]>
Commit | Line | Data |
---|---|---|
b91fb912 | 1 | === libs/libmythtv/dvbrecorder.cpp |
2 | Index: libs/libmythtv/dvbrecorder.h | |
3 | =================================================================== | |
4 | --- libs/libmythtv/dvbrecorder.h.orig 2006-04-10 21:26:56.000000000 -0400 | |
5 | +++ libs/libmythtv/dvbrecorder.h 2006-06-20 17:36:06.000000000 -0400 | |
6 | @@ -28,6 +28,15 @@ | |
7 | class ProgramMapTable; | |
8 | class TSPacket; | |
9 | ||
10 | +#define MAX_CSA_PIDS 8192 | |
11 | +struct decsaKey { | |
12 | + bool use_decsa; | |
13 | + unsigned char pidmap[MAX_CSA_PIDS]; //max # of pids | |
14 | + void *keys[16]; | |
15 | + unsigned char valid_keys; | |
16 | + pthread_mutex_t keylock; | |
17 | +}; | |
18 | + | |
19 | class PIDInfo | |
20 | { | |
21 | public: | |
22 | @@ -60,6 +69,10 @@ | |
23 | DVBRecorder(TVRec *rec, DVBChannel* dvbchannel); | |
24 | ~DVBRecorder(); | |
25 | ||
26 | + static struct decsaKey *DVBRecorder::UpdateDeCSAKeys( | |
27 | + int cardnum, unsigned char keytype, | |
28 | + int index, unsigned char *key, int pid); | |
29 | + | |
30 | void SetOption(const QString &name, int value); | |
31 | ||
32 | void SetOptionsFromProfile(RecordingProfile *profile, | |
33 | @@ -159,6 +172,18 @@ | |
34 | static const int TSPACKETS_BETWEEN_PSIP_SYNC; | |
35 | static const int POLL_INTERVAL; | |
36 | static const int POLL_WARNING_TIMEOUT; | |
37 | + | |
38 | + void initDeCSA(); | |
39 | + uint PostProcessDataTS(unsigned char *buffer, uint offset, uint len); | |
40 | + int PreProcessDataTS(unsigned char *buffer, uint len); | |
41 | + int _csa_pkt_buf_count; | |
42 | + unsigned char **_csa_cluster; | |
43 | + unsigned char **_csa_clusterptr; | |
44 | + int _csa_cluster_size; | |
45 | + | |
46 | + struct decsaKey *_csa_softKey; | |
47 | + unsigned char _csa_even_ck[8], _csa_odd_ck[8]; | |
48 | + | |
49 | }; | |
50 | ||
51 | inline void PIDInfo::Close(void) | |
52 | Index: libs/libmythtv/libmythtv.pro | |
53 | =================================================================== | |
54 | --- libs/libmythtv/libmythtv.pro.orig 2006-04-10 21:26:56.000000000 -0400 | |
55 | +++ libs/libmythtv/libmythtv.pro 2006-06-20 17:36:06.000000000 -0400 | |
56 | @@ -14,6 +14,10 @@ | |
57 | DEPENDPATH += ../libmythmpeg2 | |
58 | DEPENDPATH += ./dvbdev ./mpeg | |
59 | ||
60 | +# (begin)softcam-0.4 | |
61 | +LIBS += FFdecsa/FFdecsa.o | |
62 | +# (end)softcam-0.4 | |
63 | + | |
64 | LIBS += -L../libmyth -L../libavutil -L../libavcodec -L../libavformat -L../libmythmpeg2 | |
65 | LIBS += -lmyth-$${LIBVERSION} -lmythavutil-$${LIBVERSION} \ | |
66 | -lmythavcodec-$${LIBVERSION} \ | |
67 | Index: libs/libmythtv/FFdecsa/FFdecsa_test_testcases.h | |
68 | =================================================================== | |
69 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
70 | +++ libs/libmythtv/FFdecsa/FFdecsa_test_testcases.h 2006-06-20 17:36:06.000000000 -0400 | |
71 | @@ -0,0 +1,279 @@ | |
72 | +/* FFdecsa -- fast decsa algorithm | |
73 | + * | |
74 | + * Copyright (C) 2003-2004 fatih89r | |
75 | + * | |
76 | + * This program is free software; you can redistribute it and/or modify | |
77 | + * it under the terms of the GNU General Public License as published by | |
78 | + * the Free Software Foundation; either version 2 of the License, or | |
79 | + * (at your option) any later version. | |
80 | + * | |
81 | + * This program is distributed in the hope that it will be useful, | |
82 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
83 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
84 | + * GNU General Public License for more details. | |
85 | + * | |
86 | + * You should have received a copy of the GNU General Public License | |
87 | + * along with this program; if not, write to the Free Software | |
88 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
89 | + */ | |
90 | + | |
91 | + | |
92 | +// TEST DATA | |
93 | + | |
94 | +////////// used as a wrong key | |
95 | +unsigned char test_invalid_key[0x08] = { | |
96 | + 0x0f, 0x1e, 0x2d, 0x3c, 0x4b, 0x5a, 0x69, 0x78 | |
97 | +}; | |
98 | + | |
99 | + | |
100 | +////////// test 1: odd key | |
101 | +unsigned char test_1_key[0x8] = { | |
102 | + 0x07, 0xe0, 0x1b, 0x02, 0xc9, 0xe0, 0x45, 0xee | |
103 | +}; | |
104 | +unsigned char test_1_encrypted[0x100] = { | |
105 | + 0x47, 0x00, 0x00, 0xd0, | |
106 | + 0xde, 0xcf, 0x0a, 0x0d, 0xb2, 0xd7, 0xc4, 0x40, 0xde, 0x5d, 0x63, 0x18, 0x5a, 0x98, 0x17, 0xaa, | |
107 | + 0xc9, 0xbc, 0x27, 0xc6, 0xcb, 0x49, 0x40, 0x48, 0xfd, 0x20, 0xb7, 0x05, 0x5b, 0x27, 0xcb, 0xeb, | |
108 | + 0x9a, 0xf0, 0xac, 0x45, 0x6d, 0x56, 0xf4, 0x7b, 0x6f, 0xa0, 0x57, 0xf3, 0x9b, 0xf7, 0xa2, 0xc7, | |
109 | + 0xd4, 0x68, 0x24, 0x00, 0x2f, 0x28, 0x13, 0x96, 0x94, 0xa8, 0x7c, 0xf4, 0x6f, 0x07, 0x2a, 0x0e, | |
110 | + 0xe8, 0xa1, 0xeb, 0xc7, 0x80, 0xac, 0x1f, 0x79, 0xbf, 0x5d, 0xb6, 0x10, 0x7c, 0x2e, 0x52, 0xe9, | |
111 | + 0x34, 0x2c, 0xa8, 0x39, 0x01, 0x73, 0x04, 0x24, 0xa8, 0x1e, 0xdb, 0x5b, 0xcb, 0x24, 0xf6, 0x31, | |
112 | + 0xab, 0x02, 0x6b, 0xf9, 0xf6, 0xf7, 0xe9, 0x52, 0xad, 0xcf, 0x62, 0x0f, 0x42, 0xf6, 0x66, 0x5d, | |
113 | + 0xc0, 0x86, 0xf2, 0x7b, 0x40, 0x20, 0xa9, 0xbd, 0x1f, 0xfd, 0x16, 0xad, 0x2e, 0x75, 0xa6, 0xa0, | |
114 | + 0x85, 0xf3, 0x9c, 0x31, 0x20, 0x4e, 0xfb, 0x95, 0x61, 0x78, 0xce, 0x10, 0xc1, 0x48, 0x5f, 0xd3, | |
115 | + 0x61, 0x05, 0x12, 0xf4, 0xe2, 0x04, 0xae, 0xe0, 0x86, 0x01, 0x56, 0x55, 0xb1, 0x0f, 0xa6, 0x33, | |
116 | + 0x95, 0x20, 0x92, 0xf0, 0xbe, 0x39, 0x31, 0xe1, 0x2a, 0xf7, 0x93, 0xb4, 0xf7, 0xe4, 0xf1, 0x85, | |
117 | + 0xae, 0x50, 0xf1, 0x63, 0xd4, 0x5d, 0x9c, 0x6c | |
118 | +}; | |
119 | +unsigned char test_1_expected[0x100] = { | |
120 | + 0x47, 0x00, 0x00, 0xd0, | |
121 | + 0xaf, 0xbe, 0xfb, 0xef, 0xbe, 0xfb, 0xef, 0xbe, 0xfb, 0xef, 0xbe, 0xfb, 0xe6, 0xb5, 0xad, 0x7c, | |
122 | + 0xf9, 0xf3, 0xe5, 0xb1, 0x6c, 0x7c, 0xf9, 0xf3, 0xe6, 0xb5, 0xad, 0x6b, 0x5f, 0x3e, 0x7c, 0xf9, | |
123 | + 0x6c, 0x5b, 0x1f, 0x3e, 0x7c, 0xf9, 0xad, 0x6b, 0x5a, 0xd7, 0xcf, 0x9f, 0x3e, 0x5b, 0x16, 0xc7, | |
124 | + 0xcf, 0x9f, 0x3e, 0x6b, 0x5a, 0xd6, 0xb5, 0xf3, 0xe7, 0xcf, 0x96, 0xc5, 0xb1, 0xf3, 0xe7, 0xcf, | |
125 | + 0x9a, 0xd6, 0xb5, 0xad, 0x7c, 0xf9, 0xf3, 0xe5, 0xb1, 0x6c, 0x7c, 0xf9, 0xf3, 0xe6, 0xb5, 0xad, | |
126 | + 0x6b, 0x5f, 0x3e, 0x7c, 0xf9, 0x6c, 0x5b, 0x1f, 0x3e, 0x7c, 0xf9, 0xad, 0x6b, 0x5a, 0xd7, 0xcf, | |
127 | + 0x9f, 0x3e, 0x5b, 0x16, 0xc7, 0xcf, 0x9f, 0x3e, 0x6b, 0x5a, 0xd6, 0xb5, 0xf3, 0xe7, 0xcf, 0x96, | |
128 | + 0xc5, 0xb1, 0xf3, 0xe7, 0xcf, 0x9a, 0xd6, 0xb5, 0xad, 0x7c, 0xf9, 0xf3, 0xe5, 0xb1, 0x6c, 0x7c, | |
129 | + 0xf9, 0xf3, 0xe6, 0xb5, 0xad, 0x6b, 0x5f, 0x3e, 0x7c, 0xf9, 0x6c, 0x5b, 0x1f, 0x3e, 0x7c, 0xf9, | |
130 | + 0xad, 0x6b, 0x5a, 0xd7, 0xcf, 0x9f, 0x3e, 0x5b, 0x16, 0xc7, 0xcf, 0x9f, 0x3e, 0x6b, 0x5a, 0xd6, | |
131 | + 0xb5, 0xf3, 0xe7, 0xcf, 0x96, 0xc5, 0xb1, 0xf3, 0xe7, 0xcf, 0x9a, 0xd0, 0x00, 0x00, 0x00, 0x00, | |
132 | + 0xff, 0xfc, 0x44, 0x00, 0x66, 0xb1, 0x11, 0x11 | |
133 | +}; | |
134 | +unsigned char test_1_expected_stream[0x100] = { | |
135 | + 0xdc, 0x15, 0xde, 0xf1, 0x4a, 0xf1, 0xf8, 0x2c, | |
136 | + 0x75, 0xc8, 0x3a, 0x1f, 0xbf, 0x67, 0x19, 0xe1, | |
137 | + 0xf4, 0x6c, 0x78, 0x99, 0x48, 0xaf, 0xef, 0x94, | |
138 | + 0x71, 0x6b, 0x23, 0x9e, 0x29, 0x69, 0x2d, 0xa1, | |
139 | + 0x8a, 0xbb, 0xf4, 0x16, 0x68, 0xa5, 0x7f, 0x14, | |
140 | + 0xa9, 0x37, 0x24, 0x05, 0x5e, 0xdd, 0xec, 0x4b, | |
141 | + 0xb5, 0xcb, 0x7f, 0x1d, 0xa7, 0x09, 0x2a, 0xce, | |
142 | + 0xc4, 0x30, 0x83, 0xfd, 0xd9, 0x88, 0xa9, 0xf3, | |
143 | + 0x85, 0x9c, 0x38, 0x31, 0x88, 0xac, 0x74, 0x02, | |
144 | + 0x44, 0xdc, 0xb7, 0x81, 0x07, 0xc8, 0x1b, 0x03, | |
145 | + 0x9c, 0x76, 0xbe, 0xe9, 0x4d, 0x3e, 0x19, 0xad, | |
146 | + 0xe1, 0xf1, 0xa5, 0x13, 0xe8, 0xc0, 0x12, 0x57, | |
147 | + 0x68, 0xb1, 0x9c, 0x6c, 0x9f, 0x58, 0x78, 0xee, | |
148 | + 0x4f, 0x5b, 0x33, 0x1e, 0xc6, 0x29, 0xfc, 0x40, | |
149 | + 0x58, 0x22, 0xa2, 0xd8, 0x32, 0xdd, 0x29, 0x4f, | |
150 | + 0x2b, 0xe1, 0xef, 0xe4, 0xbb, 0xf2, 0x60, 0x94, | |
151 | + 0x6c, 0xc5, 0x51, 0xec, 0x35, 0x4c, 0x27, 0xc6, | |
152 | + 0x9d, 0x73, 0xe0, 0xf4, 0x2b, 0xfa, 0x62, 0x12, | |
153 | + 0xcd, 0x44, 0xbe, 0x57, 0xfe, 0x80, 0xe7, 0xa9, | |
154 | + 0x3c, 0x49, 0x42, 0xb6, 0xed, 0x05, 0x57, 0x00, | |
155 | + 0xd2, 0x25, 0x90, 0xb3, 0xe4, 0x65, 0x8f, 0xd6, | |
156 | + 0x4e, 0x0c, 0x73, 0x30, 0x3b, 0x68, 0x48, 0xdd, | |
157 | +// stream ^ sb | |
158 | +// 0x02, 0x48, 0xbd, 0xe9, 0x10, 0x69, 0xef, 0x86, | |
159 | +// 0xbc, 0x74, 0x1d, 0xd9, 0x74, 0x2e, 0x59, 0xa9, | |
160 | +// 0x09, 0x4c, 0xcf, 0x9c, 0x13, 0x88, 0x24, 0x7f, | |
161 | +// 0xeb, 0x9b, 0x8f, 0xdb, 0x44, 0x3f, 0xd9, 0xda, | |
162 | +}; | |
163 | +unsigned char test_1_expected_block[0x100] = { | |
164 | + 0xad, 0xf6, 0x46, 0x06, 0xae, 0x92, 0x00, 0x38, | |
165 | + 0x47, 0x9b, 0xa3, 0x22, 0x92, 0x9b, 0xf4, 0xd5, | |
166 | + 0xf0, 0xbf, 0x2a, 0x2d, 0x7f, 0xf4, 0xdd, 0x8c, | |
167 | + 0x0d, 0x2e, 0x22, 0xb0, 0x1b, 0x01, 0xa5, 0x23, | |
168 | + 0x89, 0x40, 0xbc, 0xdb, 0x8f, 0xab, 0x70, 0xb8, | |
169 | + 0x27, 0x88, 0xcf, 0x9a, 0x4f, 0xae, 0xe9, 0x1a, | |
170 | + 0xee, 0xfc, 0x3d, 0x82, 0x92, 0xd8, 0xb5, 0x33, | |
171 | + 0xcb, 0x5e, 0xfe, 0xff, 0xe8, 0xd7, 0x51, 0x45, | |
172 | + 0xa0, 0x17, 0x3b, 0x8c, 0x88, 0x7b, 0xd5, 0x0e, | |
173 | + 0xc1, 0x9c, 0x63, 0x41, 0xf5, 0x5d, 0xaa, 0x8a, | |
174 | + 0x5f, 0x37, 0x5b, 0xce, 0x7f, 0x76, 0xb4, 0x83, | |
175 | + 0x74, 0x8f, 0x37, 0x47, 0x75, 0x6d, 0x2c, 0xca, | |
176 | + 0x5a, 0x40, 0xa5, 0x75, 0x1a, 0x61, 0x81, 0x8d, | |
177 | + 0xe4, 0x87, 0x17, 0xd0, 0x75, 0xee, 0x9a, 0x6b, | |
178 | + 0x82, 0x6e, 0x47, 0x92, 0xd3, 0x32, 0x59, 0x5a, | |
179 | + 0x03, 0x6e, 0x8a, 0x26, 0x7e, 0x0d, 0xf7, 0x7d, | |
180 | + 0xf4, 0x4e, 0x79, 0x49, 0x59, 0x6f, 0x27, 0x2b, | |
181 | + 0x80, 0x8f, 0x9e, 0x5b, 0xd6, 0xc0, 0xb0, 0x0b, | |
182 | + 0xe6, 0x2e, 0xb2, 0xd5, 0x80, 0x10, 0x7f, 0xc1, | |
183 | + 0xbf, 0xae, 0x1f, 0xd9, 0x6d, 0x57, 0x3c, 0x37, | |
184 | + 0x4d, 0x21, 0xe4, 0xc8, 0x85, 0x44, 0xcf, 0xa0, | |
185 | + 0x07, 0x93, 0x18, 0x83, 0xef, 0x35, 0xd4, 0xb1, | |
186 | + 0xff, 0xfc, 0x44, 0x00, 0x66, 0xb1, 0x11, 0x11 | |
187 | +}; | |
188 | +unsigned char test_1_expected_kb[] = { | |
189 | + 0xEE, 0x45, 0xE0, 0xC9, 0x02, 0x1B, 0xE0, 0x07, | |
190 | + 0x46, 0xA4, 0x1C, 0x26, 0x7B, 0x0C, 0x01, 0xED, | |
191 | + 0x93, 0x99, 0xC3, 0x14, 0xC4, 0x4A, 0x8D, 0x54, | |
192 | + 0x19, 0x82, 0x39, 0xD1, 0x33, 0xB0, 0x33, 0x52, | |
193 | + 0x75, 0x62, 0x80, 0x3A, 0xC8, 0x83, 0x5E, 0x23, | |
194 | + 0xA2, 0x57, 0x0C, 0xC4, 0x2C, 0x2D, 0xD2, 0x98, | |
195 | + 0xA0, 0x6C, 0x77, 0x29, 0x11, 0x42, 0x49, 0xCE, | |
196 | +}; | |
197 | +unsigned char test_1_expected_kk[] = { | |
198 | + 0x5e, 0x9d, 0xff, 0x2e, 0xbb, 0xaa, 0xa8, 0xe9, | |
199 | + 0xf6, 0x0e, 0xff, 0x7c, 0xda, 0xce, 0x55, 0x03, | |
200 | + 0xd9, 0xde, 0x79, 0xf5, 0x2c, 0xaf, 0x06, 0xf8, | |
201 | + 0xb2, 0xc9, 0xf8, 0x78, 0x54, 0xf9, 0xd1, 0xe7, | |
202 | + 0xeb, 0xbe, 0xd7, 0xeb, 0x25, 0xe9, 0x17, 0x99, | |
203 | + 0xbf, 0x24, 0xce, 0x2a, 0x73, 0xfe, 0xf9, 0xbc, | |
204 | + 0xd9, 0x55, 0x91, 0xcf, 0xe0, 0xc9, 0xdf, 0x88, | |
205 | +}; | |
206 | + | |
207 | + | |
208 | +////////// test 2: even key | |
209 | +unsigned char test_2_key[0x8] = { | |
210 | + 0x07, 0x06, 0x05, 0x04, 0x03, 0x02, 0x01, 0x00 | |
211 | +}; | |
212 | +unsigned char test_2_encrypted[0x100] = { | |
213 | + 0x47, 0x00, 0x00, 0x90, | |
214 | + 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, | |
215 | + 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, | |
216 | + 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, | |
217 | + 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, | |
218 | + 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f, | |
219 | + 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, | |
220 | + 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, | |
221 | + 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, | |
222 | + 0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f, | |
223 | + 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, | |
224 | + 0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, | |
225 | + 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, | |
226 | +}; | |
227 | +unsigned char test_2_expected[0x100] = { | |
228 | + 0x47, 0x00, 0x00, 0x90, | |
229 | + 0x2d, 0x0a, 0x47, 0x20, 0x18, 0x11, 0x9c, 0x8a, 0xd1, 0x2a, 0x65, 0x6b, 0x89, 0xe4, 0x35, 0x2b, | |
230 | + 0xc2, 0xb5, 0x90, 0x61, 0xd1, 0x7e, 0x02, 0xe1, 0x3f, 0x46, 0x70, 0xcf, 0x77, 0x91, 0x2f, 0x22, | |
231 | + 0x93, 0xc1, 0x6c, 0xfe, 0x49, 0xad, 0x7c, 0xc2, 0xaf, 0x86, 0x1b, 0xa3, 0x29, 0xbe, 0xaa, 0x64, | |
232 | + 0xf0, 0x22, 0xb9, 0x5e, 0x98, 0xaa, 0x60, 0xef, 0xdf, 0xd6, 0x44, 0x77, 0xe6, 0xbf, 0xbb, 0x94, | |
233 | + 0xb2, 0x0a, 0x63, 0x0e, 0x5c, 0xf2, 0xac, 0xb4, 0x49, 0xcc, 0x9e, 0x4f, 0x94, 0x4c, 0x30, 0x12, | |
234 | + 0xe8, 0x55, 0xc2, 0x44, 0xa4, 0x52, 0xcb, 0x61, 0x81, 0xc9, 0xb6, 0xa6, 0x6b, 0xef, 0xaf, 0xa6, | |
235 | + 0x71, 0x1d, 0x7b, 0x58, 0x2f, 0xfa, 0xd1, 0x0c, 0x07, 0x9d, 0x1f, 0x35, 0x87, 0xbe, 0x02, 0x9f, | |
236 | + 0x20, 0xc6, 0x60, 0x8f, 0x1c, 0x30, 0x0f, 0x96, 0xd0, 0x71, 0xd6, 0x51, 0x10, 0xdf, 0x5b, 0xf6, | |
237 | + 0x44, 0x2f, 0x80, 0x28, 0xb7, 0xec, 0x23, 0x59, 0x4b, 0x94, 0x0b, 0x9a, 0x74, 0xa1, 0x1f, 0xf7, | |
238 | + 0x9e, 0x76, 0xb4, 0xdf, 0xbb, 0x3c, 0x8c, 0x88, 0x97, 0x22, 0x56, 0x73, 0x16, 0x05, 0xac, 0xf9, | |
239 | + 0x4f, 0x77, 0x9d, 0x38, 0xa0, 0x6b, 0x05, 0xd2, 0xe6, 0x15, 0x01, 0xb1, 0x5c, 0xc9, 0x62, 0xa9, | |
240 | + 0x9b, 0x1a, 0x6a, 0x1a, 0xcf, 0xe6, 0xa8, 0xba, | |
241 | +}; | |
242 | + | |
243 | + | |
244 | +////////// test 3: even key | |
245 | +unsigned char test_3_key[0x8] = { | |
246 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff | |
247 | +}; | |
248 | +unsigned char test_3_encrypted[0x100] = { | |
249 | + 0x47, 0x00, 0x00, 0x90, | |
250 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
251 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
252 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
253 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
254 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
255 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
256 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
257 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
258 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
259 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
260 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
261 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff | |
262 | + | |
263 | +}; | |
264 | +unsigned char test_3_expected[0x100] = { | |
265 | + 0x47, 0x00, 0x00, 0x90, | |
266 | + 0xfe, 0x91, 0xa7, 0x2f, 0xbf, 0xb0, 0x6a, 0x54, 0xc1, 0xe4, 0x33, 0x27, 0x18, 0xd5, 0x9c, 0x43, | |
267 | + 0xea, 0xaa, 0x6b, 0x38, 0x5c, 0xe7, 0xae, 0xc9, 0xac, 0xec, 0xef, 0xc3, 0x51, 0x7d, 0x53, 0x47, | |
268 | + 0xa0, 0xa7, 0x6d, 0x73, 0x8a, 0x9d, 0x16, 0x7d, 0x05, 0x2d, 0xd6, 0x6b, 0xf4, 0x8d, 0x4b, 0x81, | |
269 | + 0x98, 0x2f, 0x46, 0xa5, 0x34, 0x84, 0xf3, 0x70, 0xa4, 0xe9, 0x04, 0x84, 0x7b, 0x87, 0x79, 0x3c, | |
270 | + 0x01, 0x25, 0xb5, 0xfc, 0x3d, 0xd0, 0x25, 0xea, 0x2f, 0x91, 0xf0, 0x3f, 0x7f, 0xd4, 0x8e, 0x1e, | |
271 | + 0x36, 0x83, 0x22, 0x91, 0x57, 0x92, 0x36, 0x0b, 0x44, 0xa5, 0xcc, 0x5e, 0xef, 0x44, 0x3e, 0xf8, | |
272 | + 0xe9, 0x7b, 0x5e, 0x0c, 0xea, 0xb2, 0x50, 0x39, 0xb7, 0xea, 0xc4, 0xfb, 0xe4, 0x37, 0xf8, 0x85, | |
273 | + 0xc2, 0xdc, 0x01, 0x98, 0x01, 0x2a, 0x44, 0xd3, 0x75, 0x10, 0x38, 0xf4, 0x85, 0x3e, 0xc9, 0xf7, | |
274 | + 0xe7, 0xe4, 0xec, 0x40, 0x3d, 0x8f, 0xa5, 0xd2, 0x8a, 0xca, 0x62, 0x03, 0x3f, 0x65, 0x28, 0x8d, | |
275 | + 0xf5, 0x56, 0xa7, 0xea, 0xd1, 0x0d, 0x70, 0x82, 0xbc, 0x90, 0x59, 0xf8, 0x3e, 0x08, 0xc9, 0xe1, | |
276 | + 0x97, 0xef, 0x82, 0x43, 0x35, 0x41, 0x3e, 0x7f, 0x00, 0x96, 0x3f, 0x90, 0xe5, 0x1e, 0x96, 0xba, | |
277 | + 0xce, 0x6d, 0xd2, 0x54, 0xce, 0x84, 0x76, 0x3c | |
278 | +}; | |
279 | + | |
280 | + | |
281 | +////////// odd key, only 80 (0x50) bytes of payload (10 groups of 8 bytes + 0 byte residue) | |
282 | +unsigned char test_p_10_0_key[0x8] = { | |
283 | + 0x2d, 0x11, 0x5f, 0x9d, 0x29, 0xbf, 0x7f, 0x67 | |
284 | +}; | |
285 | +unsigned char test_p_10_0_encrypted[0x100] = { | |
286 | + 0x47, 0x00, 0x7a, 0xbe, | |
287 | + 0x67, 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
288 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
289 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
290 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
291 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
292 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
293 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x71, 0xa5, 0x7b, 0x8f, 0xf9, 0x87, 0xcb, 0xac, | |
294 | + 0xea, 0x08, 0x0c, 0x02, 0x87, 0x7b, 0xad, 0x10, 0x40, 0x28, 0x8e, 0xd4, 0x4e, 0x62, 0xc7, 0x74, | |
295 | + 0xd6, 0xbb, 0x3a, 0xaa, 0xb0, 0x7b, 0x70, 0xbe, 0x06, 0xc9, 0xdc, 0x07, 0xd2, 0x2d, 0xab, 0x2d, | |
296 | + 0xe2, 0xc6, 0x36, 0xa6, 0xda, 0x64, 0x61, 0x15, 0xd1, 0x6a, 0x40, 0xc0, 0xa9, 0xfb, 0x3f, 0xb2, | |
297 | + 0x6d, 0xa5, 0x59, 0xae, 0x57, 0x88, 0x6b, 0x0e, 0x00, 0xae, 0xce, 0x64, 0xee, 0xfd, 0xb1, 0x7f, | |
298 | + 0x78, 0x9c, 0x12, 0x42, 0xbe, 0x30, 0x8a, 0xa3 | |
299 | +}; | |
300 | +unsigned char test_p_10_0_expected[0x100] = { | |
301 | + 0x47, 0x00, 0x7a, 0xbe, | |
302 | + 0x67, 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
303 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
304 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
305 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
306 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
307 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
308 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xa7, 0xca, 0x32, 0xaf, 0x2e, 0x6a, 0xea, 0x05, | |
309 | + 0x39, 0x33, 0x67, 0x5d, 0xa3, 0x61, 0x0f, 0x34, 0x40, 0x6c, 0x1a, 0xb3, 0xee, 0x54, 0x64, 0xd5, | |
310 | + 0xa3, 0x01, 0x95, 0x87, 0x9d, 0x3d, 0x38, 0xc5, 0x82, 0x8b, 0x8d, 0xab, 0xad, 0x93, 0x0f, 0xe8, | |
311 | + 0xf9, 0xbd, 0x52, 0x98, 0x59, 0xb2, 0x41, 0x95, 0xcd, 0xae, 0x9b, 0x3e, 0xdf, 0xdb, 0x14, 0x9b, | |
312 | + 0xa9, 0x22, 0x0d, 0x2d, 0x61, 0xf5, 0xf2, 0x52, 0x83, 0x20, 0xae, 0xb8, 0x83, 0x52, 0x02, 0xee, | |
313 | + 0xbd, 0xd2, 0x94, 0x6c, 0x27, 0x58, 0x55, 0xd0 | |
314 | +}; | |
315 | + | |
316 | + | |
317 | +////////// odd key, only 14 (0x0e) bytes of payload (1 group of 8 bytes + 6 byte residue) | |
318 | +unsigned char test_p_1_6_key[0x8] = { | |
319 | + 0x2d, 0x11, 0x5f, 0x9d, 0x29, 0xbf, 0x7f, 0x67 | |
320 | +}; | |
321 | +unsigned char test_p_1_6_encrypted[0x100] = { | |
322 | + 0x47, 0x00, 0x7a, 0xb7, | |
323 | + 0xa9, 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
324 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
325 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
326 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
327 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
328 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
329 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
330 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
331 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
332 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
333 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xc0, 0x5e, 0xfb, 0xc8, 0x4a, 0x63, | |
334 | + 0xe3, 0x3c, 0x11, 0xd9, 0xe0, 0x75, 0x8e, 0xf2 | |
335 | +}; | |
336 | +unsigned char test_p_1_6_expected[0x100] = { | |
337 | + 0x47, 0x00, 0x7a, 0xb7, | |
338 | + 0xa9, 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
339 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
340 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
341 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
342 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
343 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
344 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
345 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
346 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
347 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, | |
348 | + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x5a, 0x2c, 0xee, 0xb3, 0xde, 0x92, | |
349 | + 0xe7, 0xa6, 0x6c, 0xaa, 0x99, 0x84, 0xe4, 0x00 | |
350 | +}; | |
351 | Index: libs/libmythtv/FFdecsa/COPYING | |
352 | =================================================================== | |
353 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
354 | +++ libs/libmythtv/FFdecsa/COPYING 2006-06-20 17:36:06.000000000 -0400 | |
355 | @@ -0,0 +1,339 @@ | |
356 | + GNU GENERAL PUBLIC LICENSE | |
357 | + Version 2, June 1991 | |
358 | + | |
359 | + Copyright (C) 1989, 1991 Free Software Foundation, Inc. | |
360 | + 675 Mass Ave, Cambridge, MA 02139, USA | |
361 | + Everyone is permitted to copy and distribute verbatim copies | |
362 | + of this license document, but changing it is not allowed. | |
363 | + | |
364 | + Preamble | |
365 | + | |
366 | + The licenses for most software are designed to take away your | |
367 | +freedom to share and change it. By contrast, the GNU General Public | |
368 | +License is intended to guarantee your freedom to share and change free | |
369 | +software--to make sure the software is free for all its users. This | |
370 | +General Public License applies to most of the Free Software | |
371 | +Foundation's software and to any other program whose authors commit to | |
372 | +using it. (Some other Free Software Foundation software is covered by | |
373 | +the GNU Library General Public License instead.) You can apply it to | |
374 | +your programs, too. | |
375 | + | |
376 | + When we speak of free software, we are referring to freedom, not | |
377 | +price. Our General Public Licenses are designed to make sure that you | |
378 | +have the freedom to distribute copies of free software (and charge for | |
379 | +this service if you wish), that you receive source code or can get it | |
380 | +if you want it, that you can change the software or use pieces of it | |
381 | +in new free programs; and that you know you can do these things. | |
382 | + | |
383 | + To protect your rights, we need to make restrictions that forbid | |
384 | +anyone to deny you these rights or to ask you to surrender the rights. | |
385 | +These restrictions translate to certain responsibilities for you if you | |
386 | +distribute copies of the software, or if you modify it. | |
387 | + | |
388 | + For example, if you distribute copies of such a program, whether | |
389 | +gratis or for a fee, you must give the recipients all the rights that | |
390 | +you have. You must make sure that they, too, receive or can get the | |
391 | +source code. And you must show them these terms so they know their | |
392 | +rights. | |
393 | + | |
394 | + We protect your rights with two steps: (1) copyright the software, and | |
395 | +(2) offer you this license which gives you legal permission to copy, | |
396 | +distribute and/or modify the software. | |
397 | + | |
398 | + Also, for each author's protection and ours, we want to make certain | |
399 | +that everyone understands that there is no warranty for this free | |
400 | +software. If the software is modified by someone else and passed on, we | |
401 | +want its recipients to know that what they have is not the original, so | |
402 | +that any problems introduced by others will not reflect on the original | |
403 | +authors' reputations. | |
404 | + | |
405 | + Finally, any free program is threatened constantly by software | |
406 | +patents. We wish to avoid the danger that redistributors of a free | |
407 | +program will individually obtain patent licenses, in effect making the | |
408 | +program proprietary. To prevent this, we have made it clear that any | |
409 | +patent must be licensed for everyone's free use or not licensed at all. | |
410 | + | |
411 | + The precise terms and conditions for copying, distribution and | |
412 | +modification follow. | |
413 | +\f | |
414 | + GNU GENERAL PUBLIC LICENSE | |
415 | + TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION | |
416 | + | |
417 | + 0. This License applies to any program or other work which contains | |
418 | +a notice placed by the copyright holder saying it may be distributed | |
419 | +under the terms of this General Public License. The "Program", below, | |
420 | +refers to any such program or work, and a "work based on the Program" | |
421 | +means either the Program or any derivative work under copyright law: | |
422 | +that is to say, a work containing the Program or a portion of it, | |
423 | +either verbatim or with modifications and/or translated into another | |
424 | +language. (Hereinafter, translation is included without limitation in | |
425 | +the term "modification".) Each licensee is addressed as "you". | |
426 | + | |
427 | +Activities other than copying, distribution and modification are not | |
428 | +covered by this License; they are outside its scope. The act of | |
429 | +running the Program is not restricted, and the output from the Program | |
430 | +is covered only if its contents constitute a work based on the | |
431 | +Program (independent of having been made by running the Program). | |
432 | +Whether that is true depends on what the Program does. | |
433 | + | |
434 | + 1. You may copy and distribute verbatim copies of the Program's | |
435 | +source code as you receive it, in any medium, provided that you | |
436 | +conspicuously and appropriately publish on each copy an appropriate | |
437 | +copyright notice and disclaimer of warranty; keep intact all the | |
438 | +notices that refer to this License and to the absence of any warranty; | |
439 | +and give any other recipients of the Program a copy of this License | |
440 | +along with the Program. | |
441 | + | |
442 | +You may charge a fee for the physical act of transferring a copy, and | |
443 | +you may at your option offer warranty protection in exchange for a fee. | |
444 | + | |
445 | + 2. You may modify your copy or copies of the Program or any portion | |
446 | +of it, thus forming a work based on the Program, and copy and | |
447 | +distribute such modifications or work under the terms of Section 1 | |
448 | +above, provided that you also meet all of these conditions: | |
449 | + | |
450 | + a) You must cause the modified files to carry prominent notices | |
451 | + stating that you changed the files and the date of any change. | |
452 | + | |
453 | + b) You must cause any work that you distribute or publish, that in | |
454 | + whole or in part contains or is derived from the Program or any | |
455 | + part thereof, to be licensed as a whole at no charge to all third | |
456 | + parties under the terms of this License. | |
457 | + | |
458 | + c) If the modified program normally reads commands interactively | |
459 | + when run, you must cause it, when started running for such | |
460 | + interactive use in the most ordinary way, to print or display an | |
461 | + announcement including an appropriate copyright notice and a | |
462 | + notice that there is no warranty (or else, saying that you provide | |
463 | + a warranty) and that users may redistribute the program under | |
464 | + these conditions, and telling the user how to view a copy of this | |
465 | + License. (Exception: if the Program itself is interactive but | |
466 | + does not normally print such an announcement, your work based on | |
467 | + the Program is not required to print an announcement.) | |
468 | +\f | |
469 | +These requirements apply to the modified work as a whole. If | |
470 | +identifiable sections of that work are not derived from the Program, | |
471 | +and can be reasonably considered independent and separate works in | |
472 | +themselves, then this License, and its terms, do not apply to those | |
473 | +sections when you distribute them as separate works. But when you | |
474 | +distribute the same sections as part of a whole which is a work based | |
475 | +on the Program, the distribution of the whole must be on the terms of | |
476 | +this License, whose permissions for other licensees extend to the | |
477 | +entire whole, and thus to each and every part regardless of who wrote it. | |
478 | + | |
479 | +Thus, it is not the intent of this section to claim rights or contest | |
480 | +your rights to work written entirely by you; rather, the intent is to | |
481 | +exercise the right to control the distribution of derivative or | |
482 | +collective works based on the Program. | |
483 | + | |
484 | +In addition, mere aggregation of another work not based on the Program | |
485 | +with the Program (or with a work based on the Program) on a volume of | |
486 | +a storage or distribution medium does not bring the other work under | |
487 | +the scope of this License. | |
488 | + | |
489 | + 3. You may copy and distribute the Program (or a work based on it, | |
490 | +under Section 2) in object code or executable form under the terms of | |
491 | +Sections 1 and 2 above provided that you also do one of the following: | |
492 | + | |
493 | + a) Accompany it with the complete corresponding machine-readable | |
494 | + source code, which must be distributed under the terms of Sections | |
495 | + 1 and 2 above on a medium customarily used for software interchange; or, | |
496 | + | |
497 | + b) Accompany it with a written offer, valid for at least three | |
498 | + years, to give any third party, for a charge no more than your | |
499 | + cost of physically performing source distribution, a complete | |
500 | + machine-readable copy of the corresponding source code, to be | |
501 | + distributed under the terms of Sections 1 and 2 above on a medium | |
502 | + customarily used for software interchange; or, | |
503 | + | |
504 | + c) Accompany it with the information you received as to the offer | |
505 | + to distribute corresponding source code. (This alternative is | |
506 | + allowed only for noncommercial distribution and only if you | |
507 | + received the program in object code or executable form with such | |
508 | + an offer, in accord with Subsection b above.) | |
509 | + | |
510 | +The source code for a work means the preferred form of the work for | |
511 | +making modifications to it. For an executable work, complete source | |
512 | +code means all the source code for all modules it contains, plus any | |
513 | +associated interface definition files, plus the scripts used to | |
514 | +control compilation and installation of the executable. However, as a | |
515 | +special exception, the source code distributed need not include | |
516 | +anything that is normally distributed (in either source or binary | |
517 | +form) with the major components (compiler, kernel, and so on) of the | |
518 | +operating system on which the executable runs, unless that component | |
519 | +itself accompanies the executable. | |
520 | + | |
521 | +If distribution of executable or object code is made by offering | |
522 | +access to copy from a designated place, then offering equivalent | |
523 | +access to copy the source code from the same place counts as | |
524 | +distribution of the source code, even though third parties are not | |
525 | +compelled to copy the source along with the object code. | |
526 | +\f | |
527 | + 4. You may not copy, modify, sublicense, or distribute the Program | |
528 | +except as expressly provided under this License. Any attempt | |
529 | +otherwise to copy, modify, sublicense or distribute the Program is | |
530 | +void, and will automatically terminate your rights under this License. | |
531 | +However, parties who have received copies, or rights, from you under | |
532 | +this License will not have their licenses terminated so long as such | |
533 | +parties remain in full compliance. | |
534 | + | |
535 | + 5. You are not required to accept this License, since you have not | |
536 | +signed it. However, nothing else grants you permission to modify or | |
537 | +distribute the Program or its derivative works. These actions are | |
538 | +prohibited by law if you do not accept this License. Therefore, by | |
539 | +modifying or distributing the Program (or any work based on the | |
540 | +Program), you indicate your acceptance of this License to do so, and | |
541 | +all its terms and conditions for copying, distributing or modifying | |
542 | +the Program or works based on it. | |
543 | + | |
544 | + 6. Each time you redistribute the Program (or any work based on the | |
545 | +Program), the recipient automatically receives a license from the | |
546 | +original licensor to copy, distribute or modify the Program subject to | |
547 | +these terms and conditions. You may not impose any further | |
548 | +restrictions on the recipients' exercise of the rights granted herein. | |
549 | +You are not responsible for enforcing compliance by third parties to | |
550 | +this License. | |
551 | + | |
552 | + 7. If, as a consequence of a court judgment or allegation of patent | |
553 | +infringement or for any other reason (not limited to patent issues), | |
554 | +conditions are imposed on you (whether by court order, agreement or | |
555 | +otherwise) that contradict the conditions of this License, they do not | |
556 | +excuse you from the conditions of this License. If you cannot | |
557 | +distribute so as to satisfy simultaneously your obligations under this | |
558 | +License and any other pertinent obligations, then as a consequence you | |
559 | +may not distribute the Program at all. For example, if a patent | |
560 | +license would not permit royalty-free redistribution of the Program by | |
561 | +all those who receive copies directly or indirectly through you, then | |
562 | +the only way you could satisfy both it and this License would be to | |
563 | +refrain entirely from distribution of the Program. | |
564 | + | |
565 | +If any portion of this section is held invalid or unenforceable under | |
566 | +any particular circumstance, the balance of the section is intended to | |
567 | +apply and the section as a whole is intended to apply in other | |
568 | +circumstances. | |
569 | + | |
570 | +It is not the purpose of this section to induce you to infringe any | |
571 | +patents or other property right claims or to contest validity of any | |
572 | +such claims; this section has the sole purpose of protecting the | |
573 | +integrity of the free software distribution system, which is | |
574 | +implemented by public license practices. Many people have made | |
575 | +generous contributions to the wide range of software distributed | |
576 | +through that system in reliance on consistent application of that | |
577 | +system; it is up to the author/donor to decide if he or she is willing | |
578 | +to distribute software through any other system and a licensee cannot | |
579 | +impose that choice. | |
580 | + | |
581 | +This section is intended to make thoroughly clear what is believed to | |
582 | +be a consequence of the rest of this License. | |
583 | +\f | |
584 | + 8. If the distribution and/or use of the Program is restricted in | |
585 | +certain countries either by patents or by copyrighted interfaces, the | |
586 | +original copyright holder who places the Program under this License | |
587 | +may add an explicit geographical distribution limitation excluding | |
588 | +those countries, so that distribution is permitted only in or among | |
589 | +countries not thus excluded. In such case, this License incorporates | |
590 | +the limitation as if written in the body of this License. | |
591 | + | |
592 | + 9. The Free Software Foundation may publish revised and/or new versions | |
593 | +of the General Public License from time to time. Such new versions will | |
594 | +be similar in spirit to the present version, but may differ in detail to | |
595 | +address new problems or concerns. | |
596 | + | |
597 | +Each version is given a distinguishing version number. If the Program | |
598 | +specifies a version number of this License which applies to it and "any | |
599 | +later version", you have the option of following the terms and conditions | |
600 | +either of that version or of any later version published by the Free | |
601 | +Software Foundation. If the Program does not specify a version number of | |
602 | +this License, you may choose any version ever published by the Free Software | |
603 | +Foundation. | |
604 | + | |
605 | + 10. If you wish to incorporate parts of the Program into other free | |
606 | +programs whose distribution conditions are different, write to the author | |
607 | +to ask for permission. For software which is copyrighted by the Free | |
608 | +Software Foundation, write to the Free Software Foundation; we sometimes | |
609 | +make exceptions for this. Our decision will be guided by the two goals | |
610 | +of preserving the free status of all derivatives of our free software and | |
611 | +of promoting the sharing and reuse of software generally. | |
612 | + | |
613 | + NO WARRANTY | |
614 | + | |
615 | + 11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY | |
616 | +FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN | |
617 | +OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES | |
618 | +PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED | |
619 | +OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF | |
620 | +MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS | |
621 | +TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE | |
622 | +PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, | |
623 | +REPAIR OR CORRECTION. | |
624 | + | |
625 | + 12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING | |
626 | +WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR | |
627 | +REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, | |
628 | +INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING | |
629 | +OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED | |
630 | +TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY | |
631 | +YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER | |
632 | +PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE | |
633 | +POSSIBILITY OF SUCH DAMAGES. | |
634 | + | |
635 | + END OF TERMS AND CONDITIONS | |
636 | +\f | |
637 | + Appendix: How to Apply These Terms to Your New Programs | |
638 | + | |
639 | + If you develop a new program, and you want it to be of the greatest | |
640 | +possible use to the public, the best way to achieve this is to make it | |
641 | +free software which everyone can redistribute and change under these terms. | |
642 | + | |
643 | + To do so, attach the following notices to the program. It is safest | |
644 | +to attach them to the start of each source file to most effectively | |
645 | +convey the exclusion of warranty; and each file should have at least | |
646 | +the "copyright" line and a pointer to where the full notice is found. | |
647 | + | |
648 | + <one line to give the program's name and a brief idea of what it does.> | |
649 | + Copyright (C) 19yy <name of author> | |
650 | + | |
651 | + This program is free software; you can redistribute it and/or modify | |
652 | + it under the terms of the GNU General Public License as published by | |
653 | + the Free Software Foundation; either version 2 of the License, or | |
654 | + (at your option) any later version. | |
655 | + | |
656 | + This program is distributed in the hope that it will be useful, | |
657 | + but WITHOUT ANY WARRANTY; without even the implied warranty of | |
658 | + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
659 | + GNU General Public License for more details. | |
660 | + | |
661 | + You should have received a copy of the GNU General Public License | |
662 | + along with this program; if not, write to the Free Software | |
663 | + Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
664 | + | |
665 | +Also add information on how to contact you by electronic and paper mail. | |
666 | + | |
667 | +If the program is interactive, make it output a short notice like this | |
668 | +when it starts in an interactive mode: | |
669 | + | |
670 | + Gnomovision version 69, Copyright (C) 19yy name of author | |
671 | + Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. | |
672 | + This is free software, and you are welcome to redistribute it | |
673 | + under certain conditions; type `show c' for details. | |
674 | + | |
675 | +The hypothetical commands `show w' and `show c' should show the appropriate | |
676 | +parts of the General Public License. Of course, the commands you use may | |
677 | +be called something other than `show w' and `show c'; they could even be | |
678 | +mouse-clicks or menu items--whatever suits your program. | |
679 | + | |
680 | +You should also get your employer (if you work as a programmer) or your | |
681 | +school, if any, to sign a "copyright disclaimer" for the program, if | |
682 | +necessary. Here is a sample; alter the names: | |
683 | + | |
684 | + Yoyodyne, Inc., hereby disclaims all copyright interest in the program | |
685 | + `Gnomovision' (which makes passes at compilers) written by James Hacker. | |
686 | + | |
687 | + <signature of Ty Coon>, 1 April 1989 | |
688 | + Ty Coon, President of Vice | |
689 | + | |
690 | +This General Public License does not permit incorporating your program into | |
691 | +proprietary programs. If your program is a subroutine library, you may | |
692 | +consider it more useful to permit linking proprietary applications with the | |
693 | +library. If this is what you want to do, use the GNU Library General | |
694 | +Public License instead of this License. | |
695 | Index: libs/libmythtv/FFdecsa/parallel_032_int.h | |
696 | =================================================================== | |
697 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
698 | +++ libs/libmythtv/FFdecsa/parallel_032_int.h 2006-06-20 17:36:06.000000000 -0400 | |
699 | @@ -0,0 +1,55 @@ | |
700 | +/* FFdecsa -- fast decsa algorithm | |
701 | + * | |
702 | + * Copyright (C) 2003-2004 fatih89r | |
703 | + * | |
704 | + * This program is free software; you can redistribute it and/or modify | |
705 | + * it under the terms of the GNU General Public License as published by | |
706 | + * the Free Software Foundation; either version 2 of the License, or | |
707 | + * (at your option) any later version. | |
