tab[1] = ff_put_vc1_chroma_mc4_neon;
137 dsp->avg_no_rnd_vc1_chroma_pixels_tab[1] = ff_avg_vc1_chroma_mc4_neon;
138
139 dsp->vc1_unescape_buffer = vc1_unescape_buffer_neon;
140 }
141 }
142