708 | + * | |
709 | + * This program is distributed in the hope that it will be useful, | |
710 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
711 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
712 | + * GNU General Public License for more details. | |
713 | + * | |
714 | + * You should have received a copy of the GNU General Public License | |
715 | + * along with this program; if not, write to the Free Software | |
716 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
717 | + */ | |
718 | + | |
719 | +#include "parallel_std_def.h" | |
720 | + | |
721 | +typedef unsigned int group; | |
722 | +#define GROUP_PARALLELISM 32 | |
723 | +#define FF0() 0x0 | |
724 | +#define FF1() 0xffffffff | |
725 | + | |
726 | +/* 64 rows of 32 bits */ | |
727 | + | |
728 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
729 | + *(((int *)tab)+g)=*((int *)data); | |
730 | + *(((int *)tab)+32+g)=*(((int *)data)+1); | |
731 | +} | |
732 | + | |
733 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
734 | + *((int *)data)=*(((int *)tab)+g); | |
735 | + *(((int *)data)+1)=*(((int *)tab)+32+g); | |
736 | +} | |
737 | + | |
738 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
739 | + int j; | |
740 | + for(j=0;j<n;j++){ | |
741 | + *(data+j)^=*(tab+4*(g+(j>=4?32-1:0))+j); | |
742 | + } | |
743 | +} | |
744 | + | |
745 | +typedef unsigned int batch; | |
746 | +#define BYTES_PER_BATCH 4 | |
747 | +#define B_FFN_ALL_29() 0x29292929 | |
748 | +#define B_FFN_ALL_02() 0x02020202 | |
749 | +#define B_FFN_ALL_04() 0x04040404 | |
750 | +#define B_FFN_ALL_10() 0x10101010 | |
751 | +#define B_FFN_ALL_40() 0x40404040 | |
752 | +#define B_FFN_ALL_80() 0x80808080 | |
753 | + | |
754 | +#define M_EMPTY() | |
755 | Index: libs/libmythtv/FFdecsa/tmp_autogenerated_stuff_stream.c | |
756 | =================================================================== | |
757 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
758 | +++ libs/libmythtv/FFdecsa/tmp_autogenerated_stuff_stream.c 2006-06-20 17:36:06.000000000 -0400 | |
759 | @@ -0,0 +1,814 @@ | |
760 | +/* FFdecsa -- fast decsa algorithm | |
761 | + * | |
762 | + * Copyright (C) 2003-2004 fatih89r | |
763 | + * | |
764 | + * This program is free software; you can redistribute it and/or modify | |
765 | + * it under the terms of the GNU General Public License as published by | |
766 | + * the Free Software Foundation; either version 2 of the License, or | |
767 | + * (at your option) any later version. | |
768 | + * | |
769 | + * This program is distributed in the hope that it will be useful, | |
770 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
771 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
772 | + * GNU General Public License for more details. | |
773 | + * | |
774 | + * You should have received a copy of the GNU General Public License | |
775 | + * along with this program; if not, write to the Free Software | |
776 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
777 | + */ | |
778 | + | |
779 | + | |
780 | + | |
781 | +// define statics only once, when STREAM_INIT | |
782 | +#ifdef STREAM_INIT | |
783 | +static group A[32+10][4]; // 32 because we will move back (virtual shift register) | |
784 | +static group B[32+10][4]; // 32 because we will move back (virtual shift register) | |
785 | +static group X[4]; | |
786 | +static group Y[4]; | |
787 | +static group Z[4]; | |
788 | +static group D[4]; | |
789 | +static group E[4]; | |
790 | +static group F[4]; | |
791 | +static group p; | |
792 | +static group q; | |
793 | +static group r; | |
794 | + | |
795 | +static inline void trasp64_32_88ccw(unsigned char *data){ | |
796 | +/* 64 rows of 32 bits transposition (bytes transp. - 8x8 rotate counterclockwise)*/ | |
797 | +#define row ((unsigned int *)data) | |
798 | + int i,j; | |
799 | + for(j=0;j<64;j+=32){ | |
800 | + unsigned int t,b; | |
801 | + for(i=0;i<16;i++){ | |
802 | + t=row[j+i]; | |
803 | + b=row[j+16+i]; | |
804 | + row[j+i] = (t&0x0000ffff) | ((b )<<16); | |
805 | + row[j+16+i]=((t )>>16) | (b&0xffff0000) ; | |
806 | + } | |
807 | + } | |
808 | + for(j=0;j<64;j+=16){ | |
809 | + unsigned int t,b; | |
810 | + for(i=0;i<8;i++){ | |
811 | + t=row[j+i]; | |
812 | + b=row[j+8+i]; | |
813 | + row[j+i] = (t&0x00ff00ff) | ((b&0x00ff00ff)<<8); | |
814 | + row[j+8+i] =((t&0xff00ff00)>>8) | (b&0xff00ff00); | |
815 | + } | |
816 | + } | |
817 | + for(j=0;j<64;j+=8){ | |
818 | + unsigned int t,b; | |
819 | + for(i=0;i<4;i++){ | |
820 | + t=row[j+i]; | |
821 | + b=row[j+4+i]; | |
822 | + row[j+i] =((t&0x0f0f0f0f)<<4) | (b&0x0f0f0f0f); | |
823 | + row[j+4+i] = (t&0xf0f0f0f0) | ((b&0xf0f0f0f0)>>4); | |
824 | + } | |
825 | + } | |
826 | + for(j=0;j<64;j+=4){ | |
827 | + unsigned int t,b; | |
828 | + for(i=0;i<2;i++){ | |
829 | + t=row[j+i]; | |
830 | + b=row[j+2+i]; | |
831 | + row[j+i] =((t&0x33333333)<<2) | (b&0x33333333); | |
832 | + row[j+2+i] = (t&0xcccccccc) | ((b&0xcccccccc)>>2); | |
833 | + } | |
834 | + } | |
835 | + for(j=0;j<64;j+=2){ | |
836 | + unsigned int t,b; | |
837 | + for(i=0;i<1;i++){ | |
838 | + t=row[j+i]; | |
839 | + b=row[j+1+i]; | |
840 | + row[j+i] =((t&0x55555555)<<1) | (b&0x55555555); | |
841 | + row[j+1+i] = (t&0xaaaaaaaa) | ((b&0xaaaaaaaa)>>1); | |
842 | + } | |
843 | + } | |
844 | +#undef row | |
845 | +} | |
846 | + | |
847 | +static inline void trasp64_32_88cw(unsigned char *data){ | |
848 | +/* 64 rows of 32 bits transposition (bytes transp. - 8x8 rotate clockwise)*/ | |
849 | +#define row ((unsigned int *)data) | |
850 | + int i,j; | |
851 | + for(j=0;j<64;j+=32){ | |
852 | + unsigned int t,b; | |
853 | + for(i=0;i<16;i++){ | |
854 | + t=row[j+i]; | |
855 | + b=row[j+16+i]; | |
856 | + row[j+i] = (t&0x0000ffff) | ((b )<<16); | |
857 | + row[j+16+i]=((t )>>16) | (b&0xffff0000) ; | |
858 | + } | |
859 | + } | |
860 | + for(j=0;j<64;j+=16){ | |
861 | + unsigned int t,b; | |
862 | + for(i=0;i<8;i++){ | |
863 | + t=row[j+i]; | |
864 | + b=row[j+8+i]; | |
865 | + row[j+i] = (t&0x00ff00ff) | ((b&0x00ff00ff)<<8); | |
866 | + row[j+8+i] =((t&0xff00ff00)>>8) | (b&0xff00ff00); | |
867 | + } | |
868 | + } | |
869 | + for(j=0;j<64;j+=8){ | |
870 | + unsigned int t,b; | |
871 | + for(i=0;i<4;i++){ | |
872 | + t=row[j+i]; | |
873 | + b=row[j+4+i]; | |
874 | + row[j+i] =((t&0xf0f0f0f0)>>4) | (b&0xf0f0f0f0); | |
875 | + row[j+4+i]= (t&0x0f0f0f0f) | ((b&0x0f0f0f0f)<<4); | |
876 | + } | |
877 | + } | |
878 | + for(j=0;j<64;j+=4){ | |
879 | + unsigned int t,b; | |
880 | + for(i=0;i<2;i++){ | |
881 | + t=row[j+i]; | |
882 | + b=row[j+2+i]; | |
883 | + row[j+i] =((t&0xcccccccc)>>2) | (b&0xcccccccc); | |
884 | + row[j+2+i]= (t&0x33333333) | ((b&0x33333333)<<2); | |
885 | + } | |
886 | + } | |
887 | + for(j=0;j<64;j+=2){ | |
888 | + unsigned int t,b; | |
889 | + for(i=0;i<1;i++){ | |
890 | + t=row[j+i]; | |
891 | + b=row[j+1+i]; | |
892 | + row[j+i] =((t&0xaaaaaaaa)>>1) | (b&0xaaaaaaaa); | |
893 | + row[j+1+i]= (t&0x55555555) | ((b&0x55555555)<<1); | |
894 | + } | |
895 | + } | |
896 | +#undef row | |
897 | +} | |
898 | + | |
899 | +//64-64---------------------------------------------------------- | |
900 | +static inline void trasp64_64_88ccw(unsigned char *data){ | |
901 | +/* 64 rows of 64 bits transposition (bytes transp. - 8x8 rotate counterclockwise)*/ | |
902 | +#define row ((unsigned long long int *)data) | |
903 | + int i,j; | |
904 | + for(j=0;j<64;j+=64){ | |
905 | + unsigned long long int t,b; | |
906 | + for(i=0;i<32;i++){ | |
907 | + t=row[j+i]; | |
908 | + b=row[j+32+i]; | |
909 | + row[j+i] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
910 | + row[j+32+i]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
911 | + } | |
912 | + } | |
913 | + for(j=0;j<64;j+=32){ | |
914 | + unsigned long long int t,b; | |
915 | + for(i=0;i<16;i++){ | |
916 | + t=row[j+i]; | |
917 | + b=row[j+16+i]; | |
918 | + row[j+i] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
919 | + row[j+16+i]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
920 | + } | |
921 | + } | |
922 | + for(j=0;j<64;j+=16){ | |
923 | + unsigned long long int t,b; | |
924 | + for(i=0;i<8;i++){ | |
925 | + t=row[j+i]; | |
926 | + b=row[j+8+i]; | |
927 | + row[j+i] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
928 | + row[j+8+i] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
929 | + } | |
930 | + } | |
931 | + for(j=0;j<64;j+=8){ | |
932 | + unsigned long long int t,b; | |
933 | + for(i=0;i<4;i++){ | |
934 | + t=row[j+i]; | |
935 | + b=row[j+4+i]; | |
936 | + row[j+i] =((t&0x0f0f0f0f0f0f0f0fULL)<<4) | (b&0x0f0f0f0f0f0f0f0fULL); | |
937 | + row[j+4+i] = (t&0xf0f0f0f0f0f0f0f0ULL) | ((b&0xf0f0f0f0f0f0f0f0ULL)>>4); | |
938 | + } | |
939 | + } | |
940 | + for(j=0;j<64;j+=4){ | |
941 | + unsigned long long int t,b; | |
942 | + for(i=0;i<2;i++){ | |
943 | + t=row[j+i]; | |
944 | + b=row[j+2+i]; | |
945 | + row[j+i] =((t&0x3333333333333333ULL)<<2) | (b&0x3333333333333333ULL); | |
946 | + row[j+2+i] = (t&0xccccccccccccccccULL) | ((b&0xccccccccccccccccULL)>>2); | |
947 | + } | |
948 | + } | |
949 | + for(j=0;j<64;j+=2){ | |
950 | + unsigned long long int t,b; | |
951 | + for(i=0;i<1;i++){ | |
952 | + t=row[j+i]; | |
953 | + b=row[j+1+i]; | |
954 | + row[j+i] =((t&0x5555555555555555ULL)<<1) | (b&0x5555555555555555ULL); | |
955 | + row[j+1+i] = (t&0xaaaaaaaaaaaaaaaaULL) | ((b&0xaaaaaaaaaaaaaaaaULL)>>1); | |
956 | + } | |
957 | + } | |
958 | +#undef row | |
959 | +} | |
960 | + | |
961 | +static inline void trasp64_64_88cw(unsigned char *data){ | |
962 | +/* 64 rows of 64 bits transposition (bytes transp. - 8x8 rotate clockwise)*/ | |
963 | +#define row ((unsigned long long int *)data) | |
964 | + int i,j; | |
965 | + for(j=0;j<64;j+=64){ | |
966 | + unsigned long long int t,b; | |
967 | + for(i=0;i<32;i++){ | |
968 | + t=row[j+i]; | |
969 | + b=row[j+32+i]; | |
970 | + row[j+i] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
971 | + row[j+32+i]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
972 | + } | |
973 | + } | |
974 | + for(j=0;j<64;j+=32){ | |
975 | + unsigned long long int t,b; | |
976 | + for(i=0;i<16;i++){ | |
977 | + t=row[j+i]; | |
978 | + b=row[j+16+i]; | |
979 | + row[j+i] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
980 | + row[j+16+i]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
981 | + } | |
982 | + } | |
983 | + for(j=0;j<64;j+=16){ | |
984 | + unsigned long long int t,b; | |
985 | + for(i=0;i<8;i++){ | |
986 | + t=row[j+i]; | |
987 | + b=row[j+8+i]; | |
988 | + row[j+i] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
989 | + row[j+8+i] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
990 | + } | |
991 | + } | |
992 | + for(j=0;j<64;j+=8){ | |
993 | + unsigned long long int t,b; | |
994 | + for(i=0;i<4;i++){ | |
995 | + t=row[j+i]; | |
996 | + b=row[j+4+i]; | |
997 | + row[j+i] =((t&0xf0f0f0f0f0f0f0f0ULL)>>4) | (b&0xf0f0f0f0f0f0f0f0ULL); | |
998 | + row[j+4+i] = (t&0x0f0f0f0f0f0f0f0fULL) | ((b&0x0f0f0f0f0f0f0f0fULL)<<4); | |
999 | + } | |
1000 | + } | |
1001 | + for(j=0;j<64;j+=4){ | |
1002 | + unsigned long long int t,b; | |
1003 | + for(i=0;i<2;i++){ | |
1004 | + t=row[j+i]; | |
1005 | + b=row[j+2+i]; | |
1006 | + row[j+i] =((t&0xccccccccccccccccULL)>>2) | (b&0xccccccccccccccccULL); | |
1007 | + row[j+2+i] = (t&0x3333333333333333ULL) | ((b&0x3333333333333333ULL)<<2); | |
1008 | + } | |
1009 | + } | |
1010 | + for(j=0;j<64;j+=2){ | |
1011 | + unsigned long long int t,b; | |
1012 | + for(i=0;i<1;i++){ | |
1013 | + t=row[j+i]; | |
1014 | + b=row[j+1+i]; | |
1015 | + row[j+i] =((t&0xaaaaaaaaaaaaaaaaULL)>>1) | (b&0xaaaaaaaaaaaaaaaaULL); | |
1016 | + row[j+1+i] = (t&0x5555555555555555ULL) | ((b&0x5555555555555555ULL)<<1); | |
1017 | + } | |
1018 | + } | |
1019 | +#undef row | |
1020 | +} | |
1021 | + | |
1022 | +//64-128---------------------------------------------------------- | |
1023 | +static inline void trasp64_128_88ccw(unsigned char *data){ | |
1024 | +/* 64 rows of 128 bits transposition (bytes transp. - 8x8 rotate counterclockwise)*/ | |
1025 | +#define halfrow ((unsigned long long int *)data) | |
1026 | + int i,j; | |
1027 | + for(j=0;j<64;j+=64){ | |
1028 | + unsigned long long int t,b; | |
1029 | + for(i=0;i<32;i++){ | |
1030 | + t=halfrow[2*(j+i)]; | |
1031 | + b=halfrow[2*(j+32+i)]; | |
1032 | + halfrow[2*(j+i)] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
1033 | + halfrow[2*(j+32+i)]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
1034 | + t=halfrow[2*(j+i)+1]; | |
1035 | + b=halfrow[2*(j+32+i)+1]; | |
1036 | + halfrow[2*(j+i)+1] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
1037 | + halfrow[2*(j+32+i)+1]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
1038 | + } | |
1039 | + } | |
1040 | + for(j=0;j<64;j+=32){ | |
1041 | + unsigned long long int t,b; | |
1042 | + for(i=0;i<16;i++){ | |
1043 | + t=halfrow[2*(j+i)]; | |
1044 | + b=halfrow[2*(j+16+i)]; | |
1045 | + halfrow[2*(j+i)] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
1046 | + halfrow[2*(j+16+i)]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
1047 | + t=halfrow[2*(j+i)+1]; | |
1048 | + b=halfrow[2*(j+16+i)+1]; | |
1049 | + halfrow[2*(j+i)+1] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
1050 | + halfrow[2*(j+16+i)+1]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
1051 | + } | |
1052 | + } | |
1053 | + for(j=0;j<64;j+=16){ | |
1054 | + unsigned long long int t,b; | |
1055 | + for(i=0;i<8;i++){ | |
1056 | + t=halfrow[2*(j+i)]; | |
1057 | + b=halfrow[2*(j+8+i)]; | |
1058 | + halfrow[2*(j+i)] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
1059 | + halfrow[2*(j+8+i)] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
1060 | + t=halfrow[2*(j+i)+1]; | |
1061 | + b=halfrow[2*(j+8+i)+1]; | |
1062 | + halfrow[2*(j+i)+1] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
1063 | + halfrow[2*(j+8+i)+1] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
1064 | + } | |
1065 | + } | |
1066 | + for(j=0;j<64;j+=8){ | |
1067 | + unsigned long long int t,b; | |
1068 | + for(i=0;i<4;i++){ | |
1069 | + t=halfrow[2*(j+i)]; | |
1070 | + b=halfrow[2*(j+4+i)]; | |
1071 | + halfrow[2*(j+i)] =((t&0x0f0f0f0f0f0f0f0fULL)<<4) | (b&0x0f0f0f0f0f0f0f0fULL); | |
1072 | + halfrow[2*(j+4+i)] = (t&0xf0f0f0f0f0f0f0f0ULL) | ((b&0xf0f0f0f0f0f0f0f0ULL)>>4); | |
1073 | + t=halfrow[2*(j+i)+1]; | |
1074 | + b=halfrow[2*(j+4+i)+1]; | |
1075 | + halfrow[2*(j+i)+1] =((t&0x0f0f0f0f0f0f0f0fULL)<<4) | (b&0x0f0f0f0f0f0f0f0fULL); | |
1076 | + halfrow[2*(j+4+i)+1] = (t&0xf0f0f0f0f0f0f0f0ULL) | ((b&0xf0f0f0f0f0f0f0f0ULL)>>4); | |
1077 | + } | |
1078 | + } | |
1079 | + for(j=0;j<64;j+=4){ | |
1080 | + unsigned long long int t,b; | |
1081 | + for(i=0;i<2;i++){ | |
1082 | + t=halfrow[2*(j+i)]; | |
1083 | + b=halfrow[2*(j+2+i)]; | |
1084 | + halfrow[2*(j+i)] =((t&0x3333333333333333ULL)<<2) | (b&0x3333333333333333ULL); | |
1085 | + halfrow[2*(j+2+i)] = (t&0xccccccccccccccccULL) | ((b&0xccccccccccccccccULL)>>2); | |
1086 | + t=halfrow[2*(j+i)+1]; | |
1087 | + b=halfrow[2*(j+2+i)+1]; | |
1088 | + halfrow[2*(j+i)+1] =((t&0x3333333333333333ULL)<<2) | (b&0x3333333333333333ULL); | |
1089 | + halfrow[2*(j+2+i)+1] = (t&0xccccccccccccccccULL) | ((b&0xccccccccccccccccULL)>>2); | |
1090 | + } | |
1091 | + } | |
1092 | + for(j=0;j<64;j+=2){ | |
1093 | + unsigned long long int t,b; | |
1094 | + for(i=0;i<1;i++){ | |
1095 | + t=halfrow[2*(j+i)]; | |
1096 | + b=halfrow[2*(j+1+i)]; | |
1097 | + halfrow[2*(j+i)] =((t&0x5555555555555555ULL)<<1) | (b&0x5555555555555555ULL); | |
1098 | + halfrow[2*(j+1+i)] = (t&0xaaaaaaaaaaaaaaaaULL) | ((b&0xaaaaaaaaaaaaaaaaULL)>>1); | |
1099 | + t=halfrow[2*(j+i)+1]; | |
1100 | + b=halfrow[2*(j+1+i)+1]; | |
1101 | + halfrow[2*(j+i)+1] =((t&0x5555555555555555ULL)<<1) | (b&0x5555555555555555ULL); | |
1102 | + halfrow[2*(j+1+i)+1] = (t&0xaaaaaaaaaaaaaaaaULL) | ((b&0xaaaaaaaaaaaaaaaaULL)>>1); | |
1103 | + } | |
1104 | + } | |
1105 | +#undef halfrow | |
1106 | +} | |
1107 | + | |
1108 | +static inline void trasp64_128_88cw(unsigned char *data){ | |
1109 | +/* 64 rows of 128 bits transposition (bytes transp. - 8x8 rotate clockwise)*/ | |
1110 | +#define halfrow ((unsigned long long int *)data) | |
1111 | + int i,j; | |
1112 | + for(j=0;j<64;j+=64){ | |
1113 | + unsigned long long int t,b; | |
1114 | + for(i=0;i<32;i++){ | |
1115 | + t=halfrow[2*(j+i)]; | |
1116 | + b=halfrow[2*(j+32+i)]; | |
1117 | + halfrow[2*(j+i)] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
1118 | + halfrow[2*(j+32+i)]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
1119 | + t=halfrow[2*(j+i)+1]; | |
1120 | + b=halfrow[2*(j+32+i)+1]; | |
1121 | + halfrow[2*(j+i)+1] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
1122 | + halfrow[2*(j+32+i)+1]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
1123 | + } | |
1124 | + } | |
1125 | + for(j=0;j<64;j+=32){ | |
1126 | + unsigned long long int t,b; | |
1127 | + for(i=0;i<16;i++){ | |
1128 | + t=halfrow[2*(j+i)]; | |
1129 | + b=halfrow[2*(j+16+i)]; | |
1130 | + halfrow[2*(j+i)] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
1131 | + halfrow[2*(j+16+i)]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
1132 | + t=halfrow[2*(j+i)+1]; | |
1133 | + b=halfrow[2*(j+16+i)+1]; | |
1134 | + halfrow[2*(j+i)+1] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
1135 | + halfrow[2*(j+16+i)+1]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
1136 | + } | |
1137 | + } | |
1138 | + for(j=0;j<64;j+=16){ | |
1139 | + unsigned long long int t,b; | |
1140 | + for(i=0;i<8;i++){ | |
1141 | + t=halfrow[2*(j+i)]; | |
1142 | + b=halfrow[2*(j+8+i)]; | |
1143 | + halfrow[2*(j+i)] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
1144 | + halfrow[2*(j+8+i)] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
1145 | + t=halfrow[2*(j+i)+1]; | |
1146 | + b=halfrow[2*(j+8+i)+1]; | |
1147 | + halfrow[2*(j+i)+1] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
1148 | + halfrow[2*(j+8+i)+1] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
1149 | + } | |
1150 | + } | |
1151 | + for(j=0;j<64;j+=8){ | |
1152 | + unsigned long long int t,b; | |
1153 | + for(i=0;i<4;i++){ | |
1154 | + t=halfrow[2*(j+i)]; | |
1155 | + b=halfrow[2*(j+4+i)]; | |
1156 | + halfrow[2*(j+i)] =((t&0xf0f0f0f0f0f0f0f0ULL)>>4) | (b&0xf0f0f0f0f0f0f0f0ULL); | |
1157 | + halfrow[2*(j+4+i)] = (t&0x0f0f0f0f0f0f0f0fULL) | ((b&0x0f0f0f0f0f0f0f0fULL)<<4); | |
1158 | + t=halfrow[2*(j+i)+1]; | |
1159 | + b=halfrow[2*(j+4+i)+1]; | |
1160 | + halfrow[2*(j+i)+1] =((t&0xf0f0f0f0f0f0f0f0ULL)>>4) | (b&0xf0f0f0f0f0f0f0f0ULL); | |
1161 | + halfrow[2*(j+4+i)+1] = (t&0x0f0f0f0f0f0f0f0fULL) | ((b&0x0f0f0f0f0f0f0f0fULL)<<4); | |
1162 | + } | |
1163 | + } | |
1164 | + for(j=0;j<64;j+=4){ | |
1165 | + unsigned long long int t,b; | |
1166 | + for(i=0;i<2;i++){ | |
1167 | + t=halfrow[2*(j+i)]; | |
1168 | + b=halfrow[2*(j+2+i)]; | |
1169 | + halfrow[2*(j+i)] =((t&0xccccccccccccccccULL)>>2) | (b&0xccccccccccccccccULL); | |
1170 | + halfrow[2*(j+2+i)] = (t&0x3333333333333333ULL) | ((b&0x3333333333333333ULL)<<2); | |
1171 | + t=halfrow[2*(j+i)+1]; | |
1172 | + b=halfrow[2*(j+2+i)+1]; | |
1173 | + halfrow[2*(j+i)+1] =((t&0xccccccccccccccccULL)>>2) | (b&0xccccccccccccccccULL); | |
1174 | + halfrow[2*(j+2+i)+1] = (t&0x3333333333333333ULL) | ((b&0x3333333333333333ULL)<<2); | |
1175 | + } | |
1176 | + } | |
1177 | + for(j=0;j<64;j+=2){ | |
1178 | + unsigned long long int t,b; | |
1179 | + for(i=0;i<1;i++){ | |
1180 | + t=halfrow[2*(j+i)]; | |
1181 | + b=halfrow[2*(j+1+i)]; | |
1182 | + halfrow[2*(j+i)] =((t&0xaaaaaaaaaaaaaaaaULL)>>1) | (b&0xaaaaaaaaaaaaaaaaULL); | |
1183 | + halfrow[2*(j+1+i)] = (t&0x5555555555555555ULL) | ((b&0x5555555555555555ULL)<<1); | |
1184 | + t=halfrow[2*(j+i)+1]; | |
1185 | + b=halfrow[2*(j+1+i)+1]; | |
1186 | + halfrow[2*(j+i)+1] =((t&0xaaaaaaaaaaaaaaaaULL)>>1) | (b&0xaaaaaaaaaaaaaaaaULL); | |
1187 | + halfrow[2*(j+1+i)+1] = (t&0x5555555555555555ULL) | ((b&0x5555555555555555ULL)<<1); | |
1188 | + } | |
1189 | + } | |
1190 | +#undef halfrow | |
1191 | +} | |
1192 | +#endif | |
1193 | + | |
1194 | + | |
1195 | +#ifdef STREAM_INIT | |
1196 | +void stream_cypher_group_init( | |
1197 | + group iA[8][4], // [In] iA00,iA01,...iA73 32 groups | Derived from key. | |
1198 | + group iB[8][4], // [In] iB00,iB01,...iB73 32 groups | Derived from key. | |
1199 | + unsigned char *sb) // [In] (SB0,SB1,...SB7)...x32 32*8 bytes | Extra input. | |
1200 | +#endif | |
1201 | +#ifdef STREAM_NORMAL | |
1202 | +void stream_cypher_group_normal( | |
1203 | + unsigned char *cb) // [Out] (CB0,CB1,...CB7)...x32 32*8 bytes | Output. | |
1204 | +#endif | |
1205 | +{ | |
1206 | +#ifdef STREAM_INIT | |
1207 | + group in1[4]; | |
1208 | + group in2[4]; | |
1209 | +#endif | |
1210 | + group extra_B[4]; | |
1211 | + group fa,fb,fc,fd,fe; | |
1212 | + group s1a,s1b,s2a,s2b,s3a,s3b,s4a,s4b,s5a,s5b,s6a,s6b,s7a,s7b; | |
1213 | + group next_E[4]; | |
1214 | + group tmp0,tmp1,tmp2,tmp3,tmp4; | |
1215 | +#ifdef STREAM_INIT | |
1216 | + group *sb_g=(group *)sb; | |
1217 | +#endif | |
1218 | +#ifdef STREAM_NORMAL | |
1219 | + group *cb_g=(group *)cb; | |
1220 | +#endif | |
1221 | + int aboff; | |
1222 | + int i,j,k,b; | |
1223 | + | |
1224 | +#ifdef STREAM_INIT | |
1225 | +#endif | |
1226 | +#ifdef STREAM_NORMAL | |
1227 | +#endif | |
1228 | +#ifdef STREAM_INIT | |
1229 | + | |
1230 | +#if GROUP_PARALLELISM==32 | |
1231 | +trasp64_32_88ccw(sb); | |
1232 | +#endif | |
1233 | +#if GROUP_PARALLELISM==64 | |
1234 | +trasp64_64_88ccw(sb); | |
1235 | +#endif | |
1236 | +#if GROUP_PARALLELISM==128 | |
1237 | +trasp64_128_88ccw(sb); | |
1238 | +#endif | |
1239 | + | |
1240 | +#endif | |
1241 | + | |
1242 | + aboff=32; | |
1243 | + | |
1244 | +#ifdef STREAM_INIT | |
1245 | + // load first 32 bits of ck into A[aboff+0]..A[aboff+7] | |
1246 | + // load last 32 bits of ck into B[aboff+0]..B[aboff+7] | |
1247 | + // all other regs = 0 | |
1248 | + for(i=0;i<8;i++){ | |
1249 | + for(b=0;b<4;b++){ | |
1250 | + A[aboff+i][b]=iA[i][b]; | |
1251 | + B[aboff+i][b]=iB[i][b]; | |
1252 | + } | |
1253 | + } | |
1254 | + for(b=0;b<4;b++){ | |
1255 | + A[aboff+8][b]=FF0(); | |
1256 | + A[aboff+9][b]=FF0(); | |
1257 | + B[aboff+8][b]=FF0(); | |
1258 | + B[aboff+9][b]=FF0(); | |
1259 | + } | |
1260 | + for(b=0;b<4;b++){ | |
1261 | + X[b]=FF0(); | |
1262 | + Y[b]=FF0(); | |
1263 | + Z[b]=FF0(); | |
1264 | + D[b]=FF0(); | |
1265 | + E[b]=FF0(); | |
1266 | + F[b]=FF0(); | |
1267 | + } | |
1268 | + p=FF0(); | |
1269 | + q=FF0(); | |
1270 | + r=FF0(); | |
1271 | +#endif | |
1272 | + | |
1273 | + | |
1274 | +//////////////////////////////////////////////////////////////////////////////// | |
1275 | + | |
1276 | + // EXTERNAL LOOP - 8 bytes per operation | |
1277 | + for(i=0;i<8;i++){ | |
1278 | + | |
1279 | + | |
1280 | +#ifdef STREAM_INIT | |
1281 | + for(b=0;b<4;b++){ | |
1282 | + in1[b]=sb_g[8*i+4+b]; | |
1283 | + in2[b]=sb_g[8*i+b]; | |
1284 | + } | |
1285 | +#endif | |
1286 | + | |
1287 | + // INTERNAL LOOP - 2 bits per iteration | |
1288 | + for(j=0; j<4; j++){ | |
1289 | + | |
1290 | + | |
1291 | + // from A0..A9, 35 bits are selected as inputs to 7 s-boxes | |
1292 | + // 5 bits input per s-box, 2 bits output per s-box | |
1293 | + | |
1294 | + // we can select bits with zero masking and shifting operations | |
1295 | + // and synthetize s-boxes with optimized boolean functions. | |
1296 | + // this is the actual reason we do all the crazy transposition | |
1297 | + // stuff to switch between normal and bit slice representations. | |
1298 | + // this code really flies. | |
1299 | + | |
1300 | + fe=A[aboff+3][0];fa=A[aboff+0][2];fb=A[aboff+5][1];fc=A[aboff+6][3];fd=A[aboff+8][0]; | |
1301 | +/* 1000 1110 1110 0001 : lev 7: */ //tmp0=( fa^( fb^( ( ( ( fa|fb )^fc )|( fc^fd ) )^ALL_ONES ) ) ); | |
1302 | +/* 1110 0010 0011 0011 : lev 6: */ //tmp1=( ( fa|fb )^( ( fc&( fa|( fb^fd ) ) )^ALL_ONES ) ); | |
1303 | +/* 0011 0110 1000 1101 : lev 5: */ //tmp2=( fa^( ( fb&fd )^( ( fa&fd )|fc ) ) ); | |
1304 | +/* 0101 0101 1001 0011 : lev 5: */ //tmp3=( ( fa&fc )^( fa^( ( fa&fb )|fd ) ) ); | |
1305 | +/* 1000 1110 1110 0001 : lev 7: */ tmp0=FFXOR(fa,FFXOR(fb,FFXOR(FFOR(FFXOR(FFOR(fa,fb),fc),FFXOR(fc,fd)),FF1()))); | |
1306 | +/* 1110 0010 0011 0011 : lev 6: */ tmp1=FFXOR(FFOR(fa,fb),FFXOR(FFAND(fc,FFOR(fa,FFXOR(fb,fd))),FF1())); | |
1307 | +/* 0011 0110 1000 1101 : lev 5: */ tmp2=FFXOR(fa,FFXOR(FFAND(fb,fd),FFOR(FFAND(fa,fd),fc))); | |
1308 | +/* 0101 0101 1001 0011 : lev 5: */ tmp3=FFXOR(FFAND(fa,fc),FFXOR(fa,FFOR(FFAND(fa,fb),fd))); | |
1309 | + s1a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
1310 | + s1b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
1311 | +//dump_mem("s1as1b-fe",&fe,BYPG,BYPG); | |
1312 | +//dump_mem("s1as1b-fa",&fa,BYPG,BYPG); | |
1313 | +//dump_mem("s1as1b-fb",&fb,BYPG,BYPG); | |
1314 | +//dump_mem("s1as1b-fc",&fc,BYPG,BYPG); | |
1315 | +//dump_mem("s1as1b-fd",&fd,BYPG,BYPG); | |
1316 | + | |
1317 | + fe=A[aboff+1][1];fa=A[aboff+2][2];fb=A[aboff+5][3];fc=A[aboff+6][0];fd=A[aboff+8][1]; | |
1318 | +/* 1001 1110 0110 0001 : lev 6: */ //tmp0=( fa^( ( fb&( fc|fd ) )^( fc^( fd^ALL_ONES ) ) ) ); | |
1319 | +/* 0000 0011 0111 1011 : lev 5: */ //tmp1=( ( fa&( fb^fd ) )|( ( fa|fb )&fc ) ); | |
1320 | +/* 1100 0110 1101 0010 : lev 6: */ //tmp2=( ( fb&fd )^( ( fa&fd )|( fb^( fc^ALL_ONES ) ) ) ); | |
1321 | +/* 0001 1110 1111 0101 : lev 5: */ //tmp3=( ( fa&fd )|( fa^( fb^( fc&fd ) ) ) ); | |
1322 | +/* 1001 1110 0110 0001 : lev 6: */ tmp0=FFXOR(fa,FFXOR(FFAND(fb,FFOR(fc,fd)),FFXOR(fc,FFXOR(fd,FF1())))); | |
1323 | +/* 0000 0011 0111 1011 : lev 5: */ tmp1=FFOR(FFAND(fa,FFXOR(fb,fd)),FFAND(FFOR(fa,fb),fc)); | |
1324 | +/* 1100 0110 1101 0010 : lev 6: */ tmp2=FFXOR(FFAND(fb,fd),FFOR(FFAND(fa,fd),FFXOR(fb,FFXOR(fc,FF1())))); | |
1325 | +/* 0001 1110 1111 0101 : lev 5: */ tmp3=FFOR(FFAND(fa,fd),FFXOR(fa,FFXOR(fb,FFAND(fc,fd)))); | |
1326 | + s2a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
1327 | + s2b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
1328 | + | |
1329 | + fe=A[aboff+0][3];fa=A[aboff+1][0];fb=A[aboff+4][1];fc=A[aboff+4][3];fd=A[aboff+5][2]; | |
1330 | +/* 0100 1011 1001 0110 : lev 5: */ //tmp0=( fa^( fb^( ( fc&( fa|fd ) )^fd ) ) ); | |
1331 | +/* 1101 0101 1000 1100 : lev 7: */ //tmp1=( ( fa&fc )^( ( fa^fd )|( ( fb|fc )^( fd^ALL_ONES ) ) ) ); | |
1332 | +/* 0010 0111 1101 1000 : lev 4: */ //tmp2=( fa^( ( ( fb^fc )&fd )^fc ) ); | |
1333 | +/* 1111 1111 1111 1111 : lev 0: */ //tmp3=ALL_ONES; | |
1334 | +/* 0100 1011 1001 0110 : lev 5: */ tmp0=FFXOR(fa,FFXOR(fb,FFXOR(FFAND(fc,FFOR(fa,fd)),fd))); | |
1335 | +/* 1101 0101 1000 1100 : lev 7: */ tmp1=FFXOR(FFAND(fa,fc),FFOR(FFXOR(fa,fd),FFXOR(FFOR(fb,fc),FFXOR(fd,FF1())))); | |
1336 | +/* 0010 0111 1101 1000 : lev 4: */ tmp2=FFXOR(fa,FFXOR(FFAND(FFXOR(fb,fc),fd),fc)); | |
1337 | +/* 1111 1111 1111 1111 : lev 0: */ tmp3=FF1(); | |
1338 | + s3a=FFXOR(tmp0,FFAND(FFNOT(fe),tmp1)); | |
1339 | + s3b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
1340 | + | |
1341 | + fe=A[aboff+2][3];fa=A[aboff+0][1];fb=A[aboff+1][3];fc=A[aboff+3][2];fd=A[aboff+7][0]; | |
1342 | +/* 1011 0101 0100 1001 : lev 7: */ //tmp0=( fa^( ( fc&( fa^fd ) )|( fb^( fc|( fd^ALL_ONES ) ) ) ) ); | |
1343 | +/* 0010 1101 0110 0110 : lev 6: */ //tmp1=( ( fa&fb )^( fb^( ( ( fa|fc )&fd )^fc ) ) ); | |
1344 | +/* 0110 0111 1101 0000 : lev 7: */ //tmp2=( fa^( ( fb&fc )|( ( ( fa&( fb^fd ) )|fc )^fd ) ) ); | |
1345 | +/* 1111 1111 1111 1111 : lev 0: */ //tmp3=ALL_ONES; | |
1346 | +/* 1011 0101 0100 1001 : lev 7: */ tmp0=FFXOR(fa,FFOR(FFAND(fc,FFXOR(fa,fd)),FFXOR(fb,FFOR(fc,FFXOR(fd,FF1()))))); | |
1347 | +/* 0010 1101 0110 0110 : lev 6: */ tmp1=FFXOR(FFAND(fa,fb),FFXOR(fb,FFXOR(FFAND(FFOR(fa,fc),fd),fc))); | |
1348 | +/* 0110 0111 1101 0000 : lev 7: */ tmp2=FFXOR(fa,FFOR(FFAND(fb,fc),FFXOR(FFOR(FFAND(fa,FFXOR(fb,fd)),fc),fd))); | |
1349 | +/* 1111 1111 1111 1111 : lev 0: */ tmp3=FF1(); | |
1350 | + s4a=FFXOR(tmp0,FFAND(fe,FFXOR(tmp1,tmp0))); | |
1351 | + s4b=FFXOR(FFXOR(s4a,tmp2),FFAND(fe,tmp3)); | |
1352 | + | |
1353 | + fe=A[aboff+4][2];fa=A[aboff+3][3];fb=A[aboff+5][0];fc=A[aboff+7][1];fd=A[aboff+8][2]; | |
1354 | +/* 1000 1111 0011 0010 : lev 7: */ //tmp0=( ( ( fa&( fb|fc ) )^fb )|( ( ( fa^fc )|fd )^ALL_ONES ) ); | |
1355 | +/* 0110 1011 0000 1011 : lev 6: */ //tmp1=( fb^( ( fc^fd )&( fc^( fb|( fa^fd ) ) ) ) ); | |
1356 | +/* 0001 1010 0111 1001 : lev 6: */ //tmp2=( ( fa&fc )^( fb^( ( fb|( fa^fc ) )&fd ) ) ); | |
1357 | +/* 0101 1101 1101 0101 : lev 4: */ //tmp3=( ( ( fa^fb )&( fc^ALL_ONES ) )|fd ); | |
1358 | +/* 1000 1111 0011 0010 : lev 7: */ tmp0=FFOR(FFXOR(FFAND(fa,FFOR(fb,fc)),fb),FFXOR(FFOR(FFXOR(fa,fc),fd),FF1())); | |
1359 | +/* 0110 1011 0000 1011 : lev 6: */ tmp1=FFXOR(fb,FFAND(FFXOR(fc,fd),FFXOR(fc,FFOR(fb,FFXOR(fa,fd))))); | |
1360 | +/* 0001 1010 0111 1001 : lev 6: */ tmp2=FFXOR(FFAND(fa,fc),FFXOR(fb,FFAND(FFOR(fb,FFXOR(fa,fc)),fd))); | |
1361 | +/* 0101 1101 1101 0101 : lev 4: */ tmp3=FFOR(FFAND(FFXOR(fa,fb),FFXOR(fc,FF1())),fd); | |
1362 | + s5a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
1363 | + s5b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
1364 | + | |
1365 | + fe=A[aboff+2][1];fa=A[aboff+3][1];fb=A[aboff+4][0];fc=A[aboff+6][2];fd=A[aboff+8][3]; | |
1366 | +/* 0011 0110 0010 1101 : lev 6: */ //tmp0=( ( ( fa&fc )&fd )^( ( fb&( fa|fd ) )^fc ) ); | |
1367 | +/* 1110 1110 1011 1011 : lev 3: */ //tmp1=( ( ( fa^fc )&fd )^ALL_ONES ); | |
1368 | +/* 0101 1000 0110 0111 : lev 6: */ //tmp2=( ( fa&( fb|fc ) )^( fb^( ( fb&fc )|fd ) ) ); | |
1369 | +/* 0001 0011 0000 0001 : lev 5: */ //tmp3=( fc&( ( fa&( fb^fd ) )^( fb|fd ) ) ); | |
1370 | +/* 0011 0110 0010 1101 : lev 6: */ tmp0=FFXOR(FFAND(FFAND(fa,fc),fd),FFXOR(FFAND(fb,FFOR(fa,fd)),fc)); | |
1371 | +/* 1110 1110 1011 1011 : lev 3: */ tmp1=FFXOR(FFAND(FFXOR(fa,fc),fd),FF1()); | |
1372 | +/* 0101 1000 0110 0111 : lev 6: */ tmp2=FFXOR(FFAND(fa,FFOR(fb,fc)),FFXOR(fb,FFOR(FFAND(fb,fc),fd))); | |
1373 | +/* 0001 0011 0000 0001 : lev 5: */ tmp3=FFAND(fc,FFXOR(FFAND(fa,FFXOR(fb,fd)),FFOR(fb,fd))); | |
1374 | + s6a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
1375 | + s6b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
1376 | + | |
1377 | + fe=A[aboff+1][2];fa=A[aboff+2][0];fb=A[aboff+6][1];fc=A[aboff+7][2];fd=A[aboff+7][3]; | |
1378 | +/* 0111 1000 1001 0110 : lev 5: */ //tmp0=( fb^( ( fc&fd )|( fa^( fc^fd ) ) ) ); | |
1379 | +/* 0100 1001 0101 1011 : lev 6: */ //tmp1=( ( fb|fd )&( ( fa&fc )|( fb^( fc^fd ) ) ) ); | |
1380 | +/* 0100 1001 1011 1001 : lev 5: */ //tmp2=( ( fa|fb )^( ( fc&( fb|fd ) )^fd ) ); | |
1381 | +/* 1111 1111 1101 1101 : lev 3: */ //tmp3=( fd|( ( fa&fc )^ALL_ONES ) ); | |
1382 | +/* 0111 1000 1001 0110 : lev 5: */ tmp0=FFXOR(fb,FFOR(FFAND(fc,fd),FFXOR(fa,FFXOR(fc,fd)))); | |
1383 | +/* 0100 1001 0101 1011 : lev 6: */ tmp1=FFAND(FFOR(fb,fd),FFOR(FFAND(fa,fc),FFXOR(fb,FFXOR(fc,fd)))); | |
1384 | +/* 0100 1001 1011 1001 : lev 5: */ tmp2=FFXOR(FFOR(fa,fb),FFXOR(FFAND(fc,FFOR(fb,fd)),fd)); | |
1385 | +/* 1111 1111 1101 1101 : lev 3: */ tmp3=FFOR(fd,FFXOR(FFAND(fa,fc),FF1())); | |
1386 | + s7a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
1387 | + s7b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
1388 | + | |
1389 | + | |
1390 | +/* | |
1391 | + we have just done this: | |
1392 | + | |
1393 | + int sbox1[0x20] = {2,0,1,1,2,3,3,0, 3,2,2,0,1,1,0,3, 0,3,3,0,2,2,1,1, 2,2,0,3,1,1,3,0}; | |
1394 | + int sbox2[0x20] = {3,1,0,2,2,3,3,0, 1,3,2,1,0,0,1,2, 3,1,0,3,3,2,0,2, 0,0,1,2,2,1,3,1}; | |
1395 | + int sbox3[0x20] = {2,0,1,2,2,3,3,1, 1,1,0,3,3,0,2,0, 1,3,0,1,3,0,2,2, 2,0,1,2,0,3,3,1}; | |
1396 | + int sbox4[0x20] = {3,1,2,3,0,2,1,2, 1,2,0,1,3,0,0,3, 1,0,3,1,2,3,0,3, 0,3,2,0,1,2,2,1}; | |
1397 | + int sbox5[0x20] = {2,0,0,1,3,2,3,2, 0,1,3,3,1,0,2,1, 2,3,2,0,0,3,1,1, 1,0,3,2,3,1,0,2}; | |
1398 | + int sbox6[0x20] = {0,1,2,3,1,2,2,0, 0,1,3,0,2,3,1,3, 2,3,0,2,3,0,1,1, 2,1,1,2,0,3,3,0}; | |
1399 | + int sbox7[0x20] = {0,3,2,2,3,0,0,1, 3,0,1,3,1,2,2,1, 1,0,3,3,0,1,1,2, 2,3,1,0,2,3,0,2}; | |
1400 | + | |
1401 | + s12 = sbox1[ (((A3>>0)&1)<<4) | (((A0>>2)&1)<<3) | (((A5>>1)&1)<<2) | (((A6>>3)&1)<<1) | (((A8>>0)&1)<<0) ] | |
1402 | + |sbox2[ (((A1>>1)&1)<<4) | (((A2>>2)&1)<<3) | (((A5>>3)&1)<<2) | (((A6>>0)&1)<<1) | (((A8>>1)&1)<<0) ]; | |
1403 | + s34 = sbox3[ (((A0>>3)&1)<<4) | (((A1>>0)&1)<<3) | (((A4>>1)&1)<<2) | (((A4>>3)&1)<<1) | (((A5>>2)&1)<<0) ] | |
1404 | + |sbox4[ (((A2>>3)&1)<<4) | (((A0>>1)&1)<<3) | (((A1>>3)&1)<<2) | (((A3>>2)&1)<<1) | (((A7>>0)&1)<<0) ]; | |
1405 | + s56 = sbox5[ (((A4>>2)&1)<<4) | (((A3>>3)&1)<<3) | (((A5>>0)&1)<<2) | (((A7>>1)&1)<<1) | (((A8>>2)&1)<<0) ] | |
1406 | + |sbox6[ (((A2>>1)&1)<<4) | (((A3>>1)&1)<<3) | (((A4>>0)&1)<<2) | (((A6>>2)&1)<<1) | (((A8>>3)&1)<<0) ]; | |
1407 | + s7 = sbox7[ (((A1>>2)&1)<<4) | (((A2>>0)&1)<<3) | (((A6>>1)&1)<<2) | (((A7>>2)&1)<<1) | (((A7>>3)&1)<<0) ]; | |
1408 | +*/ | |
1409 | + | |
1410 | + // use 4x4 xor to produce extra nibble for T3 | |
1411 | + | |
1412 | + extra_B[3]=FFXOR(FFXOR(FFXOR(B[aboff+2][0],B[aboff+5][1]),B[aboff+6][2]),B[aboff+8][3]); | |
1413 | + extra_B[2]=FFXOR(FFXOR(FFXOR(B[aboff+5][0],B[aboff+7][1]),B[aboff+2][3]),B[aboff+3][2]); | |
1414 | + extra_B[1]=FFXOR(FFXOR(FFXOR(B[aboff+4][3],B[aboff+7][2]),B[aboff+3][0]),B[aboff+4][1]); | |
1415 | + extra_B[0]=FFXOR(FFXOR(FFXOR(B[aboff+8][2],B[aboff+5][3]),B[aboff+2][1]),B[aboff+7][0]); | |
1416 | + | |
1417 | + // T1 = xor all inputs | |
1418 | + // in1, in2, D are only used in T1 during initialisation, not generation | |
1419 | + for(b=0;b<4;b++){ | |
1420 | + A[aboff-1][b]=FFXOR(A[aboff+9][b],X[b]); | |
1421 | + } | |
1422 | + | |
1423 | +#ifdef STREAM_INIT | |
1424 | + for(b=0;b<4;b++){ | |
1425 | + A[aboff-1][b]=FFXOR(FFXOR(A[aboff-1][b],D[b]),((j % 2) ? in2[b] : in1[b])); | |
1426 | + } | |
1427 | +#endif | |
1428 | + | |
1429 | + | |
1430 | + // T2 = xor all inputs | |
1431 | + // in1, in2 are only used in T1 during initialisation, not generation | |
1432 | + // if p=0, use this, if p=1, rotate the result left | |
1433 | + for(b=0;b<4;b++){ | |
1434 | + B[aboff-1][b]=FFXOR(FFXOR(B[aboff+6][b],B[aboff+9][b]),Y[b]); | |
1435 | + } | |
1436 | + | |
1437 | +#ifdef STREAM_INIT | |
1438 | + for(b=0;b<4;b++){ | |
1439 | + B[aboff-1][b]=FFXOR(B[aboff-1][b],((j % 2) ? in1[b] : in2[b])); | |
1440 | + } | |
1441 | +#endif | |
1442 | + | |
1443 | + | |
1444 | + // if p=1, rotate left (yes, this is what we're doing) | |
1445 | + tmp3=B[aboff-1][3]; | |
1446 | + B[aboff-1][3]=FFXOR(B[aboff-1][3],FFAND(FFXOR(B[aboff-1][3],B[aboff-1][2]),p)); | |
1447 | + B[aboff-1][2]=FFXOR(B[aboff-1][2],FFAND(FFXOR(B[aboff-1][2],B[aboff-1][1]),p)); | |
1448 | + B[aboff-1][1]=FFXOR(B[aboff-1][1],FFAND(FFXOR(B[aboff-1][1],B[aboff-1][0]),p)); | |
1449 | + B[aboff-1][0]=FFXOR(B[aboff-1][0],FFAND(FFXOR(B[aboff-1][0],tmp3),p)); | |
1450 | + | |
1451 | + | |
1452 | + // T3 = xor all inputs | |
1453 | + for(b=0;b<4;b++){ | |
1454 | + D[b]=FFXOR(FFXOR(E[b],Z[b]),extra_B[b]); | |
1455 | + } | |
1456 | + | |
1457 | + | |
1458 | + // T4 = sum, carry of Z + E + r | |
1459 | + for(b=0;b<4;b++){ | |
1460 | + next_E[b]=F[b]; | |
1461 | + } | |
1462 | + | |
1463 | + tmp0=FFXOR(Z[0],E[0]); | |
1464 | + tmp1=FFAND(Z[0],E[0]); | |
1465 | + F[0]=FFXOR(E[0],FFAND(q,FFXOR(Z[0],r))); | |
1466 | + tmp3=FFAND(tmp0,r); | |
1467 | + tmp4=FFOR(tmp1,tmp3); | |
1468 | + | |
1469 | + tmp0=FFXOR(Z[1],E[1]); | |
1470 | + tmp1=FFAND(Z[1],E[1]); | |
1471 | + F[1]=FFXOR(E[1],FFAND(q,FFXOR(Z[1],tmp4))); | |
1472 | + tmp3=FFAND(tmp0,tmp4); | |
1473 | + tmp4=FFOR(tmp1,tmp3); | |
1474 | + | |
1475 | + tmp0=FFXOR(Z[2],E[2]); | |
1476 | + tmp1=FFAND(Z[2],E[2]); | |
1477 | + F[2]=FFXOR(E[2],FFAND(q,FFXOR(Z[2],tmp4))); | |
1478 | + tmp3=FFAND(tmp0,tmp4); | |
1479 | + tmp4=FFOR(tmp1,tmp3); | |
1480 | + | |
1481 | + tmp0=FFXOR(Z[3],E[3]); | |
1482 | + tmp1=FFAND(Z[3],E[3]); | |
1483 | + F[3]=FFXOR(E[3],FFAND(q,FFXOR(Z[3],tmp4))); | |
1484 | + tmp3=FFAND(tmp0,tmp4); | |
1485 | + r=FFXOR(r,FFAND(q,FFXOR(FFOR(tmp1,tmp3),r))); // ultimate carry | |
1486 | + | |
1487 | +/* | |
1488 | + we have just done this: (believe it or not) | |
1489 | + | |
1490 | + if (q) { | |
1491 | + F = Z + E + r; | |
1492 | + r = (F >> 4) & 1; | |
1493 | + F = F & 0x0f; | |
1494 | + } | |
1495 | + else { | |
1496 | + F = E; | |
1497 | + } | |
1498 | +*/ | |
1499 | + for(b=0;b<4;b++){ | |
1500 | + E[b]=next_E[b]; | |
1501 | + } | |
1502 | + | |
1503 | + // this simple instruction is virtually shifting all the shift registers | |
1504 | + aboff--; | |
1505 | + | |
1506 | +/* | |
1507 | + we've just done this: | |
1508 | + | |
1509 | + A9=A8;A8=A7;A7=A6;A6=A5;A5=A4;A4=A3;A3=A2;A2=A1;A1=A0;A0=next_A0; | |
1510 | + B9=B8;B8=B7;B7=B6;B6=B5;B5=B4;B4=B3;B3=B2;B2=B1;B1=B0;B0=next_B0; | |
1511 | +*/ | |
1512 | + | |
1513 | + X[0]=s1a; | |
1514 | + X[1]=s2a; | |
1515 | + X[2]=s3b; | |
1516 | + X[3]=s4b; | |
1517 | + Y[0]=s3a; | |
1518 | + Y[1]=s4a; | |
1519 | + Y[2]=s5b; | |
1520 | + Y[3]=s6b; | |
1521 | + Z[0]=s5a; | |
1522 | + Z[1]=s6a; | |
1523 | + Z[2]=s1b; | |
1524 | + Z[3]=s2b; | |
1525 | + p=s7a; | |
1526 | + q=s7b; | |
1527 | + | |
1528 | +#ifdef STREAM_NORMAL | |
1529 | + // require 4 loops per output byte | |
1530 | + // 2 output bits are a function of the 4 bits of D | |
1531 | + // xor 2 by 2 | |
1532 | + cb_g[8*i+7-2*j]=FFXOR(D[2],D[3]); | |
1533 | + cb_g[8*i+6-2*j]=FFXOR(D[0],D[1]); | |
1534 | +#endif | |
1535 | + | |
1536 | + | |
1537 | + } // INTERNAL LOOP | |
1538 | + | |
1539 | + | |
1540 | + } // EXTERNAL LOOP | |
1541 | + | |
1542 | + // move 32 steps forward, ready for next call | |
1543 | + for(k=0;k<10;k++){ | |
1544 | + for(b=0;b<4;b++){ | |
1545 | + A[32+k][b]=A[k][b]; | |
1546 | + B[32+k][b]=B[k][b]; | |
1547 | + } | |
1548 | + } | |
1549 | + | |
1550 | + | |
1551 | +//////////////////////////////////////////////////////////////////////////////// | |
1552 | + | |
1553 | +#ifdef STREAM_NORMAL | |
1554 | + | |
1555 | +#if GROUP_PARALLELISM==32 | |
1556 | +trasp64_32_88cw(cb); | |
1557 | +#endif | |
1558 | +#if GROUP_PARALLELISM==64 | |
1559 | +trasp64_64_88cw(cb); | |
1560 | +#endif | |
1561 | +#if GROUP_PARALLELISM==128 | |
1562 | +trasp64_128_88cw(cb); | |
1563 | +#endif | |
1564 | + | |
1565 | +#endif | |
1566 | + | |
1567 | +#ifdef STREAM_INIT | |
1568 | +#endif | |
1569 | +#ifdef STREAM_NORMAL | |
1570 | +#endif | |
1571 | + | |
1572 | +} | |
1573 | + | |
1574 | Index: libs/libmythtv/FFdecsa/parallel_064_8charA.h | |
1575 | =================================================================== | |
1576 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
1577 | +++ libs/libmythtv/FFdecsa/parallel_064_8charA.h 2006-06-20 17:36:06.000000000 -0400 | |
1578 | @@ -0,0 +1,171 @@ | |
1579 | +/* FFdecsa -- fast decsa algorithm | |
1580 | + * | |
1581 | + * Copyright (C) 2003-2004 fatih89r | |
1582 | + * | |
1583 | + * This program is free software; you can redistribute it and/or modify | |
1584 | + * it under the terms of the GNU General Public License as published by | |
1585 | + * the Free Software Foundation; either version 2 of the License, or | |
1586 | + * (at your option) any later version. | |
1587 | + * | |
1588 | + * This program is distributed in the hope that it will be useful, | |
1589 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
1590 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
1591 | + * GNU General Public License for more details. | |
1592 | + * | |
1593 | + * You should have received a copy of the GNU General Public License | |
1594 | + * along with this program; if not, write to the Free Software | |
1595 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
1596 | + */ | |
1597 | + | |
1598 | + | |
1599 | +struct group_t{ | |
1600 | + unsigned char s1[8]; | |
1601 | +}; | |
1602 | +typedef struct group_t group; | |
1603 | + | |
1604 | +#define GROUP_PARALLELISM 64 | |
1605 | + | |
1606 | +group static inline FF0(){ | |
1607 | + group res; | |
1608 | + int i; | |
1609 | + for(i=0;i<8;i++) res.s1[i]=0x0; | |
1610 | + return res; | |
1611 | +} | |
1612 | + | |
1613 | +group static inline FF1(){ | |
1614 | + group res; | |
1615 | + int i; | |
1616 | + for(i=0;i<8;i++) res.s1[i]=0xff; | |
1617 | + return res; | |
1618 | +} | |
1619 | + | |
1620 | +group static inline FFAND(group a,group b){ | |
1621 | + group res; | |
1622 | + int i; | |
1623 | + for(i=0;i<8;i++) res.s1[i]=a.s1[i]&b.s1[i]; | |
1624 | + return res; | |
1625 | +} | |
1626 | + | |
1627 | +group static inline FFOR(group a,group b){ | |
1628 | + group res; | |
1629 | + int i; | |
1630 | + for(i=0;i<8;i++) res.s1[i]=a.s1[i]|b.s1[i]; | |
1631 | + return res; | |
1632 | +} | |
1633 | + | |
1634 | +group static inline FFXOR(group a,group b){ | |
1635 | + group res; | |
1636 | + int i; | |
1637 | + for(i=0;i<8;i++) res.s1[i]=a.s1[i]^b.s1[i]; | |
1638 | + return res; | |
1639 | +} | |
1640 | + | |
1641 | +group static inline FFNOT(group a){ | |
1642 | + group res; | |
1643 | + int i; | |
1644 | + for(i=0;i<8;i++) res.s1[i]=~a.s1[i]; | |
1645 | + return res; | |
1646 | +} | |
1647 | + | |
1648 | + | |
1649 | +/* 64 rows of 64 bits */ | |
1650 | + | |
1651 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
1652 | + *(((int *)tab)+2*g)=*((int *)data); | |
1653 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
1654 | +} | |
1655 | + | |
1656 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
1657 | + *((int *)data)=*(((int *)tab)+2*g); | |
1658 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
1659 | +} | |
1660 | + | |
1661 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
1662 | + int j; | |
1663 | + for(j=0;j<n;j++){ | |
1664 | + *(data+j)^=*(tab+8*g+j); | |
1665 | + } | |
1666 | +} | |
1667 | + | |
1668 | +struct batch_t{ | |
1669 | + unsigned char s1[8]; | |
1670 | +}; | |
1671 | +typedef struct batch_t batch; | |
1672 | + | |
1673 | +#define BYTES_PER_BATCH 8 | |
1674 | + | |
1675 | +batch static inline B_FFAND(batch a,batch b){ | |
1676 | + batch res; | |
1677 | + int i; | |
1678 | + for(i=0;i<8;i++) res.s1[i]=a.s1[i]&b.s1[i]; | |
1679 | + return res; | |
1680 | +} | |
1681 | + | |
1682 | +batch static inline B_FFOR(batch a,batch b){ | |
1683 | + batch res; | |
1684 | + int i; | |
1685 | + for(i=0;i<8;i++) res.s1[i]=a.s1[i]|b.s1[i]; | |
1686 | + return res; | |
1687 | +} | |
1688 | + | |
1689 | +batch static inline B_FFXOR(batch a,batch b){ | |
1690 | + batch res; | |
1691 | + int i; | |
1692 | + for(i=0;i<8;i++) res.s1[i]=a.s1[i]^b.s1[i]; | |
1693 | + return res; | |
1694 | +} | |
1695 | + | |
1696 | + | |
1697 | +batch static inline B_FFN_ALL_29(){ | |
1698 | + batch res; | |
1699 | + int i; | |
1700 | + for(i=0;i<8;i++) res.s1[i]=0x29; | |
1701 | + return res; | |
1702 | +} | |
1703 | +batch static inline B_FFN_ALL_02(){ | |
1704 | + batch res; | |
1705 | + int i; | |
1706 | + for(i=0;i<8;i++) res.s1[i]=0x02; | |
1707 | + return res; | |
1708 | +} | |
1709 | +batch static inline B_FFN_ALL_04(){ | |
1710 | + batch res; | |
1711 | + int i; | |
1712 | + for(i=0;i<8;i++) res.s1[i]=0x04; | |
1713 | + return res; | |
1714 | +} | |
1715 | +batch static inline B_FFN_ALL_10(){ | |
1716 | + batch res; | |
1717 | + int i; | |
1718 | + for(i=0;i<8;i++) res.s1[i]=0x10; | |
1719 | + return res; | |
1720 | +} | |
1721 | +batch static inline B_FFN_ALL_40(){ | |
1722 | + batch res; | |
1723 | + int i; | |
1724 | + for(i=0;i<8;i++) res.s1[i]=0x40; | |
1725 | + return res; | |
1726 | +} | |
1727 | +batch static inline B_FFN_ALL_80(){ | |
1728 | + batch res; | |
1729 | + int i; | |
1730 | + for(i=0;i<8;i++) res.s1[i]=0x80; | |
1731 | + return res; | |
1732 | +} | |
1733 | + | |
1734 | +batch static inline B_FFSH8L(batch a,int n){ | |
1735 | + batch res; | |
1736 | + int i; | |
1737 | + for(i=0;i<8;i++) res.s1[i]=a.s1[i]<<n; | |
1738 | + return res; | |
1739 | +} | |
1740 | + | |
1741 | +batch static inline B_FFSH8R(batch a,int n){ | |
1742 | + batch res; | |
1743 | + int i; | |
1744 | + for(i=0;i<8;i++) res.s1[i]=a.s1[i]>>n; | |
1745 | + return res; | |
1746 | +} | |
1747 | + | |
1748 | +void static inline M_EMPTY(void){ | |
1749 | +} | |
1750 | Index: libs/libmythtv/FFdecsa/parallel_128_sse.h | |
1751 | =================================================================== | |
1752 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
1753 | +++ libs/libmythtv/FFdecsa/parallel_128_sse.h 2006-06-20 17:36:06.000000000 -0400 | |
1754 | @@ -0,0 +1,184 @@ | |
1755 | +/* FFdecsa -- fast decsa algorithm | |
1756 | + * | |
1757 | + * Copyright (C) 2003-2004 fatih89r | |
1758 | + * | |
1759 | + * This program is free software; you can redistribute it and/or modify | |
1760 | + * it under the terms of the GNU General Public License as published by | |
1761 | + * the Free Software Foundation; either version 2 of the License, or | |
1762 | + * (at your option) any later version. | |
1763 | + * | |
1764 | + * This program is distributed in the hope that it will be useful, | |
1765 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
1766 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
1767 | + * GNU General Public License for more details. | |
1768 | + * | |
1769 | + * You should have received a copy of the GNU General Public License | |
1770 | + * along with this program; if not, write to the Free Software | |
1771 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
1772 | + */ | |
1773 | + | |
1774 | + | |
1775 | +#include <xmmintrin.h> | |
1776 | + | |
1777 | +struct group_t{ | |
1778 | + __m128 s1; | |
1779 | +}; | |
1780 | +typedef struct group_t group; | |
1781 | + | |
1782 | +#define GROUP_PARALLELISM 128 | |
1783 | + | |
1784 | +group static inline FF0(){ | |
1785 | + group res; | |
1786 | + static unsigned long long l[2]={0x0000000000000000ULL,0x0000000000000000ULL}; | |
1787 | + res.s1=*(__m128*)l; | |
1788 | + return res; | |
1789 | +} | |
1790 | + | |
1791 | +group static inline FF1(){ | |
1792 | + group res; | |
1793 | + static unsigned long long l[2]={0xffffffffffffffffULL,0xffffffffffffffffULL}; | |
1794 | + res.s1=*(__m128*)l; | |
1795 | + return res; | |
1796 | +} | |
1797 | + | |
1798 | +group static inline FFAND(group a,group b){ | |
1799 | + group res; | |
1800 | + res.s1=_mm_and_ps(a.s1,b.s1); | |
1801 | + return res; | |
1802 | +} | |
1803 | + | |
1804 | +group static inline FFOR(group a,group b){ | |
1805 | + group res; | |
1806 | + res.s1=_mm_or_ps(a.s1,b.s1); | |
1807 | + return res; | |
1808 | +} | |
1809 | + | |
1810 | +group static inline FFXOR(group a,group b){ | |
1811 | + group res; | |
1812 | + res.s1=_mm_xor_ps(a.s1,b.s1); | |
1813 | + return res; | |
1814 | +} | |
1815 | + | |
1816 | +group static inline FFNOT(group a){ | |
1817 | + group res; | |
1818 | + res.s1=_mm_xor_ps(a.s1,FF1().s1); | |
1819 | + return res; | |
1820 | +} | |
1821 | + | |
1822 | + | |
1823 | +/* 64 rows of 128 bits */ | |
1824 | + | |
1825 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
1826 | + *(((int *)tab)+2*g)=*((int *)data); | |
1827 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
1828 | +} | |
1829 | + | |
1830 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
1831 | + *((int *)data)=*(((int *)tab)+2*g); | |
1832 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
1833 | +} | |
1834 | + | |
1835 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
1836 | + int j; | |
1837 | + for(j=0;j<n;j++){ | |
1838 | + *(data+j)^=*(tab+8*g+j); | |
1839 | + } | |
1840 | +} | |
1841 | + | |
1842 | + | |
1843 | +struct batch_t{ | |
1844 | + __m64 s1; | |
1845 | +}; | |
1846 | +typedef struct batch_t batch; | |
1847 | + | |
1848 | +#define BYTES_PER_BATCH 8 | |
1849 | + | |
1850 | +batch static inline B_FFAND(batch a,batch b){ | |
1851 | + batch res; | |
1852 | + res.s1=_m_pand(a.s1,b.s1); | |
1853 | + return res; | |
1854 | +} | |
1855 | + | |
1856 | +batch static inline B_FFOR(batch a,batch b){ | |
1857 | + batch res; | |
1858 | + res.s1=_m_por(a.s1,b.s1); | |
1859 | + return res; | |
1860 | +} | |
1861 | + | |
1862 | +batch static inline B_FFXOR(batch a,batch b){ | |
1863 | + batch res; | |
1864 | + res.s1=_m_pxor(a.s1,b.s1); | |
1865 | + return res; | |
1866 | +} | |
1867 | + | |
1868 | +batch static inline B_FFN_ALL_29(){ | |
1869 | + batch res; | |
1870 | + res.s1=(__m64)0x2929292929292929ULL; | |
1871 | + return res; | |
1872 | +} | |
1873 | +batch static inline B_FFN_ALL_02(){ | |
1874 | + batch res; | |
1875 | + res.s1=(__m64)0x0202020202020202ULL; | |
1876 | + return res; | |
1877 | +} | |
1878 | +batch static inline B_FFN_ALL_04(){ | |
1879 | + batch res; | |
1880 | + res.s1=(__m64)0x0404040404040404ULL; | |
1881 | + return res; | |
1882 | +} | |
1883 | +batch static inline B_FFN_ALL_10(){ | |
1884 | + batch res; | |
1885 | + res.s1=(__m64)0x1010101010101010ULL; | |
1886 | + return res; | |
1887 | +} | |
1888 | +batch static inline B_FFN_ALL_40(){ | |
1889 | + batch res; | |
1890 | + res.s1=(__m64)0x4040404040404040ULL; | |
1891 | + return res; | |
1892 | +} | |
1893 | +batch static inline B_FFN_ALL_80(){ | |
1894 | + batch res; | |
1895 | + res.s1=(__m64)0x8080808080808080ULL; | |
1896 | + return res; | |
1897 | +} | |
1898 | + | |
1899 | +batch static inline B_FFSH8L(batch a,int n){ | |
1900 | + batch res; | |
1901 | + res.s1=_m_psllqi(a.s1,n); | |
1902 | + return res; | |
1903 | +} | |
1904 | + | |
1905 | +batch static inline B_FFSH8R(batch a,int n){ | |
1906 | + batch res; | |
1907 | + res.s1=_m_psrlqi(a.s1,n); | |
1908 | + return res; | |
1909 | +} | |
1910 | + | |
1911 | +void static inline M_EMPTY(void){ | |
1912 | + _m_empty(); | |
1913 | +} | |
1914 | + | |
1915 | + | |
1916 | +#undef XOR_8_BY | |
1917 | +#define XOR_8_BY(d,s1,s2) do{ __m64 *pd=(__m64 *)(d), *ps1=(__m64 *)(s1), *ps2=(__m64 *)(s2); \ | |
1918 | + *pd = _m_pxor( *ps1 , *ps2 ); }while(0) | |
1919 | + | |
1920 | +#undef XOREQ_8_BY | |
1921 | +#define XOREQ_8_BY(d,s) do{ __m64 *pd=(__m64 *)(d), *ps=(__m64 *)(s); \ | |
1922 | + *pd = _m_pxor( *ps, *pd ); }while(0) | |
1923 | + | |
1924 | +#undef COPY_8_BY | |
1925 | +#define COPY_8_BY(d,s) do{ __m64 *pd=(__m64 *)(d), *ps=(__m64 *)(s); \ | |
1926 | + *pd = *ps; }while(0) | |
1927 | + | |
1928 | +#undef BEST_SPAN | |
1929 | +#define BEST_SPAN 8 | |
1930 | + | |
1931 | +#undef XOR_BEST_BY | |
1932 | +#define XOR_BEST_BY(d,s1,s2) do{ XOR_8_BY(d,s1,s2); }while(0); | |
1933 | + | |
1934 | +#undef XOREQ_BEST_BY | |
1935 | +#define XOREQ_BEST_BY(d,s) do{ XOREQ_8_BY(d,s); }while(0); | |
1936 | + | |
1937 | +#undef COPY_BEST_BY | |
1938 | +#define COPY_BEST_BY(d,s) do{ COPY_8_BY(d,s); }while(0); | |
1939 | Index: libs/libmythtv/FFdecsa/docs/how_to_compile.txt | |
1940 | =================================================================== | |
1941 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
1942 | +++ libs/libmythtv/FFdecsa/docs/how_to_compile.txt 2006-06-20 17:36:06.000000000 -0400 | |
1943 | @@ -0,0 +1,114 @@ | |
1944 | +------- | |
1945 | +FFdecsa | |
1946 | +------- | |
1947 | + | |
1948 | +Compiling is as easy as running a make command, if you have gcc and are | |
1949 | +using a little endian machine. 64 bit machines have not been tested but | |
1950 | +may work with little or no changes; big endian machines will certainly | |
1951 | +give incorrect results (read the technical_background.txt to know where | |
1952 | +the problem is). | |
1953 | + | |
1954 | +Before compiling you could edit the Makefile to tweak compiler flags for | |
1955 | +optimal performance. If you want to play with different bit-grouping | |
1956 | +strategies you have to edit FFdecsa_DBG.c and change the "our choice" | |
1957 | +definition. This is highly critical for performance. | |
1958 | + | |
1959 | +After compilation run the FFdecsa_test application. It will test correct | |
1960 | +decryption and print the meausered speed (use "nice --19 ./FFdecsa_test" | |
1961 | +on an idle machine for better results). Or just use "make test". | |
1962 | + | |
1963 | +gcc >=3.3.3 is highly recommended. Older versions could give performance | |
1964 | +problems. | |
1965 | + | |
1966 | +icc is currently unusable. In the initial phases of development of | |
1967 | +FFdecsa icc was able to compile the code and gave interesting speed | |
1968 | +results when using the 8charA grouping mode (array of 8 characters are | |
1969 | +automatically manipulated through MMX instructions). At some point the | |
1970 | +code began to work incorrectly because of a compiler bug (but I found a | |
1971 | +workaround). Then, the performance dropped with no reason; I found a | |
1972 | +workaround by adding an unused variable (alignment problem, grep for icc | |
1973 | +in the code to see where it happens). Then, with the introduction of | |
1974 | +group modes based on intrinsics, gcc was finally able to go beyond the | |
1975 | +speed record originally set by icc. Additional code tweaks added more | |
1976 | +speed to gcc, while icc started to segfault on compilation (both version | |
1977 | +7 and 8). In conclusion, icc is bugged and this code is too hard for it. | |
1978 | +gcc on the other hand is great. I tried to inspect generated assembler | |
1979 | +to find weak spots, and the generated code is very good indeed. | |
1980 | + | |
1981 | +Note: the code can be compiled with gcc or g++. g++ is 3% faster for | |
1982 | +some reason. | |
1983 | + | |
1984 | +You should not get any errors or warnings. I only get two "inlining | |
1985 | +failed" warnings on two functions I asked to be inlined but gcc doesn't | |
1986 | +want to inline. | |
1987 | + | |
1988 | +The build process creates additional temp files by running grep | |
1989 | +commands. This is how debugging output is handled. All the lines | |
1990 | +containing DBG are removed and the temp file is compiled (so the line | |
1991 | +numbers change between temp and original files). Don't edit the temp | |
1992 | +files, they will be overwritten. If you don't remove the DBG lines (for | |
1993 | +example, by changing "grep -v DBG" into "grep -v aaDBG" in Makefile) a | |
1994 | +lot of output will be generated. This is useful to understand what's | |
1995 | +wrong when the FFdecsa_test is failing. I included a reference "known | |
1996 | +good" output in the debug_output directory. Extra debug output is | |
1997 | +commented out in the code. | |
1998 | + | |
1999 | +The debug output functionality could be... bugged. This is because I | |
2000 | +tested everything using hard coded int grouping mode and then | |
2001 | +generalized the debug output to abstract grouping modes. A bug where 4 | |
2002 | +bytes are printed instead of 8 could be present somewhere. I think it | |
2003 | +isn't, but you've been warned. | |
2004 | + | |
2005 | +This code was only tried on Linux. | |
2006 | +It should work on Windows or other platforms, but you may encounter | |
2007 | +problems related to the compiler quality. If you want to try, begin with | |
2008 | +the int grouping mode. It is only 30% slower then the best (MMX) and it | |
2009 | +should be easily portable because no intrinsics are used. I'm | |
2010 | +particularly interested in hearing what kind of performance can be | |
2011 | +obtained on x86_64 processors in int, long long int, mmx, 2mmx, sse | |
2012 | +modes. | |
2013 | + | |
2014 | + | |
2015 | +As a reference, here are the results I get on an Athlon XP 2400+ (this | |
2016 | +processor runs at 2000MHz); other processors belonging to the Athlon XP | |
2017 | +architecture, including Durons, should have the same speed per MHz. | |
2018 | +Cache size and bus speed don't matter. | |
2019 | + | |
2020 | +CPU: AMD Athlon XP 2400+ | |
2021 | + | |
2022 | +Compiler: g++ (gcc version 3.3.3 20040412 (Red Hat Linux 3.3.3-7)) | |
2023 | + | |
2024 | +Flags: -O3 -march=athlon-xp -fexpensive-optimizations -funroll-loops | |
2025 | + --param max-unrolled-insns=500 | |
2026 | + | |
2027 | +grouping mode speed (Mbit/s) notes | |
2028 | +--------------------------------------------------------------------- | |
2029 | +PARALLEL_32_4CHAR 14 | |
2030 | +PARALLEL_32_4CHARA 12 | |
2031 | +PARALLEL_32_INT 125 very good and very portable | |
2032 | +PARALLEL_64_8CHAR 17 | |
2033 | +PARALLEL_64_8CHARA 15 needs a vectorizing compiler | |
2034 | +PARALLEL_64_2INT 75 x86 has too few registers | |
2035 | +PARALLEL_64_LONG 97 try this on x86_64 | |
2036 | +PARALLEL_64_MMX 165 the best | |
2037 | +PARALLEL_128_16CHAR 6 | |
2038 | +PARALLEL_128_16CHARA 7 | |
2039 | +PARALLEL_128_4INT 69 | |
2040 | +PARALLEL_128_2LONG 52 | |
2041 | +PARALLEL_128_2MMX 36 slower than expected | |
2042 | +PARALLEL_128_SSE 156 just slower than 64_MMX | |
2043 | + | |
2044 | +Best speeds are obtained with native data types: int, mmx, sse (this | |
2045 | +could be a compiler artifact). | |
2046 | + | |
2047 | +64 bit processors should try 64_LONG. | |
2048 | + | |
2049 | +Vectorizing compilers should like *CHARA. | |
2050 | + | |
2051 | +64_MMX is faster than 128_SSE on the Athlon; perhaps SSE instruction are | |
2052 | +internally split into 64 bit chunks. Could be different on x86_64 or | |
2053 | +Intel processors. | |
2054 | + | |
2055 | +128_SSE has a 64 bit (MMX) batch type because SSE has no shifting | |
2056 | +instructions, they are only available on SSE2. As the Athlon XP doesn't | |
2057 | +support SSE2, I couldn't experiment with that. | |
2058 | Index: libs/libmythtv/FFdecsa/docs/FAQ.txt | |
2059 | =================================================================== | |
2060 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
2061 | +++ libs/libmythtv/FFdecsa/docs/FAQ.txt 2006-06-20 17:36:06.000000000 -0400 | |
2062 | @@ -0,0 +1,77 @@ | |
2063 | +------- | |
2064 | +FFdecsa | |
2065 | +------- | |
2066 | + | |
2067 | +FFdecsa is a fast implementation of the CSA decryption algorithm for MPEG | |
2068 | +TS packets. | |
2069 | + | |
2070 | +Q: What does FF stands for? | |
2071 | +A: FFdecsa means "Fucking Fast decsa". | |
2072 | + | |
2073 | +Q: Why would you use such a rude name? | |
2074 | +A: Because this code is fucking fast, more than 800% the speed of the best | |
2075 | + implementation I'm able to find around at the moment. | |
2076 | + | |
2077 | +Q: How it that possible? Are all other programmers stupid? | |
2078 | +A: No, they just tried to save a cycle or two tweaking a fundamentally wrong | |
2079 | + implementation. The algorithm has to be implemented in a totally different | |
2080 | + way to achieve good speed. | |
2081 | + | |
2082 | +Q: Do you use multimedia instructions? | |
2083 | +A: I use every trick I could come up with, including multimedia instructions. | |
2084 | + They are not fundamental in achieving speed, a version without them runs | |
2085 | + at 6x the speed of the best implementation around (which uses MMX). | |
2086 | + | |
2087 | +Q: So how did you do that? | |
2088 | +A: By using a different approach for the implementation. This code is not | |
2089 | + exploiting some new CSA vulnerability, it is just doing the same | |
2090 | + calculations better. Think about replacing bubble sort with quick sort. | |
2091 | + | |
2092 | +Q: You're joking, it's impossible to gain so much speed. | |
2093 | +A: Speed test are available, technical documentation is available, source | |
2094 | + code is available. Try it yourself. | |
2095 | + If you want details, these are some of the documented tricks I used | |
2096 | + (more details in the docs directory): | |
2097 | + TRICK NUMBER 0: emulate the hardware | |
2098 | + TRICK NUMBER 1: virtual shift registers | |
2099 | + TRICK NUMBER 2: parallel bitslice | |
2100 | + TRICK NUMBER 3: multimedia instructions | |
2101 | + TRICK NUMBER 4: parallel byteslice | |
2102 | + TRICK NUMBER 5: efficient bit permutation | |
2103 | + TRICK NUMBER 6: efficient normal<->slice conversion | |
2104 | + TRICK NUMBER 7: try hard to process packets together | |
2105 | + TRICK NUMBER 8: try to avoid doing the same thing many times | |
2106 | + TRICK NUMBER 9: compiler | |
2107 | + TRICK NUMBER a: a lot of brain work | |
2108 | + | |
2109 | +Q: How can be this code useful? | |
2110 | +A: You can use this code in place of the old slow implementations and save a | |
2111 | + lot of CPU power. | |
2112 | + | |
2113 | +Q: Just that? | |
2114 | +A: Well, new applications are possible. | |
2115 | + Decrypting a whole transponder is easily doable now. Well, a $50 CPU can | |
2116 | + decrypt four transponder at the same time if you have four DVB boards (but | |
2117 | + I couldn't test that). | |
2118 | + | |
2119 | +Q: You're cheating, this code is fake, I don't believe one word. | |
2120 | +A: Go away. This is technical stuff for people with brains. | |
2121 | + | |
2122 | +Q: This code is great, may I distribute your code in original or modified | |
2123 | + form? | |
2124 | +A: Only if you respect the license. | |
2125 | + | |
2126 | +Q: May I use your code in my player/library/plugin...? | |
2127 | +A: Again, you have to respect the license. | |
2128 | + | |
2129 | +Q: Are you an extraterrestrial programmer? | |
2130 | +A: No, just a Turkish guy with a PC to play with :-) | |
2131 | + | |
2132 | +Q: Why did you spend your time doing this? | |
2133 | +A: Because I thought that my approach was doable and I was sure it would | |
2134 | + have been much faster, so I had to implement it to confirm I was right. | |
2135 | + I got 8x the speed and that's enough to be proud of it. And I could not | |
2136 | + just keep the code for myself only. | |
2137 | + | |
2138 | +Q: What is the answer to the meaning of the universe? | |
2139 | +A: 42,43,71,5f,65,85,f6,76,0d,13,28,96,... | |
2140 | Index: libs/libmythtv/FFdecsa/parallel_032_4charA.h | |
2141 | =================================================================== | |
2142 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
2143 | +++ libs/libmythtv/FFdecsa/parallel_032_4charA.h 2006-06-20 17:36:06.000000000 -0400 | |
2144 | @@ -0,0 +1,171 @@ | |
2145 | +/* FFdecsa -- fast decsa algorithm | |
2146 | + * | |
2147 | + * Copyright (C) 2003-2004 fatih89r | |
2148 | + * | |
2149 | + * This program is free software; you can redistribute it and/or modify | |
2150 | + * it under the terms of the GNU General Public License as published by | |
2151 | + * the Free Software Foundation; either version 2 of the License, or | |
2152 | + * (at your option) any later version. | |
2153 | + * | |
2154 | + * This program is distributed in the hope that it will be useful, | |
2155 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
2156 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
2157 | + * GNU General Public License for more details. | |
2158 | + * | |
2159 | + * You should have received a copy of the GNU General Public License | |
2160 | + * along with this program; if not, write to the Free Software | |
2161 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
2162 | + */ | |
2163 | + | |
2164 | + | |
2165 | +struct group_t{ | |
2166 | + unsigned char s1[4]; | |
2167 | +}; | |
2168 | +typedef struct group_t group; | |
2169 | + | |
2170 | +#define GROUP_PARALLELISM 32 | |
2171 | + | |
2172 | +group static inline FF0(){ | |
2173 | + group res; | |
2174 | + int i; | |
2175 | + for(i=0;i<4;i++) res.s1[i]=0x0; | |
2176 | + return res; | |
2177 | +} | |
2178 | + | |
2179 | +group static inline FF1(){ | |
2180 | + group res; | |
2181 | + int i; | |
2182 | + for(i=0;i<4;i++) res.s1[i]=0xff; | |
2183 | + return res; | |
2184 | +} | |
2185 | + | |
2186 | +group static inline FFAND(group a,group b){ | |
2187 | + group res; | |
2188 | + int i; | |
2189 | + for(i=0;i<4;i++) res.s1[i]=a.s1[i]&b.s1[i]; | |
2190 | + return res; | |
2191 | +} | |
2192 | + | |
2193 | +group static inline FFOR(group a,group b){ | |
2194 | + group res; | |
2195 | + int i; | |
2196 | + for(i=0;i<4;i++) res.s1[i]=a.s1[i]|b.s1[i]; | |
2197 | + return res; | |
2198 | +} | |
2199 | + | |
2200 | +group static inline FFXOR(group a,group b){ | |
2201 | + group res; | |
2202 | + int i; | |
2203 | + for(i=0;i<4;i++) res.s1[i]=a.s1[i]^b.s1[i]; | |
2204 | + return res; | |
2205 | +} | |
2206 | + | |
2207 | +group static inline FFNOT(group a){ | |
2208 | + group res; | |
2209 | + int i; | |
2210 | + for(i=0;i<4;i++) res.s1[i]=~a.s1[i]; | |
2211 | + return res; | |
2212 | +} | |
2213 | + | |
2214 | + | |
2215 | +/* 64 rows of 32 bits */ | |
2216 | + | |
2217 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
2218 | + *(((int *)tab)+g)=*((int *)data); | |
2219 | + *(((int *)tab)+32+g)=*(((int *)data)+1); | |
2220 | +} | |
2221 | + | |
2222 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
2223 | + *((int *)data)=*(((int *)tab)+g); | |
2224 | + *(((int *)data)+1)=*(((int *)tab)+32+g); | |
2225 | +} | |
2226 | + | |
2227 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
2228 | + int j; | |
2229 | + for(j=0;j<n;j++){ | |
2230 | + *(data+j)^=*(tab+4*(g+(j>=4?32-1:0))+j); | |
2231 | + } | |
2232 | +} | |
2233 | + | |
2234 | +struct batch_t{ | |
2235 | + unsigned char s1[4]; | |
2236 | +}; | |
2237 | +typedef struct batch_t batch; | |
2238 | + | |
2239 | +#define BYTES_PER_BATCH 4 | |
2240 | + | |
2241 | +batch static inline B_FFAND(batch a,batch b){ | |
2242 | + batch res; | |
2243 | + int i; | |
2244 | + for(i=0;i<4;i++) res.s1[i]=a.s1[i]&b.s1[i]; | |
2245 | + return res; | |
2246 | +} | |
2247 | + | |
2248 | +batch static inline B_FFOR(batch a,batch b){ | |
2249 | + batch res; | |
2250 | + int i; | |
2251 | + for(i=0;i<4;i++) res.s1[i]=a.s1[i]|b.s1[i]; | |
2252 | + return res; | |
2253 | +} | |
2254 | + | |
2255 | +batch static inline B_FFXOR(batch a,batch b){ | |
2256 | + batch res; | |
2257 | + int i; | |
2258 | + for(i=0;i<4;i++) res.s1[i]=a.s1[i]^b.s1[i]; | |
2259 | + return res; | |
2260 | +} | |
2261 | + | |
2262 | + | |
2263 | +batch static inline B_FFN_ALL_29(){ | |
2264 | + batch res; | |
2265 | + int i; | |
2266 | + for(i=0;i<4;i++) res.s1[i]=0x29; | |
2267 | + return res; | |
2268 | +} | |
2269 | +batch static inline B_FFN_ALL_02(){ | |
2270 | + batch res; | |
2271 | + int i; | |
2272 | + for(i=0;i<4;i++) res.s1[i]=0x02; | |
2273 | + return res; | |
2274 | +} | |
2275 | +batch static inline B_FFN_ALL_04(){ | |
2276 | + batch res; | |
2277 | + int i; | |
2278 | + for(i=0;i<4;i++) res.s1[i]=0x04; | |
2279 | + return res; | |
2280 | +} | |
2281 | +batch static inline B_FFN_ALL_10(){ | |
2282 | + batch res; | |
2283 | + int i; | |
2284 | + for(i=0;i<4;i++) res.s1[i]=0x10; | |
2285 | + return res; | |
2286 | +} | |
2287 | +batch static inline B_FFN_ALL_40(){ | |
2288 | + batch res; | |
2289 | + int i; | |
2290 | + for(i=0;i<4;i++) res.s1[i]=0x40; | |
2291 | + return res; | |
2292 | +} | |
2293 | +batch static inline B_FFN_ALL_80(){ | |
2294 | + batch res; | |
2295 | + int i; | |
2296 | + for(i=0;i<4;i++) res.s1[i]=0x80; | |
2297 | + return res; | |
2298 | +} | |
2299 | + | |
2300 | +batch static inline B_FFSH8L(batch a,int n){ | |
2301 | + batch res; | |
2302 | + int i; | |
2303 | + for(i=0;i<4;i++) res.s1[i]=a.s1[i]<<n; | |
2304 | + return res; | |
2305 | +} | |
2306 | + | |
2307 | +batch static inline B_FFSH8R(batch a,int n){ | |
2308 | + batch res; | |
2309 | + int i; | |
2310 | + for(i=0;i<4;i++) res.s1[i]=a.s1[i]>>n; | |
2311 | + return res; | |
2312 | +} | |
2313 | + | |
2314 | +void static inline M_EMPTY(void){ | |
2315 | +} | |
2316 | Index: libs/libmythtv/FFdecsa/parallel_128_16char.h | |
2317 | =================================================================== | |
2318 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
2319 | +++ libs/libmythtv/FFdecsa/parallel_128_16char.h 2006-06-20 17:36:06.000000000 -0400 | |
2320 | @@ -0,0 +1,411 @@ | |
2321 | +/* FFdecsa -- fast decsa algorithm | |
2322 | + * | |
2323 | + * Copyright (C) 2003-2004 fatih89r | |
2324 | + * | |
2325 | + * This program is free software; you can redistribute it and/or modify | |
2326 | + * it under the terms of the GNU General Public License as published by | |
2327 | + * the Free Software Foundation; either version 2 of the License, or | |
2328 | + * (at your option) any later version. | |
2329 | + * | |
2330 | + * This program is distributed in the hope that it will be useful, | |
2331 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
2332 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
2333 | + * GNU General Public License for more details. | |
2334 | + * | |
2335 | + * You should have received a copy of the GNU General Public License | |
2336 | + * along with this program; if not, write to the Free Software | |
2337 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
2338 | + */ | |
2339 | + | |
2340 | + | |
2341 | +struct group_t{ | |
2342 | + unsigned char s1,s2,s3,s4,s5,s6,s7,s8,s9,s10,s11,s12,s13,s14,s15,s16; | |
2343 | +}; | |
2344 | +typedef struct group_t group; | |
2345 | + | |
2346 | +#define GROUP_PARALLELISM 128 | |
2347 | + | |
2348 | +group static inline FF0(){ | |
2349 | + group res; | |
2350 | + res.s1=0x0; | |
2351 | + res.s2=0x0; | |
2352 | + res.s3=0x0; | |
2353 | + res.s4=0x0; | |
2354 | + res.s5=0x0; | |
2355 | + res.s6=0x0; | |
2356 | + res.s7=0x0; | |
2357 | + res.s8=0x0; | |
2358 | + res.s9=0x0; | |
2359 | + res.s10=0x0; | |
2360 | + res.s11=0x0; | |
2361 | + res.s12=0x0; | |
2362 | + res.s13=0x0; | |
2363 | + res.s14=0x0; | |
2364 | + res.s15=0x0; | |
2365 | + res.s16=0x0; | |
2366 | + return res; | |
2367 | +} | |
2368 | + | |
2369 | +group static inline FF1(){ | |
2370 | + group res; | |
2371 | + res.s1=0xff; | |
2372 | + res.s2=0xff; | |
2373 | + res.s3=0xff; | |
2374 | + res.s4=0xff; | |
2375 | + res.s5=0xff; | |
2376 | + res.s6=0xff; | |
2377 | + res.s7=0xff; | |
2378 | + res.s8=0xff; | |
2379 | + res.s9=0xff; | |
2380 | + res.s10=0xff; | |
2381 | + res.s11=0xff; | |
2382 | + res.s12=0xff; | |
2383 | + res.s13=0xff; | |
2384 | + res.s14=0xff; | |
2385 | + res.s15=0xff; | |
2386 | + res.s16=0xff; | |
2387 | + return res; | |
2388 | +} | |
2389 | + | |
2390 | +group static inline FFAND(group a,group b){ | |
2391 | + group res; | |
2392 | + res.s1=a.s1&b.s1; | |
2393 | + res.s2=a.s2&b.s2; | |
2394 | + res.s3=a.s3&b.s3; | |
2395 | + res.s4=a.s4&b.s4; | |
2396 | + res.s5=a.s5&b.s5; | |
2397 | + res.s6=a.s6&b.s6; | |
2398 | + res.s7=a.s7&b.s7; | |
2399 | + res.s8=a.s8&b.s8; | |
2400 | + res.s9=a.s9&b.s9; | |
2401 | + res.s10=a.s10&b.s10; | |
2402 | + res.s11=a.s11&b.s11; | |
2403 | + res.s12=a.s12&b.s12; | |
2404 | + res.s13=a.s13&b.s13; | |
2405 | + res.s14=a.s14&b.s14; | |
2406 | + res.s15=a.s15&b.s15; | |
2407 | + res.s16=a.s16&b.s16; | |
2408 | + return res; | |
2409 | +} | |
2410 | + | |
2411 | +group static inline FFOR(group a,group b){ | |
2412 | + group res; | |
2413 | + res.s1=a.s1|b.s1; | |
2414 | + res.s2=a.s2|b.s2; | |
2415 | + res.s3=a.s3|b.s3; | |
2416 | + res.s4=a.s4|b.s4; | |
2417 | + res.s5=a.s5|b.s5; | |
2418 | + res.s6=a.s6|b.s6; | |
2419 | + res.s7=a.s7|b.s7; | |
2420 | + res.s8=a.s8|b.s8; | |
2421 | + res.s9=a.s9|b.s9; | |
2422 | + res.s10=a.s10|b.s10; | |
2423 | + res.s11=a.s11|b.s11; | |
2424 | + res.s12=a.s12|b.s12; | |
2425 | + res.s13=a.s13|b.s13; | |
2426 | + res.s14=a.s14|b.s14; | |
2427 | + res.s15=a.s15|b.s15; | |
2428 | + res.s16=a.s16|b.s16; | |
2429 | + return res; | |
2430 | +} | |
2431 | + | |
2432 | +group static inline FFXOR(group a,group b){ | |
2433 | + group res; | |
2434 | + res.s1=a.s1^b.s1; | |
2435 | + res.s2=a.s2^b.s2; | |
2436 | + res.s3=a.s3^b.s3; | |
2437 | + res.s4=a.s4^b.s4; | |
2438 | + res.s5=a.s5^b.s5; | |
2439 | + res.s6=a.s6^b.s6; | |
2440 | + res.s7=a.s7^b.s7; | |
2441 | + res.s8=a.s8^b.s8; | |
2442 | + res.s9=a.s9^b.s9; | |
2443 | + res.s10=a.s10^b.s10; | |
2444 | + res.s11=a.s11^b.s11; | |
2445 | + res.s12=a.s12^b.s12; | |
2446 | + res.s13=a.s13^b.s13; | |
2447 | + res.s14=a.s14^b.s14; | |
2448 | + res.s15=a.s15^b.s15; | |
2449 | + res.s16=a.s16^b.s16; | |
2450 | + return res; | |
2451 | +} | |
2452 | + | |
2453 | +group static inline FFNOT(group a){ | |
2454 | + group res; | |
2455 | + res.s1=~a.s1; | |
2456 | + res.s2=~a.s2; | |
2457 | + res.s3=~a.s3; | |
2458 | + res.s4=~a.s4; | |
2459 | + res.s5=~a.s5; | |
2460 | + res.s6=~a.s6; | |
2461 | + res.s7=~a.s7; | |
2462 | + res.s8=~a.s8; | |
2463 | + res.s9=~a.s9; | |
2464 | + res.s10=~a.s10; | |
2465 | + res.s11=~a.s11; | |
2466 | + res.s12=~a.s12; | |
2467 | + res.s13=~a.s13; | |
2468 | + res.s14=~a.s14; | |
2469 | + res.s15=~a.s15; | |
2470 | + res.s16=~a.s16; | |
2471 | + return res; | |
2472 | +} | |
2473 | + | |
2474 | + | |
2475 | +/* 64 rows of 128 bits */ | |
2476 | + | |
2477 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
2478 | + *(((int *)tab)+2*g)=*((int *)data); | |
2479 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
2480 | +} | |
2481 | + | |
2482 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
2483 | + *((int *)data)=*(((int *)tab)+2*g); | |
2484 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
2485 | +} | |
2486 | + | |
2487 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
2488 | + int j; | |
2489 | + for(j=0;j<n;j++){ | |
2490 | + *(data+j)^=*(tab+8*g+j); | |
2491 | + } | |
2492 | +} | |
2493 | + | |
2494 | + | |
2495 | +struct batch_t{ | |
2496 | + unsigned char s1,s2,s3,s4,s5,s6,s7,s8,s9,s10,s11,s12,s13,s14,s15,s16; | |
2497 | +}; | |
2498 | +typedef struct batch_t batch; | |
2499 | + | |
2500 | +#define BYTES_PER_BATCH 16 | |
2501 | + | |
2502 | +batch static inline B_FFAND(batch a,batch b){ | |
2503 | + batch res; | |
2504 | + res.s1=a.s1&b.s1; | |
2505 | + res.s2=a.s2&b.s2; | |
2506 | + res.s3=a.s3&b.s3; | |
2507 | + res.s4=a.s4&b.s4; | |
2508 | + res.s5=a.s5&b.s5; | |
2509 | + res.s6=a.s6&b.s6; | |
2510 | + res.s7=a.s7&b.s7; | |
2511 | + res.s8=a.s8&b.s8; | |
2512 | + res.s9=a.s9&b.s9; | |
2513 | + res.s10=a.s10&b.s10; | |
2514 | + res.s11=a.s11&b.s11; | |
2515 | + res.s12=a.s12&b.s12; | |
2516 | + res.s13=a.s13&b.s13; | |
2517 | + res.s14=a.s14&b.s14; | |
2518 | + res.s15=a.s15&b.s15; | |
2519 | + res.s16=a.s16&b.s16; | |
2520 | + return res; | |
2521 | +} | |
2522 | + | |
2523 | +batch static inline B_FFOR(batch a,batch b){ | |
2524 | + batch res; | |
2525 | + res.s1=a.s1|b.s1; | |
2526 | + res.s2=a.s2|b.s2; | |
2527 | + res.s3=a.s3|b.s3; | |
2528 | + res.s4=a.s4|b.s4; | |
2529 | + res.s5=a.s5|b.s5; | |
2530 | + res.s6=a.s6|b.s6; | |
2531 | + res.s7=a.s7|b.s7; | |
2532 | + res.s8=a.s8|b.s8; | |
2533 | + res.s9=a.s9|b.s9; | |
2534 | + res.s10=a.s10|b.s10; | |
2535 | + res.s11=a.s11|b.s11; | |
2536 | + res.s12=a.s12|b.s12; | |
2537 | + res.s13=a.s13|b.s13; | |
2538 | + res.s14=a.s14|b.s14; | |
2539 | + res.s15=a.s15|b.s15; | |
2540 | + res.s16=a.s16|b.s16; | |
2541 | + return res; | |
2542 | +} | |
2543 | + | |
2544 | +batch static inline B_FFXOR(batch a,batch b){ | |
2545 | + batch res; | |
2546 | + res.s1=a.s1^b.s1; | |
2547 | + res.s2=a.s2^b.s2; | |
2548 | + res.s3=a.s3^b.s3; | |
2549 | + res.s4=a.s4^b.s4; | |
2550 | + res.s5=a.s5^b.s5; | |
2551 | + res.s6=a.s6^b.s6; | |
2552 | + res.s7=a.s7^b.s7; | |
2553 | + res.s8=a.s8^b.s8; | |
2554 | + res.s9=a.s9^b.s9; | |
2555 | + res.s10=a.s10^b.s10; | |
2556 | + res.s11=a.s11^b.s11; | |
2557 | + res.s12=a.s12^b.s12; | |
2558 | + res.s13=a.s13^b.s13; | |
2559 | + res.s14=a.s14^b.s14; | |
2560 | + res.s15=a.s15^b.s15; | |
2561 | + res.s16=a.s16^b.s16; | |
2562 | + return res; | |
2563 | +} | |
2564 | + | |
2565 | + | |
2566 | +batch static inline B_FFN_ALL_29(){ | |
2567 | + batch res; | |
2568 | + res.s1=0x29; | |
2569 | + res.s2=0x29; | |
2570 | + res.s3=0x29; | |
2571 | + res.s4=0x29; | |
2572 | + res.s5=0x29; | |
2573 | + res.s6=0x29; | |
2574 | + res.s7=0x29; | |
2575 | + res.s8=0x29; | |
2576 | + res.s9=0x29; | |
2577 | + res.s10=0x29; | |
2578 | + res.s11=0x29; | |
2579 | + res.s12=0x29; | |
2580 | + res.s13=0x29; | |
2581 | + res.s14=0x29; | |
2582 | + res.s15=0x29; | |
2583 | + res.s16=0x29; | |
2584 | + return res; | |
2585 | +} | |
2586 | +batch static inline B_FFN_ALL_02(){ | |
2587 | + batch res; | |
2588 | + res.s1=0x02; | |
2589 | + res.s2=0x02; | |
2590 | + res.s3=0x02; | |
2591 | + res.s4=0x02; | |
2592 | + res.s5=0x02; | |
2593 | + res.s6=0x02; | |
2594 | + res.s7=0x02; | |
2595 | + res.s8=0x02; | |
2596 | + res.s9=0x02; | |
2597 | + res.s10=0x02; | |
2598 | + res.s11=0x02; | |
2599 | + res.s12=0x02; | |
2600 | + res.s13=0x02; | |
2601 | + res.s14=0x02; | |
2602 | + res.s15=0x02; | |
2603 | + res.s16=0x02; | |
2604 | + return res; | |
2605 | +} | |
2606 | +batch static inline B_FFN_ALL_04(){ | |
2607 | + batch res; | |
2608 | + res.s1=0x04; | |
2609 | + res.s2=0x04; | |
2610 | + res.s3=0x04; | |
2611 | + res.s4=0x04; | |
2612 | + res.s5=0x04; | |
2613 | + res.s6=0x04; | |
2614 | + res.s7=0x04; | |
2615 | + res.s8=0x04; | |
2616 | + res.s9=0x04; | |
2617 | + res.s10=0x04; | |
2618 | + res.s11=0x04; | |
2619 | + res.s12=0x04; | |
2620 | + res.s13=0x04; | |
2621 | + res.s14=0x04; | |
2622 | + res.s15=0x04; | |
2623 | + res.s16=0x04; | |
2624 | + return res; | |
2625 | +} | |
2626 | +batch static inline B_FFN_ALL_10(){ | |
2627 | + batch res; | |
2628 | + res.s1=0x10; | |
2629 | + res.s2=0x10; | |
2630 | + res.s3=0x10; | |
2631 | + res.s4=0x10; | |
2632 | + res.s5=0x10; | |
2633 | + res.s6=0x10; | |
2634 | + res.s7=0x10; | |
2635 | + res.s8=0x10; | |
2636 | + res.s9=0x10; | |
2637 | + res.s10=0x10; | |
2638 | + res.s11=0x10; | |
2639 | + res.s12=0x10; | |
2640 | + res.s13=0x10; | |
2641 | + res.s14=0x10; | |
2642 | + res.s15=0x10; | |
2643 | + res.s16=0x10; | |
2644 | + return res; | |
2645 | +} | |
2646 | +batch static inline B_FFN_ALL_40(){ | |
2647 | + batch res; | |
2648 | + res.s1=0x40; | |
2649 | + res.s2=0x40; | |
2650 | + res.s3=0x40; | |
2651 | + res.s4=0x40; | |
2652 | + res.s5=0x40; | |
2653 | + res.s6=0x40; | |
2654 | + res.s7=0x40; | |
2655 | + res.s8=0x40; | |
2656 | + res.s9=0x40; | |
2657 | + res.s10=0x40; | |
2658 | + res.s11=0x40; | |
2659 | + res.s12=0x40; | |
2660 | + res.s13=0x40; | |
2661 | + res.s14=0x40; | |
2662 | + res.s15=0x40; | |
2663 | + res.s16=0x40; | |
2664 | + return res; | |
2665 | +} | |
2666 | +batch static inline B_FFN_ALL_80(){ | |
2667 | + batch res; | |
2668 | + res.s1=0x80; | |
2669 | + res.s2=0x80; | |
2670 | + res.s3=0x80; | |
2671 | + res.s4=0x80; | |
2672 | + res.s5=0x80; | |
2673 | + res.s6=0x80; | |
2674 | + res.s7=0x80; | |
2675 | + res.s8=0x80; | |
2676 | + res.s9=0x80; | |
2677 | + res.s10=0x80; | |
2678 | + res.s11=0x80; | |
2679 | + res.s12=0x80; | |
2680 | + res.s13=0x80; | |
2681 | + res.s14=0x80; | |
2682 | + res.s15=0x80; | |
2683 | + res.s16=0x80; | |
2684 | + return res; | |
2685 | +} | |
2686 | + | |
2687 | +batch static inline B_FFSH8L(batch a,int n){ | |
2688 | + batch res; | |
2689 | + res.s1=a.s1<<n; | |
2690 | + res.s2=a.s2<<n; | |
2691 | + res.s3=a.s3<<n; | |
2692 | + res.s4=a.s4<<n; | |
2693 | + res.s5=a.s5<<n; | |
2694 | + res.s6=a.s6<<n; | |
2695 | + res.s7=a.s7<<n; | |
2696 | + res.s8=a.s8<<n; | |
2697 | + res.s9=a.s9<<n; | |
2698 | + res.s10=a.s10<<n; | |
2699 | + res.s11=a.s11<<n; | |
2700 | + res.s12=a.s12<<n; | |
2701 | + res.s13=a.s13<<n; | |
2702 | + res.s14=a.s14<<n; | |
2703 | + res.s15=a.s15<<n; | |
2704 | + res.s16=a.s16<<n; | |
2705 | + return res; | |
2706 | +} | |
2707 | + | |
2708 | +batch static inline B_FFSH8R(batch a,int n){ | |
2709 | + batch res; | |
2710 | + res.s1=a.s1>>n; | |
2711 | + res.s2=a.s2>>n; | |
2712 | + res.s3=a.s3>>n; | |
2713 | + res.s4=a.s4>>n; | |
2714 | + res.s5=a.s5>>n; | |
2715 | + res.s6=a.s6>>n; | |
2716 | + res.s7=a.s7>>n; | |
2717 | + res.s8=a.s8>>n; | |
2718 | + res.s9=a.s9>>n; | |
2719 | + res.s10=a.s10>>n; | |
2720 | + res.s11=a.s11>>n; | |
2721 | + res.s12=a.s12>>n; | |
2722 | + res.s13=a.s13>>n; | |
2723 | + res.s14=a.s14>>n; | |
2724 | + res.s15=a.s15>>n; | |
2725 | + res.s16=a.s16>>n; | |
2726 | + return res; | |
2727 | +} | |
2728 | + | |
2729 | + | |
2730 | +void static inline M_EMPTY(void){ | |
2731 | +} | |
2732 | Index: libs/libmythtv/FFdecsa/FFdecsa_test.c | |
2733 | =================================================================== | |
2734 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
2735 | +++ libs/libmythtv/FFdecsa/FFdecsa_test.c 2006-06-20 17:36:06.000000000 -0400 | |
2736 | @@ -0,0 +1,174 @@ | |
2737 | +/* FFdecsa -- fast decsa algorithm | |
2738 | + * | |
2739 | + * Copyright (C) 2003-2004 fatih89r | |
2740 | + * | |
2741 | + * This program is free software; you can redistribute it and/or modify | |
2742 | + * it under the terms of the GNU General Public License as published by | |
2743 | + * the Free Software Foundation; either version 2 of the License, or | |
2744 | + * (at your option) any later version. | |
2745 | + * | |
2746 | + * This program is distributed in the hope that it will be useful, | |
2747 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
2748 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
2749 | + * GNU General Public License for more details. | |
2750 | + * | |
2751 | + * You should have received a copy of the GNU General Public License | |
2752 | + * along with this program; if not, write to the Free Software | |
2753 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
2754 | + */ | |
2755 | + | |
2756 | + | |
2757 | +#include <string.h> | |
2758 | +#include <stdio.h> | |
2759 | +#include <sys/time.h> | |
2760 | + | |
2761 | +#include "FFdecsa.h" | |
2762 | + | |
2763 | +#ifndef NULL | |
2764 | +#define NULL 0 | |
2765 | +#endif | |
2766 | + | |
2767 | +#include "FFdecsa_test_testcases.h" | |
2768 | + | |
2769 | +int compare(unsigned char *p1, unsigned char *p2, int n, int silently){ | |
2770 | + int i; | |
2771 | + int ok=1; | |
2772 | + for(i=0;i<n;i++){ | |
2773 | + if(i==3) continue; // tolerate this | |
2774 | + if(p1[i]!=p2[i]){ | |
2775 | + fprintf(stderr,"at pos 0x%02x, got 0x%02x instead of 0x%02x\n",i,p1[i],p2[i]); | |
2776 | + ok=0; | |
2777 | + } | |
2778 | + } | |
2779 | + if(!silently){ | |
2780 | + if(ok){ | |
2781 | + fprintf(stderr,"CORRECT!\n"); | |
2782 | + } | |
2783 | + else{ | |
2784 | + fprintf(stderr,"FAILED!\n"); | |
2785 | + } | |
2786 | + } | |
2787 | + return ok; | |
2788 | +} | |
2789 | + | |
2790 | + | |
2791 | +//MAIN | |
2792 | + | |
2793 | +#define TS_PKTS_FOR_TEST 30*1000 | |
2794 | +//#define TS_PKTS_FOR_TEST 1000*1000 | |
2795 | +unsigned char megabuf[188*TS_PKTS_FOR_TEST]; | |
2796 | +unsigned char onebuf[188]; | |
2797 | + | |
2798 | +unsigned char *cluster[10]; | |
2799 | + | |
2800 | +int main(void){ | |
2801 | + int i; | |
2802 | + struct timeval tvs,tve; | |
2803 | + void *keys=get_key_struct(); | |
2804 | + | |
2805 | + fprintf(stderr,"FFdecsa 1.0: testing correctness and speed\n"); | |
2806 | + | |
2807 | +/* begin correctness testing */ | |
2808 | + | |
2809 | + set_control_words(keys,test_invalid_key,test_1_key); | |
2810 | + memcpy(onebuf,test_1_encrypted,188); | |
2811 | + cluster[0]=onebuf;cluster[1]=onebuf+188;cluster[2]=NULL; | |
2812 | + decrypt_packets(keys,cluster); | |
2813 | + compare(onebuf,test_1_expected,188,0); | |
2814 | + | |
2815 | + set_control_words(keys,test_2_key,test_invalid_key); | |
2816 | + memcpy(onebuf,test_2_encrypted,188); | |
2817 | + cluster[0]=onebuf;cluster[1]=onebuf+188;cluster[2]=NULL; | |
2818 | + decrypt_packets(keys,cluster); | |
2819 | + compare(onebuf,test_2_expected,188,0); | |
2820 | + | |
2821 | + set_control_words(keys,test_3_key,test_invalid_key); | |
2822 | + memcpy(onebuf,test_3_encrypted,188); | |
2823 | + cluster[0]=onebuf;cluster[1]=onebuf+188;cluster[2]=NULL; | |
2824 | + decrypt_packets(keys,cluster); | |
2825 | + compare(onebuf,test_3_expected,188,0); | |
2826 | + | |
2827 | + set_control_words(keys,test_p_10_0_key,test_invalid_key); | |
2828 | + memcpy(onebuf,test_p_10_0_encrypted,188); | |
2829 | + cluster[0]=onebuf;cluster[1]=onebuf+188;cluster[2]=NULL; | |
2830 | + decrypt_packets(keys,cluster); | |
2831 | + compare(onebuf,test_p_10_0_expected,188,0); | |
2832 | + | |
2833 | + set_control_words(keys,test_p_1_6_key,test_invalid_key); | |
2834 | + memcpy(onebuf,test_p_1_6_encrypted,188); | |
2835 | + cluster[0]=onebuf;cluster[1]=onebuf+188;cluster[2]=NULL; | |
2836 | + decrypt_packets(keys,cluster); | |
2837 | + compare(onebuf,test_p_1_6_expected,188,0); | |
2838 | + | |
2839 | +/* begin speed testing */ | |
2840 | + | |
2841 | +#if 0 | |
2842 | +// test on short packets | |
2843 | +#define s_encrypted test_p_1_6_encrypted | |
2844 | +#define s_key_e test_p_1_6_key | |
2845 | +#define s_key_o test_invalid_key | |
2846 | +#define s_expected test_p_1_6_expected | |
2847 | + | |
2848 | +#else | |
2849 | +//test on full packets | |
2850 | +#define s_encrypted test_2_encrypted | |
2851 | +#define s_key_e test_2_key | |
2852 | +#define s_key_o test_invalid_key | |
2853 | +#define s_expected test_2_expected | |
2854 | + | |
2855 | +#endif | |
2856 | + | |
2857 | + for(i=0;i<TS_PKTS_FOR_TEST;i++){ | |
2858 | + memcpy(&megabuf[188*i],s_encrypted,188); | |
2859 | + } | |
2860 | +// test that packets are not shuffled around | |
2861 | +// so, let's put an undecryptable packet somewhere in the middle (we will use a wrong key) | |
2862 | +#define noONE_POISONED_PACKET | |
2863 | +#ifdef ONE_POISONED_PACKET | |
2864 | + memcpy(&megabuf[188*(TS_PKTS_FOR_TEST*2/3)],test_3_encrypted,188); | |
2865 | +#endif | |
2866 | + | |
2867 | + // start decryption | |
2868 | + set_control_words(keys,s_key_e,s_key_o); | |
2869 | + gettimeofday(&tvs,NULL); | |
2870 | +#if 0 | |
2871 | +// force one by one | |
2872 | + for(i=0;i<TS_PKTS_FOR_TEST;i++){ | |
2873 | + cluster[0]=megabuf+188*i;cluster[1]=onebuf+188*i+188;cluster[2]=NULL; | |
2874 | + decrypt_packets(keys,cluster); | |
2875 | + } | |
2876 | +#else | |
2877 | + { | |
2878 | + int done=0; | |
2879 | + while(done<TS_PKTS_FOR_TEST){ | |
2880 | + //fprintf(stderr,"done=%i\n",done); | |
2881 | + cluster[0]=megabuf+188*done;cluster[1]=megabuf+188*TS_PKTS_FOR_TEST;cluster[2]=NULL; | |
2882 | + done+=decrypt_packets(keys,cluster); | |
2883 | + } | |
2884 | + } | |
2885 | +#endif | |
2886 | + gettimeofday(&tve,NULL); | |
2887 | + //end decryption | |
2888 | + | |
2889 | + fprintf(stderr,"speed=%f Mbit/s\n",(184*TS_PKTS_FOR_TEST*8)/((tve.tv_sec-tvs.tv_sec)+1e-6*(tve.tv_usec-tvs.tv_usec))/1000000); | |
2890 | + fprintf(stderr,"speed=%f pkts/s\n",TS_PKTS_FOR_TEST/((tve.tv_sec-tvs.tv_sec)+1e-6*(tve.tv_usec-tvs.tv_usec))); | |
2891 | + | |
2892 | + // this packet couldn't be decrypted correctly | |
2893 | +#ifdef ONE_POISONED_PACKET | |
2894 | + compare(megabuf+188*(TS_PKTS_FOR_TEST*2/3),test_3_expected,188,0); /* will fail because we used a wrong key */ | |
2895 | +#endif | |
2896 | + // these should be ok | |
2897 | + compare(megabuf,s_expected,188,0); | |
2898 | + compare(megabuf+188*511,s_expected,188,0); | |
2899 | + compare(megabuf+188*512,s_expected,188,0); | |
2900 | + compare(megabuf+188*319,s_expected,188,0); | |
2901 | + compare(megabuf+188*(TS_PKTS_FOR_TEST-1),s_expected,188,0); | |
2902 | + | |
2903 | + for(i=0;i<TS_PKTS_FOR_TEST;i++){ | |
2904 | + if(!compare(megabuf+188*i,s_expected,188,1)){ | |
2905 | + fprintf(stderr,"FAILED COMPARISON OF PACKET %10i\n",i); | |
2906 | + }; | |
2907 | + } | |
2908 | + | |
2909 | + return 0; | |
2910 | +} | |
2911 | Index: libs/libmythtv/dvbcam.cpp | |
2912 | =================================================================== | |
2913 | --- libs/libmythtv/dvbcam.cpp.orig 2006-06-20 17:36:05.000000000 -0400 | |
2914 | +++ libs/libmythtv/dvbcam.cpp 2006-06-20 17:36:06.000000000 -0400 | |
2915 | @@ -374,6 +374,12 @@ | |
2916 | QString keyFifo = QDir::homeDirPath() + | |
2917 | QString("/.mythtv/externcam_%1/keyfifo").arg(cardnum); | |
2918 | unlink(keyFifo.ascii()); | |
2919 | + bool spawn_thread = false; | |
2920 | + if(numslots==0) { | |
2921 | + cmd += " --writekey " + keyFifo; | |
2922 | + mkfifo(keyFifo.ascii(), 0600); | |
2923 | + spawn_thread = true; | |
2924 | + } | |
2925 | ||
2926 | int tmppid[2]; | |
2927 | if (pipe(tmppid) < 0) | |
2928 | @@ -399,6 +405,9 @@ | |
2929 | } | |
2930 | close(tmppid[0]); | |
2931 | external_cam_fd=tmppid[1]; | |
2932 | + if(spawn_thread) | |
2933 | + pthread_create(&externSoftkeyThread, NULL, | |
2934 | + ExternalCamSoftkeyHelper, this); | |
2935 | ||
2936 | //we need to fake ciThreadRunning in oreder to get PMTs | |
2937 | ciThreadRunning = true; | |
2938 | @@ -415,7 +424,6 @@ | |
2939 | unsigned char newpmt[512]; | |
2940 | int hdr[34]; //Allow for up to 16 PIDs | |
2941 | int pmtlen=0, hdrlen = 0; | |
2942 | - bool need_pcrpid = false; | |
2943 | ||
2944 | hdr[hdrlen++]=pmt->ServiceID; | |
2945 | hdr[hdrlen++]=1; //Place holder for transportID (future) | |
2946 | @@ -489,6 +497,43 @@ | |
2947 | } | |
2948 | } | |
2949 | ||
2950 | +void *DVBCam::ExternalCamSoftkeyHelper(void*self) | |
2951 | +{ | |
2952 | + ((DVBCam*)self)->ExternalCamSoftkeyLoop(); | |
2953 | + return NULL; | |
2954 | +} | |
2955 | +void DVBCam::ExternalCamSoftkeyLoop() | |
2956 | +{ | |
2957 | + int keyfd; | |
2958 | + unsigned char key[8]; | |
2959 | + int index, pid; | |
2960 | + skThreadRunning = true; | |
2961 | + QString keyFifo = QDir::homeDirPath() + | |
2962 | + QString("/.mythtv/externcam_%1/keyfifo").arg(cardnum); | |
2963 | + if ((keyfd=open(keyFifo.ascii(),O_RDONLY))<0) { | |
2964 | + ERROR("Could not open keyfifo!"); | |
2965 | + skThreadRunning = false; | |
2966 | + return; | |
2967 | + } | |
2968 | + while (!exitSkThread) | |
2969 | + { | |
2970 | + unsigned char keytype; | |
2971 | + read(keyfd, &keytype,1); | |
2972 | + index = 0; | |
2973 | + pid = 0; | |
2974 | + read(keyfd, &index, sizeof(int)); | |
2975 | + if(keytype == 'E' || keytype == 'O') | |
2976 | + read(keyfd, key, 8); | |
2977 | + else if (keytype == 'P') | |
2978 | + read(keyfd, &pid, sizeof(int)); | |
2979 | + | |
2980 | + DVBRecorder::UpdateDeCSAKeys(cardnum, keytype, index, key, pid); | |
2981 | + } | |
2982 | + skThreadRunning = false; | |
2983 | + GENERAL(QString("CA: External Softkey thread stopped")); | |
2984 | + return; | |
2985 | +} | |
2986 | + | |
2987 | void DVBCam::stopExternalCam() | |
2988 | { | |
2989 | if(external_cam_fd != -1) { | |
2990 | Index: libs/libmythtv/FFdecsa/parallel_064_2int.h | |
2991 | =================================================================== | |
2992 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
2993 | +++ libs/libmythtv/FFdecsa/parallel_064_2int.h 2006-06-20 17:36:06.000000000 -0400 | |
2994 | @@ -0,0 +1,175 @@ | |
2995 | +/* FFdecsa -- fast decsa algorithm | |
2996 | + * | |
2997 | + * Copyright (C) 2003-2004 fatih89r | |
2998 | + * | |
2999 | + * This program is free software; you can redistribute it and/or modify | |
3000 | + * it under the terms of the GNU General Public License as published by | |
3001 | + * the Free Software Foundation; either version 2 of the License, or | |
3002 | + * (at your option) any later version. | |
3003 | + * | |
3004 | + * This program is distributed in the hope that it will be useful, | |
3005 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
3006 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
3007 | + * GNU General Public License for more details. | |
3008 | + * | |
3009 | + * You should have received a copy of the GNU General Public License | |
3010 | + * along with this program; if not, write to the Free Software | |
3011 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
3012 | + */ | |
3013 | + | |
3014 | + | |
3015 | +struct group_t{ | |
3016 | + unsigned int s1; | |
3017 | + unsigned int s2; | |
3018 | +}; | |
3019 | +typedef struct group_t group; | |
3020 | + | |
3021 | +#define GROUP_PARALLELISM 64 | |
3022 | + | |
3023 | +group static inline FF0(){ | |
3024 | + group res; | |
3025 | + res.s1=0x0; | |
3026 | + res.s2=0x0; | |
3027 | + return res; | |
3028 | +} | |
3029 | + | |
3030 | +group static inline FF1(){ | |
3031 | + group res; | |
3032 | + res.s1=0xffffffff; | |
3033 | + res.s2=0xffffffff; | |
3034 | + return res; | |
3035 | +} | |
3036 | + | |
3037 | +group static inline FFAND(group a,group b){ | |
3038 | + group res; | |
3039 | + res.s1=a.s1&b.s1; | |
3040 | + res.s2=a.s2&b.s2; | |
3041 | + return res; | |
3042 | +} | |
3043 | + | |
3044 | +group static inline FFOR(group a,group b){ | |
3045 | + group res; | |
3046 | + res.s1=a.s1|b.s1; | |
3047 | + res.s2=a.s2|b.s2; | |
3048 | + return res; | |
3049 | +} | |
3050 | + | |
3051 | +group static inline FFXOR(group a,group b){ | |
3052 | + group res; | |
3053 | + res.s1=a.s1^b.s1; | |
3054 | + res.s2=a.s2^b.s2; | |
3055 | + return res; | |
3056 | +} | |
3057 | + | |
3058 | +group static inline FFNOT(group a){ | |
3059 | + group res; | |
3060 | + res.s1=~a.s1; | |
3061 | + res.s2=~a.s2; | |
3062 | + return res; | |
3063 | +} | |
3064 | + | |
3065 | + | |
3066 | +/* 64 rows of 64 bits */ | |
3067 | + | |
3068 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
3069 | + *(((int *)tab)+2*g)=*((int *)data); | |
3070 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
3071 | +} | |
3072 | + | |
3073 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
3074 | + *((int *)data)=*(((int *)tab)+2*g); | |
3075 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
3076 | +} | |
3077 | + | |
3078 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
3079 | + int j; | |
3080 | + for(j=0;j<n;j++){ | |
3081 | + *(data+j)^=*(tab+8*g+j); | |
3082 | + } | |
3083 | +} | |
3084 | + | |
3085 | +struct batch_t{ | |
3086 | + unsigned int s1; | |
3087 | + unsigned int s2; | |
3088 | +}; | |
3089 | +typedef struct batch_t batch; | |
3090 | + | |
3091 | +#define BYTES_PER_BATCH 8 | |
3092 | + | |
3093 | +batch static inline B_FFAND(batch a,batch b){ | |
3094 | + batch res; | |
3095 | + res.s1=a.s1&b.s1; | |
3096 | + res.s2=a.s2&b.s2; | |
3097 | + return res; | |
3098 | +} | |
3099 | + | |
3100 | +batch static inline B_FFOR(batch a,batch b){ | |
3101 | + batch res; | |
3102 | + res.s1=a.s1|b.s1; | |
3103 | + res.s2=a.s2|b.s2; | |
3104 | + return res; | |
3105 | +} | |
3106 | + | |
3107 | +batch static inline B_FFXOR(batch a,batch b){ | |
3108 | + batch res; | |
3109 | + res.s1=a.s1^b.s1; | |
3110 | + res.s2=a.s2^b.s2; | |
3111 | + return res; | |
3112 | +} | |
3113 | + | |
3114 | + | |
3115 | +batch static inline B_FFN_ALL_29(){ | |
3116 | + batch res; | |
3117 | + res.s1=0x29292929; | |
3118 | + res.s2=0x29292929; | |
3119 | + return res; | |
3120 | +} | |
3121 | +batch static inline B_FFN_ALL_02(){ | |
3122 | + batch res; | |
3123 | + res.s1=0x02020202; | |
3124 | + res.s2=0x02020202; | |
3125 | + return res; | |
3126 | +} | |
3127 | +batch static inline B_FFN_ALL_04(){ | |
3128 | + batch res; | |
3129 | + res.s1=0x04040404; | |
3130 | + res.s2=0x04040404; | |
3131 | + return res; | |
3132 | +} | |
3133 | +batch static inline B_FFN_ALL_10(){ | |
3134 | + batch res; | |
3135 | + res.s1=0x10101010; | |
3136 | + res.s2=0x10101010; | |
3137 | + return res; | |
3138 | +} | |
3139 | +batch static inline B_FFN_ALL_40(){ | |
3140 | + batch res; | |
3141 | + res.s1=0x40404040; | |
3142 | + res.s2=0x40404040; | |
3143 | + return res; | |
3144 | +} | |
3145 | +batch static inline B_FFN_ALL_80(){ | |
3146 | + batch res; | |
3147 | + res.s1=0x80808080; | |
3148 | + res.s2=0x80808080; | |
3149 | + return res; | |
3150 | +} | |
3151 | + | |
3152 | + | |
3153 | +batch static inline B_FFSH8L(batch a,int n){ | |
3154 | + batch res; | |
3155 | + res.s1=a.s1<<n; | |
3156 | + res.s2=a.s2<<n; | |
3157 | + return res; | |
3158 | +} | |
3159 | + | |
3160 | +batch static inline B_FFSH8R(batch a,int n){ | |
3161 | + batch res; | |
3162 | + res.s1=a.s1>>n; | |
3163 | + res.s2=a.s2>>n; | |
3164 | + return res; | |
3165 | +} | |
3166 | + | |
3167 | + | |
3168 | +void static inline M_EMPTY(void){ | |
3169 | +} | |
3170 | Index: libs/libmythtv/FFdecsa/logic/logic.c | |
3171 | =================================================================== | |
3172 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
3173 | +++ libs/libmythtv/FFdecsa/logic/logic.c 2006-06-20 17:36:06.000000000 -0400 | |
3174 | @@ -0,0 +1,330 @@ | |
3175 | +/* logic -- synthetize logic functions with 4 inputs | |
3176 | + * | |
3177 | + * Copyright (C) 2003-2004 fatih89r | |
3178 | + * | |
3179 | + * This program is free software; you can redistribute it and/or modify | |
3180 | + * it under the terms of the GNU General Public License as published by | |
3181 | + * the Free Software Foundation; either version 2 of the License, or | |
3182 | + * (at your option) any later version. | |
3183 | + * | |
3184 | + * This program is distributed in the hope that it will be useful, | |
3185 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
3186 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
3187 | + * GNU General Public License for more details. | |
3188 | + * | |
3189 | + * You should have received a copy of the GNU General Public License | |
3190 | + * along with this program; if not, write to the Free Software | |
3191 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
3192 | + */ | |
3193 | + | |
3194 | + | |
3195 | + | |
3196 | + | |
3197 | +/* Can we use negated inputs? */ | |
3198 | +#define noNEGATEDTOO | |
3199 | + | |
3200 | + | |
3201 | +#include <stdio.h> | |
3202 | + | |
3203 | + | |
3204 | +/* | |
3205 | + * abcd | |
3206 | + */ | |
3207 | + | |
3208 | +#define BINARY(b15,b14,b13,b12,b11,b10,b9,b8,b7,b6,b5,b4,b3,b2,b1,b0) \ | |
3209 | + ((b15)<<15)|((b14)<<14)|((b13)<<13)|((b12)<<12)| \ | |
3210 | + ((b11)<<11)|((b10)<<10)|((b9) << 9)|((b8) << 8)| \ | |
3211 | + ((b7) << 7)|((b6) << 6)|((b5) << 5)|((b4) << 4)| \ | |
3212 | + ((b3) << 3)|((b2) << 2)|((b1) << 1)|((b0) << 0) | |
3213 | + | |
3214 | +struct fun{ | |
3215 | + int level; | |
3216 | + int op_type; | |
3217 | + int op1; | |
3218 | + int op2; | |
3219 | +}; | |
3220 | + | |
3221 | +struct fun db[65536]; | |
3222 | +int n_fun; | |
3223 | + | |
3224 | +#define LEVEL_ALOT 1000000 | |
3225 | + | |
3226 | +#define OP_FALSE 0 | |
3227 | +#define OP_TRUE 1 | |
3228 | +#define OP_SRC 2 | |
3229 | +#define OP_AND 3 | |
3230 | +#define OP_OR 4 | |
3231 | +#define OP_XOR 5 | |
3232 | + | |
3233 | +#define SRC_A 10 | |
3234 | +#define SRC_B 20 | |
3235 | +#define SRC_C 30 | |
3236 | +#define SRC_D 40 | |
3237 | +#define SRC_AN 11 | |
3238 | +#define SRC_BN 21 | |
3239 | +#define SRC_CN 31 | |
3240 | +#define SRC_DN 41 | |
3241 | + | |
3242 | +void dump_element_prefix(int); | |
3243 | +void dump_element_infix(int); | |
3244 | + | |
3245 | +int main(void){ | |
3246 | + int i,j; | |
3247 | + int l,p1,p2; | |
3248 | + int candidate; | |
3249 | + int max_p2_lev; | |
3250 | + | |
3251 | + for(i=0;i<65536;i++){ | |
3252 | + db[i].level=LEVEL_ALOT; | |
3253 | + } | |
3254 | + n_fun=0; | |
3255 | + | |
3256 | + db[0].level=0; | |
3257 | + db[0].op_type=OP_FALSE; | |
3258 | + n_fun++; | |
3259 | + | |
3260 | + db[65535].level=0; | |
3261 | + db[65535].op_type=OP_TRUE; | |
3262 | + n_fun++; | |
3263 | + | |
3264 | + db[BINARY(0,0,0,0, 0,0,0,0, 1,1,1,1, 1,1,1,1)].level=0; | |
3265 | + db[BINARY(0,0,0,0, 0,0,0,0, 1,1,1,1, 1,1,1,1)].op_type=OP_SRC; | |
3266 | + db[BINARY(0,0,0,0, 0,0,0,0, 1,1,1,1, 1,1,1,1)].op1=SRC_A; | |
3267 | + n_fun++; | |
3268 | + | |
3269 | + db[BINARY(0,0,0,0, 1,1,1,1, 0,0,0,0, 1,1,1,1)].level=0; | |
3270 | + db[BINARY(0,0,0,0, 1,1,1,1, 0,0,0,0, 1,1,1,1)].op_type=OP_SRC; | |
3271 | + db[BINARY(0,0,0,0, 1,1,1,1, 0,0,0,0, 1,1,1,1)].op1=SRC_B; | |
3272 | + n_fun++; | |
3273 | + | |
3274 | + db[BINARY(0,0,1,1, 0,0,1,1, 0,0,1,1, 0,0,1,1)].level=0; | |
3275 | + db[BINARY(0,0,1,1, 0,0,1,1, 0,0,1,1, 0,0,1,1)].op_type=OP_SRC; | |
3276 | + db[BINARY(0,0,1,1, 0,0,1,1, 0,0,1,1, 0,0,1,1)].op1=SRC_C; | |
3277 | + n_fun++; | |
3278 | + | |
3279 | + db[BINARY(0,1,0,1, 0,1,0,1, 0,1,0,1, 0,1,0,1)].level=0; | |
3280 | + db[BINARY(0,1,0,1, 0,1,0,1, 0,1,0,1, 0,1,0,1)].op_type=OP_SRC; | |
3281 | + db[BINARY(0,1,0,1, 0,1,0,1, 0,1,0,1, 0,1,0,1)].op1=SRC_D; | |
3282 | + n_fun++; | |
3283 | +#ifdef NEGATEDTOO | |
3284 | + db[BINARY(1,1,1,1, 1,1,1,1, 0,0,0,0, 0,0,0,0)].level=0; | |
3285 | + db[BINARY(1,1,1,1, 1,1,1,1, 0,0,0,0, 0,0,0,0)].op_type=OP_SRC; | |
3286 | + db[BINARY(1,1,1,1, 1,1,1,1, 0,0,0,0, 0,0,0,0)].op1=SRC_AN; | |
3287 | + n_fun++; | |
3288 | + | |
3289 | + db[BINARY(1,1,1,1, 0,0,0,0, 1,1,1,1, 0,0,0,0)].level=0; | |
3290 | + db[BINARY(1,1,1,1, 0,0,0,0, 1,1,1,1, 0,0,0,0)].op_type=OP_SRC; | |
3291 | + db[BINARY(1,1,1,1, 0,0,0,0, 1,1,1,1, 0,0,0,0)].op1=SRC_BN; | |
3292 | + n_fun++; | |
3293 | + | |
3294 | + db[BINARY(1,1,0,0, 1,1,0,0, 1,1,0,0, 1,1,0,0)].level=0; | |
3295 | + db[BINARY(1,1,0,0, 1,1,0,0, 1,1,0,0, 1,1,0,0)].op_type=OP_SRC; | |
3296 | + db[BINARY(1,1,0,0, 1,1,0,0, 1,1,0,0, 1,1,0,0)].op1=SRC_CN; | |
3297 | + n_fun++; | |
3298 | + | |
3299 | + db[BINARY(1,0,1,0, 1,0,1,0, 1,0,1,0, 1,0,1,0)].level=0; | |
3300 | + db[BINARY(1,0,1,0, 1,0,1,0, 1,0,1,0, 1,0,1,0)].op_type=OP_SRC; | |
3301 | + db[BINARY(1,0,1,0, 1,0,1,0, 1,0,1,0, 1,0,1,0)].op1=SRC_DN; | |
3302 | + n_fun++; | |
3303 | +#endif | |
3304 | + | |
3305 | + for(l=0;l<100;l++){ | |
3306 | + printf("calculating level %i\n",l); | |
3307 | + for(p1=1;p1<65536;p1++){ | |
3308 | + if(db[p1].level==LEVEL_ALOT) continue; | |
3309 | + max_p2_lev=l-db[p1].level-1; | |
3310 | + for(p2=p1+1;p2<65536;p2++){ | |
3311 | + if(db[p2].level>max_p2_lev) continue; | |
3312 | + | |
3313 | + candidate=p1&p2; | |
3314 | + if(db[candidate].level==LEVEL_ALOT){ | |
3315 | + //found new | |
3316 | + db[candidate].level=db[p1].level+db[p2].level+1; | |
3317 | + db[candidate].op_type=OP_AND; | |
3318 | + db[candidate].op1=p1; | |
3319 | + db[candidate].op2=p2; | |
3320 | + n_fun++; | |
3321 | + } | |
3322 | + | |
3323 | + candidate=p1|p2; | |
3324 | + if(db[candidate].level==LEVEL_ALOT){ | |
3325 | + //found new | |
3326 | + db[candidate].level=db[p1].level+db[p2].level+1; | |
3327 | + db[candidate].op_type=OP_OR; | |
3328 | + db[candidate].op1=p1; | |
3329 | + db[candidate].op2=p2; | |
3330 | + n_fun++; | |
3331 | + } | |
3332 | + | |
3333 | + candidate=p1^p2; | |
3334 | + if(db[candidate].level==LEVEL_ALOT){ | |
3335 | + //found new | |
3336 | + db[candidate].level=db[p1].level+db[p2].level+1; | |
3337 | + db[candidate].op_type=OP_XOR; | |
3338 | + db[candidate].op1=p1; | |
3339 | + db[candidate].op2=p2; | |
3340 | + n_fun++; | |
3341 | + } | |
3342 | + | |
3343 | + } | |
3344 | + } | |
3345 | + printf("num fun=%i\n\n",n_fun); | |
3346 | + fflush(stdout); | |
3347 | + if(n_fun>=65536) break; | |
3348 | + } | |
3349 | + | |
3350 | + | |
3351 | + for(i=0;i<65536;i++){ | |
3352 | + if(db[i].level==LEVEL_ALOT) continue; | |
3353 | + | |
3354 | + printf("PREFIX "); | |
3355 | + for(j=15;j>=0;j--){ | |
3356 | + printf("%i",i&(1<<j)?1:0); | |
3357 | + if(j%4==0) printf(" "); | |
3358 | + if(j%8==0) printf(" "); | |
3359 | + } | |
3360 | + printf(" : lev %2i: ",db[i].level); | |
3361 | + dump_element_prefix(i); | |
3362 | + printf("\n"); | |
3363 | + | |
3364 | + printf("INFIX "); | |
3365 | + for(j=15;j>=0;j--){ | |
3366 | + printf("%i",i&(1<<j)?1:0); | |
3367 | + if(j%4==0) printf(" "); | |
3368 | + if(j%8==0) printf(" "); | |
3369 | + } | |
3370 | + printf(" : lev %2i: ",db[i].level); | |
3371 | + dump_element_infix(i); | |
3372 | + printf("\n"); | |
3373 | + } | |
3374 | + | |
3375 | + return 0; | |
3376 | +} | |
3377 | + | |
3378 | +void dump_element_prefix(int e){ | |
3379 | + if(db[e].level==LEVEL_ALOT){ | |
3380 | + printf("PANIC!\n"); | |
3381 | + return; | |
3382 | + }; | |
3383 | + switch(db[e].op_type){ | |
3384 | + case OP_FALSE: | |
3385 | + printf("0"); | |
3386 | + break; | |
3387 | + case OP_TRUE: | |
3388 | + printf("1"); | |
3389 | + break; | |
3390 | + case OP_SRC: | |
3391 | + switch(db[e].op1){ | |
3392 | + case SRC_A: | |
3393 | + printf("a"); | |
3394 | + break; | |
3395 | + case SRC_B: | |
3396 | + printf("b"); | |
3397 | + break; | |
3398 | + case SRC_C: | |
3399 | + printf("c"); | |
3400 | + break; | |
3401 | + case SRC_D: | |
3402 | + printf("d"); | |
3403 | + break; | |
3404 | + case SRC_AN: | |
3405 | + printf("an"); | |
3406 | + break; | |
3407 | + case SRC_BN: | |
3408 | + printf("bn"); | |
3409 | + break; | |
3410 | + case SRC_CN: | |
3411 | + printf("cn"); | |
3412 | + break; | |
3413 | + case SRC_DN: | |
3414 | + printf("dn"); | |
3415 | + break; | |
3416 | + } | |
3417 | + break; | |
3418 | + case OP_AND: | |
3419 | + printf("FFAND("); | |
3420 | + dump_element_prefix(db[e].op1); | |
3421 | + printf(","); | |
3422 | + dump_element_prefix(db[e].op2); | |
3423 | + printf(")"); | |
3424 | + break; | |
3425 | + case OP_OR: | |
3426 | + printf("FFOR("); | |
3427 | + dump_element_prefix(db[e].op1); | |
3428 | + printf(","); | |
3429 | + dump_element_prefix(db[e].op2); | |
3430 | + printf(")"); | |
3431 | + break; | |
3432 | + case OP_XOR: | |
3433 | + printf("FFXOR("); | |
3434 | + dump_element_prefix(db[e].op1); | |
3435 | + printf(","); | |
3436 | + dump_element_prefix(db[e].op2); | |
3437 | + printf(")"); | |
3438 | + break; | |
3439 | + } | |
3440 | +} | |
3441 | + | |
3442 | +void dump_element_infix(int e){ | |
3443 | + if(db[e].level==LEVEL_ALOT){ | |
3444 | + printf("PANIC!\n"); | |
3445 | + return; | |
3446 | + }; | |
3447 | + switch(db[e].op_type){ | |
3448 | + case OP_FALSE: | |
3449 | + printf("0"); | |
3450 | + break; | |
3451 | + case OP_TRUE: | |
3452 | + printf("1"); | |
3453 | + break; | |
3454 | + case OP_SRC: | |
3455 | + switch(db[e].op1){ | |
3456 | + case SRC_A: | |
3457 | + printf("a"); | |
3458 | + break; | |
3459 | + case SRC_B: | |
3460 | + printf("b"); | |
3461 | + break; | |
3462 | + case SRC_C: | |
3463 | + printf("c"); | |
3464 | + break; | |
3465 | + case SRC_D: | |
3466 | + printf("d"); | |
3467 | + break; | |
3468 | + case SRC_AN: | |
3469 | + printf("an"); | |
3470 | + break; | |
3471 | + case SRC_BN: | |
3472 | + printf("bn"); | |
3473 | + break; | |
3474 | + case SRC_CN: | |
3475 | + printf("cn"); | |
3476 | + break; | |
3477 | + case SRC_DN: | |
3478 | + printf("dn"); | |
3479 | + break; | |
3480 | + } | |
3481 | + break; | |
3482 | + case OP_AND: | |
3483 | + printf("( "); | |
3484 | + dump_element_infix(db[e].op1); | |
3485 | + printf("&"); | |
3486 | + dump_element_infix(db[e].op2); | |
3487 | + printf(" )"); | |
3488 | + break; | |
3489 | + case OP_OR: | |
3490 | + printf("( "); | |
3491 | + dump_element_infix(db[e].op1); | |
3492 | + printf("|"); | |
3493 | + dump_element_infix(db[e].op2); | |
3494 | + printf(" )"); | |
3495 | + break; | |
3496 | + case OP_XOR: | |
3497 | + printf("( "); | |
3498 | + dump_element_infix(db[e].op1); | |
3499 | + printf("^"); | |
3500 | + dump_element_infix(db[e].op2); | |
3501 | + printf(" )"); | |
3502 | + break; | |
3503 | + } | |
3504 | +} | |
3505 | Index: libs/libmythtv/FFdecsa/parallel_128_4int.h | |
3506 | =================================================================== | |
3507 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
3508 | +++ libs/libmythtv/FFdecsa/parallel_128_4int.h 2006-06-20 17:36:06.000000000 -0400 | |
3509 | @@ -0,0 +1,207 @@ | |
3510 | +/* FFdecsa -- fast decsa algorithm | |
3511 | + * | |
3512 | + * Copyright (C) 2003-2004 fatih89r | |
3513 | + * | |
3514 | + * This program is free software; you can redistribute it and/or modify | |
3515 | + * it under the terms of the GNU General Public License as published by | |
3516 | + * the Free Software Foundation; either version 2 of the License, or | |
3517 | + * (at your option) any later version. | |
3518 | + * | |
3519 | + * This program is distributed in the hope that it will be useful, | |
3520 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
3521 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
3522 | + * GNU General Public License for more details. | |
3523 | + * | |
3524 | + * You should have received a copy of the GNU General Public License | |
3525 | + * along with this program; if not, write to the Free Software | |
3526 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
3527 | + */ | |
3528 | + | |
3529 | + | |
3530 | +struct group_t{ | |
3531 | + unsigned int s1,s2,s3,s4; | |
3532 | +}; | |
3533 | +typedef struct group_t group; | |
3534 | + | |
3535 | +#define GROUP_PARALLELISM 128 | |
3536 | + | |
3537 | +group static inline FF0(){ | |
3538 | + group res; | |
3539 | + res.s1=0x0; | |
3540 | + res.s2=0x0; | |
3541 | + res.s3=0x0; | |
3542 | + res.s4=0x0; | |
3543 | + return res; | |
3544 | +} | |
3545 | + | |
3546 | +group static inline FF1(){ | |
3547 | + group res; | |
3548 | + res.s1=0xffffffff; | |
3549 | + res.s2=0xffffffff; | |
3550 | + res.s3=0xffffffff; | |
3551 | + res.s4=0xffffffff; | |
3552 | + return res; | |
3553 | +} | |
3554 | + | |
3555 | +group static inline FFAND(group a,group b){ | |
3556 | + group res; | |
3557 | + res.s1=a.s1&b.s1; | |
3558 | + res.s2=a.s2&b.s2; | |
3559 | + res.s3=a.s3&b.s3; | |
3560 | + res.s4=a.s4&b.s4; | |
3561 | + return res; | |
3562 | +} | |
3563 | + | |
3564 | +group static inline FFOR(group a,group b){ | |
3565 | + group res; | |
3566 | + res.s1=a.s1|b.s1; | |
3567 | + res.s2=a.s2|b.s2; | |
3568 | + res.s3=a.s3|b.s3; | |
3569 | + res.s4=a.s4|b.s4; | |
3570 | + return res; | |
3571 | +} | |
3572 | + | |
3573 | +group static inline FFXOR(group a,group b){ | |
3574 | + group res; | |
3575 | + res.s1=a.s1^b.s1; | |
3576 | + res.s2=a.s2^b.s2; | |
3577 | + res.s3=a.s3^b.s3; | |
3578 | + res.s4=a.s4^b.s4; | |
3579 | + return res; | |
3580 | +} | |
3581 | + | |
3582 | +group static inline FFNOT(group a){ | |
3583 | + group res; | |
3584 | + res.s1=~a.s1; | |
3585 | + res.s2=~a.s2; | |
3586 | + res.s3=~a.s3; | |
3587 | + res.s4=~a.s4; | |
3588 | + return res; | |
3589 | +} | |
3590 | + | |
3591 | + | |
3592 | +/* 64 rows of 128 bits */ | |
3593 | + | |
3594 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
3595 | + *(((int *)tab)+2*g)=*((int *)data); | |
3596 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
3597 | +} | |
3598 | + | |
3599 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
3600 | + *((int *)data)=*(((int *)tab)+2*g); | |
3601 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
3602 | +} | |
3603 | + | |
3604 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
3605 | + int j; | |
3606 | + for(j=0;j<n;j++){ | |
3607 | + *(data+j)^=*(tab+8*g+j); | |
3608 | + } | |
3609 | +} | |
3610 | + | |
3611 | + | |
3612 | +struct batch_t{ | |
3613 | + unsigned int s1,s2,s3,s4; | |
3614 | +}; | |
3615 | +typedef struct batch_t batch; | |
3616 | + | |
3617 | +#define BYTES_PER_BATCH 16 | |
3618 | + | |
3619 | +batch static inline B_FFAND(batch a,batch b){ | |
3620 | + batch res; | |
3621 | + res.s1=a.s1&b.s1; | |
3622 | + res.s2=a.s2&b.s2; | |
3623 | + res.s3=a.s3&b.s3; | |
3624 | + res.s4=a.s4&b.s4; | |
3625 | + return res; | |
3626 | +} | |
3627 | + | |
3628 | +batch static inline B_FFOR(batch a,batch b){ | |
3629 | + batch res; | |
3630 | + res.s1=a.s1|b.s1; | |
3631 | + res.s2=a.s2|b.s2; | |
3632 | + res.s3=a.s3|b.s3; | |
3633 | + res.s4=a.s4|b.s4; | |
3634 | + return res; | |
3635 | +} | |
3636 | + | |
3637 | +batch static inline B_FFXOR(batch a,batch b){ | |
3638 | + batch res; | |
3639 | + res.s1=a.s1^b.s1; | |
3640 | + res.s2=a.s2^b.s2; | |
3641 | + res.s3=a.s3^b.s3; | |
3642 | + res.s4=a.s4^b.s4; | |
3643 | + return res; | |
3644 | +} | |
3645 | + | |
3646 | + | |
3647 | +batch static inline B_FFN_ALL_29(){ | |
3648 | + batch res; | |
3649 | + res.s1=0x29292929; | |
3650 | + res.s2=0x29292929; | |
3651 | + res.s3=0x29292929; | |
3652 | + res.s4=0x29292929; | |
3653 | + return res; | |
3654 | +} | |
3655 | +batch static inline B_FFN_ALL_02(){ | |
3656 | + batch res; | |
3657 | + res.s1=0x02020202; | |
3658 | + res.s2=0x02020202; | |
3659 | + res.s3=0x02020202; | |
3660 | + res.s4=0x02020202; | |
3661 | + return res; | |
3662 | +} | |
3663 | +batch static inline B_FFN_ALL_04(){ | |
3664 | + batch res; | |
3665 | + res.s1=0x04040404; | |
3666 | + res.s2=0x04040404; | |
3667 | + res.s3=0x04040404; | |
3668 | + res.s4=0x04040404; | |
3669 | + return res; | |
3670 | +} | |
3671 | +batch static inline B_FFN_ALL_10(){ | |
3672 | + batch res; | |
3673 | + res.s1=0x10101010; | |
3674 | + res.s2=0x10101010; | |
3675 | + res.s3=0x10101010; | |
3676 | + res.s4=0x10101010; | |
3677 | + return res; | |
3678 | +} | |
3679 | +batch static inline B_FFN_ALL_40(){ | |
3680 | + batch res; | |
3681 | + res.s1=0x40404040; | |
3682 | + res.s2=0x40404040; | |
3683 | + res.s3=0x40404040; | |
3684 | + res.s4=0x40404040; | |
3685 | + return res; | |
3686 | +} | |
3687 | +batch static inline B_FFN_ALL_80(){ | |
3688 | + batch res; | |
3689 | + res.s1=0x80808080; | |
3690 | + res.s2=0x80808080; | |
3691 | + res.s3=0x80808080; | |
3692 | + res.s4=0x80808080; | |
3693 | + return res; | |
3694 | +} | |
3695 | + | |
3696 | +batch static inline B_FFSH8L(batch a,int n){ | |
3697 | + batch res; | |
3698 | + res.s1=a.s1<<n; | |
3699 | + res.s2=a.s2<<n; | |
3700 | + res.s3=a.s3<<n; | |
3701 | + res.s4=a.s4<<n; | |
3702 | + return res; | |
3703 | +} | |
3704 | + | |
3705 | +batch static inline B_FFSH8R(batch a,int n){ | |
3706 | + batch res; | |
3707 | + res.s1=a.s1>>n; | |
3708 | + res.s2=a.s2>>n; | |
3709 | + res.s3=a.s3>>n; | |
3710 | + res.s4=a.s4>>n; | |
3711 | + return res; | |
3712 | +} | |
3713 | + | |
3714 | + | |
3715 | +void static inline M_EMPTY(void){ | |
3716 | +} | |
3717 | Index: libs/libmythtv/FFdecsa/parallel_128_2mmx.h | |
3718 | =================================================================== | |
3719 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
3720 | +++ libs/libmythtv/FFdecsa/parallel_128_2mmx.h 2006-06-20 17:36:06.000000000 -0400 | |
3721 | @@ -0,0 +1,199 @@ | |
3722 | +/* FFdecsa -- fast decsa algorithm | |
3723 | + * | |
3724 | + * Copyright (C) 2003-2004 fatih89r | |
3725 | + * | |
3726 | + * This program is free software; you can redistribute it and/or modify | |
3727 | + * it under the terms of the GNU General Public License as published by | |
3728 | + * the Free Software Foundation; either version 2 of the License, or | |
3729 | + * (at your option) any later version. | |
3730 | + * | |
3731 | + * This program is distributed in the hope that it will be useful, | |
3732 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
3733 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
3734 | + * GNU General Public License for more details. | |
3735 | + * | |
3736 | + * You should have received a copy of the GNU General Public License | |
3737 | + * along with this program; if not, write to the Free Software | |
3738 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
3739 | + */ | |
3740 | + | |
3741 | + | |
3742 | +#include <mmintrin.h> | |
3743 | + | |
3744 | +struct group_t{ | |
3745 | + __m64 s1,s2; | |
3746 | +}; | |
3747 | +typedef struct group_t group; | |
3748 | + | |
3749 | +#define GROUP_PARALLELISM 128 | |
3750 | + | |
3751 | +group static inline FF0(){ | |
3752 | + group res; | |
3753 | + res.s1=(__m64)0x0ULL; | |
3754 | + res.s2=(__m64)0x0ULL; | |
3755 | + return res; | |
3756 | +} | |
3757 | + | |
3758 | +group static inline FF1(){ | |
3759 | + group res; | |
3760 | + res.s1=(__m64)0xffffffffffffffffULL; | |
3761 | + res.s2=(__m64)0xffffffffffffffffULL; | |
3762 | + return res; | |
3763 | +} | |
3764 | + | |
3765 | +group static inline FFAND(group a,group b){ | |
3766 | + group res; | |
3767 | + res.s1=_m_pand(a.s1,b.s1); | |
3768 | + res.s2=_m_pand(a.s2,b.s2); | |
3769 | + return res; | |
3770 | +} | |
3771 | + | |
3772 | +group static inline FFOR(group a,group b){ | |
3773 | + group res; | |
3774 | + res.s1=_m_por(a.s1,b.s1); | |
3775 | + res.s2=_m_por(a.s2,b.s2); | |
3776 | + return res; | |
3777 | +} | |
3778 | + | |
3779 | +group static inline FFXOR(group a,group b){ | |
3780 | + group res; | |
3781 | + res.s1=_m_pxor(a.s1,b.s1); | |
3782 | + res.s2=_m_pxor(a.s2,b.s2); | |
3783 | + return res; | |
3784 | +} | |
3785 | + | |
3786 | +group static inline FFNOT(group a){ | |
3787 | + group res; | |
3788 | + res.s1=_m_pxor(a.s1,FF1().s1); | |
3789 | + res.s2=_m_pxor(a.s2,FF1().s2); | |
3790 | + return res; | |
3791 | +} | |
3792 | + | |
3793 | + | |
3794 | +/* 64 rows of 128 bits */ | |
3795 | + | |
3796 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
3797 | + *(((int *)tab)+2*g)=*((int *)data); | |
3798 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
3799 | +} | |
3800 | + | |
3801 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
3802 | + *((int *)data)=*(((int *)tab)+2*g); | |
3803 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
3804 | +} | |
3805 | + | |
3806 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
3807 | + int j; | |
3808 | + for(j=0;j<n;j++){ | |
3809 | + *(data+j)^=*(tab+8*g+j); | |
3810 | + } | |
3811 | +} | |
3812 | + | |
3813 | + | |
3814 | +struct batch_t{ | |
3815 | + __m64 s1,s2; | |
3816 | +}; | |
3817 | +typedef struct batch_t batch; | |
3818 | + | |
3819 | +#define BYTES_PER_BATCH 16 | |
3820 | + | |
3821 | +batch static inline B_FFAND(batch a,batch b){ | |
3822 | + batch res; | |
3823 | + res.s1=_m_pand(a.s1,b.s1); | |
3824 | + res.s2=_m_pand(a.s2,b.s2); | |
3825 | + return res; | |
3826 | +} | |
3827 | + | |
3828 | +batch static inline B_FFOR(batch a,batch b){ | |
3829 | + batch res; | |
3830 | + res.s1=_m_por(a.s1,b.s1); | |
3831 | + res.s2=_m_por(a.s2,b.s2); | |
3832 | + return res; | |
3833 | +} | |
3834 | + | |
3835 | +batch static inline B_FFXOR(batch a,batch b){ | |
3836 | + batch res; | |
3837 | + res.s1=_m_pxor(a.s1,b.s1); | |
3838 | + res.s2=_m_pxor(a.s2,b.s2); | |
3839 | + return res; | |
3840 | +} | |
3841 | + | |
3842 | +batch static inline B_FFN_ALL_29(){ | |
3843 | + batch res; | |
3844 | + res.s1=(__m64)0x2929292929292929ULL; | |
3845 | + res.s2=(__m64)0x2929292929292929ULL; | |
3846 | + return res; | |
3847 | +} | |
3848 | +batch static inline B_FFN_ALL_02(){ | |
3849 | + batch res; | |
3850 | + res.s1=(__m64)0x0202020202020202ULL; | |
3851 | + res.s2=(__m64)0x0202020202020202ULL; | |
3852 | + return res; | |
3853 | +} | |
3854 | +batch static inline B_FFN_ALL_04(){ | |
3855 | + batch res; | |
3856 | + res.s1=(__m64)0x0404040404040404ULL; | |
3857 | + res.s2=(__m64)0x0404040404040404ULL; | |
3858 | + return res; | |
3859 | +} | |
3860 | +batch static inline B_FFN_ALL_10(){ | |
3861 | + batch res; | |
3862 | + res.s1=(__m64)0x1010101010101010ULL; | |
3863 | + res.s2=(__m64)0x1010101010101010ULL; | |
3864 | + return res; | |
3865 | +} | |
3866 | +batch static inline B_FFN_ALL_40(){ | |
3867 | + batch res; | |
3868 | + res.s1=(__m64)0x4040404040404040ULL; | |
3869 | + res.s2=(__m64)0x4040404040404040ULL; | |
3870 | + return res; | |
3871 | +} | |
3872 | +batch static inline B_FFN_ALL_80(){ | |
3873 | + batch res; | |
3874 | + res.s1=(__m64)0x8080808080808080ULL; | |
3875 | + res.s2=(__m64)0x8080808080808080ULL; | |
3876 | + return res; | |
3877 | +} | |
3878 | + | |
3879 | +batch static inline B_FFSH8L(batch a,int n){ | |
3880 | + batch res; | |
3881 | + res.s1=_m_psllqi(a.s1,n); | |
3882 | + res.s2=_m_psllqi(a.s2,n); | |
3883 | + return res; | |
3884 | +} | |
3885 | + | |
3886 | +batch static inline B_FFSH8R(batch a,int n){ | |
3887 | + batch res; | |
3888 | + res.s1=_m_psrlqi(a.s1,n); | |
3889 | + res.s2=_m_psrlqi(a.s2,n); | |
3890 | + return res; | |
3891 | +} | |
3892 | + | |
3893 | +void static inline M_EMPTY(void){ | |
3894 | + _m_empty(); | |
3895 | +} | |
3896 | + | |
3897 | + | |
3898 | +#undef XOR_8_BY | |
3899 | +#define XOR_8_BY(d,s1,s2) do{ __m64 *pd=(__m64 *)(d), *ps1=(__m64 *)(s1), *ps2=(__m64 *)(s2); \ | |
3900 | + *pd = _m_pxor( *ps1 , *ps2 ); }while(0) | |
3901 | + | |
3902 | +#undef XOREQ_8_BY | |
3903 | +#define XOREQ_8_BY(d,s) do{ __m64 *pd=(__m64 *)(d), *ps=(__m64 *)(s); \ | |
3904 | + *pd = _m_pxor( *ps, *pd ); }while(0) | |
3905 | + | |
3906 | +#undef COPY_8_BY | |
3907 | +#define COPY_8_BY(d,s) do{ __m64 *pd=(__m64 *)(d), *ps=(__m64 *)(s); \ | |
3908 | + *pd = *ps; }while(0) | |
3909 | + | |
3910 | +#undef BEST_SPAN | |
3911 | +#define BEST_SPAN 8 | |
3912 | + | |
3913 | +#undef XOR_BEST_BY | |
3914 | +#define XOR_BEST_BY(d,s1,s2) do{ XOR_8_BY(d,s1,s2); }while(0); | |
3915 | + | |
3916 | +#undef XOREQ_BEST_BY | |
3917 | +#define XOREQ_BEST_BY(d,s) do{ XOREQ_8_BY(d,s); }while(0); | |
3918 | + | |
3919 | +#undef COPY_BEST_BY | |
3920 | +#define COPY_BEST_BY(d,s) do{ COPY_8_BY(d,s); }while(0); | |
3921 | Index: libs/libmythtv/FFdecsa/Makefile | |
3922 | =================================================================== | |
3923 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
3924 | +++ libs/libmythtv/FFdecsa/Makefile 2006-06-20 17:36:06.000000000 -0400 | |
3925 | @@ -0,0 +1,54 @@ | |
3926 | +##### compiling with g++ gives a little more speed | |
3927 | +#COMPILER=gcc | |
3928 | +#COMPILER=g++ | |
3929 | + | |
3930 | +###there are two functions which apparently don't want to be inlined | |
3931 | +#FLAGS=-O3 -march=athlon-xp -fexpensive-optimizations -funroll-loops -finline-limit=6000000 --param max-unrolled-insns=500 | |
3932 | +#FLAGS=-O3 -march=athlon-xp -fexpensive-optimizations -funroll-loops --param max-unrolled-insns=500 | |
3933 | +#FLAGS=-O3 -march=pentium3 -fexpensive-optimizations -funroll-loops | |
3934 | + | |
3935 | +###icc crashes for unknown reasons | |
3936 | +#COMPILER=/opt/intel_cc_80/bin/icc | |
3937 | +#FLAGS=-O3 -march=pentiumiii | |
3938 | + | |
3939 | +#FLAGS += -g | |
3940 | +#FLAGS += -fno-alias | |
3941 | +#FLAGS += -vec_report3 | |
3942 | +#FLAGS += -Wall -Winline | |
3943 | +#FLAGS += -fomit-frame-pointer | |
3944 | +#FLAGS += -pg | |
3945 | + | |
3946 | +COMPILER ?= g++ | |
3947 | +FLAGS ?= -Wall -O3 -march=pentium -mmmx -fomit-frame-pointer -fexpensive-optimizations -funroll-loops | |
3948 | + | |
3949 | +H_FILES = FFdecsa.h parallel_generic.h parallel_std_def.h \ | |
3950 | + parallel_032_4char.h \ | |
3951 | + parallel_032_int.h \ | |
3952 | + parallel_064_2int.h \ | |
3953 | + parallel_064_8charA.h \ | |
3954 | + parallel_064_8char.h \ | |
3955 | + parallel_064_long.h \ | |
3956 | + parallel_064_mmx.h \ | |
3957 | + parallel_128_16charA.h \ | |
3958 | + parallel_128_16char.h \ | |
3959 | + parallel_128_2long.h \ | |
3960 | + parallel_128_2mmx.h \ | |
3961 | + parallel_128_4int.h \ | |
3962 | + parallel_128_sse.h | |
3963 | + | |
3964 | +all: FFdecsa.o | |
3965 | + | |
3966 | +%.o: %.c | |
3967 | + $(COMPILER) $(FLAGS) -c $< | |
3968 | + | |
3969 | +FFdecsa_test: FFdecsa_test.o FFdecsa.o | |
3970 | + $(COMPILER) $(FLAGS) -o FFdecsa_test FFdecsa_test.o FFdecsa.o | |
3971 | + | |
3972 | +FFdecsa_test.o: FFdecsa_test.c FFdecsa.h FFdecsa_test_testcases.h | |
3973 | +FFdecsa.o: FFdecsa.c stream.c $(H_FILES) | |
3974 | + | |
3975 | +clean: | |
3976 | + rm -f FFdecsa_test *.o | |
3977 | + | |
3978 | +test: FFdecsa_test | |
3979 | + sync;usleep 200000;nice --19 ./FFdecsa_test | |
3980 | Index: libs/libmythtv/FFdecsa/vdr_patches/README_vdr.txt | |
3981 | =================================================================== | |
3982 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
3983 | +++ libs/libmythtv/FFdecsa/vdr_patches/README_vdr.txt 2006-06-20 17:36:06.000000000 -0400 | |
3984 | @@ -0,0 +1,58 @@ | |
3985 | +------- | |
3986 | +FFdecsa | |
3987 | +------- | |
3988 | + | |
3989 | +This directory contains patches to use FFdecsa with vdr, by means of a | |
3990 | +new FFdecsa-based SoftCSA. | |
3991 | + | |
3992 | +You don't need a SoftCSA patch!!! | |
3993 | + | |
3994 | +Step by step instructions: | |
3995 | + | |
3996 | +- create a directory somewhere, we will call this dir $BASE | |
3997 | + | |
3998 | +- download vdr-1.3.11.tar.bz2 and put it in $BASE | |
3999 | + | |
4000 | +- download vdr-sc-0.3.15.tar.gz and put it in $BASE | |
4001 | + | |
4002 | +- download FFdecsa-1.0.0.tar.bz2 and put it in $BASE | |
4003 | + | |
4004 | +- cd $BASE | |
4005 | + | |
4006 | +- tar xvjf vdr-1.3.11.tar.bz2 | |
4007 | + | |
4008 | +- cd vdr-1.3.11/PLUGINS/src/ | |
4009 | + | |
4010 | +- tar xvzf ../../../vdr-sc-0.3.15.tar.gz | |
4011 | + | |
4012 | +- ln -s sc-0.3.15 sc | |
4013 | + | |
4014 | +- cd $BASE/vdr-1.3.11 | |
4015 | + | |
4016 | +- tar xvjf ../FFdecsa-1.0.0.tar.bz2 | |
4017 | + | |
4018 | +- ln -s FFdecsa-1.0.0 FFdecsa | |
4019 | + | |
4020 | +- patch -p1 <PLUGINS/src/sc-0.3.15/patches/vdr-1.3.10-sc.diff | |
4021 | + | |
4022 | +- patch -p1 <FFdecsa/vdr_patches/vdr-1.3.11-FFdecsa.diff | |
4023 | + | |
4024 | +- cd FFdecsa | |
4025 | + | |
4026 | +- optional: edit Makefile | |
4027 | + | |
4028 | +- make | |
4029 | + | |
4030 | +- ./FFdecsa_test | |
4031 | + | |
4032 | +- cd $BASE/vdr-1.3.11 | |
4033 | + | |
4034 | +- cp Make.config.template Make.config | |
4035 | + | |
4036 | +- optional: edit Make.config | |
4037 | + | |
4038 | +- make | |
4039 | + | |
4040 | +- make plugins | |
4041 | + | |
4042 | +Good luck! | |
4043 | Index: libs/libmythtv/FFdecsa/tmp_autogenerated_stuff_FFdecsa.c | |
4044 | =================================================================== | |
4045 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
4046 | +++ libs/libmythtv/FFdecsa/tmp_autogenerated_stuff_FFdecsa.c 2006-06-20 17:36:06.000000000 -0400 | |
4047 | @@ -0,0 +1,790 @@ | |
4048 | +/* FFdecsa -- fast decsa algorithm | |
4049 | + * | |
4050 | + * Copyright (C) 2003-2004 fatih89r | |
4051 | + * | |
4052 | + * This program is free software; you can redistribute it and/or modify | |
4053 | + * it under the terms of the GNU General Public License as published by | |
4054 | + * the Free Software Foundation; either version 2 of the License, or | |
4055 | + * (at your option) any later version. | |
4056 | + * | |
4057 | + * This program is distributed in the hope that it will be useful, | |
4058 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
4059 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
4060 | + * GNU General Public License for more details. | |
4061 | + * | |
4062 | + * You should have received a copy of the GNU General Public License | |
4063 | + * along with this program; if not, write to the Free Software | |
4064 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
4065 | + */ | |
4066 | + | |
4067 | + | |
4068 | +#include <sys/types.h> | |
4069 | +#include <string.h> | |
4070 | +#include <stdio.h> | |
4071 | + | |
4072 | +#include "FFdecsa.h" | |
4073 | + | |
4074 | +#ifndef NULL | |
4075 | +#define NULL 0 | |
4076 | +#endif | |
4077 | + | |
4078 | +// activate debug by changing the grep command there. | |
4079 | +// don't edit autogenerated files (name beginning with "_"). | |
4080 | + | |
4081 | +//// parallelization stuff, large speed differences are possible | |
4082 | +// possible choices | |
4083 | +#define PARALLEL_32_4CHAR 320 | |
4084 | +#define PARALLEL_32_4CHARA 321 | |
4085 | +#define PARALLEL_32_INT 322 | |
4086 | +#define PARALLEL_64_8CHAR 640 | |
4087 | +#define PARALLEL_64_8CHARA 641 | |
4088 | +#define PARALLEL_64_2INT 642 | |
4089 | +#define PARALLEL_64_LONG 643 | |
4090 | +#define PARALLEL_64_MMX 644 | |
4091 | +#define PARALLEL_128_16CHAR 1280 | |
4092 | +#define PARALLEL_128_16CHARA 1281 | |
4093 | +#define PARALLEL_128_4INT 1282 | |
4094 | +#define PARALLEL_128_2LONG 1283 | |
4095 | +#define PARALLEL_128_2MMX 1284 | |
4096 | +#define PARALLEL_128_SSE 1285 | |
4097 | + | |
4098 | +//////// our choice //////////////// our choice //////////////// our choice //////////////// our choice //////// | |
4099 | +#define PARALLEL_MODE PARALLEL_64_MMX | |
4100 | +//////// our choice //////////////// our choice //////////////// our choice //////////////// our choice //////// | |
4101 | + | |
4102 | +#include "parallel_generic.h" | |
4103 | +//// conditionals | |
4104 | +#if PARALLEL_MODE==PARALLEL_32_4CHAR | |
4105 | +#include "parallel_032_4char.h" | |
4106 | +#elif PARALLEL_MODE==PARALLEL_32_4CHARA | |
4107 | +#include "parallel_032_4charA.h" | |
4108 | +#elif PARALLEL_MODE==PARALLEL_32_INT | |
4109 | +#include "parallel_032_int.h" | |
4110 | +#elif PARALLEL_MODE==PARALLEL_64_8CHAR | |
4111 | +#include "parallel_064_8char.h" | |
4112 | +#elif PARALLEL_MODE==PARALLEL_64_8CHARA | |
4113 | +#include "parallel_064_8charA.h" | |
4114 | +#elif PARALLEL_MODE==PARALLEL_64_2INT | |
4115 | +#include "parallel_064_2int.h" | |
4116 | +#elif PARALLEL_MODE==PARALLEL_64_LONG | |
4117 | +#include "parallel_064_long.h" | |
4118 | +#elif PARALLEL_MODE==PARALLEL_64_MMX | |
4119 | +#include "parallel_064_mmx.h" | |
4120 | +#elif PARALLEL_MODE==PARALLEL_128_16CHAR | |
4121 | +#include "parallel_128_16char.h" | |
4122 | +#elif PARALLEL_MODE==PARALLEL_128_16CHARA | |
4123 | +#include "parallel_128_16charA.h" | |
4124 | +#elif PARALLEL_MODE==PARALLEL_128_4INT | |
4125 | +#include "parallel_128_4int.h" | |
4126 | +#elif PARALLEL_MODE==PARALLEL_128_2LONG | |
4127 | +#include "parallel_128_2long.h" | |
4128 | +#elif PARALLEL_MODE==PARALLEL_128_2MMX | |
4129 | +#include "parallel_128_2mmx.h" | |
4130 | +#elif PARALLEL_MODE==PARALLEL_128_SSE | |
4131 | +#include "parallel_128_sse.h" | |
4132 | +#else | |
4133 | +#error "unknown/undefined parallel mode" | |
4134 | +#endif | |
4135 | + | |
4136 | +// stuff depending on conditionals | |
4137 | + | |
4138 | +#define BYTES_PER_GROUP (GROUP_PARALLELISM/8) | |
4139 | +#define BYPG BYTES_PER_GROUP | |
4140 | +#define BITS_PER_GROUP GROUP_PARALLELISM | |
4141 | +#define BIPG BITS_PER_GROUP | |
4142 | + | |
4143 | + | |
4144 | +//// debug tool | |
4145 | + | |
4146 | +static void dump_mem(unsigned char *string, unsigned char *p, int len, int linelen){ | |
4147 | + int i; | |
4148 | + for(i=0;i<len;i++){ | |
4149 | + if(i%linelen==0&&i) fprintf(stderr,"\n"); | |
4150 | + if(i%linelen==0) fprintf(stderr,"%s %08x:",string,i); | |
4151 | + else{ | |
4152 | + if(i%8==0) fprintf(stderr," "); | |
4153 | + if(i%4==0) fprintf(stderr," "); | |
4154 | + } | |
4155 | + fprintf(stderr," %02x",p[i]); | |
4156 | + } | |
4157 | + if(i%linelen==0) fprintf(stderr,"\n"); | |
4158 | +} | |
4159 | + | |
4160 | +////////////////////////////////////////////////////////////////////////////////// | |
4161 | + | |
4162 | +struct csa_key_t{ | |
4163 | + unsigned char ck[8]; | |
4164 | +// used by stream | |
4165 | + int iA[8]; // iA[0] is for A1, iA[7] is for A8 | |
4166 | + int iB[8]; // iB[0] is for B1, iB[7] is for B8 | |
4167 | +// used by stream (group) | |
4168 | + group ck_g[8][8]; // [byte][bit:0=LSB,7=MSB] | |
4169 | + group iA_g[8][4]; // [0 for A1][0 for LSB] | |
4170 | + group iB_g[8][4]; // [0 for B1][0 for LSB] | |
4171 | +// used by block | |
4172 | + unsigned char kk[56]; | |
4173 | +// used by block (group) | |
4174 | + __attribute__((aligned(16))) batch kkmulti[56]; // many times the same byte in every batch | |
4175 | +}; | |
4176 | + | |
4177 | +static struct csa_keys_t{ | |
4178 | + struct csa_key_t even; | |
4179 | + struct csa_key_t odd; | |
4180 | +} keys; | |
4181 | + | |
4182 | + | |
4183 | +//-----stream cypher | |
4184 | + | |
4185 | +//-----key schedule for stream decypher | |
4186 | +static void key_schedule_stream( | |
4187 | + unsigned char *ck, // [In] ck[0]-ck[7] 8 bytes | Key. | |
4188 | + int *iA, // [Out] iA[0]-iA[7] 8 nibbles | Key schedule. | |
4189 | + int *iB) // [Out] iB[0]-iB[7] 8 nibbles | Key schedule. | |
4190 | +{ | |
4191 | + iA[0]=(ck[0]>>4)&0xf; | |
4192 | + iA[1]=(ck[0] )&0xf; | |
4193 | + iA[2]=(ck[1]>>4)&0xf; | |
4194 | + iA[3]=(ck[1] )&0xf; | |
4195 | + iA[4]=(ck[2]>>4)&0xf; | |
4196 | + iA[5]=(ck[2] )&0xf; | |
4197 | + iA[6]=(ck[3]>>4)&0xf; | |
4198 | + iA[7]=(ck[3] )&0xf; | |
4199 | + iB[0]=(ck[4]>>4)&0xf; | |
4200 | + iB[1]=(ck[4] )&0xf; | |
4201 | + iB[2]=(ck[5]>>4)&0xf; | |
4202 | + iB[3]=(ck[5] )&0xf; | |
4203 | + iB[4]=(ck[6]>>4)&0xf; | |
4204 | + iB[5]=(ck[6] )&0xf; | |
4205 | + iB[6]=(ck[7]>>4)&0xf; | |
4206 | + iB[7]=(ck[7] )&0xf; | |
4207 | +} | |
4208 | + | |
4209 | +//----- stream main function | |
4210 | + | |
4211 | +#define STREAM_INIT | |
4212 | +#include "tmp_autogenerated_stuff_stream.c" | |
4213 | +#undef STREAM_INIT | |
4214 | + | |
4215 | +#define STREAM_NORMAL | |
4216 | +#include "tmp_autogenerated_stuff_stream.c" | |
4217 | +#undef STREAM_NORMAL | |
4218 | + | |
4219 | + | |
4220 | +//-----block decypher | |
4221 | + | |
4222 | +//-----key schedule for block decypher | |
4223 | + | |
4224 | +static void key_schedule_block( | |
4225 | + unsigned char *ck, // [In] ck[0]-ck[7] 8 bytes | Key. | |
4226 | + unsigned char *kk) // [Out] kk[0]-kk[55] 56 bytes | Key schedule. | |
4227 | +{ | |
4228 | + static const unsigned char key_perm[0x40] = { | |
4229 | + 0x12,0x24,0x09,0x07,0x2A,0x31,0x1D,0x15, 0x1C,0x36,0x3E,0x32,0x13,0x21,0x3B,0x40, | |
4230 | + 0x18,0x14,0x25,0x27,0x02,0x35,0x1B,0x01, 0x22,0x04,0x0D,0x0E,0x39,0x28,0x1A,0x29, | |
4231 | + 0x33,0x23,0x34,0x0C,0x16,0x30,0x1E,0x3A, 0x2D,0x1F,0x08,0x19,0x17,0x2F,0x3D,0x11, | |
4232 | + 0x3C,0x05,0x38,0x2B,0x0B,0x06,0x0A,0x2C, 0x20,0x3F,0x2E,0x0F,0x03,0x26,0x10,0x37, | |
4233 | + }; | |
4234 | + | |
4235 | + int i,j,k; | |
4236 | + int bit[64]; | |
4237 | + int newbit[64]; | |
4238 | + int kb[7][8]; | |
4239 | + | |
4240 | + // 56 steps | |
4241 | + // 56 key bytes kk(55)..kk(0) by key schedule from ck | |
4242 | + | |
4243 | + // kb(6,0) .. kb(6,7) = ck(0) .. ck(7) | |
4244 | + kb[6][0] = ck[0]; | |
4245 | + kb[6][1] = ck[1]; | |
4246 | + kb[6][2] = ck[2]; | |
4247 | + kb[6][3] = ck[3]; | |
4248 | + kb[6][4] = ck[4]; | |
4249 | + kb[6][5] = ck[5]; | |
4250 | + kb[6][6] = ck[6]; | |
4251 | + kb[6][7] = ck[7]; | |
4252 | + | |
4253 | + // calculate kb[5] .. kb[0] | |
4254 | + for(i=5; i>=0; i--){ | |
4255 | + // 64 bit perm on kb | |
4256 | + for(j=0; j<8; j++){ | |
4257 | + for(k=0; k<8; k++){ | |
4258 | + bit[j*8+k] = (kb[i+1][j] >> (7-k)) & 1; | |
4259 | + newbit[key_perm[j*8+k]-1] = bit[j*8+k]; | |
4260 | + } | |
4261 | + } | |
4262 | + for(j=0; j<8; j++){ | |
4263 | + kb[i][j] = 0; | |
4264 | + for(k=0; k<8; k++){ | |
4265 | + kb[i][j] |= newbit[j*8+k] << (7-k); | |
4266 | + } | |
4267 | + } | |
4268 | + } | |
4269 | + | |
4270 | + // xor to give kk | |
4271 | + for(i=0; i<7; i++){ | |
4272 | + for(j=0; j<8; j++){ | |
4273 | + kk[i*8+j] = kb[i][j] ^ i; | |
4274 | + } | |
4275 | + } | |
4276 | + | |
4277 | +} | |
4278 | + | |
4279 | +//-----block utils | |
4280 | + | |
4281 | +static inline __attribute__((always_inline)) void trasp_N_8 (unsigned char *in,unsigned char* out,int count){ | |
4282 | + int *ri=(int *)in; | |
4283 | + int *ibi=(int *)out; | |
4284 | + int j,i,k,g; | |
4285 | + // copy and first step | |
4286 | + for(g=0;g<count;g++){ | |
4287 | + ri[g]=ibi[2*g]; | |
4288 | + ri[GROUP_PARALLELISM+g]=ibi[2*g+1]; | |
4289 | + } | |
4290 | +//dump_mem("NE1 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
4291 | +// now 01230123 | |
4292 | +#define INTS_PER_ROW (GROUP_PARALLELISM/8*2) | |
4293 | + for(j=0;j<8;j+=4){ | |
4294 | + for(i=0;i<2;i++){ | |
4295 | + for(k=0;k<INTS_PER_ROW;k++){ | |
4296 | + unsigned int t,b; | |
4297 | + t=ri[INTS_PER_ROW*(j+i)+k]; | |
4298 | + b=ri[INTS_PER_ROW*(j+i+2)+k]; | |
4299 | + ri[INTS_PER_ROW*(j+i)+k]= (t&0x0000ffff) | ((b )<<16); | |
4300 | + ri[INTS_PER_ROW*(j+i+2)+k]= ((t )>>16) | (b&0xffff0000) ; | |
4301 | + } | |
4302 | + } | |
4303 | + } | |
4304 | +//dump_mem("NE2 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
4305 | +// now 01010101 | |
4306 | + for(j=0;j<8;j+=2){ | |
4307 | + for(i=0;i<1;i++){ | |
4308 | + for(k=0;k<INTS_PER_ROW;k++){ | |
4309 | + unsigned int t,b; | |
4310 | + t=ri[INTS_PER_ROW*(j+i)+k]; | |
4311 | + b=ri[INTS_PER_ROW*(j+i+1)+k]; | |
4312 | + ri[INTS_PER_ROW*(j+i)+k]= (t&0x00ff00ff) | ((b&0x00ff00ff)<<8); | |
4313 | + ri[INTS_PER_ROW*(j+i+1)+k]= ((t&0xff00ff00)>>8) | (b&0xff00ff00); | |
4314 | + } | |
4315 | + } | |
4316 | + } | |
4317 | +//dump_mem("NE3 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
4318 | +// now 00000000 | |
4319 | +} | |
4320 | + | |
4321 | +static inline __attribute__((always_inline)) void trasp_8_N (unsigned char *in,unsigned char* out,int count){ | |
4322 | + int *ri=(int *)in; | |
4323 | + int *bdi=(int *)out; | |
4324 | + int j,i,k,g; | |
4325 | +#define INTS_PER_ROW (GROUP_PARALLELISM/8*2) | |
4326 | +//dump_mem("NE1 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
4327 | +// now 00000000 | |
4328 | + for(j=0;j<8;j+=2){ | |
4329 | + for(i=0;i<1;i++){ | |
4330 | + for(k=0;k<INTS_PER_ROW;k++){ | |
4331 | + unsigned int t,b; | |
4332 | + t=ri[INTS_PER_ROW*(j+i)+k]; | |
4333 | + b=ri[INTS_PER_ROW*(j+i+1)+k]; | |
4334 | + ri[INTS_PER_ROW*(j+i)+k]= (t&0x00ff00ff) | ((b&0x00ff00ff)<<8); | |
4335 | + ri[INTS_PER_ROW*(j+i+1)+k]= ((t&0xff00ff00)>>8) | (b&0xff00ff00); | |
4336 | + } | |
4337 | + } | |
4338 | + } | |
4339 | +//dump_mem("NE2 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
4340 | +// now 01010101 | |
4341 | + for(j=0;j<8;j+=4){ | |
4342 | + for(i=0;i<2;i++){ | |
4343 | + for(k=0;k<INTS_PER_ROW;k++){ | |
4344 | + unsigned int t,b; | |
4345 | + t=ri[INTS_PER_ROW*(j+i)+k]; | |
4346 | + b=ri[INTS_PER_ROW*(j+i+2)+k]; | |
4347 | + ri[INTS_PER_ROW*(j+i)+k]= (t&0x0000ffff) | ((b )<<16); | |
4348 | + ri[INTS_PER_ROW*(j+i+2)+k]= ((t )>>16) | (b&0xffff0000) ; | |
4349 | + } | |
4350 | + } | |
4351 | + } | |
4352 | +//dump_mem("NE3 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
4353 | +// now 01230123 | |
4354 | + for(g=0;g<count;g++){ | |
4355 | + bdi[2*g]=ri[g]; | |
4356 | + bdi[2*g+1]=ri[GROUP_PARALLELISM+g]; | |
4357 | + } | |
4358 | +} | |
4359 | + | |
4360 | +//-----block main function | |
4361 | + | |
4362 | +// block group | |
4363 | +static void block_decypher_group( | |
4364 | + batch *kkmulti, // [In] kkmulti[0]-kkmulti[55] 56 batches | Key schedule (each batch has repeated equal bytes). | |
4365 | + unsigned char *ib, // [In] (ib0,ib1,...ib7)...x32 32*8 bytes | Initialization vector. | |
4366 | + unsigned char *bd, // [Out] (bd0,bd1,...bd7)...x32 32*8 bytes | Block decipher. | |
4367 | + int count) | |
4368 | +{ | |
4369 | + // int is faster than unsigned char. apparently not | |
4370 | + static const unsigned char block_sbox[0x100] = { | |
4371 | + 0x3A,0xEA,0x68,0xFE,0x33,0xE9,0x88,0x1A, 0x83,0xCF,0xE1,0x7F,0xBA,0xE2,0x38,0x12, | |
4372 | + 0xE8,0x27,0x61,0x95,0x0C,0x36,0xE5,0x70, 0xA2,0x06,0x82,0x7C,0x17,0xA3,0x26,0x49, | |
4373 | + 0xBE,0x7A,0x6D,0x47,0xC1,0x51,0x8F,0xF3, 0xCC,0x5B,0x67,0xBD,0xCD,0x18,0x08,0xC9, | |
4374 | + 0xFF,0x69,0xEF,0x03,0x4E,0x48,0x4A,0x84, 0x3F,0xB4,0x10,0x04,0xDC,0xF5,0x5C,0xC6, | |
4375 | + 0x16,0xAB,0xAC,0x4C,0xF1,0x6A,0x2F,0x3C, 0x3B,0xD4,0xD5,0x94,0xD0,0xC4,0x63,0x62, | |
4376 | + 0x71,0xA1,0xF9,0x4F,0x2E,0xAA,0xC5,0x56, 0xE3,0x39,0x93,0xCE,0x65,0x64,0xE4,0x58, | |
4377 | + 0x6C,0x19,0x42,0x79,0xDD,0xEE,0x96,0xF6, 0x8A,0xEC,0x1E,0x85,0x53,0x45,0xDE,0xBB, | |
4378 | + 0x7E,0x0A,0x9A,0x13,0x2A,0x9D,0xC2,0x5E, 0x5A,0x1F,0x32,0x35,0x9C,0xA8,0x73,0x30, | |
4379 | + | |
4380 | + 0x29,0x3D,0xE7,0x92,0x87,0x1B,0x2B,0x4B, 0xA5,0x57,0x97,0x40,0x15,0xE6,0xBC,0x0E, | |
4381 | + 0xEB,0xC3,0x34,0x2D,0xB8,0x44,0x25,0xA4, 0x1C,0xC7,0x23,0xED,0x90,0x6E,0x50,0x00, | |
4382 | + 0x99,0x9E,0x4D,0xD9,0xDA,0x8D,0x6F,0x5F, 0x3E,0xD7,0x21,0x74,0x86,0xDF,0x6B,0x05, | |
4383 | + 0x8E,0x5D,0x37,0x11,0xD2,0x28,0x75,0xD6, 0xA7,0x77,0x24,0xBF,0xF0,0xB0,0x02,0xB7, | |
4384 | + 0xF8,0xFC,0x81,0x09,0xB1,0x01,0x76,0x91, 0x7D,0x0F,0xC8,0xA0,0xF2,0xCB,0x78,0x60, | |
4385 | + 0xD1,0xF7,0xE0,0xB5,0x98,0x22,0xB3,0x20, 0x1D,0xA6,0xDB,0x7B,0x59,0x9F,0xAE,0x31, | |
4386 | + 0xFB,0xD3,0xB6,0xCA,0x43,0x72,0x07,0xF4, 0xD8,0x41,0x14,0x55,0x0D,0x54,0x8B,0xB9, | |
4387 | + 0xAD,0x46,0x0B,0xAF,0x80,0x52,0x2C,0xFA, 0x8C,0x89,0x66,0xFD,0xB2,0xA9,0x9B,0xC0, | |
4388 | + }; | |
4389 | + unsigned char r[GROUP_PARALLELISM*(8+56)]; /* 56 because we will move back in memory while looping */ | |
4390 | + unsigned char sbox_in[GROUP_PARALLELISM],sbox_out[GROUP_PARALLELISM],perm_out[GROUP_PARALLELISM]; | |
4391 | + int roff; | |
4392 | + int i,g,count_all=GROUP_PARALLELISM; | |
4393 | + | |
4394 | + roff=GROUP_PARALLELISM*56; | |
4395 | + | |
4396 | +#define FASTTRASP1 | |
4397 | +#ifndef FASTTRASP1 | |
4398 | + for(g=0;g<count;g++){ | |
4399 | + // Init registers | |
4400 | + int j; | |
4401 | + for(j=0;j<8;j++){ | |
4402 | + r[roff+GROUP_PARALLELISM*j+g]=ib[8*g+j]; | |
4403 | + } | |
4404 | + } | |
4405 | +#else | |
4406 | + trasp_N_8((unsigned char *)&r[roff],(unsigned char *)ib,count); | |
4407 | +#endif | |
4408 | +//dump_mem("OLD r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
4409 | + | |
4410 | + // loop over kk[55]..kk[0] | |
4411 | + for(i=55;i>=0;i--){ | |
4412 | + { | |
4413 | + batch tkkmulti=kkmulti[i]; | |
4414 | + batch *si=(batch *)sbox_in; | |
4415 | + batch *r6_N=(batch *)(r+roff+GROUP_PARALLELISM*6); | |
4416 | + for(g=0;g<count_all/BYTES_PER_BATCH;g++){ | |
4417 | + si[g]=B_FFXOR(tkkmulti,r6_N[g]); //FIXME: introduce FASTBATCH? | |
4418 | + } | |
4419 | + } | |
4420 | + | |
4421 | + // table lookup, this works on only one byte at a time | |
4422 | + // most difficult part of all | |
4423 | + // - can't be parallelized | |
4424 | + // - can't be synthetized through boolean terms (8 input bits are too many) | |
4425 | + for(g=0;g<count_all;g++){ | |
4426 | + sbox_out[g]=block_sbox[sbox_in[g]]; | |
4427 | + } | |
4428 | + | |
4429 | + // bit permutation | |
4430 | + { | |
4431 | + unsigned char *po=(unsigned char *)perm_out; | |
4432 | + unsigned char *so=(unsigned char *)sbox_out; | |
4433 | +//dump_mem("pre perm ",(unsigned char *)so,GROUP_PARALLELISM,GROUP_PARALLELISM); | |
4434 | + for(g=0;g<count_all;g+=BYTES_PER_BATCH){ | |
4435 | + batch in,out; | |
4436 | + in=*(batch *)&so[g]; | |
4437 | + | |
4438 | + out=B_FFOR( | |
4439 | + B_FFOR( | |
4440 | + B_FFOR( | |
4441 | + B_FFOR( | |
4442 | + B_FFOR( | |
4443 | + B_FFSH8L(B_FFAND(in,B_FFN_ALL_29()),1), | |
4444 | + B_FFSH8L(B_FFAND(in,B_FFN_ALL_02()),6)), | |
4445 | + B_FFSH8L(B_FFAND(in,B_FFN_ALL_04()),3)), | |
4446 | + B_FFSH8R(B_FFAND(in,B_FFN_ALL_10()),2)), | |
4447 | + B_FFSH8R(B_FFAND(in,B_FFN_ALL_40()),6)), | |
4448 | + B_FFSH8R(B_FFAND(in,B_FFN_ALL_80()),4)); | |
4449 | + | |
4450 | + *(batch *)&po[g]=out; | |
4451 | + } | |
4452 | +//dump_mem("post perm",(unsigned char *)po,GROUP_PARALLELISM,GROUP_PARALLELISM); | |
4453 | + } | |
4454 | + | |
4455 | + roff-=GROUP_PARALLELISM; /* virtual shift of registers */ | |
4456 | + | |
4457 | +#if 0 | |
4458 | +/* one by one */ | |
4459 | + for(g=0;g<count_all;g++){ | |
4460 | + r[roff+GROUP_PARALLELISM*0+g]=r[roff+GROUP_PARALLELISM*8+g]^sbox_out[g]; | |
4461 | + r[roff+GROUP_PARALLELISM*6+g]^=perm_out[g]; | |
4462 | + r[roff+GROUP_PARALLELISM*4+g]^=r[roff+GROUP_PARALLELISM*0+g]; | |
4463 | + r[roff+GROUP_PARALLELISM*3+g]^=r[roff+GROUP_PARALLELISM*0+g]; | |
4464 | + r[roff+GROUP_PARALLELISM*2+g]^=r[roff+GROUP_PARALLELISM*0+g]; | |
4465 | + } | |
4466 | +#else | |
4467 | + for(g=0;g<count_all;g+=BEST_SPAN){ | |
4468 | + XOR_BEST_BY(&r[roff+GROUP_PARALLELISM*0+g],&r[roff+GROUP_PARALLELISM*8+g],&sbox_out[g]); | |
4469 | + XOREQ_BEST_BY(&r[roff+GROUP_PARALLELISM*6+g],&perm_out[g]); | |
4470 | + XOREQ_BEST_BY(&r[roff+GROUP_PARALLELISM*4+g],&r[roff+GROUP_PARALLELISM*0+g]); | |
4471 | + XOREQ_BEST_BY(&r[roff+GROUP_PARALLELISM*3+g],&r[roff+GROUP_PARALLELISM*0+g]); | |
4472 | + XOREQ_BEST_BY(&r[roff+GROUP_PARALLELISM*2+g],&r[roff+GROUP_PARALLELISM*0+g]); | |
4473 | + } | |
4474 | +#endif | |
4475 | + } | |
4476 | + | |
4477 | +#define FASTTRASP2 | |
4478 | +#ifndef FASTTRASP2 | |
4479 | + for(g=0;g<count;g++){ | |
4480 | + // Copy results | |
4481 | + int j; | |
4482 | + for(j=0;j<8;j++){ | |
4483 | + bd[8*g+j]=r[roff+GROUP_PARALLELISM*j+g]; | |
4484 | + } | |
4485 | + } | |
4486 | +#else | |
4487 | + trasp_8_N((unsigned char *)&r[roff],(unsigned char *)bd,count); | |
4488 | +#endif | |
4489 | +} | |
4490 | + | |
4491 | +//-----------------------------------EXTERNAL INTERFACE | |
4492 | + | |
4493 | +//-----get internal parallelism | |
4494 | + | |
4495 | +int get_internal_parallelism(void){ | |
4496 | + return GROUP_PARALLELISM; | |
4497 | +} | |
4498 | + | |
4499 | +//-----get suggested cluster size | |
4500 | + | |
4501 | +int get_suggested_cluster_size(void){ | |
4502 | + int r; | |
4503 | + r=GROUP_PARALLELISM+GROUP_PARALLELISM/10; | |
4504 | + if(r<GROUP_PARALLELISM+5) r=GROUP_PARALLELISM+5; | |
4505 | + return r; | |
4506 | +} | |
4507 | + | |
4508 | +//-----set control words | |
4509 | + | |
4510 | +void set_control_words(unsigned char *ev, unsigned char *od){ | |
4511 | + // could be made faster, but is not run often | |
4512 | + int bi,by; | |
4513 | + int i,j; | |
4514 | +// key | |
4515 | + memcpy(keys.even.ck,ev,8); | |
4516 | + memcpy(keys.odd.ck,od,8); | |
4517 | +// precalculations for stream | |
4518 | + key_schedule_stream(keys.even.ck,keys.even.iA,keys.even.iB); | |
4519 | + key_schedule_stream(keys.odd.ck,keys.odd.iA,keys.odd.iB); | |
4520 | + for(by=0;by<8;by++){ | |
4521 | + for(bi=0;bi<8;bi++){ | |
4522 | + keys.even.ck_g[by][bi]=(keys.even.ck[by]&(1<<bi))?FF1():FF0(); | |
4523 | + keys.odd.ck_g[by][bi]=(keys.odd.ck[by]&(1<<bi))?FF1():FF0(); | |
4524 | + } | |
4525 | + } | |
4526 | + for(by=0;by<8;by++){ | |
4527 | + for(bi=0;bi<4;bi++){ | |
4528 | + keys.even.iA_g[by][bi]=(keys.even.iA[by]&(1<<bi))?FF1():FF0(); | |
4529 | + keys.odd.iA_g[by][bi]=(keys.odd.iA[by]&(1<<bi))?FF1():FF0(); | |
4530 | + keys.even.iB_g[by][bi]=(keys.even.iB[by]&(1<<bi))?FF1():FF0(); | |
4531 | + keys.odd.iB_g[by][bi]=(keys.odd.iB[by]&(1<<bi))?FF1():FF0(); | |
4532 | + } | |
4533 | + } | |
4534 | +// precalculations for block | |
4535 | + key_schedule_block(keys.even.ck,keys.even.kk); | |
4536 | + key_schedule_block(keys.odd.ck,keys.odd.kk); | |
4537 | + for(i=0;i<56;i++){ | |
4538 | + for(j=0;j<BYTES_PER_BATCH;j++){ | |
4539 | + *(((unsigned char *)&keys.even.kkmulti[i])+j)=keys.even.kk[i]; | |
4540 | + *(((unsigned char *)&keys.odd.kkmulti[i])+j)=keys.odd.kk[i]; | |
4541 | + } | |
4542 | + } | |
4543 | +} | |
4544 | + | |
4545 | +//-----get control words | |
4546 | + | |
4547 | +void get_control_words(unsigned char *even, unsigned char *odd){ | |
4548 | + memcpy(even,keys.even.ck,8); | |
4549 | + memcpy(odd,keys.odd.ck,8); | |
4550 | +} | |
4551 | + | |
4552 | +//----- decrypt | |
4553 | + | |
4554 | +int decrypt_packets(unsigned char **cluster){ | |
4555 | + // statistics, currently unused | |
4556 | + int stat_no_scramble=0; | |
4557 | + int stat_reserved=0; | |
4558 | + int stat_decrypted[2]={0,0}; | |
4559 | + int stat_decrypted_mini=0; | |
4560 | + unsigned char **clst; | |
4561 | + unsigned char **clst2; | |
4562 | + int grouped; | |
4563 | + int group_ev_od; | |
4564 | + int advanced; | |
4565 | + int can_advance; | |
4566 | + unsigned char *g_pkt[GROUP_PARALLELISM]; | |
4567 | + int g_len[GROUP_PARALLELISM]; | |
4568 | + int g_offset[GROUP_PARALLELISM]; | |
4569 | + int g_n[GROUP_PARALLELISM]; | |
4570 | + int g_residue[GROUP_PARALLELISM]; | |
4571 | + unsigned char *pkt; | |
4572 | + int xc0,ev_od,len,offset,n,residue; | |
4573 | + struct csa_key_t* k; | |
4574 | + int i,j,iter,g; | |
4575 | + int t23,tsmall; | |
4576 | + int alive[24]; | |
4577 | +//icc craziness int pad1=0; //////////align! FIXME | |
4578 | + unsigned char *encp[GROUP_PARALLELISM]; | |
4579 | + unsigned char stream_in[GROUP_PARALLELISM*8]; | |
4580 | + unsigned char stream_out[GROUP_PARALLELISM*8]; | |
4581 | + unsigned char ib[GROUP_PARALLELISM*8]; | |
4582 | + unsigned char block_out[GROUP_PARALLELISM*8]; | |
4583 | + | |
4584 | +//icc craziness i=(int)&pad1;//////////align!!! FIXME | |
4585 | + | |
4586 | + // build a list of packets to be processed | |
4587 | + clst=cluster; | |
4588 | + grouped=0; | |
4589 | + advanced=0; | |
4590 | + can_advance=1; | |
4591 | + group_ev_od=-1; // silence incorrect compiler warning | |
4592 | + pkt=*clst; | |
4593 | + do{ // find a new packet | |
4594 | + if(grouped==GROUP_PARALLELISM){ | |
4595 | + // full | |
4596 | + break; | |
4597 | + } | |
4598 | + if(pkt==NULL){ | |
4599 | + // no more ranges | |
4600 | + break; | |
4601 | + } | |
4602 | + if(pkt>=*(clst+1)){ | |
4603 | + // out of this range, try next | |
4604 | + clst++;clst++; | |
4605 | + pkt=*clst; | |
4606 | + continue; | |
4607 | + } | |
4608 | + | |
4609 | + do{ // handle this packet | |
4610 | + xc0=pkt[3]&0xc0; | |
4611 | + if(xc0==0x00){ | |
4612 | + advanced+=can_advance; | |
4613 | + stat_no_scramble++; | |
4614 | + break; | |
4615 | + } | |
4616 | + if(xc0==0x40){ | |
4617 | + advanced+=can_advance; | |
4618 | + stat_reserved++; | |
4619 | + break; | |
4620 | + } | |
4621 | + if(xc0==0x80||xc0==0xc0){ // encrypted | |
4622 | + ev_od=(xc0&0x40)>>6; // 0 even, 1 odd | |
4623 | + if(grouped==0) group_ev_od=ev_od; // this group will be all even (or odd) | |
4624 | + if(group_ev_od==ev_od){ // could be added to group | |
4625 | + pkt[3]&=0x3f; // consider it decrypted now | |
4626 | + if(pkt[3]&0x20){ // incomplete packet | |
4627 | + offset=4+pkt[4]+1; | |
4628 | + len=188-offset; | |
4629 | + n=len>>3; | |
4630 | + residue=len-(n<<3); | |
4631 | + if(n==0){ // decrypted==encrypted! | |
4632 | + advanced+=can_advance; | |
4633 | + stat_decrypted_mini++; | |
4634 | + break; // this doesn't need more processing | |
4635 | + } | |
4636 | + }else{ | |
4637 | + len=184; | |
4638 | + offset=4; | |
4639 | + n=23; | |
4640 | + residue=0; | |
4641 | + } | |
4642 | + g_pkt[grouped]=pkt; | |
4643 | + g_len[grouped]=len; | |
4644 | + g_offset[grouped]=offset; | |
4645 | + g_n[grouped]=n; | |
4646 | + g_residue[grouped]=residue; | |
4647 | + grouped++; | |
4648 | + advanced+=can_advance; | |
4649 | + stat_decrypted[ev_od]++; | |
4650 | + } | |
4651 | + else{ | |
4652 | + can_advance=0; | |
4653 | + break; // skip and go on | |
4654 | + } | |
4655 | + } | |
4656 | + } while(0); | |
4657 | + | |
4658 | + if(can_advance){ | |
4659 | + // move range start forward | |
4660 | + *clst+=188; | |
4661 | + } | |
4662 | + // next packet, if there is one | |
4663 | + pkt+=188; | |
4664 | + } while(1); | |
4665 | + | |
4666 | + // delete empty ranges and compact list | |
4667 | + clst2=cluster; | |
4668 | + for(clst=cluster;*clst!=NULL;clst+=2){ | |
4669 | + // if not empty | |
4670 | + if(*clst<*(clst+1)){ | |
4671 | + // it will remain | |
4672 | + *clst2=*clst; | |
4673 | + *(clst2+1)=*(clst+1); | |
4674 | + clst2+=2; | |
4675 | + } | |
4676 | + } | |
4677 | + *clst2=NULL; | |
4678 | + | |
4679 | + if(grouped==0){ | |
4680 | + // no processing needed | |
4681 | + return advanced; | |
4682 | + } | |
4683 | + | |
4684 | + // sort them, longest payload first | |
4685 | + // we expect many n=23 packets and a few n<23 | |
4686 | + // grouped is always <= GROUP_PARALLELISM | |
4687 | + | |
4688 | +#define g_swap(a,b) \ | |
4689 | + pkt=g_pkt[a]; \ | |
4690 | + g_pkt[a]=g_pkt[b]; \ | |
4691 | + g_pkt[b]=pkt; \ | |
4692 | +\ | |
4693 | + len=g_len[a]; \ | |
4694 | + g_len[a]=g_len[b]; \ | |
4695 | + g_len[b]=len; \ | |
4696 | +\ | |
4697 | + offset=g_offset[a]; \ | |
4698 | + g_offset[a]=g_offset[b]; \ | |
4699 | + g_offset[b]=offset; \ | |
4700 | +\ | |
4701 | + n=g_n[a]; \ | |
4702 | + g_n[a]=g_n[b]; \ | |
4703 | + g_n[b]=n; \ | |
4704 | +\ | |
4705 | + residue=g_residue[a]; \ | |
4706 | + g_residue[a]=g_residue[b]; \ | |
4707 | + g_residue[b]=residue; | |
4708 | + | |
4709 | + // step 1: move n=23 packets before small packets | |
4710 | + t23=0; | |
4711 | + tsmall=grouped-1; | |
4712 | + for(;;){ | |
4713 | + for(;t23<grouped;t23++){ | |
4714 | + if(g_n[t23]!=23) break; | |
4715 | + } | |
4716 | + | |
4717 | + for(;tsmall>=0;tsmall--){ | |
4718 | + if(g_n[tsmall]==23) break; | |
4719 | + } | |
4720 | + | |
4721 | + if(tsmall-t23<1) break; | |
4722 | + | |
4723 | + | |
4724 | + g_swap(t23,tsmall); | |
4725 | + | |
4726 | + t23++; | |
4727 | + tsmall--; | |
4728 | + } | |
4729 | + | |
4730 | + // step 2: sort small packets in decreasing order of n (bubble sort is enough) | |
4731 | + for(i=t23;i<grouped;i++){ | |
4732 | + for(j=i+1;j<grouped;j++){ | |
4733 | + if(g_n[j]>g_n[i]){ | |
4734 | + g_swap(i,j); | |
4735 | + } | |
4736 | + } | |
4737 | + } | |
4738 | + | |
4739 | + // we need to know how many packets need 23 iterations, how many 22... | |
4740 | + for(i=0;i<=23;i++){ | |
4741 | + alive[i]=0; | |
4742 | + } | |
4743 | + // count | |
4744 | + alive[23-1]=t23; | |
4745 | + for(i=t23;i<grouped;i++){ | |
4746 | + alive[g_n[i]-1]++; | |
4747 | + } | |
4748 | + // integrate | |
4749 | + for(i=22;i>=0;i--){ | |
4750 | + alive[i]+=alive[i+1]; | |
4751 | + } | |
4752 | + | |
4753 | + // choose key | |
4754 | + if(group_ev_od==0){ | |
4755 | + k=&keys.even; | |
4756 | + } | |
4757 | + else{ | |
4758 | + k=&keys.odd; | |
4759 | + } | |
4760 | + | |
4761 | + //INIT | |
4762 | +#define INITIALIZE_UNUSED_INPUT | |
4763 | +#ifdef INITIALIZE_UNUSED_INPUT | |
4764 | +// unnecessary zeroing. | |
4765 | +// without this, we operate on uninitialized memory | |
4766 | +// when grouped<GROUP_PARALLELISM, but it's not a problem, | |
4767 | +// as final results will be discarded. | |
4768 | +// random data makes debugging sessions difficult. | |
4769 | + for(j=0;j<GROUP_PARALLELISM*8;j++) stream_in[j]=0; | |
4770 | +#else | |
4771 | +#endif | |
4772 | + | |
4773 | + for(g=0;g<grouped;g++){ | |
4774 | + encp[g]=g_pkt[g]; | |
4775 | + encp[g]+=g_offset[g]; // skip header | |
4776 | + FFTABLEIN(stream_in,g,encp[g]); | |
4777 | + } | |
4778 | +//dump_mem("stream_in",stream_in,GROUP_PARALLELISM*8,BYPG); | |
4779 | + | |
4780 | + | |
4781 | + // ITER 0 | |
4782 | + iter=0; | |
4783 | + stream_cypher_group_init(k->iA_g,k->iB_g,stream_in); | |
4784 | + // fill first ib | |
4785 | + for(g=0;g<alive[iter];g++){ | |
4786 | + COPY_8_BY(ib+8*g,encp[g]); | |
4787 | + } | |
4788 | + // ITER 1..N-1 | |
4789 | + for (iter=1;iter<23&&alive[iter-1]>0;iter++){ | |
4790 | + // alive and just dead packets: calc block | |
4791 | + block_decypher_group(k->kkmulti,ib,block_out,alive[iter-1]); | |
4792 | + // all packets (dead too): calc stream | |
4793 | + stream_cypher_group_normal(stream_out); | |
4794 | +//dump_mem("stream_out",stream_out,GROUP_PARALLELISM*8,BYPG); | |
4795 | + | |
4796 | + // alive packets: calc ib | |
4797 | + for(g=0;g<alive[iter];g++){ | |
4798 | + FFTABLEOUT(ib+8*g,stream_out,g); | |
4799 | +// XOREQ8BY gcc bug? 2x4 ok, 8 ko UPDATE: result ok but speed 1-2% slower (!!!???) | |
4800 | +#if 1 | |
4801 | + XOREQ_4_BY(ib+8*g,encp[g]+8); | |
4802 | + XOREQ_4_BY(ib+8*g+4,encp[g]+8+4); | |
4803 | +#else | |
4804 | + XOREQ_8_BY(ib+8*g,encp[g]+8); | |
4805 | +#endif | |
4806 | + } | |
4807 | + // alive packets: decrypt data | |
4808 | + for(g=0;g<alive[iter];g++){ | |
4809 | + XOR_8_BY(encp[g],ib+8*g,block_out+8*g); | |
4810 | + } | |
4811 | + // just dead packets: write decrypted data | |
4812 | + for(g=alive[iter];g<alive[iter-1];g++){ | |
4813 | + COPY_8_BY(encp[g],block_out+8*g); | |
4814 | + } | |
4815 | + // just dead packets: decrypt residue | |
4816 | + for(g=alive[iter];g<alive[iter-1];g++){ | |
4817 | + FFTABLEOUTXORNBY(g_residue[g],encp[g]+8,stream_out,g); | |
4818 | + } | |
4819 | + // alive packets: pointers++ | |
4820 | + for(g=0;g<alive[iter];g++) encp[g]+=8; | |
4821 | + }; | |
4822 | + // ITER N | |
4823 | + iter=23; | |
4824 | + // calc block | |
4825 | + block_decypher_group(k->kkmulti,ib,block_out,alive[iter-1]); | |
4826 | + // just dead packets: write decrypted data | |
4827 | + for(g=alive[iter];g<alive[iter-1];g++){ | |
4828 | + COPY_8_BY(encp[g],block_out+8*g); | |
4829 | + } | |
4830 | + // no residue possible | |
4831 | + // so do nothing | |
4832 | + | |
4833 | + | |
4834 | + M_EMPTY(); // restore CPU multimedia state | |
4835 | + | |
4836 | + return advanced; | |
4837 | +} | |
4838 | Index: libs/libmythtv/FFdecsa/FFdecsa.c | |
4839 | =================================================================== | |
4840 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
4841 | +++ libs/libmythtv/FFdecsa/FFdecsa.c 2006-06-20 17:36:06.000000000 -0400 | |
4842 | @@ -0,0 +1,866 @@ | |
4843 | +/* FFdecsa -- fast decsa algorithm | |
4844 | + * | |
4845 | + * Copyright (C) 2003-2004 fatih89r | |
4846 | + * | |
4847 | + * This program is free software; you can redistribute it and/or modify | |
4848 | + * it under the terms of the GNU General Public License as published by | |
4849 | + * the Free Software Foundation; either version 2 of the License, or | |
4850 | + * (at your option) any later version. | |
4851 | + * | |
4852 | + * This program is distributed in the hope that it will be useful, | |
4853 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
4854 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
4855 | + * GNU General Public License for more details. | |
4856 | + * | |
4857 | + * You should have received a copy of the GNU General Public License | |
4858 | + * along with this program; if not, write to the Free Software | |
4859 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
4860 | + */ | |
4861 | + | |
4862 | + | |
4863 | +#include <sys/types.h> | |
4864 | +#include <string.h> | |
4865 | +#include <stdio.h> | |
4866 | +#include <stdlib.h> | |
4867 | + | |
4868 | +#include "FFdecsa.h" | |
4869 | + | |
4870 | +#ifndef NULL | |
4871 | +#define NULL 0 | |
4872 | +#endif | |
4873 | + | |
4874 | +//#define DEBUG | |
4875 | +#ifdef DEBUG | |
4876 | +#define DBG(a) a | |
4877 | +#else | |
4878 | +#define DBG(a) | |
4879 | +#endif | |
4880 | + | |
4881 | +//// parallelization stuff, large speed differences are possible | |
4882 | +// possible choices | |
4883 | +#define PARALLEL_32_4CHAR 320 | |
4884 | +#define PARALLEL_32_4CHARA 321 | |
4885 | +#define PARALLEL_32_INT 322 | |
4886 | +#define PARALLEL_64_8CHAR 640 | |
4887 | +#define PARALLEL_64_8CHARA 641 | |
4888 | +#define PARALLEL_64_2INT 642 | |
4889 | +#define PARALLEL_64_LONG 643 | |
4890 | +#define PARALLEL_64_MMX 644 | |
4891 | +#define PARALLEL_128_16CHAR 1280 | |
4892 | +#define PARALLEL_128_16CHARA 1281 | |
4893 | +#define PARALLEL_128_4INT 1282 | |
4894 | +#define PARALLEL_128_2LONG 1283 | |
4895 | +#define PARALLEL_128_2MMX 1284 | |
4896 | +#define PARALLEL_128_SSE 1285 | |
4897 | + | |
4898 | +//////// our choice //////////////// our choice //////////////// our choice //////////////// our choice //////// | |
4899 | +#ifndef PARALLEL_MODE | |
4900 | +#define PARALLEL_MODE PARALLEL_32_INT | |
4901 | +#endif | |
4902 | +//////// our choice //////////////// our choice //////////////// our choice //////////////// our choice //////// | |
4903 | + | |
4904 | +#include "parallel_generic.h" | |
4905 | +//// conditionals | |
4906 | +#if PARALLEL_MODE==PARALLEL_32_4CHAR | |
4907 | +#include "parallel_032_4char.h" | |
4908 | +#elif PARALLEL_MODE==PARALLEL_32_4CHARA | |
4909 | +#include "parallel_032_4charA.h" | |
4910 | +#elif PARALLEL_MODE==PARALLEL_32_INT | |
4911 | +#include "parallel_032_int.h" | |
4912 | +#elif PARALLEL_MODE==PARALLEL_64_8CHAR | |
4913 | +#include "parallel_064_8char.h" | |
4914 | +#elif PARALLEL_MODE==PARALLEL_64_8CHARA | |
4915 | +#include "parallel_064_8charA.h" | |
4916 | +#elif PARALLEL_MODE==PARALLEL_64_2INT | |
4917 | +#include "parallel_064_2int.h" | |
4918 | +#elif PARALLEL_MODE==PARALLEL_64_LONG | |
4919 | +#include "parallel_064_long.h" | |
4920 | +#elif PARALLEL_MODE==PARALLEL_64_MMX | |
4921 | +#include "parallel_064_mmx.h" | |
4922 | +#elif PARALLEL_MODE==PARALLEL_128_16CHAR | |
4923 | +#include "parallel_128_16char.h" | |
4924 | +#elif PARALLEL_MODE==PARALLEL_128_16CHARA | |
4925 | +#include "parallel_128_16charA.h" | |
4926 | +#elif PARALLEL_MODE==PARALLEL_128_4INT | |
4927 | +#include "parallel_128_4int.h" | |
4928 | +#elif PARALLEL_MODE==PARALLEL_128_2LONG | |
4929 | +#include "parallel_128_2long.h" | |
4930 | +#elif PARALLEL_MODE==PARALLEL_128_2MMX | |
4931 | +#include "parallel_128_2mmx.h" | |
4932 | +#elif PARALLEL_MODE==PARALLEL_128_SSE | |
4933 | +#include "parallel_128_sse.h" | |
4934 | +#else | |
4935 | +#error "unknown/undefined parallel mode" | |
4936 | +#endif | |
4937 | + | |
4938 | +// stuff depending on conditionals | |
4939 | + | |
4940 | +#define BYTES_PER_GROUP (GROUP_PARALLELISM/8) | |
4941 | +#define BYPG BYTES_PER_GROUP | |
4942 | +#define BITS_PER_GROUP GROUP_PARALLELISM | |
4943 | +#define BIPG BITS_PER_GROUP | |
4944 | + | |
4945 | + | |
4946 | +//// debug tool | |
4947 | + | |
4948 | +static void dump_mem(const char *string, const unsigned char *p, int len, int linelen){ | |
4949 | + int i; | |
4950 | + for(i=0;i<len;i++){ | |
4951 | + if(i%linelen==0&&i) fprintf(stderr,"\n"); | |
4952 | + if(i%linelen==0) fprintf(stderr,"%s %08x:",string,i); | |
4953 | + else{ | |
4954 | + if(i%8==0) fprintf(stderr," "); | |
4955 | + if(i%4==0) fprintf(stderr," "); | |
4956 | + } | |
4957 | + fprintf(stderr," %02x",p[i]); | |
4958 | + } | |
4959 | + if(i%linelen==0) fprintf(stderr,"\n"); | |
4960 | +} | |
4961 | + | |
4962 | +////////////////////////////////////////////////////////////////////////////////// | |
4963 | + | |
4964 | +struct csa_key_t{ | |
4965 | + unsigned char ck[8]; | |
4966 | +// used by stream | |
4967 | + int iA[8]; // iA[0] is for A1, iA[7] is for A8 | |
4968 | + int iB[8]; // iB[0] is for B1, iB[7] is for B8 | |
4969 | +// used by stream (group) | |
4970 | + group ck_g[8][8]; // [byte][bit:0=LSB,7=MSB] | |
4971 | + group iA_g[8][4]; // [0 for A1][0 for LSB] | |
4972 | + group iB_g[8][4]; // [0 for B1][0 for LSB] | |
4973 | +// used by block | |
4974 | + unsigned char kk[56]; | |
4975 | +// used by block (group) | |
4976 | + __attribute__((aligned(16))) batch kkmulti[56]; // many times the same byte in every batch | |
4977 | +}; | |
4978 | + | |
4979 | +struct csa_keys_t{ | |
4980 | + struct csa_key_t even; | |
4981 | + struct csa_key_t odd; | |
4982 | +}; | |
4983 | + | |
4984 | +//-----stream cypher | |
4985 | + | |
4986 | +//-----key schedule for stream decypher | |
4987 | +static void key_schedule_stream( | |
4988 | + unsigned char *ck, // [In] ck[0]-ck[7] 8 bytes | Key. | |
4989 | + int *iA, // [Out] iA[0]-iA[7] 8 nibbles | Key schedule. | |
4990 | + int *iB) // [Out] iB[0]-iB[7] 8 nibbles | Key schedule. | |
4991 | +{ | |
4992 | + iA[0]=(ck[0]>>4)&0xf; | |
4993 | + iA[1]=(ck[0] )&0xf; | |
4994 | + iA[2]=(ck[1]>>4)&0xf; | |
4995 | + iA[3]=(ck[1] )&0xf; | |
4996 | + iA[4]=(ck[2]>>4)&0xf; | |
4997 | + iA[5]=(ck[2] )&0xf; | |
4998 | + iA[6]=(ck[3]>>4)&0xf; | |
4999 | + iA[7]=(ck[3] )&0xf; | |
5000 | + iB[0]=(ck[4]>>4)&0xf; | |
5001 | + iB[1]=(ck[4] )&0xf; | |
5002 | + iB[2]=(ck[5]>>4)&0xf; | |
5003 | + iB[3]=(ck[5] )&0xf; | |
5004 | + iB[4]=(ck[6]>>4)&0xf; | |
5005 | + iB[5]=(ck[6] )&0xf; | |
5006 | + iB[6]=(ck[7]>>4)&0xf; | |
5007 | + iB[7]=(ck[7] )&0xf; | |
5008 | +} | |
5009 | + | |
5010 | +//----- stream main function | |
5011 | + | |
5012 | +#define STREAM_INIT | |
5013 | +#include "stream.c" | |
5014 | +#undef STREAM_INIT | |
5015 | + | |
5016 | +#define STREAM_NORMAL | |
5017 | +#include "stream.c" | |
5018 | +#undef STREAM_NORMAL | |
5019 | + | |
5020 | + | |
5021 | +//-----block decypher | |
5022 | + | |
5023 | +//-----key schedule for block decypher | |
5024 | + | |
5025 | +static void key_schedule_block( | |
5026 | + unsigned char *ck, // [In] ck[0]-ck[7] 8 bytes | Key. | |
5027 | + unsigned char *kk) // [Out] kk[0]-kk[55] 56 bytes | Key schedule. | |
5028 | +{ | |
5029 | + static const unsigned char key_perm[0x40] = { | |
5030 | + 0x12,0x24,0x09,0x07,0x2A,0x31,0x1D,0x15, 0x1C,0x36,0x3E,0x32,0x13,0x21,0x3B,0x40, | |
5031 | + 0x18,0x14,0x25,0x27,0x02,0x35,0x1B,0x01, 0x22,0x04,0x0D,0x0E,0x39,0x28,0x1A,0x29, | |
5032 | + 0x33,0x23,0x34,0x0C,0x16,0x30,0x1E,0x3A, 0x2D,0x1F,0x08,0x19,0x17,0x2F,0x3D,0x11, | |
5033 | + 0x3C,0x05,0x38,0x2B,0x0B,0x06,0x0A,0x2C, 0x20,0x3F,0x2E,0x0F,0x03,0x26,0x10,0x37, | |
5034 | + }; | |
5035 | + | |
5036 | + int i,j,k; | |
5037 | + int bit[64]; | |
5038 | + int newbit[64]; | |
5039 | + int kb[7][8]; | |
5040 | + | |
5041 | + // 56 steps | |
5042 | + // 56 key bytes kk(55)..kk(0) by key schedule from ck | |
5043 | + | |
5044 | + // kb(6,0) .. kb(6,7) = ck(0) .. ck(7) | |
5045 | + kb[6][0] = ck[0]; | |
5046 | + kb[6][1] = ck[1]; | |
5047 | + kb[6][2] = ck[2]; | |
5048 | + kb[6][3] = ck[3]; | |
5049 | + kb[6][4] = ck[4]; | |
5050 | + kb[6][5] = ck[5]; | |
5051 | + kb[6][6] = ck[6]; | |
5052 | + kb[6][7] = ck[7]; | |
5053 | + | |
5054 | + // calculate kb[5] .. kb[0] | |
5055 | + for(i=5; i>=0; i--){ | |
5056 | + // 64 bit perm on kb | |
5057 | + for(j=0; j<8; j++){ | |
5058 | + for(k=0; k<8; k++){ | |
5059 | + bit[j*8+k] = (kb[i+1][j] >> (7-k)) & 1; | |
5060 | + newbit[key_perm[j*8+k]-1] = bit[j*8+k]; | |
5061 | + } | |
5062 | + } | |
5063 | + for(j=0; j<8; j++){ | |
5064 | + kb[i][j] = 0; | |
5065 | + for(k=0; k<8; k++){ | |
5066 | + kb[i][j] |= newbit[j*8+k] << (7-k); | |
5067 | + } | |
5068 | + } | |
5069 | + } | |
5070 | + | |
5071 | + // xor to give kk | |
5072 | + for(i=0; i<7; i++){ | |
5073 | + for(j=0; j<8; j++){ | |
5074 | + kk[i*8+j] = kb[i][j] ^ i; | |
5075 | + } | |
5076 | + } | |
5077 | + | |
5078 | +} | |
5079 | + | |
5080 | +//-----block utils | |
5081 | + | |
5082 | +static inline __attribute__((always_inline)) void trasp_N_8 (unsigned char *in,unsigned char* out,int count){ | |
5083 | + int *ri=(int *)in; | |
5084 | + int *ibi=(int *)out; | |
5085 | + int j,i,k,g; | |
5086 | + // copy and first step | |
5087 | + for(g=0;g<count;g++){ | |
5088 | + ri[g]=ibi[2*g]; | |
5089 | + ri[GROUP_PARALLELISM+g]=ibi[2*g+1]; | |
5090 | + } | |
5091 | +//dump_mem("NE1 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
5092 | +// now 01230123 | |
5093 | +#define INTS_PER_ROW (GROUP_PARALLELISM/8*2) | |
5094 | + for(j=0;j<8;j+=4){ | |
5095 | + for(i=0;i<2;i++){ | |
5096 | + for(k=0;k<INTS_PER_ROW;k++){ | |
5097 | + unsigned int t,b; | |
5098 | + t=ri[INTS_PER_ROW*(j+i)+k]; | |
5099 | + b=ri[INTS_PER_ROW*(j+i+2)+k]; | |
5100 | + ri[INTS_PER_ROW*(j+i)+k]= (t&0x0000ffff) | ((b )<<16); | |
5101 | + ri[INTS_PER_ROW*(j+i+2)+k]= ((t )>>16) | (b&0xffff0000) ; | |
5102 | + } | |
5103 | + } | |
5104 | + } | |
5105 | +//dump_mem("NE2 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
5106 | +// now 01010101 | |
5107 | + for(j=0;j<8;j+=2){ | |
5108 | + for(i=0;i<1;i++){ | |
5109 | + for(k=0;k<INTS_PER_ROW;k++){ | |
5110 | + unsigned int t,b; | |
5111 | + t=ri[INTS_PER_ROW*(j+i)+k]; | |
5112 | + b=ri[INTS_PER_ROW*(j+i+1)+k]; | |
5113 | + ri[INTS_PER_ROW*(j+i)+k]= (t&0x00ff00ff) | ((b&0x00ff00ff)<<8); | |
5114 | + ri[INTS_PER_ROW*(j+i+1)+k]= ((t&0xff00ff00)>>8) | (b&0xff00ff00); | |
5115 | + } | |
5116 | + } | |
5117 | + } | |
5118 | +//dump_mem("NE3 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
5119 | +// now 00000000 | |
5120 | +} | |
5121 | + | |
5122 | +static inline __attribute__((always_inline)) void trasp_8_N (unsigned char *in,unsigned char* out,int count){ | |
5123 | + int *ri=(int *)in; | |
5124 | + int *bdi=(int *)out; | |
5125 | + int j,i,k,g; | |
5126 | +#define INTS_PER_ROW (GROUP_PARALLELISM/8*2) | |
5127 | +//dump_mem("NE1 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
5128 | +// now 00000000 | |
5129 | + for(j=0;j<8;j+=2){ | |
5130 | + for(i=0;i<1;i++){ | |
5131 | + for(k=0;k<INTS_PER_ROW;k++){ | |
5132 | + unsigned int t,b; | |
5133 | + t=ri[INTS_PER_ROW*(j+i)+k]; | |
5134 | + b=ri[INTS_PER_ROW*(j+i+1)+k]; | |
5135 | + ri[INTS_PER_ROW*(j+i)+k]= (t&0x00ff00ff) | ((b&0x00ff00ff)<<8); | |
5136 | + ri[INTS_PER_ROW*(j+i+1)+k]= ((t&0xff00ff00)>>8) | (b&0xff00ff00); | |
5137 | + } | |
5138 | + } | |
5139 | + } | |
5140 | +//dump_mem("NE2 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
5141 | +// now 01010101 | |
5142 | + for(j=0;j<8;j+=4){ | |
5143 | + for(i=0;i<2;i++){ | |
5144 | + for(k=0;k<INTS_PER_ROW;k++){ | |
5145 | + unsigned int t,b; | |
5146 | + t=ri[INTS_PER_ROW*(j+i)+k]; | |
5147 | + b=ri[INTS_PER_ROW*(j+i+2)+k]; | |
5148 | + ri[INTS_PER_ROW*(j+i)+k]= (t&0x0000ffff) | ((b )<<16); | |
5149 | + ri[INTS_PER_ROW*(j+i+2)+k]= ((t )>>16) | (b&0xffff0000) ; | |
5150 | + } | |
5151 | + } | |
5152 | + } | |
5153 | +//dump_mem("NE3 r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
5154 | +// now 01230123 | |
5155 | + for(g=0;g<count;g++){ | |
5156 | + bdi[2*g]=ri[g]; | |
5157 | + bdi[2*g+1]=ri[GROUP_PARALLELISM+g]; | |
5158 | + } | |
5159 | +} | |
5160 | + | |
5161 | +//-----block main function | |
5162 | + | |
5163 | +// block group | |
5164 | +static void block_decypher_group( | |
5165 | + batch *kkmulti, // [In] kkmulti[0]-kkmulti[55] 56 batches | Key schedule (each batch has repeated equal bytes). | |
5166 | + unsigned char *ib, // [In] (ib0,ib1,...ib7)...x32 32*8 bytes | Initialization vector. | |
5167 | + unsigned char *bd, // [Out] (bd0,bd1,...bd7)...x32 32*8 bytes | Block decipher. | |
5168 | + int count) | |
5169 | +{ | |
5170 | + // int is faster than unsigned char. apparently not | |
5171 | + static const unsigned char block_sbox[0x100] = { | |
5172 | + 0x3A,0xEA,0x68,0xFE,0x33,0xE9,0x88,0x1A, 0x83,0xCF,0xE1,0x7F,0xBA,0xE2,0x38,0x12, | |
5173 | + 0xE8,0x27,0x61,0x95,0x0C,0x36,0xE5,0x70, 0xA2,0x06,0x82,0x7C,0x17,0xA3,0x26,0x49, | |
5174 | + 0xBE,0x7A,0x6D,0x47,0xC1,0x51,0x8F,0xF3, 0xCC,0x5B,0x67,0xBD,0xCD,0x18,0x08,0xC9, | |
5175 | + 0xFF,0x69,0xEF,0x03,0x4E,0x48,0x4A,0x84, 0x3F,0xB4,0x10,0x04,0xDC,0xF5,0x5C,0xC6, | |
5176 | + 0x16,0xAB,0xAC,0x4C,0xF1,0x6A,0x2F,0x3C, 0x3B,0xD4,0xD5,0x94,0xD0,0xC4,0x63,0x62, | |
5177 | + 0x71,0xA1,0xF9,0x4F,0x2E,0xAA,0xC5,0x56, 0xE3,0x39,0x93,0xCE,0x65,0x64,0xE4,0x58, | |
5178 | + 0x6C,0x19,0x42,0x79,0xDD,0xEE,0x96,0xF6, 0x8A,0xEC,0x1E,0x85,0x53,0x45,0xDE,0xBB, | |
5179 | + 0x7E,0x0A,0x9A,0x13,0x2A,0x9D,0xC2,0x5E, 0x5A,0x1F,0x32,0x35,0x9C,0xA8,0x73,0x30, | |
5180 | + | |
5181 | + 0x29,0x3D,0xE7,0x92,0x87,0x1B,0x2B,0x4B, 0xA5,0x57,0x97,0x40,0x15,0xE6,0xBC,0x0E, | |
5182 | + 0xEB,0xC3,0x34,0x2D,0xB8,0x44,0x25,0xA4, 0x1C,0xC7,0x23,0xED,0x90,0x6E,0x50,0x00, | |
5183 | + 0x99,0x9E,0x4D,0xD9,0xDA,0x8D,0x6F,0x5F, 0x3E,0xD7,0x21,0x74,0x86,0xDF,0x6B,0x05, | |
5184 | + 0x8E,0x5D,0x37,0x11,0xD2,0x28,0x75,0xD6, 0xA7,0x77,0x24,0xBF,0xF0,0xB0,0x02,0xB7, | |
5185 | + 0xF8,0xFC,0x81,0x09,0xB1,0x01,0x76,0x91, 0x7D,0x0F,0xC8,0xA0,0xF2,0xCB,0x78,0x60, | |
5186 | + 0xD1,0xF7,0xE0,0xB5,0x98,0x22,0xB3,0x20, 0x1D,0xA6,0xDB,0x7B,0x59,0x9F,0xAE,0x31, | |
5187 | + 0xFB,0xD3,0xB6,0xCA,0x43,0x72,0x07,0xF4, 0xD8,0x41,0x14,0x55,0x0D,0x54,0x8B,0xB9, | |
5188 | + 0xAD,0x46,0x0B,0xAF,0x80,0x52,0x2C,0xFA, 0x8C,0x89,0x66,0xFD,0xB2,0xA9,0x9B,0xC0, | |
5189 | + }; | |
5190 | + unsigned char r[GROUP_PARALLELISM*(8+56)]; /* 56 because we will move back in memory while looping */ | |
5191 | + unsigned char sbox_in[GROUP_PARALLELISM],sbox_out[GROUP_PARALLELISM],perm_out[GROUP_PARALLELISM]; | |
5192 | + int roff; | |
5193 | + int i,g,count_all=GROUP_PARALLELISM; | |
5194 | + | |
5195 | + roff=GROUP_PARALLELISM*56; | |
5196 | + | |
5197 | +#define FASTTRASP1 | |
5198 | +#ifndef FASTTRASP1 | |
5199 | + for(g=0;g<count;g++){ | |
5200 | + // Init registers | |
5201 | + int j; | |
5202 | + for(j=0;j<8;j++){ | |
5203 | + r[roff+GROUP_PARALLELISM*j+g]=ib[8*g+j]; | |
5204 | + } | |
5205 | + } | |
5206 | +#else | |
5207 | + trasp_N_8((unsigned char *)&r[roff],(unsigned char *)ib,count); | |
5208 | +#endif | |
5209 | +//dump_mem("OLD r[roff]",&r[roff],GROUP_PARALLELISM*8,GROUP_PARALLELISM); | |
5210 | + | |
5211 | + // loop over kk[55]..kk[0] | |
5212 | + for(i=55;i>=0;i--){ | |
5213 | + { | |
5214 | + batch tkkmulti=kkmulti[i]; | |
5215 | + batch *si=(batch *)sbox_in; | |
5216 | + batch *r6_N=(batch *)(r+roff+GROUP_PARALLELISM*6); | |
5217 | + for(g=0;g<count_all/BYTES_PER_BATCH;g++){ | |
5218 | + si[g]=B_FFXOR(tkkmulti,r6_N[g]); //FIXME: introduce FASTBATCH? | |
5219 | + } | |
5220 | + } | |
5221 | + | |
5222 | + // table lookup, this works on only one byte at a time | |
5223 | + // most difficult part of all | |
5224 | + // - can't be parallelized | |
5225 | + // - can't be synthetized through boolean terms (8 input bits are too many) | |
5226 | + for(g=0;g<count_all;g++){ | |
5227 | + sbox_out[g]=block_sbox[sbox_in[g]]; | |
5228 | + } | |
5229 | + | |
5230 | + // bit permutation | |
5231 | + { | |
5232 | + unsigned char *po=(unsigned char *)perm_out; | |
5233 | + unsigned char *so=(unsigned char *)sbox_out; | |
5234 | +//dump_mem("pre perm ",(unsigned char *)so,GROUP_PARALLELISM,GROUP_PARALLELISM); | |
5235 | + for(g=0;g<count_all;g+=BYTES_PER_BATCH){ | |
5236 | + batch in,out; | |
5237 | + in=*(batch *)&so[g]; | |
5238 | + | |
5239 | + out=B_FFOR( | |
5240 | + B_FFOR( | |
5241 | + B_FFOR( | |
5242 | + B_FFOR( | |
5243 | + B_FFOR( | |
5244 | + B_FFSH8L(B_FFAND(in,B_FFN_ALL_29()),1), | |
5245 | + B_FFSH8L(B_FFAND(in,B_FFN_ALL_02()),6)), | |
5246 | + B_FFSH8L(B_FFAND(in,B_FFN_ALL_04()),3)), | |
5247 | + B_FFSH8R(B_FFAND(in,B_FFN_ALL_10()),2)), | |
5248 | + B_FFSH8R(B_FFAND(in,B_FFN_ALL_40()),6)), | |
5249 | + B_FFSH8R(B_FFAND(in,B_FFN_ALL_80()),4)); | |
5250 | + | |
5251 | + *(batch *)&po[g]=out; | |
5252 | + } | |
5253 | +//dump_mem("post perm",(unsigned char *)po,GROUP_PARALLELISM,GROUP_PARALLELISM); | |
5254 | + } | |
5255 | + | |
5256 | + roff-=GROUP_PARALLELISM; /* virtual shift of registers */ | |
5257 | + | |
5258 | +#if 0 | |
5259 | +/* one by one */ | |
5260 | + for(g=0;g<count_all;g++){ | |
5261 | + r[roff+GROUP_PARALLELISM*0+g]=r[roff+GROUP_PARALLELISM*8+g]^sbox_out[g]; | |
5262 | + r[roff+GROUP_PARALLELISM*6+g]^=perm_out[g]; | |
5263 | + r[roff+GROUP_PARALLELISM*4+g]^=r[roff+GROUP_PARALLELISM*0+g]; | |
5264 | + r[roff+GROUP_PARALLELISM*3+g]^=r[roff+GROUP_PARALLELISM*0+g]; | |
5265 | + r[roff+GROUP_PARALLELISM*2+g]^=r[roff+GROUP_PARALLELISM*0+g]; | |
5266 | + } | |
5267 | +#else | |
5268 | + for(g=0;g<count_all;g+=BEST_SPAN){ | |
5269 | + XOR_BEST_BY(&r[roff+GROUP_PARALLELISM*0+g],&r[roff+GROUP_PARALLELISM*8+g],&sbox_out[g]); | |
5270 | + XOREQ_BEST_BY(&r[roff+GROUP_PARALLELISM*6+g],&perm_out[g]); | |
5271 | + XOREQ_BEST_BY(&r[roff+GROUP_PARALLELISM*4+g],&r[roff+GROUP_PARALLELISM*0+g]); | |
5272 | + XOREQ_BEST_BY(&r[roff+GROUP_PARALLELISM*3+g],&r[roff+GROUP_PARALLELISM*0+g]); | |
5273 | + XOREQ_BEST_BY(&r[roff+GROUP_PARALLELISM*2+g],&r[roff+GROUP_PARALLELISM*0+g]); | |
5274 | + } | |
5275 | +#endif | |
5276 | + } | |
5277 | + | |
5278 | +#define FASTTRASP2 | |
5279 | +#ifndef FASTTRASP2 | |
5280 | + for(g=0;g<count;g++){ | |
5281 | + // Copy results | |
5282 | + int j; | |
5283 | + for(j=0;j<8;j++){ | |
5284 | + bd[8*g+j]=r[roff+GROUP_PARALLELISM*j+g]; | |
5285 | + } | |
5286 | + } | |
5287 | +#else | |
5288 | + trasp_8_N((unsigned char *)&r[roff],(unsigned char *)bd,count); | |
5289 | +#endif | |
5290 | +} | |
5291 | + | |
5292 | +//-----------------------------------EXTERNAL INTERFACE | |
5293 | + | |
5294 | +//-----get internal parallelism | |
5295 | + | |
5296 | +int get_internal_parallelism(void){ | |
5297 | + return GROUP_PARALLELISM; | |
5298 | +} | |
5299 | + | |
5300 | +//-----get suggested cluster size | |
5301 | + | |
5302 | +int get_suggested_cluster_size(void){ | |
5303 | + int r; | |
5304 | + r=GROUP_PARALLELISM+GROUP_PARALLELISM/10; | |
5305 | + if(r<GROUP_PARALLELISM+5) r=GROUP_PARALLELISM+5; | |
5306 | + return r; | |
5307 | +} | |
5308 | + | |
5309 | +//-----key structure | |
5310 | + | |
5311 | +void *get_key_struct(void){ | |
5312 | + struct csa_keys_t *keys=(struct csa_keys_t *)malloc(sizeof(struct csa_keys_t)); | |
5313 | + if(keys) { | |
5314 | + static const unsigned char pk[8] = { 0,0,0,0,0,0,0,0 }; | |
5315 | + set_control_words(keys,pk,pk); | |
5316 | + } | |
5317 | + return keys; | |
5318 | +} | |
5319 | + | |
5320 | +void free_key_struct(void *keys){ | |
5321 | + return free(keys); | |
5322 | +} | |
5323 | + | |
5324 | +//-----set control words | |
5325 | + | |
5326 | +static void schedule_key(struct csa_key_t *key, const unsigned char *pk){ | |
5327 | + // could be made faster, but is not run often | |
5328 | + int bi,by; | |
5329 | + int i,j; | |
5330 | +// key | |
5331 | + memcpy(key->ck,pk,8); | |
5332 | +// precalculations for stream | |
5333 | + key_schedule_stream(key->ck,key->iA,key->iB); | |
5334 | + for(by=0;by<8;by++){ | |
5335 | + for(bi=0;bi<8;bi++){ | |
5336 | + key->ck_g[by][bi]=(key->ck[by]&(1<<bi))?FF1():FF0(); | |
5337 | + } | |
5338 | + } | |
5339 | + for(by=0;by<8;by++){ | |
5340 | + for(bi=0;bi<4;bi++){ | |
5341 | + key->iA_g[by][bi]=(key->iA[by]&(1<<bi))?FF1():FF0(); | |
5342 | + key->iB_g[by][bi]=(key->iB[by]&(1<<bi))?FF1():FF0(); | |
5343 | + } | |
5344 | + } | |
5345 | +// precalculations for block | |
5346 | + key_schedule_block(key->ck,key->kk); | |
5347 | + for(i=0;i<56;i++){ | |
5348 | + for(j=0;j<BYTES_PER_BATCH;j++){ | |
5349 | + *(((unsigned char *)&key->kkmulti[i])+j)=key->kk[i]; | |
5350 | + } | |
5351 | + } | |
5352 | +} | |
5353 | + | |
5354 | +void set_control_words(void *keys, const unsigned char *ev, const unsigned char *od){ | |
5355 | + schedule_key(&((struct csa_keys_t *)keys)->even,ev); | |
5356 | + schedule_key(&((struct csa_keys_t *)keys)->odd,od); | |
5357 | +} | |
5358 | + | |
5359 | +void set_even_control_word(void *keys, const unsigned char *pk){ | |
5360 | + schedule_key(&((struct csa_keys_t *)keys)->even,pk); | |
5361 | +} | |
5362 | + | |
5363 | +void set_odd_control_word(void *keys, const unsigned char *pk){ | |
5364 | + schedule_key(&((struct csa_keys_t *)keys)->odd,pk); | |
5365 | +} | |
5366 | + | |
5367 | +//-----get control words | |
5368 | + | |
5369 | +void get_control_words(void *keys, unsigned char *even, unsigned char *odd){ | |
5370 | + memcpy(even,&((struct csa_keys_t *)keys)->even.ck,8); | |
5371 | + memcpy(odd,&((struct csa_keys_t *)keys)->odd.ck,8); | |
5372 | +} | |
5373 | + | |
5374 | +//----- decrypt | |
5375 | + | |
5376 | +int decrypt_packets(void *keys, unsigned char **cluster){ | |
5377 | + // statistics, currently unused | |
5378 | + int stat_no_scramble=0; | |
5379 | + int stat_reserved=0; | |
5380 | + int stat_decrypted[2]={0,0}; | |
5381 | + int stat_decrypted_mini=0; | |
5382 | + unsigned char **clst; | |
5383 | + unsigned char **clst2; | |
5384 | + int grouped; | |
5385 | + int group_ev_od; | |
5386 | + int advanced; | |
5387 | + int can_advance; | |
5388 | + unsigned char *g_pkt[GROUP_PARALLELISM]; | |
5389 | + int g_len[GROUP_PARALLELISM]; | |
5390 | + int g_offset[GROUP_PARALLELISM]; | |
5391 | + int g_n[GROUP_PARALLELISM]; | |
5392 | + int g_residue[GROUP_PARALLELISM]; | |
5393 | + unsigned char *pkt; | |
5394 | + int xc0,ev_od,len,offset,n,residue; | |
5395 | + struct csa_key_t* k; | |
5396 | + int i,j,iter,g; | |
5397 | + int t23,tsmall; | |
5398 | + int alive[24]; | |
5399 | +//icc craziness int pad1=0; //////////align! FIXME | |
5400 | + unsigned char *encp[GROUP_PARALLELISM]; | |
5401 | + unsigned char stream_in[GROUP_PARALLELISM*8]; | |
5402 | + unsigned char stream_out[GROUP_PARALLELISM*8]; | |
5403 | + unsigned char ib[GROUP_PARALLELISM*8]; | |
5404 | + unsigned char block_out[GROUP_PARALLELISM*8]; | |
5405 | + struct stream_regs regs; | |
5406 | + | |
5407 | +//icc craziness i=(int)&pad1;//////////align!!! FIXME | |
5408 | + | |
5409 | + // build a list of packets to be processed | |
5410 | + clst=cluster; | |
5411 | + grouped=0; | |
5412 | + advanced=0; | |
5413 | + can_advance=1; | |
5414 | + group_ev_od=-1; // silence incorrect compiler warning | |
5415 | + pkt=*clst; | |
5416 | + do{ // find a new packet | |
5417 | + if(grouped==GROUP_PARALLELISM){ | |
5418 | + // full | |
5419 | + break; | |
5420 | + } | |
5421 | + if(pkt==NULL){ | |
5422 | + // no more ranges | |
5423 | + break; | |
5424 | + } | |
5425 | + if(pkt>=*(clst+1)){ | |
5426 | + // out of this range, try next | |
5427 | + clst++;clst++; | |
5428 | + pkt=*clst; | |
5429 | + continue; | |
5430 | + } | |
5431 | + | |
5432 | + do{ // handle this packet | |
5433 | + xc0=pkt[3]&0xc0; | |
5434 | + DBG(fprintf(stderr," exam pkt=%p, xc0=%02x, can_adv=%i\n",pkt,xc0,can_advance)); | |
5435 | + if(xc0==0x00){ | |
5436 | + DBG(fprintf(stderr,"skip clear pkt %p (can_advance is %i)\n",pkt,can_advance)); | |
5437 | + advanced+=can_advance; | |
5438 | + stat_no_scramble++; | |
5439 | + break; | |
5440 | + } | |
5441 | + if(xc0==0x40){ | |
5442 | + DBG(fprintf(stderr,"skip reserved pkt %p (can_advance is %i)\n",pkt,can_advance)); | |
5443 | + advanced+=can_advance; | |
5444 | + stat_reserved++; | |
5445 | + break; | |
5446 | + } | |
5447 | + if(xc0==0x80||xc0==0xc0){ // encrypted | |
5448 | + ev_od=(xc0&0x40)>>6; // 0 even, 1 odd | |
5449 | + if(grouped==0) group_ev_od=ev_od; // this group will be all even (or odd) | |
5450 | + if(group_ev_od==ev_od){ // could be added to group | |
5451 | + pkt[3]&=0x3f; // consider it decrypted now | |
5452 | + if(pkt[3]&0x20){ // incomplete packet | |
5453 | + offset=4+pkt[4]+1; | |
5454 | + len=188-offset; | |
5455 | + n=len>>3; | |
5456 | + residue=len-(n<<3); | |
5457 | + if(n==0){ // decrypted==encrypted! | |
5458 | + DBG(fprintf(stderr,"DECRYPTED MINI! (can_advance is %i)\n",can_advance)); | |
5459 | + advanced+=can_advance; | |
5460 | + stat_decrypted_mini++; | |
5461 | + break; // this doesn't need more processing | |
5462 | + } | |
5463 | + }else{ | |
5464 | + len=184; | |
5465 | + offset=4; | |
5466 | + n=23; | |
5467 | + residue=0; | |
5468 | + } | |
5469 | + g_pkt[grouped]=pkt; | |
5470 | + g_len[grouped]=len; | |
5471 | + g_offset[grouped]=offset; | |
5472 | + g_n[grouped]=n; | |
5473 | + g_residue[grouped]=residue; | |
5474 | + DBG(fprintf(stderr,"%2i: eo=%i pkt=%p len=%03i n=%2i residue=%i\n",grouped,ev_od,pkt,len,n,residue)); | |
5475 | + grouped++; | |
5476 | + advanced+=can_advance; | |
5477 | + stat_decrypted[ev_od]++; | |
5478 | + } | |
5479 | + else{ | |
5480 | + can_advance=0; | |
5481 | + DBG(fprintf(stderr,"skip pkt %p and can_advance set to 0\n",pkt)); | |
5482 | + break; // skip and go on | |
5483 | + } | |
5484 | + } | |
5485 | + } while(0); | |
5486 | + | |
5487 | + if(can_advance){ | |
5488 | + // move range start forward | |
5489 | + *clst+=188; | |
5490 | + } | |
5491 | + // next packet, if there is one | |
5492 | + pkt+=188; | |
5493 | + } while(1); | |
5494 | + DBG(fprintf(stderr,"-- result: grouped %i pkts, advanced %i pkts\n",grouped,advanced)); | |
5495 | + | |
5496 | + // delete empty ranges and compact list | |
5497 | + clst2=cluster; | |
5498 | + for(clst=cluster;*clst!=NULL;clst+=2){ | |
5499 | + // if not empty | |
5500 | + if(*clst<*(clst+1)){ | |
5501 | + // it will remain | |
5502 | + *clst2=*clst; | |
5503 | + *(clst2+1)=*(clst+1); | |
5504 | + clst2+=2; | |
5505 | + } | |
5506 | + } | |
5507 | + *clst2=NULL; | |
5508 | + | |
5509 | + if(grouped==0){ | |
5510 | + // no processing needed | |
5511 | + return advanced; | |
5512 | + } | |
5513 | + | |
5514 | + // sort them, longest payload first | |
5515 | + // we expect many n=23 packets and a few n<23 | |
5516 | + DBG(fprintf(stderr,"PRESORTING\n")); | |
5517 | + for(i=0;i<grouped;i++){ | |
5518 | + DBG(fprintf(stderr,"%2i of %2i: pkt=%p len=%03i n=%2i residue=%i\n",i,grouped,g_pkt[i],g_len[i],g_n[i],g_residue[i])); | |
5519 | + } | |
5520 | + // grouped is always <= GROUP_PARALLELISM | |
5521 | + | |
5522 | +#define g_swap(a,b) \ | |
5523 | + pkt=g_pkt[a]; \ | |
5524 | + g_pkt[a]=g_pkt[b]; \ | |
5525 | + g_pkt[b]=pkt; \ | |
5526 | +\ | |
5527 | + len=g_len[a]; \ | |
5528 | + g_len[a]=g_len[b]; \ | |
5529 | + g_len[b]=len; \ | |
5530 | +\ | |
5531 | + offset=g_offset[a]; \ | |
5532 | + g_offset[a]=g_offset[b]; \ | |
5533 | + g_offset[b]=offset; \ | |
5534 | +\ | |
5535 | + n=g_n[a]; \ | |
5536 | + g_n[a]=g_n[b]; \ | |
5537 | + g_n[b]=n; \ | |
5538 | +\ | |
5539 | + residue=g_residue[a]; \ | |
5540 | + g_residue[a]=g_residue[b]; \ | |
5541 | + g_residue[b]=residue; | |
5542 | + | |
5543 | + // step 1: move n=23 packets before small packets | |
5544 | + t23=0; | |
5545 | + tsmall=grouped-1; | |
5546 | + for(;;){ | |
5547 | + for(;t23<grouped;t23++){ | |
5548 | + if(g_n[t23]!=23) break; | |
5549 | + } | |
5550 | +DBG(fprintf(stderr,"t23 after for =%i\n",t23)); | |
5551 | + | |
5552 | + for(;tsmall>=0;tsmall--){ | |
5553 | + if(g_n[tsmall]==23) break; | |
5554 | + } | |
5555 | +DBG(fprintf(stderr,"tsmall after for =%i\n",tsmall)); | |
5556 | + | |
5557 | + if(tsmall-t23<1) break; | |
5558 | + | |
5559 | +DBG(fprintf(stderr,"swap t23=%i,tsmall=%i\n",t23,tsmall)); | |
5560 | + | |
5561 | + g_swap(t23,tsmall); | |
5562 | + | |
5563 | + t23++; | |
5564 | + tsmall--; | |
5565 | +DBG(fprintf(stderr,"new t23=%i,tsmall=%i\n\n",t23,tsmall)); | |
5566 | + } | |
5567 | + DBG(fprintf(stderr,"packets with n=23, t23=%i grouped=%i\n",t23,grouped)); | |
5568 | + DBG(fprintf(stderr,"MIDSORTING\n")); | |
5569 | + for(i=0;i<grouped;i++){ | |
5570 | + DBG(fprintf(stderr,"%2i of %2i: pkt=%p len=%03i n=%2i residue=%i\n",i,grouped,g_pkt[i],g_len[i],g_n[i],g_residue[i])); | |
5571 | + } | |
5572 | + | |
5573 | + // step 2: sort small packets in decreasing order of n (bubble sort is enough) | |
5574 | + for(i=t23;i<grouped;i++){ | |
5575 | + for(j=i+1;j<grouped;j++){ | |
5576 | + if(g_n[j]>g_n[i]){ | |
5577 | + g_swap(i,j); | |
5578 | + } | |
5579 | + } | |
5580 | + } | |
5581 | + DBG(fprintf(stderr,"POSTSORTING\n")); | |
5582 | + for(i=0;i<grouped;i++){ | |
5583 | + DBG(fprintf(stderr,"%2i of %2i: pkt=%p len=%03i n=%2i residue=%i\n",i,grouped,g_pkt[i],g_len[i],g_n[i],g_residue[i])); | |
5584 | + } | |
5585 | + | |
5586 | + // we need to know how many packets need 23 iterations, how many 22... | |
5587 | + for(i=0;i<=23;i++){ | |
5588 | + alive[i]=0; | |
5589 | + } | |
5590 | + // count | |
5591 | + alive[23-1]=t23; | |
5592 | + for(i=t23;i<grouped;i++){ | |
5593 | + alive[g_n[i]-1]++; | |
5594 | + } | |
5595 | + // integrate | |
5596 | + for(i=22;i>=0;i--){ | |
5597 | + alive[i]+=alive[i+1]; | |
5598 | + } | |
5599 | + DBG(fprintf(stderr,"ALIVE\n")); | |
5600 | + for(i=0;i<=23;i++){ | |
5601 | + DBG(fprintf(stderr,"alive%2i=%i\n",i,alive[i])); | |
5602 | + } | |
5603 | + | |
5604 | + // choose key | |
5605 | + if(group_ev_od==0){ | |
5606 | + k=&((struct csa_keys_t *)keys)->even; | |
5607 | + } | |
5608 | + else{ | |
5609 | + k=&((struct csa_keys_t *)keys)->odd; | |
5610 | + } | |
5611 | + | |
5612 | + //INIT | |
5613 | +//#define INITIALIZE_UNUSED_INPUT | |
5614 | +#ifdef INITIALIZE_UNUSED_INPUT | |
5615 | +// unnecessary zeroing. | |
5616 | +// without this, we operate on uninitialized memory | |
5617 | +// when grouped<GROUP_PARALLELISM, but it's not a problem, | |
5618 | +// as final results will be discarded. | |
5619 | +// random data makes debugging sessions difficult. | |
5620 | + for(j=0;j<GROUP_PARALLELISM*8;j++) stream_in[j]=0; | |
5621 | +DBG(fprintf(stderr,"--- WARNING: you could gain speed by not initializing unused memory ---\n")); | |
5622 | +#else | |
5623 | +DBG(fprintf(stderr,"--- WARNING: DEBUGGING IS MORE DIFFICULT WHEN PROCESSING RANDOM DATA CHANGING AT EVERY RUN! ---\n")); | |
5624 | +#endif | |
5625 | + | |
5626 | + for(g=0;g<grouped;g++){ | |
5627 | + encp[g]=g_pkt[g]; | |
5628 | + DBG(fprintf(stderr,"header[%i]=%p (%02x)\n",g,encp[g],*(encp[g]))); | |
5629 | + encp[g]+=g_offset[g]; // skip header | |
5630 | + FFTABLEIN(stream_in,g,encp[g]); | |
5631 | + } | |
5632 | +//dump_mem("stream_in",stream_in,GROUP_PARALLELISM*8,BYPG); | |
5633 | + | |
5634 | + | |
5635 | + // ITER 0 | |
5636 | +DBG(fprintf(stderr,">>>>>ITER 0\n")); | |
5637 | + iter=0; | |
5638 | + stream_cypher_group_init(®s,k->iA_g,k->iB_g,stream_in); | |
5639 | + // fill first ib | |
5640 | + for(g=0;g<alive[iter];g++){ | |
5641 | + COPY_8_BY(ib+8*g,encp[g]); | |
5642 | + } | |
5643 | +DBG(dump_mem("IB ",ib,8*alive[iter],8)); | |
5644 | + // ITER 1..N-1 | |
5645 | + for (iter=1;iter<23&&alive[iter-1]>0;iter++){ | |
5646 | +DBG(fprintf(stderr,">>>>>ITER %i\n",iter)); | |
5647 | + // alive and just dead packets: calc block | |
5648 | + block_decypher_group(k->kkmulti,ib,block_out,alive[iter-1]); | |
5649 | +DBG(dump_mem("BLO_ib ",block_out,8*alive[iter-1],8)); | |
5650 | + // all packets (dead too): calc stream | |
5651 | + stream_cypher_group_normal(®s,stream_out); | |
5652 | +//dump_mem("stream_out",stream_out,GROUP_PARALLELISM*8,BYPG); | |
5653 | + | |
5654 | + // alive packets: calc ib | |
5655 | + for(g=0;g<alive[iter];g++){ | |
5656 | + FFTABLEOUT(ib+8*g,stream_out,g); | |
5657 | +DBG(dump_mem("stream_out_ib ",ib+8*g,8,8)); | |
5658 | +// XOREQ8BY gcc bug? 2x4 ok, 8 ko UPDATE: result ok but speed 1-2% slower (!!!???) | |
5659 | +#if 1 | |
5660 | + XOREQ_4_BY(ib+8*g,encp[g]+8); | |
5661 | + XOREQ_4_BY(ib+8*g+4,encp[g]+8+4); | |
5662 | +#else | |
5663 | + XOREQ_8_BY(ib+8*g,encp[g]+8); | |
5664 | +#endif | |
5665 | +DBG(dump_mem("after_stream_xor_ib ",ib+8*g,8,8)); | |
5666 | + } | |
5667 | + // alive packets: decrypt data | |
5668 | + for(g=0;g<alive[iter];g++){ | |
5669 | +DBG(dump_mem("before_ib_decrypt_data ",encp[g],8,8)); | |
5670 | + XOR_8_BY(encp[g],ib+8*g,block_out+8*g); | |
5671 | +DBG(dump_mem("after_ib_decrypt_data ",encp[g],8,8)); | |
5672 | + } | |
5673 | + // just dead packets: write decrypted data | |
5674 | + for(g=alive[iter];g<alive[iter-1];g++){ | |
5675 | +DBG(dump_mem("jd_before_ib_decrypt_data ",encp[g],8,8)); | |
5676 | + COPY_8_BY(encp[g],block_out+8*g); | |
5677 | +DBG(dump_mem("jd_after_ib_decrypt_data ",encp[g],8,8)); | |
5678 | + } | |
5679 | + // just dead packets: decrypt residue | |
5680 | + for(g=alive[iter];g<alive[iter-1];g++){ | |
5681 | +DBG(dump_mem("jd_before_decrypt_residue ",encp[g]+8,g_residue[g],g_residue[g])); | |
5682 | + FFTABLEOUTXORNBY(g_residue[g],encp[g]+8,stream_out,g); | |
5683 | +DBG(dump_mem("jd_after_decrypt_residue ",encp[g]+8,g_residue[g],g_residue[g])); | |
5684 | + } | |
5685 | + // alive packets: pointers++ | |
5686 | + for(g=0;g<alive[iter];g++) encp[g]+=8; | |
5687 | + }; | |
5688 | + // ITER N | |
5689 | +DBG(fprintf(stderr,">>>>>ITER 23\n")); | |
5690 | + iter=23; | |
5691 | + // calc block | |
5692 | + block_decypher_group(k->kkmulti,ib,block_out,alive[iter-1]); | |
5693 | +DBG(dump_mem("23BLO_ib ",block_out,8*alive[iter-1],8)); | |
5694 | + // just dead packets: write decrypted data | |
5695 | + for(g=alive[iter];g<alive[iter-1];g++){ | |
5696 | +DBG(dump_mem("23jd_before_ib_decrypt_data ",encp[g],8,8)); | |
5697 | + COPY_8_BY(encp[g],block_out+8*g); | |
5698 | +DBG(dump_mem("23jd_after_ib_decrypt_data ",encp[g],8,8)); | |
5699 | + } | |
5700 | + // no residue possible | |
5701 | + // so do nothing | |
5702 | + | |
5703 | + DBG(fprintf(stderr,"returning advanced=%i\n",advanced)); | |
5704 | + | |
5705 | + M_EMPTY(); // restore CPU multimedia state | |
5706 | + | |
5707 | + return advanced; | |
5708 | +} | |
5709 | Index: libs/libmythtv/FFdecsa/README | |
5710 | =================================================================== | |
5711 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
5712 | +++ libs/libmythtv/FFdecsa/README 2006-06-20 17:36:06.000000000 -0400 | |
5713 | @@ -0,0 +1,50 @@ | |
5714 | +------- | |
5715 | +FFdecsa | |
5716 | +------- | |
5717 | +version 1.0 | |
5718 | +Copyright 2003-2004 fatih89r | |
5719 | +released under GPL | |
5720 | + | |
5721 | + | |
5722 | +FFdecsa is a fast implementation of a CSA decryption algorithm for MPEG | |
5723 | +TS packets. It is shockingly fast, more than 800% the speed of the | |
5724 | +fastest implementation I can find around. (read the docs to know what FF | |
5725 | +stands for) | |
5726 | + | |
5727 | +On an AthlonXP 2400 (2000MHz) it achieves 165Mbit/s; the previous record | |
5728 | +was around 20Mbit/s. | |
5729 | + | |
5730 | +This means that: | |
5731 | +- decrypting a 8Mbit/s stream takes 5% of CPU instead of 40% | |
5732 | +- decrypting a full transponder (with all its channels or with a big | |
5733 | + HDTV stream) carrying 38Mbit/s takes 23% of CPU instead of 190% | |
5734 | + (>100%, so undecryptable in real time) | |
5735 | +- a very slow processor can decrypt one channel with no problems | |
5736 | +- offline decoding of one hour of a 5Mbit/s channel takes less than | |
5737 | + two minutes (30x than realtime) | |
5738 | +- offline decoding will work at more than 20MB/s (megabytes/s), | |
5739 | + nearly as fast as a file copy | |
5740 | + | |
5741 | +The docs directory contains useful stuff: | |
5742 | + | |
5743 | + FAQ.txt | |
5744 | + to know something more about this software | |
5745 | + | |
5746 | + how_to_compile.txt | |
5747 | + if you want to compile this code (and get optimal speed) | |
5748 | + | |
5749 | + how_to_use.txt | |
5750 | + if you want to use this code | |
5751 | + | |
5752 | + technical_background.txt | |
5753 | + if you want to understand how this code works or you want to | |
5754 | + modify/improve it | |
5755 | + | |
5756 | + how_to_understand.txt | |
5757 | + if you want to understand the code to make modifications | |
5758 | + | |
5759 | + how_to_release.txt | |
5760 | + if you want to release modified versions of the code | |
5761 | + | |
5762 | + | |
5763 | +fatih89r | |
5764 | Index: libs/libmythtv/FFdecsa/FFdecsa.h | |
5765 | =================================================================== | |
5766 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
5767 | +++ libs/libmythtv/FFdecsa/FFdecsa.h 2006-06-20 17:36:06.000000000 -0400 | |
5768 | @@ -0,0 +1,62 @@ | |
5769 | +/* FFdecsa -- fast decsa algorithm | |
5770 | + * | |
5771 | + * Copyright (C) 2003-2004 fatih89r | |
5772 | + * | |
5773 | + * This program is free software; you can redistribute it and/or modify | |
5774 | + * it under the terms of the GNU General Public License as published by | |
5775 | + * the Free Software Foundation; either version 2 of the License, or | |
5776 | + * (at your option) any later version. | |
5777 | + * | |
5778 | + * This program is distributed in the hope that it will be useful, | |
5779 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
5780 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
5781 | + * GNU General Public License for more details. | |
5782 | + * | |
5783 | + * You should have received a copy of the GNU General Public License | |
5784 | + * along with this program; if not, write to the Free Software | |
5785 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
5786 | + */ | |
5787 | + | |
5788 | + | |
5789 | +#ifndef FFDECSA_H | |
5790 | +#define FFDECSA_H | |
5791 | + | |
5792 | +//----- public interface | |
5793 | + | |
5794 | +// -- how many packets can be decrypted at the same time | |
5795 | +// This is an info about internal decryption parallelism. | |
5796 | +// You should try to call decrypt_packets with more packets than the number | |
5797 | +// returned here for performance reasons (use get_suggested_cluster_size to know | |
5798 | +// how many). | |
5799 | +int get_internal_parallelism(void); | |
5800 | + | |
5801 | +// -- how many packets you should have in a cluster when calling decrypt_packets | |
5802 | +// This is a suggestion to achieve optimal performance; typically a little | |
5803 | +// higher than what get_internal_parallelism returns. | |
5804 | +// Passing less packets could slow down the decryption. | |
5805 | +// Passing more packets is never bad (if you don't spend a lot of time building | |
5806 | +// the list). | |
5807 | +int get_suggested_cluster_size(void); | |
5808 | + | |
5809 | +// -- alloc & free the key structure | |
5810 | +void *get_key_struct(void); | |
5811 | +void free_key_struct(void *keys); | |
5812 | + | |
5813 | +// -- set control words, 8 bytes each | |
5814 | +void set_control_words(void *keys, const unsigned char *even, const unsigned char *odd); | |
5815 | + | |
5816 | +// -- set even control word, 8 bytes | |
5817 | +void set_even_control_word(void *keys, const unsigned char *even); | |
5818 | + | |
5819 | +// -- set odd control word, 8 bytes | |
5820 | +void set_odd_control_word(void *keys, const unsigned char *odd); | |
5821 | + | |
5822 | +// -- get control words, 8 bytes each | |
5823 | +//void get_control_words(void *keys, unsigned char *even, unsigned char *odd); | |
5824 | + | |
5825 | +// -- decrypt many TS packets | |
5826 | +// This interface is a bit complicated because it is designed for maximum speed. | |
5827 | +// Please read doc/how_to_use.txt. | |
5828 | +int decrypt_packets(void *keys, unsigned char **cluster); | |
5829 | + | |
5830 | +#endif | |
5831 | Index: libs/libmythtv/FFdecsa/docs/how_to_understand.txt | |
5832 | =================================================================== | |
5833 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
5834 | +++ libs/libmythtv/FFdecsa/docs/how_to_understand.txt 2006-06-20 17:36:06.000000000 -0400 | |
5835 | @@ -0,0 +1,15 @@ | |
5836 | +------- | |
5837 | +FFdecsa | |
5838 | +------- | |
5839 | + | |
5840 | +First, you need to know how decsa works, study the source of a classical | |
5841 | +implementation. Then you have to understand how things are done in | |
5842 | +slicing mode. Read all the documentation and have a working classical | |
5843 | +implementation to compare partial results. There are comments spread | |
5844 | +around the code. Some things are difficult to understand without paper | |
5845 | +notes; for example the matrix transpositions and meaning of array | |
5846 | +indices. | |
5847 | + | |
5848 | +Sorry, it is hard to understand and modify ... | |
5849 | + | |
5850 | +... but it was harder to design and implement!!! | |
5851 | Index: libs/libmythtv/FFdecsa/parallel_064_mmx.h | |
5852 | =================================================================== | |
5853 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
5854 | +++ libs/libmythtv/FFdecsa/parallel_064_mmx.h 2006-06-20 17:36:06.000000000 -0400 | |
5855 | @@ -0,0 +1,106 @@ | |
5856 | +/* FFdecsa -- fast decsa algorithm | |
5857 | + * | |
5858 | + * Copyright (C) 2003-2004 fatih89r | |
5859 | + * | |
5860 | + * This program is free software; you can redistribute it and/or modify | |
5861 | + * it under the terms of the GNU General Public License as published by | |
5862 | + * the Free Software Foundation; either version 2 of the License, or | |
5863 | + * (at your option) any later version. | |
5864 | + * | |
5865 | + * This program is distributed in the hope that it will be useful, | |
5866 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
5867 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
5868 | + * GNU General Public License for more details. | |
5869 | + * | |
5870 | + * You should have received a copy of the GNU General Public License | |
5871 | + * along with this program; if not, write to the Free Software | |
5872 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
5873 | + */ | |
5874 | + | |
5875 | +#include <mmintrin.h> | |
5876 | + | |
5877 | +#if 0 // some older gcc version need this | |
5878 | +#define _m_pand _mm_and_si64 | |
5879 | +#define _m_por _mm_or_si64 | |
5880 | +#define _m_pxor _mm_xor_si64 | |
5881 | +#define _m_psllqi _mm_slli_si64 | |
5882 | +#define _m_psrlqi _mm_srli_si64 | |
5883 | +#define _m_empty _mm_empty | |
5884 | +#endif | |
5885 | + | |
5886 | +typedef __m64 group; | |
5887 | +#define GROUP_PARALLELISM 64 | |
5888 | +#define FF0() ((__m64)0x0ULL) | |
5889 | +#define FF1() ((__m64)0xffffffffffffffffULL) | |
5890 | +#define FFAND(a,b) _m_pand((a),(b)) | |
5891 | +#define FFOR(a,b) _m_por((a),(b)) | |
5892 | +#define FFXOR(a,b) _m_pxor((a),(b)) | |
5893 | +#define FFNOT(a) _m_pxor((a),FF1()) | |
5894 | + | |
5895 | +/* 64 rows of 64 bits */ | |
5896 | + | |
5897 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
5898 | +#if 1 | |
5899 | + *(((int *)tab)+2*g)=*((int *)data); | |
5900 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
5901 | +#else | |
5902 | + *(((long long int *)tab)+g)=*((long long int *)data); | |
5903 | +#endif | |
5904 | +} | |
5905 | + | |
5906 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
5907 | +#if 1 | |
5908 | + *((int *)data)=*(((int *)tab)+2*g); | |
5909 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
5910 | +#else | |
5911 | + *((long long int *)data)=*(((long long int *)tab)+g); | |
5912 | +#endif | |
5913 | +} | |
5914 | + | |
5915 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
5916 | + int j; | |
5917 | + for(j=0;j<n;j++){ | |
5918 | + *(data+j)^=*(tab+8*g+j); | |
5919 | + } | |
5920 | +} | |
5921 | + | |
5922 | +typedef __m64 batch; | |
5923 | +#define BYTES_PER_BATCH 8 | |
5924 | +#define B_FFAND(a,b) _m_pand((a),(b)) | |
5925 | +#define B_FFOR(a,b) _m_por((a),(b)) | |
5926 | +#define B_FFXOR(a,b) _m_pxor((a),(b)) | |
5927 | +#define B_FFN_ALL_29() ((__m64)0x2929292929292929ULL) | |
5928 | +#define B_FFN_ALL_02() ((__m64)0x0202020202020202ULL) | |
5929 | +#define B_FFN_ALL_04() ((__m64)0x0404040404040404ULL) | |
5930 | +#define B_FFN_ALL_10() ((__m64)0x1010101010101010ULL) | |
5931 | +#define B_FFN_ALL_40() ((__m64)0x4040404040404040ULL) | |
5932 | +#define B_FFN_ALL_80() ((__m64)0x8080808080808080ULL) | |
5933 | +#define B_FFSH8L(a,n) _m_psllqi((a),(n)) | |
5934 | +#define B_FFSH8R(a,n) _m_psrlqi((a),(n)) | |
5935 | + | |
5936 | +#define M_EMPTY() _m_empty() | |
5937 | + | |
5938 | + | |
5939 | +#undef XOR_8_BY | |
5940 | +#define XOR_8_BY(d,s1,s2) do{ __m64 *pd=(__m64 *)(d), *ps1=(__m64 *)(s1), *ps2=(__m64 *)(s2); \ | |
5941 | + *pd = _m_pxor( *ps1 , *ps2 ); }while(0) | |
5942 | + | |
5943 | +#undef XOREQ_8_BY | |
5944 | +#define XOREQ_8_BY(d,s) do{ __m64 *pd=(__m64 *)(d), *ps=(__m64 *)(s); \ | |
5945 | + *pd = _m_pxor( *ps, *pd ); }while(0) | |
5946 | + | |
5947 | +#undef COPY_8_BY | |
5948 | +#define COPY_8_BY(d,s) do{ __m64 *pd=(__m64 *)(d), *ps=(__m64 *)(s); \ | |
5949 | + *pd = *ps; }while(0) | |
5950 | + | |
5951 | +#undef BEST_SPAN | |
5952 | +#define BEST_SPAN 8 | |
5953 | + | |
5954 | +#undef XOR_BEST_BY | |
5955 | +#define XOR_BEST_BY(d,s1,s2) do{ XOR_8_BY(d,s1,s2); }while(0); | |
5956 | + | |
5957 | +#undef XOREQ_BEST_BY | |
5958 | +#define XOREQ_BEST_BY(d,s) do{ XOREQ_8_BY(d,s); }while(0); | |
5959 | + | |
5960 | +#undef COPY_BEST_BY | |
5961 | +#define COPY_BEST_BY(d,s) do{ COPY_8_BY(d,s); }while(0); | |
5962 | Index: libs/libmythtv/dvbcam.h | |
5963 | =================================================================== | |
5964 | --- libs/libmythtv/dvbcam.h.orig 2006-06-20 17:36:05.000000000 -0400 | |
5965 | +++ libs/libmythtv/dvbcam.h 2006-06-20 17:36:06.000000000 -0400 | |
5966 | @@ -50,6 +50,8 @@ | |
5967 | void sendExternalCam(const PMTObject *pmt); | |
5968 | bool useExternalCam(); | |
5969 | void stopExternalCam(); | |
5970 | + void ExternalCamSoftkeyLoop(); | |
5971 | + static void *ExternalCamSoftkeyHelper(void*self); | |
5972 | ||
5973 | }; | |
5974 | ||
5975 | Index: libs/libmythtv/FFdecsa/logic/Makefile | |
5976 | =================================================================== | |
5977 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
5978 | +++ libs/libmythtv/FFdecsa/logic/Makefile 2006-06-20 17:36:06.000000000 -0400 | |
5979 | @@ -0,0 +1,10 @@ | |
5980 | +all: logic | |
5981 | + | |
5982 | +logic: logic.o | |
5983 | + gcc -o logic logic.o | |
5984 | + | |
5985 | +logic.o: logic.c | |
5986 | + gcc -O3 -march=athlon-xp -c logic.c | |
5987 | + | |
5988 | +clean: | |
5989 | + rm logic *.o | |
5990 | Index: libs/libmythtv/FFdecsa/ChangeLog | |
5991 | =================================================================== | |
5992 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
5993 | +++ libs/libmythtv/FFdecsa/ChangeLog 2006-06-20 17:36:06.000000000 -0400 | |
5994 | @@ -0,0 +1,206 @@ | |
5995 | +- created | |
5996 | + | |
5997 | +- released 0.0.1 | |
5998 | + | |
5999 | +- simplified s, A, B | |
6000 | + | |
6001 | +- released 0.0.2 | |
6002 | + | |
6003 | +- simplified nxt= | |
6004 | + | |
6005 | +- released 0.0.3 | |
6006 | + | |
6007 | +- removed commented code | |
6008 | +- code formatting | |
6009 | + | |
6010 | +- released 0.0.4 | |
6011 | + | |
6012 | +- kk now unsigned char | |
6013 | +- removed 64 bit ints | |
6014 | + | |
6015 | +- released 0.0.5 | |
6016 | + | |
6017 | +- created decrypt_2ts | |
6018 | + | |
6019 | +- released 0.0.6 | |
6020 | + | |
6021 | +- renamed files | |
6022 | +- created decrypt_many_ts, removed others | |
6023 | +- external interface has 2 functions only: set_cws() and decrypt_many_ts() | |
6024 | +- reformatted code | |
6025 | +- reimplemented s12,s34,s56,s7 | |
6026 | +- unsigned char become int for table optimization | |
6027 | + | |
6028 | +- released 0.0.7 | |
6029 | + | |
6030 | +- optional icc compiler | |
6031 | +- kk now 0..55 | |
6032 | +- decrypt_many_ts really works (no parallelism yet) | |
6033 | +- added get_cws() to interface | |
6034 | +- created stream.c | |
6035 | +- created key_schedule_stream, using iA[] and iB[] | |
6036 | + | |
6037 | +- released 0.0.8 | |
6038 | + | |
6039 | +- decrypt_many_ts() makes a group, sorts the packets, processes them | |
6040 | +- preliminar stream_cypher_group() created | |
6041 | +- parallel computing activated | |
6042 | +- huge speed increase (+500%) thanks to stream_cypher_group() | |
6043 | + | |
6044 | +- released 0.0.9 | |
6045 | + | |
6046 | +- block_cypher_group() created (no parallelism yet) | |
6047 | + | |
6048 | +- released 0.0.10 | |
6049 | + | |
6050 | +- block_cypher_group() has 56 simple iterations | |
6051 | +- block_cypher_group() doesn't shift registers anymore | |
6052 | + | |
6053 | +- released 0.0.11 | |
6054 | + | |
6055 | +- some parallelization on block_cypher_group() | |
6056 | + | |
6057 | +- released 0.0.12 | |
6058 | + | |
6059 | +- better parallelization of block_cypher_group() | |
6060 | + | |
6061 | +- released 0.0.13 | |
6062 | + | |
6063 | +- block_cypher() was still called by error when N=23 | |
6064 | +- speed is now 109Mbit/s on AMD XP2000+ CPU | |
6065 | + | |
6066 | +- released 0.0.14 | |
6067 | + | |
6068 | +- stream_cypher_group() has a init and normal variant | |
6069 | +- A[0]-A[9] instead of A[1]-A[10], same for B | |
6070 | +- implemented virtual shift of A and B | |
6071 | +- speed is now 117Mbit/s on AMD XP2000+ CPU | |
6072 | + | |
6073 | +- released 0.0.15 | |
6074 | + | |
6075 | +- better optimization of E and F in the stream cypher | |
6076 | +- speed is now 119Mbit/s on AMD XP2000+ CPU | |
6077 | + | |
6078 | +- released 0.0.16 | |
6079 | + | |
6080 | +- removed some debug overhead | |
6081 | +- speed is now 120Mbit/s on AMD XP2000+ CPU | |
6082 | + | |
6083 | +- released 0.0.17 | |
6084 | + | |
6085 | +- don't move packets with residue anymore | |
6086 | +- speed is now 123Mbit/s on AMD XP2000+ CPU | |
6087 | + | |
6088 | +- released 0.0.18 | |
6089 | + | |
6090 | +- solved alignment problems | |
6091 | +- search groupable packets even beyond ungroupable ones | |
6092 | + (more speed in some real world cases) | |
6093 | +- created decrypt_many_ts2(), useful with circular buffers | |
6094 | + | |
6095 | +- released 0.0.19 | |
6096 | + | |
6097 | +- removed old code | |
6098 | + | |
6099 | +- released 0.0.20 | |
6100 | + | |
6101 | +- partially converted code to size-independent group | |
6102 | +- icc doesn't work with optimizations on | |
6103 | + | |
6104 | +- released 0.1.1 | |
6105 | + | |
6106 | +- merge loops on block_decypher (speed++ gcc, speed-- icc) | |
6107 | +- transposition are now functions (speed-- icc) | |
6108 | +- icc works again (compiler bug work around?) | |
6109 | + | |
6110 | +- released 0.1.2 | |
6111 | + | |
6112 | +- better use of COPY8 &co | |
6113 | +- better flags for gcc | |
6114 | +- removed old code | |
6115 | + | |
6116 | +- released 0.1.3 | |
6117 | + | |
6118 | +- int and not char in block cypher (speed++++++ gcc, speed-- icc) | |
6119 | + | |
6120 | +- released 0.1.4 | |
6121 | + | |
6122 | +- group abstraction finally implemented | |
6123 | +- support for group width 64 | |
6124 | + | |
6125 | +- released 0.1.5 | |
6126 | + | |
6127 | +- group 64 mmx implemented (speed++ gcc) | |
6128 | + | |
6129 | +- released 0.1.6 | |
6130 | + | |
6131 | +- more parallelism in block cypher (speed++ gcc) | |
6132 | +- transposition before and after block (disabled because of no speed gain yet) | |
6133 | + | |
6134 | +- released 0.1.7 | |
6135 | + | |
6136 | +- more parallelism in block cypher (speed++ gcc) | |
6137 | +- transposition before and after block enabled (speed++ gcc) | |
6138 | +- gcc options (unrolled 500) speed gcc++ | |
6139 | + | |
6140 | +- released 0.1.8 | |
6141 | + | |
6142 | +- reworked FFN_ALL_* constants (speed++++ gcc) | |
6143 | + | |
6144 | +- released 0.1.9 | |
6145 | + | |
6146 | +- transposition in block as inlined functions | |
6147 | +- group abstraction working well | |
6148 | + | |
6149 | +- released 0.1.10 | |
6150 | + | |
6151 | +- group 128 sse implemented, but batch is 64 mmx (not faster than group 64 mmx) | |
6152 | + | |
6153 | +- released 0.1.11 | |
6154 | + | |
6155 | +- lot of code polishing and dead code elimination | |
6156 | +- better and more debug output | |
6157 | + | |
6158 | +- released 0.1.12 | |
6159 | + | |
6160 | +- name change: FFdecsa | |
6161 | + | |
6162 | +- released 0.2.0 | |
6163 | + | |
6164 | +- separated test cases | |
6165 | +- corrected all group_modes (now called parallel_modes) | |
6166 | +- parallel 128 8 char implemented | |
6167 | +- parallel 64 long implemented | |
6168 | +- parallel 128 2 long implemented | |
6169 | +- parallel 128 2 mmx implemented (incredibly slow, the compiler is very confused) | |
6170 | +- parallel 128 16 charA implemented (very slow compilation) | |
6171 | +- parallel 128 16 char implemented | |
6172 | +- renamed softcsa* to FFdecsa* | |
6173 | + | |
6174 | +- released 0.2.1 | |
6175 | + | |
6176 | +- new external interface (based on ranges) | |
6177 | + | |
6178 | +- released 0.2.2 | |
6179 | + | |
6180 | +- can be compiled with g++ too | |
6181 | +- using g++ the code is 3% faster! | |
6182 | +- external interface: function name changing and new functions | |
6183 | +- a group of ranges is now called a cluster | |
6184 | +- renamed autogenerated files | |
6185 | + | |
6186 | +- released 0.2.3 | |
6187 | + | |
6188 | +- written docs | |
6189 | +- removed unneeded files | |
6190 | +- added Copyright and license notes | |
6191 | +- reworked "logic" | |
6192 | + | |
6193 | +- released 0.3.0 | |
6194 | + | |
6195 | +- Makefile reworked | |
6196 | +- misc fixes | |
6197 | +- added vdr patch | |
6198 | + | |
6199 | +- released 1.0.0 (public release) | |
6200 | + | |
6201 | Index: libs/libmythtv/FFdecsa/parallel_064_8char.h | |
6202 | =================================================================== | |
6203 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
6204 | +++ libs/libmythtv/FFdecsa/parallel_064_8char.h 2006-06-20 17:36:06.000000000 -0400 | |
6205 | @@ -0,0 +1,274 @@ | |
6206 | +/* FFdecsa -- fast decsa algorithm | |
6207 | + * | |
6208 | + * Copyright (C) 2003-2004 fatih89r | |
6209 | + * | |
6210 | + * This program is free software; you can redistribute it and/or modify | |
6211 | + * it under the terms of the GNU General Public License as published by | |
6212 | + * the Free Software Foundation; either version 2 of the License, or | |
6213 | + * (at your option) any later version. | |
6214 | + * | |
6215 | + * This program is distributed in the hope that it will be useful, | |
6216 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
6217 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
6218 | + * GNU General Public License for more details. | |
6219 | + * | |
6220 | + * You should have received a copy of the GNU General Public License | |
6221 | + * along with this program; if not, write to the Free Software | |
6222 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
6223 | + */ | |
6224 | + | |
6225 | + | |
6226 | +struct group_t{ | |
6227 | + unsigned char s1,s2,s3,s4,s5,s6,s7,s8; | |
6228 | +}; | |
6229 | +typedef struct group_t group; | |
6230 | + | |
6231 | +#define GROUP_PARALLELISM 64 | |
6232 | + | |
6233 | +group static inline FF0(){ | |
6234 | + group res; | |
6235 | + res.s1=0x0; | |
6236 | + res.s2=0x0; | |
6237 | + res.s3=0x0; | |
6238 | + res.s4=0x0; | |
6239 | + res.s5=0x0; | |
6240 | + res.s6=0x0; | |
6241 | + res.s7=0x0; | |
6242 | + res.s8=0x0; | |
6243 | + return res; | |
6244 | +} | |
6245 | + | |
6246 | +group static inline FF1(){ | |
6247 | + group res; | |
6248 | + res.s1=0xff; | |
6249 | + res.s2=0xff; | |
6250 | + res.s3=0xff; | |
6251 | + res.s4=0xff; | |
6252 | + res.s5=0xff; | |
6253 | + res.s6=0xff; | |
6254 | + res.s7=0xff; | |
6255 | + res.s8=0xff; | |
6256 | + return res; | |
6257 | +} | |
6258 | + | |
6259 | +group static inline FFAND(group a,group b){ | |
6260 | + group res; | |
6261 | + res.s1=a.s1&b.s1; | |
6262 | + res.s2=a.s2&b.s2; | |
6263 | + res.s3=a.s3&b.s3; | |
6264 | + res.s4=a.s4&b.s4; | |
6265 | + res.s5=a.s5&b.s5; | |
6266 | + res.s6=a.s6&b.s6; | |
6267 | + res.s7=a.s7&b.s7; | |
6268 | + res.s8=a.s8&b.s8; | |
6269 | + return res; | |
6270 | +} | |
6271 | + | |
6272 | +group static inline FFOR(group a,group b){ | |
6273 | + group res; | |
6274 | + res.s1=a.s1|b.s1; | |
6275 | + res.s2=a.s2|b.s2; | |
6276 | + res.s3=a.s3|b.s3; | |
6277 | + res.s4=a.s4|b.s4; | |
6278 | + res.s5=a.s5|b.s5; | |
6279 | + res.s6=a.s6|b.s6; | |
6280 | + res.s7=a.s7|b.s7; | |
6281 | + res.s8=a.s8|b.s8; | |
6282 | + return res; | |
6283 | +} | |
6284 | + | |
6285 | +group static inline FFXOR(group a,group b){ | |
6286 | + group res; | |
6287 | + res.s1=a.s1^b.s1; | |
6288 | + res.s2=a.s2^b.s2; | |
6289 | + res.s3=a.s3^b.s3; | |
6290 | + res.s4=a.s4^b.s4; | |
6291 | + res.s5=a.s5^b.s5; | |
6292 | + res.s6=a.s6^b.s6; | |
6293 | + res.s7=a.s7^b.s7; | |
6294 | + res.s8=a.s8^b.s8; | |
6295 | + return res; | |
6296 | +} | |
6297 | + | |
6298 | +group static inline FFNOT(group a){ | |
6299 | + group res; | |
6300 | + res.s1=~a.s1; | |
6301 | + res.s2=~a.s2; | |
6302 | + res.s3=~a.s3; | |
6303 | + res.s4=~a.s4; | |
6304 | + res.s5=~a.s5; | |
6305 | + res.s6=~a.s6; | |
6306 | + res.s7=~a.s7; | |
6307 | + res.s8=~a.s8; | |
6308 | + return res; | |
6309 | +} | |
6310 | + | |
6311 | + | |
6312 | +/* 64 rows of 64 bits */ | |
6313 | + | |
6314 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
6315 | + *(((int *)tab)+2*g)=*((int *)data); | |
6316 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
6317 | +} | |
6318 | + | |
6319 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
6320 | + *((int *)data)=*(((int *)tab)+2*g); | |
6321 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
6322 | +} | |
6323 | + | |
6324 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
6325 | + int j; | |
6326 | + for(j=0;j<n;j++){ | |
6327 | + *(data+j)^=*(tab+8*g+j); | |
6328 | + } | |
6329 | +} | |
6330 | + | |
6331 | +struct batch_t{ | |
6332 | + unsigned char s1,s2,s3,s4,s5,s6,s7,s8; | |
6333 | +}; | |
6334 | +typedef struct batch_t batch; | |
6335 | + | |
6336 | +#define BYTES_PER_BATCH 8 | |
6337 | + | |
6338 | +batch static inline B_FFAND(batch a,batch b){ | |
6339 | + batch res; | |
6340 | + res.s1=a.s1&b.s1; | |
6341 | + res.s2=a.s2&b.s2; | |
6342 | + res.s3=a.s3&b.s3; | |
6343 | + res.s4=a.s4&b.s4; | |
6344 | + res.s5=a.s5&b.s5; | |
6345 | + res.s6=a.s6&b.s6; | |
6346 | + res.s7=a.s7&b.s7; | |
6347 | + res.s8=a.s8&b.s8; | |
6348 | + return res; | |
6349 | +} | |
6350 | + | |
6351 | +batch static inline B_FFOR(batch a,batch b){ | |
6352 | + batch res; | |
6353 | + res.s1=a.s1|b.s1; | |
6354 | + res.s2=a.s2|b.s2; | |
6355 | + res.s3=a.s3|b.s3; | |
6356 | + res.s4=a.s4|b.s4; | |
6357 | + res.s5=a.s5|b.s5; | |
6358 | + res.s6=a.s6|b.s6; | |
6359 | + res.s7=a.s7|b.s7; | |
6360 | + res.s8=a.s8|b.s8; | |
6361 | + return res; | |
6362 | +} | |
6363 | + | |
6364 | +batch static inline B_FFXOR(batch a,batch b){ | |
6365 | + batch res; | |
6366 | + res.s1=a.s1^b.s1; | |
6367 | + res.s2=a.s2^b.s2; | |
6368 | + res.s3=a.s3^b.s3; | |
6369 | + res.s4=a.s4^b.s4; | |
6370 | + res.s5=a.s5^b.s5; | |
6371 | + res.s6=a.s6^b.s6; | |
6372 | + res.s7=a.s7^b.s7; | |
6373 | + res.s8=a.s8^b.s8; | |
6374 | + return res; | |
6375 | +} | |
6376 | + | |
6377 | + | |
6378 | +batch static inline B_FFN_ALL_29(){ | |
6379 | + batch res; | |
6380 | + res.s1=0x29; | |
6381 | + res.s2=0x29; | |
6382 | + res.s3=0x29; | |
6383 | + res.s4=0x29; | |
6384 | + res.s5=0x29; | |
6385 | + res.s6=0x29; | |
6386 | + res.s7=0x29; | |
6387 | + res.s8=0x29; | |
6388 | + return res; | |
6389 | +} | |
6390 | +batch static inline B_FFN_ALL_02(){ | |
6391 | + batch res; | |
6392 | + res.s1=0x02; | |
6393 | + res.s2=0x02; | |
6394 | + res.s3=0x02; | |
6395 | + res.s4=0x02; | |
6396 | + res.s5=0x02; | |
6397 | + res.s6=0x02; | |
6398 | + res.s7=0x02; | |
6399 | + res.s8=0x02; | |
6400 | + return res; | |
6401 | +} | |
6402 | +batch static inline B_FFN_ALL_04(){ | |
6403 | + batch res; | |
6404 | + res.s1=0x04; | |
6405 | + res.s2=0x04; | |
6406 | + res.s3=0x04; | |
6407 | + res.s4=0x04; | |
6408 | + res.s5=0x04; | |
6409 | + res.s6=0x04; | |
6410 | + res.s7=0x04; | |
6411 | + res.s8=0x04; | |
6412 | + return res; | |
6413 | +} | |
6414 | +batch static inline B_FFN_ALL_10(){ | |
6415 | + batch res; | |
6416 | + res.s1=0x10; | |
6417 | + res.s2=0x10; | |
6418 | + res.s3=0x10; | |
6419 | + res.s4=0x10; | |
6420 | + res.s5=0x10; | |
6421 | + res.s6=0x10; | |
6422 | + res.s7=0x10; | |
6423 | + res.s8=0x10; | |
6424 | + return res; | |
6425 | +} | |
6426 | +batch static inline B_FFN_ALL_40(){ | |
6427 | + batch res; | |
6428 | + res.s1=0x40; | |
6429 | + res.s2=0x40; | |
6430 | + res.s3=0x40; | |
6431 | + res.s4=0x40; | |
6432 | + res.s5=0x40; | |
6433 | + res.s6=0x40; | |
6434 | + res.s7=0x40; | |
6435 | + res.s8=0x40; | |
6436 | + return res; | |
6437 | +} | |
6438 | +batch static inline B_FFN_ALL_80(){ | |
6439 | + batch res; | |
6440 | + res.s1=0x80; | |
6441 | + res.s2=0x80; | |
6442 | + res.s3=0x80; | |
6443 | + res.s4=0x80; | |
6444 | + res.s5=0x80; | |
6445 | + res.s6=0x80; | |
6446 | + res.s7=0x80; | |
6447 | + res.s8=0x80; | |
6448 | + return res; | |
6449 | +} | |
6450 | + | |
6451 | +batch static inline B_FFSH8L(batch a,int n){ | |
6452 | + batch res; | |
6453 | + res.s1=a.s1<<n; | |
6454 | + res.s2=a.s2<<n; | |
6455 | + res.s3=a.s3<<n; | |
6456 | + res.s4=a.s4<<n; | |
6457 | + res.s5=a.s5<<n; | |
6458 | + res.s6=a.s6<<n; | |
6459 | + res.s7=a.s7<<n; | |
6460 | + res.s8=a.s8<<n; | |
6461 | + return res; | |
6462 | +} | |
6463 | + | |
6464 | +batch static inline B_FFSH8R(batch a,int n){ | |
6465 | + batch res; | |
6466 | + res.s1=a.s1>>n; | |
6467 | + res.s2=a.s2>>n; | |
6468 | + res.s3=a.s3>>n; | |
6469 | + res.s4=a.s4>>n; | |
6470 | + res.s5=a.s5>>n; | |
6471 | + res.s6=a.s6>>n; | |
6472 | + res.s7=a.s7>>n; | |
6473 | + res.s8=a.s8>>n; | |
6474 | + return res; | |
6475 | +} | |
6476 | + | |
6477 | + | |
6478 | +void static inline M_EMPTY(void){ | |
6479 | +} | |
6480 | Index: libs/libmythtv/FFdecsa/docs/how_to_release.txt | |
6481 | =================================================================== | |
6482 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
6483 | +++ libs/libmythtv/FFdecsa/docs/how_to_release.txt 2006-06-20 17:36:06.000000000 -0400 | |
6484 | @@ -0,0 +1,21 @@ | |
6485 | +------- | |
6486 | +FFdecsa | |
6487 | +------- | |
6488 | + | |
6489 | +Please use the name of the release you're basing on as a base name and | |
6490 | +add your suffix. | |
6491 | + | |
6492 | +For example if john modifies | |
6493 | + FFdecsa-1.0.0 | |
6494 | +he should release | |
6495 | + FFdecsa-1.0.0-john_0.3 | |
6496 | +or | |
6497 | + FFdecsa-1.0.0-john_0.4 | |
6498 | + | |
6499 | +If paul modifies john's version the correct name would be like | |
6500 | + FFdecsa-1.0.0-john_0.4-paul_0.1 | |
6501 | + | |
6502 | +This is to avoid many different versions with random version numbers, as | |
6503 | +development is not centralized. | |
6504 | + | |
6505 | +Thank you. | |
6506 | Index: libs/libmythtv/FFdecsa/parallel_064_long.h | |
6507 | =================================================================== | |
6508 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
6509 | +++ libs/libmythtv/FFdecsa/parallel_064_long.h 2006-06-20 17:36:06.000000000 -0400 | |
6510 | @@ -0,0 +1,56 @@ | |
6511 | +/* FFdecsa -- fast decsa algorithm | |
6512 | + * | |
6513 | + * Copyright (C) 2003-2004 fatih89r | |
6514 | + * | |
6515 | + * This program is free software; you can redistribute it and/or modify | |
6516 | + * it under the terms of the GNU General Public License as published by | |
6517 | + * the Free Software Foundation; either version 2 of the License, or | |
6518 | + * (at your option) any later version. | |
6519 | + * | |
6520 | + * This program is distributed in the hope that it will be useful, | |
6521 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
6522 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
6523 | + * GNU General Public License for more details. | |
6524 | + * | |
6525 | + * You should have received a copy of the GNU General Public License | |
6526 | + * along with this program; if not, write to the Free Software | |
6527 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
6528 | + */ | |
6529 | + | |
6530 | +#include "parallel_std_def.h" | |
6531 | + | |
6532 | +typedef unsigned long long group; | |
6533 | +#define GROUP_PARALLELISM 64 | |
6534 | +#define FF0() 0x0ULL | |
6535 | +#define FF1() 0xffffffffffffffffULL | |
6536 | + | |
6537 | +/* 64 rows of 64 bits */ | |
6538 | + | |
6539 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
6540 | + *(((int *)tab)+2*g)=*((int *)data); | |
6541 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
6542 | +} | |
6543 | + | |
6544 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
6545 | + *((int *)data)=*(((int *)tab)+2*g); | |
6546 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
6547 | +} | |
6548 | + | |
6549 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
6550 | + int j; | |
6551 | + for(j=0;j<n;j++){ | |
6552 | + *(data+j)^=*(tab+8*g+j); | |
6553 | + } | |
6554 | +} | |
6555 | + | |
6556 | +typedef unsigned long long int batch; | |
6557 | +#define BYTES_PER_BATCH 8 | |
6558 | +#define B_FFN_ALL_29() 0x2929292929292929ULL | |
6559 | +#define B_FFN_ALL_02() 0x0202020202020202ULL | |
6560 | +#define B_FFN_ALL_04() 0x0404040404040404ULL | |
6561 | +#define B_FFN_ALL_10() 0x1010101010101010ULL | |
6562 | +#define B_FFN_ALL_40() 0x4040404040404040ULL | |
6563 | +#define B_FFN_ALL_80() 0x8080808080808080ULL | |
6564 | + | |
6565 | +#define M_EMPTY() | |
6566 | + | |
6567 | Index: libs/libmythtv/FFdecsa/parallel_128_2long.h | |
6568 | =================================================================== | |
6569 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
6570 | +++ libs/libmythtv/FFdecsa/parallel_128_2long.h 2006-06-20 17:36:06.000000000 -0400 | |
6571 | @@ -0,0 +1,175 @@ | |
6572 | +/* FFdecsa -- fast decsa algorithm | |
6573 | + * | |
6574 | + * Copyright (C) 2003-2004 fatih89r | |
6575 | + * | |
6576 | + * This program is free software; you can redistribute it and/or modify | |
6577 | + * it under the terms of the GNU General Public License as published by | |
6578 | + * the Free Software Foundation; either version 2 of the License, or | |
6579 | + * (at your option) any later version. | |
6580 | + * | |
6581 | + * This program is distributed in the hope that it will be useful, | |
6582 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
6583 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
6584 | + * GNU General Public License for more details. | |
6585 | + * | |
6586 | + * You should have received a copy of the GNU General Public License | |
6587 | + * along with this program; if not, write to the Free Software | |
6588 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
6589 | + */ | |
6590 | + | |
6591 | + | |
6592 | +struct group_t{ | |
6593 | + unsigned long long int s1; | |
6594 | + unsigned long long int s2; | |
6595 | +}; | |
6596 | +typedef struct group_t group; | |
6597 | + | |
6598 | +#define GROUP_PARALLELISM 128 | |
6599 | + | |
6600 | +group static inline FF0(){ | |
6601 | + group res; | |
6602 | + res.s1=0x0ULL; | |
6603 | + res.s2=0x0ULL; | |
6604 | + return res; | |
6605 | +} | |
6606 | + | |
6607 | +group static inline FF1(){ | |
6608 | + group res; | |
6609 | + res.s1=0xffffffffffffffffULL; | |
6610 | + res.s2=0xffffffffffffffffULL; | |
6611 | + return res; | |
6612 | +} | |
6613 | + | |
6614 | +group static inline FFAND(group a,group b){ | |
6615 | + group res; | |
6616 | + res.s1=a.s1&b.s1; | |
6617 | + res.s2=a.s2&b.s2; | |
6618 | + return res; | |
6619 | +} | |
6620 | + | |
6621 | +group static inline FFOR(group a,group b){ | |
6622 | + group res; | |
6623 | + res.s1=a.s1|b.s1; | |
6624 | + res.s2=a.s2|b.s2; | |
6625 | + return res; | |
6626 | +} | |
6627 | + | |
6628 | +group static inline FFXOR(group a,group b){ | |
6629 | + group res; | |
6630 | + res.s1=a.s1^b.s1; | |
6631 | + res.s2=a.s2^b.s2; | |
6632 | + return res; | |
6633 | +} | |
6634 | + | |
6635 | +group static inline FFNOT(group a){ | |
6636 | + group res; | |
6637 | + res.s1=~a.s1; | |
6638 | + res.s2=~a.s2; | |
6639 | + return res; | |
6640 | +} | |
6641 | + | |
6642 | + | |
6643 | +/* 64 rows of 128 bits */ | |
6644 | + | |
6645 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
6646 | + *(((int *)tab)+2*g)=*((int *)data); | |
6647 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
6648 | +} | |
6649 | + | |
6650 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
6651 | + *((int *)data)=*(((int *)tab)+2*g); | |
6652 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
6653 | +} | |
6654 | + | |
6655 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
6656 | + int j; | |
6657 | + for(j=0;j<n;j++){ | |
6658 | + *(data+j)^=*(tab+8*g+j); | |
6659 | + } | |
6660 | +} | |
6661 | + | |
6662 | + | |
6663 | +struct batch_t{ | |
6664 | + unsigned long long int s1; | |
6665 | + unsigned long long int s2; | |
6666 | +}; | |
6667 | +typedef struct batch_t batch; | |
6668 | + | |
6669 | +#define BYTES_PER_BATCH 16 | |
6670 | + | |
6671 | +batch static inline B_FFAND(batch a,batch b){ | |
6672 | + batch res; | |
6673 | + res.s1=a.s1&b.s1; | |
6674 | + res.s2=a.s2&b.s2; | |
6675 | + return res; | |
6676 | +} | |
6677 | + | |
6678 | +batch static inline B_FFOR(batch a,batch b){ | |
6679 | + batch res; | |
6680 | + res.s1=a.s1|b.s1; | |
6681 | + res.s2=a.s2|b.s2; | |
6682 | + return res; | |
6683 | +} | |
6684 | + | |
6685 | +batch static inline B_FFXOR(batch a,batch b){ | |
6686 | + batch res; | |
6687 | + res.s1=a.s1^b.s1; | |
6688 | + res.s2=a.s2^b.s2; | |
6689 | + return res; | |
6690 | +} | |
6691 | + | |
6692 | + | |
6693 | +batch static inline B_FFN_ALL_29(){ | |
6694 | + batch res; | |
6695 | + res.s1=0x2929292929292929ULL; | |
6696 | + res.s2=0x2929292929292929ULL; | |
6697 | + return res; | |
6698 | +} | |
6699 | + | |
6700 | +batch static inline B_FFN_ALL_02(){ | |
6701 | + batch res; | |
6702 | + res.s1=0x0202020202020202ULL; | |
6703 | + res.s2=0x0202020202020202ULL; | |
6704 | + return res; | |
6705 | +} | |
6706 | +batch static inline B_FFN_ALL_04(){ | |
6707 | + batch res; | |
6708 | + res.s1=0x0404040404040404ULL; | |
6709 | + res.s2=0x0404040404040404ULL; | |
6710 | + return res; | |
6711 | +} | |
6712 | +batch static inline B_FFN_ALL_10(){ | |
6713 | + batch res; | |
6714 | + res.s1=0x1010101010101010ULL; | |
6715 | + res.s2=0x1010101010101010ULL; | |
6716 | + return res; | |
6717 | +} | |
6718 | +batch static inline B_FFN_ALL_40(){ | |
6719 | + batch res; | |
6720 | + res.s1=0x4040404040404040ULL; | |
6721 | + res.s2=0x4040404040404040ULL; | |
6722 | + return res; | |
6723 | +} | |
6724 | +batch static inline B_FFN_ALL_80(){ | |
6725 | + batch res; | |
6726 | + res.s1=0x8080808080808080ULL; | |
6727 | + res.s2=0x8080808080808080ULL; | |
6728 | + return res; | |
6729 | +} | |
6730 | +batch static inline B_FFSH8L(batch a,int n){ | |
6731 | + batch res; | |
6732 | + res.s1=a.s1<<n; | |
6733 | + res.s2=a.s2<<n; | |
6734 | + return res; | |
6735 | +} | |
6736 | + | |
6737 | +batch static inline B_FFSH8R(batch a,int n){ | |
6738 | + batch res; | |
6739 | + res.s1=a.s1>>n; | |
6740 | + res.s2=a.s2>>n; | |
6741 | + return res; | |
6742 | +} | |
6743 | + | |
6744 | + | |
6745 | +void static inline M_EMPTY(void){ | |
6746 | +} | |
6747 | Index: libs/libmythtv/FFdecsa/parallel_std_def.h | |
6748 | =================================================================== | |
6749 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
6750 | +++ libs/libmythtv/FFdecsa/parallel_std_def.h 2006-06-20 17:36:06.000000000 -0400 | |
6751 | @@ -0,0 +1,29 @@ | |
6752 | +/* FFdecsa -- fast decsa algorithm | |
6753 | + * | |
6754 | + * Copyright (C) 2003-2004 fatih89r | |
6755 | + * | |
6756 | + * This program is free software; you can redistribute it and/or modify | |
6757 | + * it under the terms of the GNU General Public License as published by | |
6758 | + * the Free Software Foundation; either version 2 of the License, or | |
6759 | + * (at your option) any later version. | |
6760 | + * | |
6761 | + * This program is distributed in the hope that it will be useful, | |
6762 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
6763 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
6764 | + * GNU General Public License for more details. | |
6765 | + * | |
6766 | + * You should have received a copy of the GNU General Public License | |
6767 | + * along with this program; if not, write to the Free Software | |
6768 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
6769 | + */ | |
6770 | + | |
6771 | +#define FFXOR(a,b) ((a)^(b)) | |
6772 | +#define FFAND(a,b) ((a)&(b)) | |
6773 | +#define FFOR(a,b) ((a)|(b)) | |
6774 | +#define FFNOT(a) (~(a)) | |
6775 | + | |
6776 | +#define B_FFAND(a,b) ((a)&(b)) | |
6777 | +#define B_FFOR(a,b) ((a)|(b)) | |
6778 | +#define B_FFXOR(a,b) ((a)^(b)) | |
6779 | +#define B_FFSH8L(a,n) ((a)<<(n)) | |
6780 | +#define B_FFSH8R(a,n) ((a)>>(n)) | |
6781 | Index: libs/libmythtv/FFdecsa/docs/technical_background.txt | |
6782 | =================================================================== | |
6783 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
6784 | +++ libs/libmythtv/FFdecsa/docs/technical_background.txt 2006-06-20 17:36:06.000000000 -0400 | |
6785 | @@ -0,0 +1,341 @@ | |
6786 | +------- | |
6787 | +FFdecsa | |
6788 | +------- | |
6789 | + | |
6790 | +This doc is for people who looked into the source code and found it | |
6791 | +difficult to believe that this is a decsa algorithm, as it appears | |
6792 | +completely different from other decsa implementations. | |
6793 | + | |
6794 | +It appears different because it is different. Being different is what | |
6795 | +enables it to be a lot faster than all the others (currently it has more | |
6796 | +than 800% the speed of the best version I was able to find) | |
6797 | + | |
6798 | +The csa algo was designed to be run in hardware, but people are now | |
6799 | +running it in software. | |
6800 | + | |
6801 | +Hardware has data lines carrying bits and functional blocks doing | |
6802 | +calculations (logic operations, adders, shifters, table lookup, ...), | |
6803 | +software instead uses memory to contain data values and executes a | |
6804 | +sequence of instructions to transform the values. As a consequence, | |
6805 | +writing a software implementation of a hardware algorithm can be | |
6806 | +inefficient. | |
6807 | + | |
6808 | +For example, if you have 32 data lines, you can permutate the bits with | |
6809 | +zero cost in hardware (you just permute the physical traces), but if you | |
6810 | +have the bits in a 32 bit variable you have to use 32 "and" operations | |
6811 | +with 32 different masks, 32 shifts and 31 "or" operations (if you | |
6812 | +suggest using "if"s testing the bits one by one you know nothing about | |
6813 | +how jump prediction works in modern processors). | |
6814 | + | |
6815 | +So the approach is *emulating the hardware*. | |
6816 | + | |
6817 | +Then there are some additional cool tricks. | |
6818 | + | |
6819 | +TRICK NUMBER 0: emulate the hardware | |
6820 | +------------------------------------ | |
6821 | +We will work on bits one by one, that is a 4 bit word is now four | |
6822 | +variables. In this way we revert complex software operations into | |
6823 | +hardware emulation: | |
6824 | + | |
6825 | + software hardware | |
6826 | + ------------------------------------------- | |
6827 | + copy values copy values | |
6828 | + logic op logic op | |
6829 | + (bit permut.) ands+shifts+ors copy values | |
6830 | + additions logic op emulating adders | |
6831 | + (comparisons) if logic op selecting one of the two results | |
6832 | + lookup tables logic op synthetizing a ROM (*) | |
6833 | + | |
6834 | +(*) sometimes lookup tables can be converted to logic expressions | |
6835 | + | |
6836 | +The sbox in the stream cypher have been converted to efficient logic | |
6837 | +operations using a custom written software (look into logic directory) | |
6838 | +and is responsible for a lot of speed increase. Maybe there exists a | |
6839 | +slightly better way to express the sbox as logical expressions, but it | |
6840 | +would be a minuscule improvement. The sbox in the block cypher can't be | |
6841 | +converted to efficient logic operations (8 bits of inputs are just too | |
6842 | +much) and is implemeted with a traditional lookup in an array. | |
6843 | + | |
6844 | +But there is a problem; if we want to process bits, but our external | |
6845 | +input and output wants bytes. We need conversion routines. Conversion | |
6846 | +routines are similar to the awful permutations we described before, so | |
6847 | +this has to be done efficiently someway. | |
6848 | + | |
6849 | + | |
6850 | +TRICK NUMBER 1: virtual shift registers | |
6851 | +--------------------------------------- | |
6852 | +Shift registers are normally implemented by moving all data around. | |
6853 | +Better leave the data in the same memory locations and redefine where | |
6854 | +the start of the register is (updating a pointer). That is called | |
6855 | +virtual shift register. | |
6856 | + | |
6857 | + | |
6858 | +TRICK NUMBER 2: parallel bitslice | |
6859 | +--------------------------------- | |
6860 | +Implementing the algorithm as described in tricks 1 and 2 give us about | |
6861 | +15% of the speed of a traditional implementation. This happens because | |
6862 | +we work on only one bit, even if our CPU is 32 bit wide. But *we can | |
6863 | +process 32 different packets at the same time*. This is called | |
6864 | +"bitslice" method. It can be done only if the program flow is not | |
6865 | +dependent of the data (if, while,...). Luckily this is true. | |
6866 | +Things like | |
6867 | + if(a){ | |
6868 | + b=c&d; | |
6869 | + } | |
6870 | + else{ | |
6871 | + b=e&f; | |
6872 | + } | |
6873 | +can be coded as (think of how hardware would implement this) | |
6874 | + b1=c&d; | |
6875 | + b2=e&f; | |
6876 | + b=b2^(a&(b1^b2)); | |
6877 | +and things like | |
6878 | + if(a){ | |
6879 | + b=c&d | |
6880 | + } | |
6881 | +can be transformed in the same way, as they may be written as | |
6882 | + if(a){ | |
6883 | + b=c&d | |
6884 | + } | |
6885 | + else{ | |
6886 | + b=b; | |
6887 | + } | |
6888 | +It could look wasteful, but it is not; and destroys data dependency. | |
6889 | + | |
6890 | +Our codes takes the same time as before, but produces 32 results, so | |
6891 | +speed is now 480% the speed of a traditional implementation. | |
6892 | + | |
6893 | + | |
6894 | +TRICK NUMBER 3: multimedia instructions | |
6895 | +--------------------------------------- | |
6896 | +If our CPU is 32 bit but it can also process larger blocks of data | |
6897 | +efficiently (multimedia instructions), we can use them. We only need | |
6898 | +logic ops and these are typically available. | |
6899 | + | |
6900 | +We can use MMX and work on 64 packets, or SSE and work on 128 packets. | |
6901 | +The speed doesn't automatically double going from 32 to 64 because the | |
6902 | +integer registers of the processor are normally faster. However, some | |
6903 | +speed is gained in this way. | |
6904 | + | |
6905 | +Multimedia instructions are often used by writing assembler by hand, but | |
6906 | +compilers are very good in doing register allocation, loop unrolling and | |
6907 | +instruction scheduling, so it is better to write the code in C and use | |
6908 | +native multimedia data types (intrinsics). | |
6909 | + | |
6910 | +Depending on number of available registers, execution latency, number of | |
6911 | +execution units in the CPU, it may be good to process more than one data | |
6912 | +block at the same time, for example 2 64bit MMX values. In this case we | |
6913 | +work on 128 bits by simulating a 128 bit op with two consecutive 64 bit | |
6914 | +op. This may or may not help (apparently not because x86 architecture | |
6915 | +has a small number of registers). | |
6916 | + | |
6917 | +We can also try working on 96 bit, pairing a MMX and an int op, or 192 | |
6918 | +bit by using MMX and SSE. While this is doable in theory and could | |
6919 | +exploit different execution units in the CPU, speed doesn't improve | |
6920 | +(because of cache line handling problems inside the CPU, maybe). | |
6921 | + | |
6922 | +Besides int, MMX, SSE, we can use long long int (64 bit) and, why not, | |
6923 | +unsigned char. | |
6924 | + | |
6925 | +Using groups of unsigned chars (8 or 16) could give the compiler an | |
6926 | +opportunity to insert multimedia instructions automatically. For | |
6927 | +example, icc can use one MMX istruction to do | |
6928 | + unsigned char a[8],b[8],c[8]; | |
6929 | + for(i=0;i<8;i++){ | |
6930 | + a[i]=b[i]&c[i]; | |
6931 | + } | |
6932 | +Some compilers (like icc) are efficient in this case, but using | |
6933 | +intrinsics manually is generally faster. | |
6934 | + | |
6935 | +All these experiments can be easily done if the code is written in a way | |
6936 | +which abstracts the data type used. This is not easy but doable, all the | |
6937 | +operations on data become (inlined) function calls or preprocessor | |
6938 | +macros. Good compilers are able to simplify all the abstraction at | |
6939 | +compile time and generate perfect code (gcc is great). | |
6940 | + | |
6941 | +The data abstraction used in the code is called "group". | |
6942 | + | |
6943 | + | |
6944 | +TRICK NUMBER 4: parallel byteslice | |
6945 | +---------------------------------- | |
6946 | +The bitslice method works wonderfully on the stream cypher, but can't be | |
6947 | +applied to the block cypher because of the evil big look up table. | |
6948 | + | |
6949 | +As we have to convert input data from normal to bitslice before starting | |
6950 | +processing and from bitslice to normal before output, we convert the | |
6951 | +stream cypher output to normal before the block calculations and do the | |
6952 | +block stage in a traditional way. | |
6953 | + | |
6954 | +There are some xors in the block cypher; so we arrange bytes from | |
6955 | +different packets side by side and use multimedia instructions to work | |
6956 | +on many bytes at the same time. This is not exactly bitslice, maybe it | |
6957 | +is called byteslice. The conversion routines are similar (just a bit | |
6958 | +simpler). | |
6959 | + | |
6960 | +The data type we use to do this in the code is called "batch". | |
6961 | + | |
6962 | +The virtual shift register described in trick number 2 is useful too. | |
6963 | + | |
6964 | +The look up table is the only thing which is done serially one byte at a | |
6965 | +time. Luckily if we do it on 32 or 64 bytes the loop is heavily | |
6966 | +unrolled, and the compiler and the CPU manage to get a good speed | |
6967 | +because there is little dependency between instructions. | |
6968 | + | |
6969 | + | |
6970 | +TRICK NUMBER 5: efficient bit permutation | |
6971 | +----------------------------------------- | |
6972 | +The block cypher has a bit permutation part. As we are not in a bit | |
6973 | +sliced form at that point, permuting bits in a byte takes 8 masks, 8 | |
6974 | +and, 7 or; but three bits move in the same direction, so we make it with | |
6975 | +6 masks, 6 and, 5 or. Batch processing through multimedia instructions | |
6976 | +is applicable too. | |
6977 | + | |
6978 | + | |
6979 | +TRICK NUMBER 6: efficient normal<->slice conversion | |
6980 | +--------------------------------------------------- | |
6981 | +The bitslice<->normal conversion routines are a sort of transposition | |
6982 | +operation, that is you have bits in rows and want them in columns. This | |
6983 | +can be done efficiently. For example, transposition of 8 bytes (matrix | |
6984 | +of 8x8=64 bits) can be done this way (we want to exchange bit[i][j] with | |
6985 | +bit[j][i] and we assume bit 0 is the MSB in the byte): | |
6986 | + | |
6987 | + // untested code, may be bugged | |
6988 | + unsigned char a[8]; | |
6989 | + unsigned char b[8]; | |
6990 | + for(i=0;i<8;i++) b[i]=0; | |
6991 | + for(i=0;i<8;i++){ | |
6992 | + for(j=0;j<8;j++){ | |
6993 | + b[i]|=((a[j]>>(7-i)&1))<<(7-j); | |
6994 | + } | |
6995 | + } | |
6996 | + | |
6997 | +but it is slow (128 shifts, 64 and, 64 or), or | |
6998 | + | |
6999 | + // untested code, may be bugged | |
7000 | + unsigned char a[8]; | |
7001 | + unsigned char b[8]; | |
7002 | + for(i=0;i<8;i++) b[i]=0; | |
7003 | + for(i=0;i<8;i++){ | |
7004 | + for(j=0;j<8;j++){ | |
7005 | + if(a[j]&(1<<(7-i))) b[i]|=1<<(7-j); | |
7006 | + } | |
7007 | + } | |
7008 | + | |
7009 | +but is very very slow (128 shifts, 64 and, 64 or, 128 unpredictable | |
7010 | +if!), or using a>>=1 and b<<=1, which gains you nothing, or | |
7011 | + | |
7012 | + // untested code, may be bugged | |
7013 | + unsigned char a[8]; | |
7014 | + unsigned char b[8]; | |
7015 | + unsigned char top,bottom; | |
7016 | + for(j=0;j<1;j++){ | |
7017 | + for(i=0;i<4;i++){ | |
7018 | + top= a[8*j+i]; | |
7019 | + bottom=a[8*j+4+i]; | |
7020 | + a[8*j+i]= (top&0xf0) |((bottom&0xf0)>>4); | |
7021 | + a[8*j+4+i]=((top&0x0f)<<4)| (bottom&0x0f); | |
7022 | + } | |
7023 | + } | |
7024 | + for(j=0;j<2;j++){ | |
7025 | + for(i=0;i<2;i++){ | |
7026 | + top= a[4*j+i]; | |
7027 | + bottom=a[4*j+2+i]; | |
7028 | + a[4*j+i] = (top&0xcc) |((bottom&0xcc)>>2); | |
7029 | + a[4*j+2+i]=((top&0x33)<<2)| (bottom&0x33); | |
7030 | + } | |
7031 | + } | |
7032 | + for(j=0;j<4;j++){ | |
7033 | + for(i=0;i<1;i++){ | |
7034 | + top= a[2*j+i]; | |
7035 | + bottom=a[2*j+1+i]; | |
7036 | + a[2*j+i] = (top&0xaa) |((bottom&0xaa)>>1); | |
7037 | + a[2*j+1+i]=((top&0x55)<<1)| (bottom&0x55); | |
7038 | + } | |
7039 | + } | |
7040 | + for(i=0;i<8;i++) b[i]=a[i]; //easy to integrate into one of the stages above | |
7041 | + | |
7042 | +which is very fast (24 shifts, 48 and, 24 or) and has redundant loops | |
7043 | +and address calculations which will be optimized away by the compiler. | |
7044 | +It can be written as 3 nested loops but it becomes less readable and | |
7045 | +makes it difficult to have results in b without an extra copy. The | |
7046 | +compiler always unrolls heavily. | |
7047 | + | |
7048 | +The gain is much bigger when operating with 32 bit or 64 bit values (we | |
7049 | +are going from N^2 to Nlog(N)). This method is used for rectangular | |
7050 | +matrixes too (they have to be seen as square matrixes side by side). | |
7051 | +Warning: this code is not *endian independent* if you use ints to work | |
7052 | +on 4 bytes. Running it on a big endian processor will give you a | |
7053 | +different and strange kind of bit rotation if you don't modify masks and | |
7054 | +shifts. | |
7055 | + | |
7056 | +This is done in the code using int or long long int. It should be | |
7057 | +possible to use MMX instead of long long int and it could be faster, but | |
7058 | +this code doesn't cost a great fraction of the total time. There are | |
7059 | +problems with the shifts, as multimedia instructions do not have all | |
7060 | +possible kind of shift we need (SSE has none!). | |
7061 | + | |
7062 | + | |
7063 | +TRICK NUMBER 7: try hard to process packets together | |
7064 | +---------------------------------------------------- | |
7065 | +As we are able to process many packets together, we have to avoid | |
7066 | +running with many slots empty. Processing one packet or 64 packets takes | |
7067 | +the same time if the internal parallelism is 64! So we try hard to | |
7068 | +aggregate packets that can be processed together; for simplicity reasons | |
7069 | +we don't mix packets with even and odd parity (different keys), even if | |
7070 | +it should be doable with a little effort. Sometimes the transition from | |
7071 | +even to odd parity and viceversa is not sharp, but there are sequences | |
7072 | +like EEEEEOEEOEEOOOO. We try to group all the E together even if there | |
7073 | +are O between them. This out-of-order processing complicates the | |
7074 | +interface to the applications a bit but saves us three or four runs with | |
7075 | +many empty slots. | |
7076 | + | |
7077 | +We have also logic to process together packets with a different size of | |
7078 | +the payload, which is not always 184 bytes. This involves sorting the | |
7079 | +packets by size before processing and careful operation of the 23 | |
7080 | +iteration loop to exclude some packets from the calculations. It is not | |
7081 | +CPU heavy. | |
7082 | + | |
7083 | +Packets with payload <8 bytes are identical before and after decryption | |
7084 | +(!), so we skip them without using a slot. (according to DVB specs these | |
7085 | +kind of packets shouldn't happen, but they are used in the real world). | |
7086 | + | |
7087 | + | |
7088 | +TRICK NUMBER 8: try to avoid doing the same thing many times | |
7089 | +------------------------------------------------------------ | |
7090 | +Some calculations related to keys are only done when the keys are set, | |
7091 | +then all the values depending on keys are stored in a convenient form | |
7092 | +and used everytime we convert a group of packets. | |
7093 | + | |
7094 | + | |
7095 | +TRICK NUMBER 9: compiler | |
7096 | +------------------------ | |
7097 | + | |
7098 | +Compilers have a lot of optimization options. I used -march to target my | |
7099 | +CPU and played with unsual options. In particular | |
7100 | + "--param max-unrolled-insns=500" | |
7101 | +does a good job on the tricky table lookup in the block cypher. Bigger | |
7102 | +values unroll too much somewhere and loose speed. All the testing has | |
7103 | +been done on an AthlonXP CPU with a specific version of gcc | |
7104 | + gcc version 3.3.3 20040412 (Red Hat Linux 3.3.3-7) | |
7105 | +Other combinations of CPU and compiler can give different speeds. If the | |
7106 | +compiler is not able to simplify the group and batch structures and | |
7107 | +stores everything in memory instead of registers, performance will be | |
7108 | +low. | |
7109 | + | |
7110 | +Absolutely use a good compiler! | |
7111 | + | |
7112 | +Note: the same code can be compiled in C or C++ mode. g++ gives a 3% | |
7113 | +speed increase compared to gcc (I suppose some stricter constraint on | |
7114 | +array and pointers in C++ mode gives the optimizer more freedom). | |
7115 | + | |
7116 | + | |
7117 | +TRICK NUMBER a: a lot of brain work | |
7118 | +----------------------------------- | |
7119 | +The code started as very slow but correct implementation and was then | |
7120 | +tweaked for months with a lot of experimentation and by adding all the | |
7121 | +good ideas one after another to achieve little steps toward the best | |
7122 | +speed possible, while continously testing that nothing had been broken. | |
7123 | + | |
7124 | +Many hours were spent on this code. | |
7125 | + | |
7126 | +Enjoy the result. | |
7127 | Index: libs/libmythtv/dvbchannel.cpp | |
7128 | =================================================================== | |
7129 | --- libs/libmythtv/dvbchannel.cpp.orig 2006-06-11 12:41:25.000000000 -0400 | |
7130 | +++ libs/libmythtv/dvbchannel.cpp 2006-06-20 17:36:06.000000000 -0400 | |
7131 | @@ -658,6 +658,9 @@ | |
7132 | ||
7133 | retune_tuning = channel.tuning; | |
7134 | ||
7135 | + GENERAL(QString("invalidating DeCSA for card %1").arg(cardnum)); | |
7136 | + DVBRecorder::UpdateDeCSAKeys(cardnum, 'I', 0, NULL, 0); | |
7137 | + | |
7138 | if (fd_frontend < 0) | |
7139 | { | |
7140 | ERROR("DVBChannel::Tune: Card not open!"); | |
7141 | Index: libs/libmythtv/FFdecsa/parallel_128_16charA.h | |
7142 | =================================================================== | |
7143 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
7144 | +++ libs/libmythtv/FFdecsa/parallel_128_16charA.h 2006-06-20 17:36:06.000000000 -0400 | |
7145 | @@ -0,0 +1,172 @@ | |
7146 | +/* FFdecsa -- fast decsa algorithm | |
7147 | + * | |
7148 | + * Copyright (C) 2003-2004 fatih89r | |
7149 | + * | |
7150 | + * This program is free software; you can redistribute it and/or modify | |
7151 | + * it under the terms of the GNU General Public License as published by | |
7152 | + * the Free Software Foundation; either version 2 of the License, or | |
7153 | + * (at your option) any later version. | |
7154 | + * | |
7155 | + * This program is distributed in the hope that it will be useful, | |
7156 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
7157 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
7158 | + * GNU General Public License for more details. | |
7159 | + * | |
7160 | + * You should have received a copy of the GNU General Public License | |
7161 | + * along with this program; if not, write to the Free Software | |
7162 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
7163 | + */ | |
7164 | + | |
7165 | + | |
7166 | +struct group_t{ | |
7167 | + unsigned char s1[16]; | |
7168 | +}; | |
7169 | +typedef struct group_t group; | |
7170 | + | |
7171 | +#define GROUP_PARALLELISM 128 | |
7172 | + | |
7173 | +group static inline FF0(){ | |
7174 | + group res; | |
7175 | + int i; | |
7176 | + for(i=0;i<16;i++) res.s1[i]=0x0; | |
7177 | + return res; | |
7178 | +} | |
7179 | + | |
7180 | +group static inline FF1(){ | |
7181 | + group res; | |
7182 | + int i; | |
7183 | + for(i=0;i<16;i++) res.s1[i]=0xff; | |
7184 | + return res; | |
7185 | +} | |
7186 | + | |
7187 | +group static inline FFAND(group a,group b){ | |
7188 | + group res; | |
7189 | + int i; | |
7190 | + for(i=0;i<16;i++) res.s1[i]=a.s1[i]&b.s1[i]; | |
7191 | + return res; | |
7192 | +} | |
7193 | + | |
7194 | +group static inline FFOR(group a,group b){ | |
7195 | + group res; | |
7196 | + int i; | |
7197 | + for(i=0;i<16;i++) res.s1[i]=a.s1[i]|b.s1[i]; | |
7198 | + return res; | |
7199 | +} | |
7200 | + | |
7201 | +group static inline FFXOR(group a,group b){ | |
7202 | + group res; | |
7203 | + int i; | |
7204 | + for(i=0;i<16;i++) res.s1[i]=a.s1[i]^b.s1[i]; | |
7205 | + return res; | |
7206 | +} | |
7207 | + | |
7208 | +group static inline FFNOT(group a){ | |
7209 | + group res; | |
7210 | + int i; | |
7211 | + for(i=0;i<16;i++) res.s1[i]=~a.s1[i]; | |
7212 | + return res; | |
7213 | +} | |
7214 | + | |
7215 | + | |
7216 | +/* 64 rows of 128 bits */ | |
7217 | + | |
7218 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
7219 | + *(((int *)tab)+2*g)=*((int *)data); | |
7220 | + *(((int *)tab)+2*g+1)=*(((int *)data)+1); | |
7221 | +} | |
7222 | + | |
7223 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
7224 | + *((int *)data)=*(((int *)tab)+2*g); | |
7225 | + *(((int *)data)+1)=*(((int *)tab)+2*g+1); | |
7226 | +} | |
7227 | + | |
7228 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
7229 | + int j; | |
7230 | + for(j=0;j<n;j++){ | |
7231 | + *(data+j)^=*(tab+8*g+j); | |
7232 | + } | |
7233 | +} | |
7234 | + | |
7235 | + | |
7236 | +struct batch_t{ | |
7237 | + unsigned char s1[16]; | |
7238 | +}; | |
7239 | +typedef struct batch_t batch; | |
7240 | + | |
7241 | +#define BYTES_PER_BATCH 16 | |
7242 | + | |
7243 | +batch static inline B_FFAND(batch a,batch b){ | |
7244 | + batch res; | |
7245 | + int i; | |
7246 | + for(i=0;i<16;i++) res.s1[i]=a.s1[i]&b.s1[i]; | |
7247 | + return res; | |
7248 | +} | |
7249 | + | |
7250 | +batch static inline B_FFOR(batch a,batch b){ | |
7251 | + batch res; | |
7252 | + int i; | |
7253 | + for(i=0;i<16;i++) res.s1[i]=a.s1[i]|b.s1[i]; | |
7254 | + return res; | |
7255 | +} | |
7256 | + | |
7257 | +batch static inline B_FFXOR(batch a,batch b){ | |
7258 | + batch res; | |
7259 | + int i; | |
7260 | + for(i=0;i<16;i++) res.s1[i]=a.s1[i]^b.s1[i]; | |
7261 | + return res; | |
7262 | +} | |
7263 | + | |
7264 | + | |
7265 | +batch static inline B_FFN_ALL_29(){ | |
7266 | + batch res; | |
7267 | + int i; | |
7268 | + for(i=0;i<16;i++) res.s1[i]=0x29; | |
7269 | + return res; | |
7270 | +} | |
7271 | +batch static inline B_FFN_ALL_02(){ | |
7272 | + batch res; | |
7273 | + int i; | |
7274 | + for(i=0;i<16;i++) res.s1[i]=0x02; | |
7275 | + return res; | |
7276 | +} | |
7277 | +batch static inline B_FFN_ALL_04(){ | |
7278 | + batch res; | |
7279 | + int i; | |
7280 | + for(i=0;i<16;i++) res.s1[i]=0x04; | |
7281 | + return res; | |
7282 | +} | |
7283 | +batch static inline B_FFN_ALL_10(){ | |
7284 | + batch res; | |
7285 | + int i; | |
7286 | + for(i=0;i<16;i++) res.s1[i]=0x10; | |
7287 | + return res; | |
7288 | +} | |
7289 | +batch static inline B_FFN_ALL_40(){ | |
7290 | + batch res; | |
7291 | + int i; | |
7292 | + for(i=0;i<16;i++) res.s1[i]=0x40; | |
7293 | + return res; | |
7294 | +} | |
7295 | +batch static inline B_FFN_ALL_80(){ | |
7296 | + batch res; | |
7297 | + int i; | |
7298 | + for(i=0;i<16;i++) res.s1[i]=0x80; | |
7299 | + return res; | |
7300 | +} | |
7301 | + | |
7302 | +batch static inline B_FFSH8L(batch a,int n){ | |
7303 | + batch res; | |
7304 | + int i; | |
7305 | + for(i=0;i<16;i++) res.s1[i]=a.s1[i]<<n; | |
7306 | + return res; | |
7307 | +} | |
7308 | + | |
7309 | +batch static inline B_FFSH8R(batch a,int n){ | |
7310 | + batch res; | |
7311 | + int i; | |
7312 | + for(i=0;i<16;i++) res.s1[i]=a.s1[i]>>n; | |
7313 | + return res; | |
7314 | +} | |
7315 | + | |
7316 | +void static inline M_EMPTY(void){ | |
7317 | +} | |
7318 | Index: libs/libmythtv/FFdecsa/parallel_generic.h | |
7319 | =================================================================== | |
7320 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
7321 | +++ libs/libmythtv/FFdecsa/parallel_generic.h 2006-06-20 17:36:06.000000000 -0400 | |
7322 | @@ -0,0 +1,102 @@ | |
7323 | +/* FFdecsa -- fast decsa algorithm | |
7324 | + * | |
7325 | + * Copyright (C) 2003-2004 fatih89r | |
7326 | + * | |
7327 | + * This program is free software; you can redistribute it and/or modify | |
7328 | + * it under the terms of the GNU General Public License as published by | |
7329 | + * the Free Software Foundation; either version 2 of the License, or | |
7330 | + * (at your option) any later version. | |
7331 | + * | |
7332 | + * This program is distributed in the hope that it will be useful, | |
7333 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
7334 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
7335 | + * GNU General Public License for more details. | |
7336 | + * | |
7337 | + * You should have received a copy of the GNU General Public License | |
7338 | + * along with this program; if not, write to the Free Software | |
7339 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
7340 | + */ | |
7341 | + | |
7342 | + | |
7343 | + | |
7344 | +#if 0 | |
7345 | +//// generics | |
7346 | +#define COPY4BY(d,s) do{ int *pd=(int *)(d), *ps=(int *)(s); \ | |
7347 | + *pd = *ps; }while(0) | |
7348 | +#define COPY8BY(d,s) do{ long long int *pd=(long long int *)(d), *ps=(long long int *)(s); \ | |
7349 | + *pd = *ps; }while(0) | |
7350 | +#define COPY16BY(d,s) do{ long long int *pd=(long long int *)(d), *ps=(long long int *)(s); \ | |
7351 | + *pd = *ps; \ | |
7352 | + *(pd+1) = *(ps+1); }while(0) | |
7353 | +#define COPY32BY(d,s) do{ long long int *pd=(long long int *)(d), *ps=(long long int *)(s); \ | |
7354 | + *pd = *ps; \ | |
7355 | + *(pd+1) = *(ps+1) \ | |
7356 | + *(pd+2) = *(ps+2) \ | |
7357 | + *(pd+3) = *(ps+3); }while(0) | |
7358 | +#define XOR4BY(d,s1,s2) do{ int *pd=(int *)(d), *ps1=(int *)(s1), *ps2=(int *)(s2); \ | |
7359 | + *pd = *ps1 ^ *ps2; }while(0) | |
7360 | +#define XOR8BY(d,s1,s2) do{ long long int *pd=(long long int *)(d), *ps1=(long long int *)(s1), *ps2=(long long int *)(s2); \ | |
7361 | + *pd = *ps1 ^ *ps2; }while(0) | |
7362 | +#define XOR16BY(d,s1,s2) do{ long long int *pd=(long long int *)(d), *ps1=(long long int *)(s1), *ps2=(long long int *)(s2); \ | |
7363 | + *pd = *ps1 ^ *ps2; \ | |
7364 | + *(pd+8) = *(ps1+8) ^ *(ps2+8); }while(0) | |
7365 | +#define XOR32BY(d,s1,s2) do{ long long int *pd=(long long int *)(d), *ps1=(long long int *)(s1), *ps2=(long long int *)(s2); \ | |
7366 | + *pd = *ps1 ^ *ps2; \ | |
7367 | + *(pd+1) = *(ps1+1) ^ *(ps2+1); \ | |
7368 | + *(pd+2) = *(ps1+2) ^ *(ps2+2); \ | |
7369 | + *(pd+3) = *(ps1+3) ^ *(ps2+3); }while(0) | |
7370 | +#define XOR32BV(d,s1,s2) do{ int *const pd=(int *const)(d), *ps1=(const int *const)(s1), *ps2=(const int *const)(s2); \ | |
7371 | + int z; \ | |
7372 | + for(z=0;z<8;z++){ \ | |
7373 | + pd[z]=ps1[z]^ps2[z]; \ | |
7374 | + } \ | |
7375 | + }while(0) | |
7376 | +#define XOREQ4BY(d,s) do{ int *pd=(int *)(d), *ps=(int *)(s); \ | |
7377 | + *pd ^= *ps; }while(0) | |
7378 | +#define XOREQ8BY(d,s) do{ long long int *pd=(long long int *)(d), *ps=(long long int *)(s); \ | |
7379 | + *pd ^= *ps; }while(0) | |
7380 | +#define XOREQ16BY(d,s) do{ long long int *pd=(long long int *)(d), *ps=(long long int *)(s); \ | |
7381 | + *pd ^= *ps; \ | |
7382 | + *(pd+1) ^=*(ps+1); }while(0) | |
7383 | +#define XOREQ32BY(d,s) do{ long long int *pd=(long long int *)(d), *ps=(long long int *)(s); \ | |
7384 | + *pd ^= *ps; \ | |
7385 | + *(pd+1) ^=*(ps+1); \ | |
7386 | + *(pd+2) ^=*(ps+2); \ | |
7387 | + *(pd+3) ^=*(ps+3); }while(0) | |
7388 | +#define XOREQ32BY4(d,s) do{ int *pd=(int *)(d), *ps=(int *)(s); \ | |
7389 | + *pd ^= *ps; \ | |
7390 | + *(pd+1) ^=*(ps+1); \ | |
7391 | + *(pd+2) ^=*(ps+2); \ | |
7392 | + *(pd+3) ^=*(ps+3); \ | |
7393 | + *(pd+4) ^=*(ps+4); \ | |
7394 | + *(pd+5) ^=*(ps+5); \ | |
7395 | + *(pd+6) ^=*(ps+6); \ | |
7396 | + *(pd+7) ^=*(ps+7); }while(0) | |
7397 | +#define XOREQ32BV(d,s) do{ unsigned char *pd=(unsigned char *)(d), *ps=(unsigned char *)(s); \ | |
7398 | + int z; \ | |
7399 | + for(z=0;z<32;z++){ \ | |
7400 | + pd[z]^=ps[z]; \ | |
7401 | + } \ | |
7402 | + }while(0) | |
7403 | + | |
7404 | +#else | |
7405 | +#define XOR_4_BY(d,s1,s2) do{ int *pd=(int *)(d), *ps1=(int *)(s1), *ps2=(int *)(s2); \ | |
7406 | + *pd = *ps1 ^ *ps2; }while(0) | |
7407 | +#define XOR_8_BY(d,s1,s2) do{ long long int *pd=(long long int *)(d), *ps1=(long long int *)(s1), *ps2=(long long int *)(s2); \ | |
7408 | + *pd = *ps1 ^ *ps2; }while(0) | |
7409 | +#define XOREQ_4_BY(d,s) do{ int *pd=(int *)(d), *ps=(int *)(s); \ | |
7410 | + *pd ^= *ps; }while(0) | |
7411 | +#define XOREQ_8_BY(d,s) do{ long long int *pd=(long long int *)(d), *ps=(long long int *)(s); \ | |
7412 | + *pd ^= *ps; }while(0) | |
7413 | +#define COPY_4_BY(d,s) do{ int *pd=(int *)(d), *ps=(int *)(s); \ | |
7414 | + *pd = *ps; }while(0) | |
7415 | +#define COPY_8_BY(d,s) do{ long long int *pd=(long long int *)(d), *ps=(long long int *)(s); \ | |
7416 | + *pd = *ps; }while(0) | |
7417 | + | |
7418 | +#define BEST_SPAN 8 | |
7419 | +#define XOR_BEST_BY(d,s1,s2) do{ XOR_8_BY(d,s1,s2); }while(0); | |
7420 | +#define XOREQ_BEST_BY(d,s) do{ XOREQ_8_BY(d,s); }while(0); | |
7421 | +#define COPY_BEST_BY(d,s) do{ COPY_8_BY(d,s); }while(0); | |
7422 | + | |
7423 | +#define END_MM do{ }while(0); | |
7424 | +#endif | |
7425 | Index: libs/libmythtv/FFdecsa/docs/how_to_use.txt | |
7426 | =================================================================== | |
7427 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
7428 | +++ libs/libmythtv/FFdecsa/docs/how_to_use.txt 2006-06-20 17:36:06.000000000 -0400 | |
7429 | @@ -0,0 +1,239 @@ | |
7430 | +------- | |
7431 | +FFdecsa | |
7432 | +------- | |
7433 | + | |
7434 | +This code is able to decrypt MPEG TS packets with the CSA algorithm. To | |
7435 | +achieve high speed, the decryption core works on many packets at the | |
7436 | +same time, so the interface is more complicated than usual decsa | |
7437 | +implementations. | |
7438 | + | |
7439 | +The FFdecsa.h file defines the external interface of this code. | |
7440 | + | |
7441 | +Basically: | |
7442 | + | |
7443 | +1) you use get_suggested_cluster_size to know the optimal number of | |
7444 | +packets you have to pass for decryption | |
7445 | + | |
7446 | +2) you use set_control_words to set the decryption keys | |
7447 | + | |
7448 | +3) you use decrypt_packets to do the actual decryption | |
7449 | + | |
7450 | +You don't need to always use set_control_words before decrypt_packets, | |
7451 | +if keys aren't changed. | |
7452 | + | |
7453 | + | |
7454 | +The decrypt_packets function call decrypts many packets at the same | |
7455 | +time. The interface is complicated because the only design goal was | |
7456 | +speed, so it implements zero-copying of packets, out-of-order decryption | |
7457 | +and optimal packet aggregation for better parallelism. This part is the | |
7458 | +most difficult to understand. | |
7459 | + | |
7460 | +--- HOW TO USE int decrypt_packets(unsigned char **cluster); --- | |
7461 | + | |
7462 | +PARAMETERS | |
7463 | + cluster points to an array of pointers, representing zero or more | |
7464 | + ranges. Every range has a start and end pointer; a start pointer==NULL | |
7465 | + terminates the array. | |
7466 | + So, an array of pointers has this content: | |
7467 | + start_of_buffer_1, end_of_buffer_1, ... start_of_buffer_N, | |
7468 | + end_of_buffer_N, NULL | |
7469 | + example: | |
7470 | + 0x12340000, 0x123400bc, 0x56780a00, 0x5678b78, NULL | |
7471 | + has two ranges (0x12340000 - 0x123400bc and 0x56780a00 - 0x5678b78), | |
7472 | + for a total of three packets (starting at 0x12340000, 0x56780a00, | |
7473 | + 0x5678abc) | |
7474 | +RETURNS | |
7475 | + How many packets can now be consumed by the caller, this is always >= | |
7476 | + 1, unless the cluster contained zero packets (in that case it's | |
7477 | + obviously zero). | |
7478 | +MODIFIES | |
7479 | + The cluster is modified to try to exclude packets which shouldn't be | |
7480 | + submitted again for decryption (because just decrypted or originally | |
7481 | + not crypted). "Try to exclude" because the returned array will never | |
7482 | + be bigger than what was passed, so if you passed only a range and some | |
7483 | + packets in the middle were decrypted making "holes" into the range, | |
7484 | + the range would have to be split into several ranges, and that will | |
7485 | + not be done. If you want a strict description of what has to be passed | |
7486 | + again to decrypt_packets, you have to use ranges with only one packet | |
7487 | + inside. Note that the first packet will certainly be eliminated from | |
7488 | + the returned cluster (see also RETURNS). | |
7489 | + | |
7490 | +You can now read the detailed description of operation or just skip to | |
7491 | +the API examples. | |
7492 | + | |
7493 | + | |
7494 | +--------------------------------- | |
7495 | +DETAILED DESCRIPTION OF OPERATION | |
7496 | +--------------------------------- | |
7497 | + consider a sequence of packets like this: | |
7498 | + 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 ... | |
7499 | + E E E E E E E E E E E O E O E O O 0 0 0 0 0 0 0 0 c O O O O O O O O O O O ... | |
7500 | + where | |
7501 | + E = encrypted_even, | |
7502 | + O = encrypted_odd, | |
7503 | + e = clear_was_encrypted_even, | |
7504 | + o = clear_was_encrypted_odd, | |
7505 | + c = clear | |
7506 | + and suppose the suggested cluster size is 10 (this could be for a function with internal parallelism 8) | |
7507 | + | |
7508 | + 1) we define the cluster to include packets 0-9 and | |
7509 | + call decrypt_packets | |
7510 | + a possible result is that the function call | |
7511 | + - returns 8 (8 packets available) | |
7512 | + - the buffer contains now this | |
7513 | + ----------------------------- | |
7514 | + 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 ... | |
7515 | + e e e e e e e e E E E O E O E O O 0 0 0 0 0 0 0 0 c O O O O O O O O O O O ... | |
7516 | + ----- | |
7517 | + - the modified cluster covers 8-9 [continue reading, but then see note 1 below] | |
7518 | + so, we can use the first 8 packets of the original cluster (0-7) | |
7519 | + | |
7520 | + 2) now, we define cluster over 8-17 and call decrypt_packets | |
7521 | + a possible result is: | |
7522 | + - returns 3 (3 packets available) | |
7523 | + - the buffer contains now this (!!!) | |
7524 | + ----------------------------- | |
7525 | + 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 ... | |
7526 | + e e e e e e e e e e e O e O e O O 0 0 0 0 0 0 0 0 c O O O O O O O O O O O ... | |
7527 | + -- -- -------- | |
7528 | + - the modified cluster covers 11-11,13-13,15-17 [continue reading, but then see note 1 below] | |
7529 | + so, we can use the first 3 packets of the original cluster (8-10) | |
7530 | + | |
7531 | + 3) now, we define cluster over 11-20 and call decrypt packets (defining a cluster 11-11,13-13,15-22 would be better) | |
7532 | + a possible result is: | |
7533 | + - returns 10 (10 packets available) | |
7534 | + - the buffer contains now this | |
7535 | + ----------------------------- | |
7536 | + 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 ... | |
7537 | + e e e e e e e e e e e o e o e o o o o o o 0 0 0 0 c O O O O O O O O O O O ... | |
7538 | + | |
7539 | + - the modified cluster is empty | |
7540 | + so, we can use the first 10 packets of the original cluster (11-20) | |
7541 | + What it happened is that the second call decrypted packets 12 and 14 but they were | |
7542 | + not made available because packet 11 was still encrypted, | |
7543 | + the third call decrypted 11,13,15-20 and included 12 and 14 as available too. | |
7544 | + | |
7545 | + 4) now, we define cluster over 21-30 and call decrypt packets | |
7546 | + a possible result is: | |
7547 | + - returns 9 (9 packets available) | |
7548 | + - the buffer contains now this | |
7549 | + ----------------------------- | |
7550 | + 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 ... | |
7551 | + e e e e e e e e e e e o e o e o o o o o o o o o o c o o o o O O O O O O O ... | |
7552 | + -- | |
7553 | + - the modified cluster covers 30-30 | |
7554 | + so, we can use the first 9 packets of the original cluster (21-29) | |
7555 | + What happened is that packet 25 could be skipped because it is in clear. | |
7556 | + | |
7557 | + Note that the suggested cluster size (10) is higher than the maximum number | |
7558 | + of packets that can be really decrypted (8), but we are able to skip 12 and 14 | |
7559 | + in step 3) and run the decryption on a full 8 packets group. | |
7560 | + In the same way, we were able to skip 25 in step 4). | |
7561 | + There are three kinds of "free" packets we can skip: | |
7562 | + - packets decrypted in a previous call (as 12 and 14) | |
7563 | + - packets already in clear (as 25) | |
7564 | + - packets with a payload of less than 8 bytes (clear==encrypted!) | |
7565 | + | |
7566 | + Note also that we could have defined a better cluster in step 3 | |
7567 | + (11-11,13-13,15-22), using what step 2 had returned. The risk of not | |
7568 | + having 8 packets to decrypt would have been smaller (consider the case | |
7569 | + where 19 and 20 were "c"). | |
7570 | + | |
7571 | + Final considerations: | |
7572 | + - you can use a bigger or smaller cluster than the suggested number of packets | |
7573 | + - every call to decrypt_packets has a *fixed* CPU cost, so you should try to | |
7574 | + not run it with a few packets, when possible | |
7575 | + - decrypt_packets can't decrypt even and odd at the same time; it guarantees | |
7576 | + that the first packet will be decrypted and tries to decrypt as many packets | |
7577 | + as possible | |
7578 | + - clear packets in the middle of encrypted packets don't happen in real world, | |
7579 | + but E,E,E,O,E,O,O,O sequences do happen (audio/video muxing problems?) and | |
7580 | + small packets (<8 bytes) happen frequently; the ability to skip is useful. | |
7581 | + | |
7582 | + note 1: | |
7583 | + As the returned cluster will not have more ranges than the passed one, what it is | |
7584 | + described above is not actually true. | |
7585 | + In the step 1) the returned cluster will cover 8-9, but in step 2) it will | |
7586 | + cover 11-17 (some extra packets had to remain in); this lack of information | |
7587 | + prevents us from using an optimal 11-11,13-13,15-22 in step 3). Note that | |
7588 | + in any case step 3) will decrypt 11,13,15,16,17,18,19,20 thanks to the | |
7589 | + extra margin we use (we put ten packets (including 19 and 20) even if the | |
7590 | + parallelism was just 8, and it was a good idea; but if 19 and 20 were of | |
7591 | + type c, we would have run the decryption with only 6/8 efficiency). | |
7592 | + This problem can be prevented by using ranges with only one packet: in | |
7593 | + step 2) we would have passed | |
7594 | + 8-8,9-9,10-10,11-11,12-12,13-13,14-14,15-15,16-16,17-17 | |
7595 | + and got back | |
7596 | + 11-11,13-13,15-17. | |
7597 | + | |
7598 | + | |
7599 | +------------ | |
7600 | +API EXAMPLES | |
7601 | +------------ | |
7602 | + | |
7603 | +Some examples of how the API can be used (this is not real code, so it | |
7604 | +may have typos or other bugs). | |
7605 | + | |
7606 | + | |
7607 | +Example 1: (big linear buffer, simple use of cluster) | |
7608 | + | |
7609 | + unsigned char *p; | |
7610 | + unsigned char *cluster[3]; | |
7611 | + for(p=start;p<end;){ | |
7612 | + cluster[0]=p;cluster[1]=end; | |
7613 | + cluster[2]=NULL; | |
7614 | + p+=188*decrypt_packets(cluster); | |
7615 | + } | |
7616 | + //consume(start,end); | |
7617 | + | |
7618 | + | |
7619 | +Example 2: (circular buffer, simple use of cluster) | |
7620 | + | |
7621 | + unsigned char *p; | |
7622 | + unsigned char *cluster[5]; | |
7623 | + | |
7624 | + while(1){ | |
7625 | + if(read==write){ | |
7626 | + //buffer is empty | |
7627 | + //write=refill_buffer(write,start,end); | |
7628 | + continue; | |
7629 | + } | |
7630 | + else if(read<write){ | |
7631 | + cluster[0]=read;cluster[1]=write; | |
7632 | + cluster[2]=NULL; | |
7633 | + } | |
7634 | + else{ | |
7635 | + cluster[0]=read;cluster[1]=end; | |
7636 | + cluster[2]=start;cluster[3]=write; | |
7637 | + cluster[4]=NULL; | |
7638 | + } | |
7639 | + new_read=read+188*decrypt_packets(cluster); | |
7640 | + if(new_read<=end){ | |
7641 | + //consume(read,new_read); | |
7642 | + } | |
7643 | + else{ | |
7644 | + new_read=start+(new_read-end); | |
7645 | + //consume(read,end); | |
7646 | + //consume(start,new_read); | |
7647 | + } | |
7648 | + read=new_read; | |
7649 | + if(read==end) read=start; | |
7650 | + } | |
7651 | + | |
7652 | + | |
7653 | +Example 3: (undefined buffer structure, advanced use of cluster) | |
7654 | + | |
7655 | + unsigned char *packets[1000000]; | |
7656 | + unsigned char *cluster[142]; //if suggested packets is 70 | |
7657 | + | |
7658 | + cluster[0]=NULL; | |
7659 | + for(n=0;n<1000000;){ | |
7660 | + i=0; | |
7661 | + while(cluster[2*i]!=NULL) i++; //preserve returned ranges | |
7662 | + for(k=i;k<70&&n<1000000;k++,n++){ | |
7663 | + cluster[2*k]=packets[n];cluster[2*k+1]=packets[n]+188; | |
7664 | + } | |
7665 | + cluster[2*k]=NULL; | |
7666 | + decrypt_packets(cluster); | |
7667 | + } | |
7668 | + //consume_all_packets(); | |
7669 | Index: libs/libmythtv/FFdecsa/parallel_032_4char.h | |
7670 | =================================================================== | |
7671 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
7672 | +++ libs/libmythtv/FFdecsa/parallel_032_4char.h 2006-06-20 17:36:06.000000000 -0400 | |
7673 | @@ -0,0 +1,206 @@ | |
7674 | +/* FFdecsa -- fast decsa algorithm | |
7675 | + * | |
7676 | + * Copyright (C) 2003-2004 fatih89r | |
7677 | + * | |
7678 | + * This program is free software; you can redistribute it and/or modify | |
7679 | + * it under the terms of the GNU General Public License as published by | |
7680 | + * the Free Software Foundation; either version 2 of the License, or | |
7681 | + * (at your option) any later version. | |
7682 | + * | |
7683 | + * This program is distributed in the hope that it will be useful, | |
7684 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
7685 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
7686 | + * GNU General Public License for more details. | |
7687 | + * | |
7688 | + * You should have received a copy of the GNU General Public License | |
7689 | + * along with this program; if not, write to the Free Software | |
7690 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
7691 | + */ | |
7692 | + | |
7693 | + | |
7694 | +struct group_t{ | |
7695 | + unsigned char s1,s2,s3,s4; | |
7696 | +}; | |
7697 | +typedef struct group_t group; | |
7698 | + | |
7699 | +#define GROUP_PARALLELISM 32 | |
7700 | + | |
7701 | +group static inline FF0(){ | |
7702 | + group res; | |
7703 | + res.s1=0x0; | |
7704 | + res.s2=0x0; | |
7705 | + res.s3=0x0; | |
7706 | + res.s4=0x0; | |
7707 | + return res; | |
7708 | +} | |
7709 | + | |
7710 | +group static inline FF1(){ | |
7711 | + group res; | |
7712 | + res.s1=0xff; | |
7713 | + res.s2=0xff; | |
7714 | + res.s3=0xff; | |
7715 | + res.s4=0xff; | |
7716 | + return res; | |
7717 | +} | |
7718 | + | |
7719 | +group static inline FFAND(group a,group b){ | |
7720 | + group res; | |
7721 | + res.s1=a.s1&b.s1; | |
7722 | + res.s2=a.s2&b.s2; | |
7723 | + res.s3=a.s3&b.s3; | |
7724 | + res.s4=a.s4&b.s4; | |
7725 | + return res; | |
7726 | +} | |
7727 | + | |
7728 | +group static inline FFOR(group a,group b){ | |
7729 | + group res; | |
7730 | + res.s1=a.s1|b.s1; | |
7731 | + res.s2=a.s2|b.s2; | |
7732 | + res.s3=a.s3|b.s3; | |
7733 | + res.s4=a.s4|b.s4; | |
7734 | + return res; | |
7735 | +} | |
7736 | + | |
7737 | +group static inline FFXOR(group a,group b){ | |
7738 | + group res; | |
7739 | + res.s1=a.s1^b.s1; | |
7740 | + res.s2=a.s2^b.s2; | |
7741 | + res.s3=a.s3^b.s3; | |
7742 | + res.s4=a.s4^b.s4; | |
7743 | + return res; | |
7744 | +} | |
7745 | + | |
7746 | +group static inline FFNOT(group a){ | |
7747 | + group res; | |
7748 | + res.s1=~a.s1; | |
7749 | + res.s2=~a.s2; | |
7750 | + res.s3=~a.s3; | |
7751 | + res.s4=~a.s4; | |
7752 | + return res; | |
7753 | +} | |
7754 | + | |
7755 | + | |
7756 | +/* 64 rows of 32 bits */ | |
7757 | + | |
7758 | +void static inline FFTABLEIN(unsigned char *tab, int g, unsigned char *data){ | |
7759 | + *(((int *)tab)+g)=*((int *)data); | |
7760 | + *(((int *)tab)+32+g)=*(((int *)data)+1); | |
7761 | +} | |
7762 | + | |
7763 | +void static inline FFTABLEOUT(unsigned char *data, unsigned char *tab, int g){ | |
7764 | + *((int *)data)=*(((int *)tab)+g); | |
7765 | + *(((int *)data)+1)=*(((int *)tab)+32+g); | |
7766 | +} | |
7767 | + | |
7768 | +void static inline FFTABLEOUTXORNBY(int n, unsigned char *data, unsigned char *tab, int g){ | |
7769 | + int j; | |
7770 | + for(j=0;j<n;j++){ | |
7771 | + *(data+j)^=*(tab+4*(g+(j>=4?32-1:0))+j); | |
7772 | + } | |
7773 | +} | |
7774 | + | |
7775 | +struct batch_t{ | |
7776 | + unsigned char s1,s2,s3,s4; | |
7777 | +}; | |
7778 | +typedef struct batch_t batch; | |
7779 | + | |
7780 | +#define BYTES_PER_BATCH 4 | |
7781 | + | |
7782 | +batch static inline B_FFAND(batch a,batch b){ | |
7783 | + batch res; | |
7784 | + res.s1=a.s1&b.s1; | |
7785 | + res.s2=a.s2&b.s2; | |
7786 | + res.s3=a.s3&b.s3; | |
7787 | + res.s4=a.s4&b.s4; | |
7788 | + return res; | |
7789 | +} | |
7790 | + | |
7791 | +batch static inline B_FFOR(batch a,batch b){ | |
7792 | + batch res; | |
7793 | + res.s1=a.s1|b.s1; | |
7794 | + res.s2=a.s2|b.s2; | |
7795 | + res.s3=a.s3|b.s3; | |
7796 | + res.s4=a.s4|b.s4; | |
7797 | + return res; | |
7798 | +} | |
7799 | + | |
7800 | +batch static inline B_FFXOR(batch a,batch b){ | |
7801 | + batch res; | |
7802 | + res.s1=a.s1^b.s1; | |
7803 | + res.s2=a.s2^b.s2; | |
7804 | + res.s3=a.s3^b.s3; | |
7805 | + res.s4=a.s4^b.s4; | |
7806 | + return res; | |
7807 | +} | |
7808 | + | |
7809 | + | |
7810 | +batch static inline B_FFN_ALL_29(){ | |
7811 | + batch res; | |
7812 | + res.s1=0x29; | |
7813 | + res.s2=0x29; | |
7814 | + res.s3=0x29; | |
7815 | + res.s4=0x29; | |
7816 | + return res; | |
7817 | +} | |
7818 | +batch static inline B_FFN_ALL_02(){ | |
7819 | + batch res; | |
7820 | + res.s1=0x02; | |
7821 | + res.s2=0x02; | |
7822 | + res.s3=0x02; | |
7823 | + res.s4=0x02; | |
7824 | + return res; | |
7825 | +} | |
7826 | +batch static inline B_FFN_ALL_04(){ | |
7827 | + batch res; | |
7828 | + res.s1=0x04; | |
7829 | + res.s2=0x04; | |
7830 | + res.s3=0x04; | |
7831 | + res.s4=0x04; | |
7832 | + return res; | |
7833 | +} | |
7834 | +batch static inline B_FFN_ALL_10(){ | |
7835 | + batch res; | |
7836 | + res.s1=0x10; | |
7837 | + res.s2=0x10; | |
7838 | + res.s3=0x10; | |
7839 | + res.s4=0x10; | |
7840 | + return res; | |
7841 | +} | |
7842 | +batch static inline B_FFN_ALL_40(){ | |
7843 | + batch res; | |
7844 | + res.s1=0x40; | |
7845 | + res.s2=0x40; | |
7846 | + res.s3=0x40; | |
7847 | + res.s4=0x40; | |
7848 | + return res; | |
7849 | +} | |
7850 | +batch static inline B_FFN_ALL_80(){ | |
7851 | + batch res; | |
7852 | + res.s1=0x80; | |
7853 | + res.s2=0x80; | |
7854 | + res.s3=0x80; | |
7855 | + res.s4=0x80; | |
7856 | + return res; | |
7857 | +} | |
7858 | + | |
7859 | +batch static inline B_FFSH8L(batch a,int n){ | |
7860 | + batch res; | |
7861 | + res.s1=a.s1<<n; | |
7862 | + res.s2=a.s2<<n; | |
7863 | + res.s3=a.s3<<n; | |
7864 | + res.s4=a.s4<<n; | |
7865 | + return res; | |
7866 | +} | |
7867 | + | |
7868 | +batch static inline B_FFSH8R(batch a,int n){ | |
7869 | + batch res; | |
7870 | + res.s1=a.s1>>n; | |
7871 | + res.s2=a.s2>>n; | |
7872 | + res.s3=a.s3>>n; | |
7873 | + res.s4=a.s4>>n; | |
7874 | + return res; | |
7875 | +} | |
7876 | + | |
7877 | + | |
7878 | +void static inline M_EMPTY(void){ | |
7879 | +} | |
7880 | Index: libs/libmythtv/FFdecsa/stream.c | |
7881 | =================================================================== | |
7882 | --- /dev/null 1970-01-01 00:00:00.000000000 +0000 | |
7883 | +++ libs/libmythtv/FFdecsa/stream.c 2006-06-20 17:36:06.000000000 -0400 | |
7884 | @@ -0,0 +1,906 @@ | |
7885 | +/* FFdecsa -- fast decsa algorithm | |
7886 | + * | |
7887 | + * Copyright (C) 2003-2004 fatih89r | |
7888 | + * | |
7889 | + * This program is free software; you can redistribute it and/or modify | |
7890 | + * it under the terms of the GNU General Public License as published by | |
7891 | + * the Free Software Foundation; either version 2 of the License, or | |
7892 | + * (at your option) any later version. | |
7893 | + * | |
7894 | + * This program is distributed in the hope that it will be useful, | |
7895 | + * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
7896 | + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
7897 | + * GNU General Public License for more details. | |
7898 | + * | |
7899 | + * You should have received a copy of the GNU General Public License | |
7900 | + * along with this program; if not, write to the Free Software | |
7901 | + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
7902 | + */ | |
7903 | + | |
7904 | + | |
7905 | + | |
7906 | +// define statics only once, when STREAM_INIT | |
7907 | +#ifdef STREAM_INIT | |
7908 | +struct stream_regs { | |
7909 | + group A[32+10][4]; // 32 because we will move back (virtual shift register) | |
7910 | + group B[32+10][4]; // 32 because we will move back (virtual shift register) | |
7911 | + group X[4]; | |
7912 | + group Y[4]; | |
7913 | + group Z[4]; | |
7914 | + group D[4]; | |
7915 | + group E[4]; | |
7916 | + group F[4]; | |
7917 | + group p; | |
7918 | + group q; | |
7919 | + group r; | |
7920 | + }; | |
7921 | + | |
7922 | +static inline void trasp64_32_88ccw(unsigned char *data){ | |
7923 | +/* 64 rows of 32 bits transposition (bytes transp. - 8x8 rotate counterclockwise)*/ | |
7924 | +#define row ((unsigned int *)data) | |
7925 | + int i,j; | |
7926 | + for(j=0;j<64;j+=32){ | |
7927 | + unsigned int t,b; | |
7928 | + for(i=0;i<16;i++){ | |
7929 | + t=row[j+i]; | |
7930 | + b=row[j+16+i]; | |
7931 | + row[j+i] = (t&0x0000ffff) | ((b )<<16); | |
7932 | + row[j+16+i]=((t )>>16) | (b&0xffff0000) ; | |
7933 | + } | |
7934 | + } | |
7935 | + for(j=0;j<64;j+=16){ | |
7936 | + unsigned int t,b; | |
7937 | + for(i=0;i<8;i++){ | |
7938 | + t=row[j+i]; | |
7939 | + b=row[j+8+i]; | |
7940 | + row[j+i] = (t&0x00ff00ff) | ((b&0x00ff00ff)<<8); | |
7941 | + row[j+8+i] =((t&0xff00ff00)>>8) | (b&0xff00ff00); | |
7942 | + } | |
7943 | + } | |
7944 | + for(j=0;j<64;j+=8){ | |
7945 | + unsigned int t,b; | |
7946 | + for(i=0;i<4;i++){ | |
7947 | + t=row[j+i]; | |
7948 | + b=row[j+4+i]; | |
7949 | + row[j+i] =((t&0x0f0f0f0f)<<4) | (b&0x0f0f0f0f); | |
7950 | + row[j+4+i] = (t&0xf0f0f0f0) | ((b&0xf0f0f0f0)>>4); | |
7951 | + } | |
7952 | + } | |
7953 | + for(j=0;j<64;j+=4){ | |
7954 | + unsigned int t,b; | |
7955 | + for(i=0;i<2;i++){ | |
7956 | + t=row[j+i]; | |
7957 | + b=row[j+2+i]; | |
7958 | + row[j+i] =((t&0x33333333)<<2) | (b&0x33333333); | |
7959 | + row[j+2+i] = (t&0xcccccccc) | ((b&0xcccccccc)>>2); | |
7960 | + } | |
7961 | + } | |
7962 | + for(j=0;j<64;j+=2){ | |
7963 | + unsigned int t,b; | |
7964 | + for(i=0;i<1;i++){ | |
7965 | + t=row[j+i]; | |
7966 | + b=row[j+1+i]; | |
7967 | + row[j+i] =((t&0x55555555)<<1) | (b&0x55555555); | |
7968 | + row[j+1+i] = (t&0xaaaaaaaa) | ((b&0xaaaaaaaa)>>1); | |
7969 | + } | |
7970 | + } | |
7971 | +#undef row | |
7972 | +} | |
7973 | + | |
7974 | +static inline void trasp64_32_88cw(unsigned char *data){ | |
7975 | +/* 64 rows of 32 bits transposition (bytes transp. - 8x8 rotate clockwise)*/ | |
7976 | +#define row ((unsigned int *)data) | |
7977 | + int i,j; | |
7978 | + for(j=0;j<64;j+=32){ | |
7979 | + unsigned int t,b; | |
7980 | + for(i=0;i<16;i++){ | |
7981 | + t=row[j+i]; | |
7982 | + b=row[j+16+i]; | |
7983 | + row[j+i] = (t&0x0000ffff) | ((b )<<16); | |
7984 | + row[j+16+i]=((t )>>16) | (b&0xffff0000) ; | |
7985 | + } | |
7986 | + } | |
7987 | + for(j=0;j<64;j+=16){ | |
7988 | + unsigned int t,b; | |
7989 | + for(i=0;i<8;i++){ | |
7990 | + t=row[j+i]; | |
7991 | + b=row[j+8+i]; | |
7992 | + row[j+i] = (t&0x00ff00ff) | ((b&0x00ff00ff)<<8); | |
7993 | + row[j+8+i] =((t&0xff00ff00)>>8) | (b&0xff00ff00); | |
7994 | + } | |
7995 | + } | |
7996 | + for(j=0;j<64;j+=8){ | |
7997 | + unsigned int t,b; | |
7998 | + for(i=0;i<4;i++){ | |
7999 | + t=row[j+i]; | |
8000 | + b=row[j+4+i]; | |
8001 | + row[j+i] =((t&0xf0f0f0f0)>>4) | (b&0xf0f0f0f0); | |
8002 | + row[j+4+i]= (t&0x0f0f0f0f) | ((b&0x0f0f0f0f)<<4); | |
8003 | + } | |
8004 | + } | |
8005 | + for(j=0;j<64;j+=4){ | |
8006 | + unsigned int t,b; | |
8007 | + for(i=0;i<2;i++){ | |
8008 | + t=row[j+i]; | |
8009 | + b=row[j+2+i]; | |
8010 | + row[j+i] =((t&0xcccccccc)>>2) | (b&0xcccccccc); | |
8011 | + row[j+2+i]= (t&0x33333333) | ((b&0x33333333)<<2); | |
8012 | + } | |
8013 | + } | |
8014 | + for(j=0;j<64;j+=2){ | |
8015 | + unsigned int t,b; | |
8016 | + for(i=0;i<1;i++){ | |
8017 | + t=row[j+i]; | |
8018 | + b=row[j+1+i]; | |
8019 | + row[j+i] =((t&0xaaaaaaaa)>>1) | (b&0xaaaaaaaa); | |
8020 | + row[j+1+i]= (t&0x55555555) | ((b&0x55555555)<<1); | |
8021 | + } | |
8022 | + } | |
8023 | +#undef row | |
8024 | +} | |
8025 | + | |
8026 | +//64-64---------------------------------------------------------- | |
8027 | +static inline void trasp64_64_88ccw(unsigned char *data){ | |
8028 | +/* 64 rows of 64 bits transposition (bytes transp. - 8x8 rotate counterclockwise)*/ | |
8029 | +#define row ((unsigned long long int *)data) | |
8030 | + int i,j; | |
8031 | + for(j=0;j<64;j+=64){ | |
8032 | + unsigned long long int t,b; | |
8033 | + for(i=0;i<32;i++){ | |
8034 | + t=row[j+i]; | |
8035 | + b=row[j+32+i]; | |
8036 | + row[j+i] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
8037 | + row[j+32+i]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
8038 | + } | |
8039 | + } | |
8040 | + for(j=0;j<64;j+=32){ | |
8041 | + unsigned long long int t,b; | |
8042 | + for(i=0;i<16;i++){ | |
8043 | + t=row[j+i]; | |
8044 | + b=row[j+16+i]; | |
8045 | + row[j+i] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
8046 | + row[j+16+i]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
8047 | + } | |
8048 | + } | |
8049 | + for(j=0;j<64;j+=16){ | |
8050 | + unsigned long long int t,b; | |
8051 | + for(i=0;i<8;i++){ | |
8052 | + t=row[j+i]; | |
8053 | + b=row[j+8+i]; | |
8054 | + row[j+i] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
8055 | + row[j+8+i] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
8056 | + } | |
8057 | + } | |
8058 | + for(j=0;j<64;j+=8){ | |
8059 | + unsigned long long int t,b; | |
8060 | + for(i=0;i<4;i++){ | |
8061 | + t=row[j+i]; | |
8062 | + b=row[j+4+i]; | |
8063 | + row[j+i] =((t&0x0f0f0f0f0f0f0f0fULL)<<4) | (b&0x0f0f0f0f0f0f0f0fULL); | |
8064 | + row[j+4+i] = (t&0xf0f0f0f0f0f0f0f0ULL) | ((b&0xf0f0f0f0f0f0f0f0ULL)>>4); | |
8065 | + } | |
8066 | + } | |
8067 | + for(j=0;j<64;j+=4){ | |
8068 | + unsigned long long int t,b; | |
8069 | + for(i=0;i<2;i++){ | |
8070 | + t=row[j+i]; | |
8071 | + b=row[j+2+i]; | |
8072 | + row[j+i] =((t&0x3333333333333333ULL)<<2) | (b&0x3333333333333333ULL); | |
8073 | + row[j+2+i] = (t&0xccccccccccccccccULL) | ((b&0xccccccccccccccccULL)>>2); | |
8074 | + } | |
8075 | + } | |
8076 | + for(j=0;j<64;j+=2){ | |
8077 | + unsigned long long int t,b; | |
8078 | + for(i=0;i<1;i++){ | |
8079 | + t=row[j+i]; | |
8080 | + b=row[j+1+i]; | |
8081 | + row[j+i] =((t&0x5555555555555555ULL)<<1) | (b&0x5555555555555555ULL); | |
8082 | + row[j+1+i] = (t&0xaaaaaaaaaaaaaaaaULL) | ((b&0xaaaaaaaaaaaaaaaaULL)>>1); | |
8083 | + } | |
8084 | + } | |
8085 | +#undef row | |
8086 | +} | |
8087 | + | |
8088 | +static inline void trasp64_64_88cw(unsigned char *data){ | |
8089 | +/* 64 rows of 64 bits transposition (bytes transp. - 8x8 rotate clockwise)*/ | |
8090 | +#define row ((unsigned long long int *)data) | |
8091 | + int i,j; | |
8092 | + for(j=0;j<64;j+=64){ | |
8093 | + unsigned long long int t,b; | |
8094 | + for(i=0;i<32;i++){ | |
8095 | + t=row[j+i]; | |
8096 | + b=row[j+32+i]; | |
8097 | + row[j+i] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
8098 | + row[j+32+i]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
8099 | + } | |
8100 | + } | |
8101 | + for(j=0;j<64;j+=32){ | |
8102 | + unsigned long long int t,b; | |
8103 | + for(i=0;i<16;i++){ | |
8104 | + t=row[j+i]; | |
8105 | + b=row[j+16+i]; | |
8106 | + row[j+i] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
8107 | + row[j+16+i]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
8108 | + } | |
8109 | + } | |
8110 | + for(j=0;j<64;j+=16){ | |
8111 | + unsigned long long int t,b; | |
8112 | + for(i=0;i<8;i++){ | |
8113 | + t=row[j+i]; | |
8114 | + b=row[j+8+i]; | |
8115 | + row[j+i] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
8116 | + row[j+8+i] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
8117 | + } | |
8118 | + } | |
8119 | + for(j=0;j<64;j+=8){ | |
8120 | + unsigned long long int t,b; | |
8121 | + for(i=0;i<4;i++){ | |
8122 | + t=row[j+i]; | |
8123 | + b=row[j+4+i]; | |
8124 | + row[j+i] =((t&0xf0f0f0f0f0f0f0f0ULL)>>4) | (b&0xf0f0f0f0f0f0f0f0ULL); | |
8125 | + row[j+4+i] = (t&0x0f0f0f0f0f0f0f0fULL) | ((b&0x0f0f0f0f0f0f0f0fULL)<<4); | |
8126 | + } | |
8127 | + } | |
8128 | + for(j=0;j<64;j+=4){ | |
8129 | + unsigned long long int t,b; | |
8130 | + for(i=0;i<2;i++){ | |
8131 | + t=row[j+i]; | |
8132 | + b=row[j+2+i]; | |
8133 | + row[j+i] =((t&0xccccccccccccccccULL)>>2) | (b&0xccccccccccccccccULL); | |
8134 | + row[j+2+i] = (t&0x3333333333333333ULL) | ((b&0x3333333333333333ULL)<<2); | |
8135 | + } | |
8136 | + } | |
8137 | + for(j=0;j<64;j+=2){ | |
8138 | + unsigned long long int t,b; | |
8139 | + for(i=0;i<1;i++){ | |
8140 | + t=row[j+i]; | |
8141 | + b=row[j+1+i]; | |
8142 | + row[j+i] =((t&0xaaaaaaaaaaaaaaaaULL)>>1) | (b&0xaaaaaaaaaaaaaaaaULL); | |
8143 | + row[j+1+i] = (t&0x5555555555555555ULL) | ((b&0x5555555555555555ULL)<<1); | |
8144 | + } | |
8145 | + } | |
8146 | +#undef row | |
8147 | +} | |
8148 | + | |
8149 | +//64-128---------------------------------------------------------- | |
8150 | +static inline void trasp64_128_88ccw(unsigned char *data){ | |
8151 | +/* 64 rows of 128 bits transposition (bytes transp. - 8x8 rotate counterclockwise)*/ | |
8152 | +#define halfrow ((unsigned long long int *)data) | |
8153 | + int i,j; | |
8154 | + for(j=0;j<64;j+=64){ | |
8155 | + unsigned long long int t,b; | |
8156 | + for(i=0;i<32;i++){ | |
8157 | + t=halfrow[2*(j+i)]; | |
8158 | + b=halfrow[2*(j+32+i)]; | |
8159 | + halfrow[2*(j+i)] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
8160 | + halfrow[2*(j+32+i)]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
8161 | + t=halfrow[2*(j+i)+1]; | |
8162 | + b=halfrow[2*(j+32+i)+1]; | |
8163 | + halfrow[2*(j+i)+1] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
8164 | + halfrow[2*(j+32+i)+1]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
8165 | + } | |
8166 | + } | |
8167 | + for(j=0;j<64;j+=32){ | |
8168 | + unsigned long long int t,b; | |
8169 | + for(i=0;i<16;i++){ | |
8170 | + t=halfrow[2*(j+i)]; | |
8171 | + b=halfrow[2*(j+16+i)]; | |
8172 | + halfrow[2*(j+i)] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
8173 | + halfrow[2*(j+16+i)]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
8174 | + t=halfrow[2*(j+i)+1]; | |
8175 | + b=halfrow[2*(j+16+i)+1]; | |
8176 | + halfrow[2*(j+i)+1] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
8177 | + halfrow[2*(j+16+i)+1]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
8178 | + } | |
8179 | + } | |
8180 | + for(j=0;j<64;j+=16){ | |
8181 | + unsigned long long int t,b; | |
8182 | + for(i=0;i<8;i++){ | |
8183 | + t=halfrow[2*(j+i)]; | |
8184 | + b=halfrow[2*(j+8+i)]; | |
8185 | + halfrow[2*(j+i)] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
8186 | + halfrow[2*(j+8+i)] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
8187 | + t=halfrow[2*(j+i)+1]; | |
8188 | + b=halfrow[2*(j+8+i)+1]; | |
8189 | + halfrow[2*(j+i)+1] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
8190 | + halfrow[2*(j+8+i)+1] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
8191 | + } | |
8192 | + } | |
8193 | + for(j=0;j<64;j+=8){ | |
8194 | + unsigned long long int t,b; | |
8195 | + for(i=0;i<4;i++){ | |
8196 | + t=halfrow[2*(j+i)]; | |
8197 | + b=halfrow[2*(j+4+i)]; | |
8198 | + halfrow[2*(j+i)] =((t&0x0f0f0f0f0f0f0f0fULL)<<4) | (b&0x0f0f0f0f0f0f0f0fULL); | |
8199 | + halfrow[2*(j+4+i)] = (t&0xf0f0f0f0f0f0f0f0ULL) | ((b&0xf0f0f0f0f0f0f0f0ULL)>>4); | |
8200 | + t=halfrow[2*(j+i)+1]; | |
8201 | + b=halfrow[2*(j+4+i)+1]; | |
8202 | + halfrow[2*(j+i)+1] =((t&0x0f0f0f0f0f0f0f0fULL)<<4) | (b&0x0f0f0f0f0f0f0f0fULL); | |
8203 | + halfrow[2*(j+4+i)+1] = (t&0xf0f0f0f0f0f0f0f0ULL) | ((b&0xf0f0f0f0f0f0f0f0ULL)>>4); | |
8204 | + } | |
8205 | + } | |
8206 | + for(j=0;j<64;j+=4){ | |
8207 | + unsigned long long int t,b; | |
8208 | + for(i=0;i<2;i++){ | |
8209 | + t=halfrow[2*(j+i)]; | |
8210 | + b=halfrow[2*(j+2+i)]; | |
8211 | + halfrow[2*(j+i)] =((t&0x3333333333333333ULL)<<2) | (b&0x3333333333333333ULL); | |
8212 | + halfrow[2*(j+2+i)] = (t&0xccccccccccccccccULL) | ((b&0xccccccccccccccccULL)>>2); | |
8213 | + t=halfrow[2*(j+i)+1]; | |
8214 | + b=halfrow[2*(j+2+i)+1]; | |
8215 | + halfrow[2*(j+i)+1] =((t&0x3333333333333333ULL)<<2) | (b&0x3333333333333333ULL); | |
8216 | + halfrow[2*(j+2+i)+1] = (t&0xccccccccccccccccULL) | ((b&0xccccccccccccccccULL)>>2); | |
8217 | + } | |
8218 | + } | |
8219 | + for(j=0;j<64;j+=2){ | |
8220 | + unsigned long long int t,b; | |
8221 | + for(i=0;i<1;i++){ | |
8222 | + t=halfrow[2*(j+i)]; | |
8223 | + b=halfrow[2*(j+1+i)]; | |
8224 | + halfrow[2*(j+i)] =((t&0x5555555555555555ULL)<<1) | (b&0x5555555555555555ULL); | |
8225 | + halfrow[2*(j+1+i)] = (t&0xaaaaaaaaaaaaaaaaULL) | ((b&0xaaaaaaaaaaaaaaaaULL)>>1); | |
8226 | + t=halfrow[2*(j+i)+1]; | |
8227 | + b=halfrow[2*(j+1+i)+1]; | |
8228 | + halfrow[2*(j+i)+1] =((t&0x5555555555555555ULL)<<1) | (b&0x5555555555555555ULL); | |
8229 | + halfrow[2*(j+1+i)+1] = (t&0xaaaaaaaaaaaaaaaaULL) | ((b&0xaaaaaaaaaaaaaaaaULL)>>1); | |
8230 | + } | |
8231 | + } | |
8232 | +#undef halfrow | |
8233 | +} | |
8234 | + | |
8235 | +static inline void trasp64_128_88cw(unsigned char *data){ | |
8236 | +/* 64 rows of 128 bits transposition (bytes transp. - 8x8 rotate clockwise)*/ | |
8237 | +#define halfrow ((unsigned long long int *)data) | |
8238 | + int i,j; | |
8239 | + for(j=0;j<64;j+=64){ | |
8240 | + unsigned long long int t,b; | |
8241 | + for(i=0;i<32;i++){ | |
8242 | + t=halfrow[2*(j+i)]; | |
8243 | + b=halfrow[2*(j+32+i)]; | |
8244 | + halfrow[2*(j+i)] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
8245 | + halfrow[2*(j+32+i)]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
8246 | + t=halfrow[2*(j+i)+1]; | |
8247 | + b=halfrow[2*(j+32+i)+1]; | |
8248 | + halfrow[2*(j+i)+1] = (t&0x00000000ffffffffULL) | ((b )<<32); | |
8249 | + halfrow[2*(j+32+i)+1]=((t )>>32) | (b&0xffffffff00000000ULL) ; | |
8250 | + } | |
8251 | + } | |
8252 | + for(j=0;j<64;j+=32){ | |
8253 | + unsigned long long int t,b; | |
8254 | + for(i=0;i<16;i++){ | |
8255 | + t=halfrow[2*(j+i)]; | |
8256 | + b=halfrow[2*(j+16+i)]; | |
8257 | + halfrow[2*(j+i)] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
8258 | + halfrow[2*(j+16+i)]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
8259 | + t=halfrow[2*(j+i)+1]; | |
8260 | + b=halfrow[2*(j+16+i)+1]; | |
8261 | + halfrow[2*(j+i)+1] = (t&0x0000ffff0000ffffULL) | ((b&0x0000ffff0000ffffULL)<<16); | |
8262 | + halfrow[2*(j+16+i)+1]=((t&0xffff0000ffff0000ULL)>>16) | (b&0xffff0000ffff0000ULL) ; | |
8263 | + } | |
8264 | + } | |
8265 | + for(j=0;j<64;j+=16){ | |
8266 | + unsigned long long int t,b; | |
8267 | + for(i=0;i<8;i++){ | |
8268 | + t=halfrow[2*(j+i)]; | |
8269 | + b=halfrow[2*(j+8+i)]; | |
8270 | + halfrow[2*(j+i)] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
8271 | + halfrow[2*(j+8+i)] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
8272 | + t=halfrow[2*(j+i)+1]; | |
8273 | + b=halfrow[2*(j+8+i)+1]; | |
8274 | + halfrow[2*(j+i)+1] = (t&0x00ff00ff00ff00ffULL) | ((b&0x00ff00ff00ff00ffULL)<<8); | |
8275 | + halfrow[2*(j+8+i)+1] =((t&0xff00ff00ff00ff00ULL)>>8) | (b&0xff00ff00ff00ff00ULL); | |
8276 | + } | |
8277 | + } | |
8278 | + for(j=0;j<64;j+=8){ | |
8279 | + unsigned long long int t,b; | |
8280 | + for(i=0;i<4;i++){ | |
8281 | + t=halfrow[2*(j+i)]; | |
8282 | + b=halfrow[2*(j+4+i)]; | |
8283 | + halfrow[2*(j+i)] =((t&0xf0f0f0f0f0f0f0f0ULL)>>4) | (b&0xf0f0f0f0f0f0f0f0ULL); | |
8284 | + halfrow[2*(j+4+i)] = (t&0x0f0f0f0f0f0f0f0fULL) | ((b&0x0f0f0f0f0f0f0f0fULL)<<4); | |
8285 | + t=halfrow[2*(j+i)+1]; | |
8286 | + b=halfrow[2*(j+4+i)+1]; | |
8287 | + halfrow[2*(j+i)+1] =((t&0xf0f0f0f0f0f0f0f0ULL)>>4) | (b&0xf0f0f0f0f0f0f0f0ULL); | |
8288 | + halfrow[2*(j+4+i)+1] = (t&0x0f0f0f0f0f0f0f0fULL) | ((b&0x0f0f0f0f0f0f0f0fULL)<<4); | |
8289 | + } | |
8290 | + } | |
8291 | + for(j=0;j<64;j+=4){ | |
8292 | + unsigned long long int t,b; | |
8293 | + for(i=0;i<2;i++){ | |
8294 | + t=halfrow[2*(j+i)]; | |
8295 | + b=halfrow[2*(j+2+i)]; | |
8296 | + halfrow[2*(j+i)] =((t&0xccccccccccccccccULL)>>2) | (b&0xccccccccccccccccULL); | |
8297 | + halfrow[2*(j+2+i)] = (t&0x3333333333333333ULL) | ((b&0x3333333333333333ULL)<<2); | |
8298 | + t=halfrow[2*(j+i)+1]; | |
8299 | + b=halfrow[2*(j+2+i)+1]; | |
8300 | + halfrow[2*(j+i)+1] =((t&0xccccccccccccccccULL)>>2) | (b&0xccccccccccccccccULL); | |
8301 | + halfrow[2*(j+2+i)+1] = (t&0x3333333333333333ULL) | ((b&0x3333333333333333ULL)<<2); | |
8302 | + } | |
8303 | + } | |
8304 | + for(j=0;j<64;j+=2){ | |
8305 | + unsigned long long int t,b; | |
8306 | + for(i=0;i<1;i++){ | |
8307 | + t=halfrow[2*(j+i)]; | |
8308 | + b=halfrow[2*(j+1+i)]; | |
8309 | + halfrow[2*(j+i)] =((t&0xaaaaaaaaaaaaaaaaULL)>>1) | (b&0xaaaaaaaaaaaaaaaaULL); | |
8310 | + halfrow[2*(j+1+i)] = (t&0x5555555555555555ULL) | ((b&0x5555555555555555ULL)<<1); | |
8311 | + t=halfrow[2*(j+i)+1]; | |
8312 | + b=halfrow[2*(j+1+i)+1]; | |
8313 | + halfrow[2*(j+i)+1] =((t&0xaaaaaaaaaaaaaaaaULL)>>1) | (b&0xaaaaaaaaaaaaaaaaULL); | |
8314 | + halfrow[2*(j+1+i)+1] = (t&0x5555555555555555ULL) | ((b&0x5555555555555555ULL)<<1); | |
8315 | + } | |
8316 | + } | |
8317 | +#undef halfrow | |
8318 | +} | |
8319 | +#endif | |
8320 | + | |
8321 | + | |
8322 | +#ifdef STREAM_INIT | |
8323 | +void stream_cypher_group_init( | |
8324 | + struct stream_regs *regs, | |
8325 | + group iA[8][4], // [In] iA00,iA01,...iA73 32 groups | Derived from key. | |
8326 | + group iB[8][4], // [In] iB00,iB01,...iB73 32 groups | Derived from key. | |
8327 | + unsigned char *sb) // [In] (SB0,SB1,...SB7)...x32 32*8 bytes | Extra input. | |
8328 | +#endif | |
8329 | +#ifdef STREAM_NORMAL | |
8330 | +void stream_cypher_group_normal( | |
8331 | + struct stream_regs *regs, | |
8332 | + unsigned char *cb) // [Out] (CB0,CB1,...CB7)...x32 32*8 bytes | Output. | |
8333 | +#endif | |
8334 | +{ | |
8335 | +#ifdef STREAM_INIT | |
8336 | + group in1[4]; | |
8337 | + group in2[4]; | |
8338 | +#endif | |
8339 | + group extra_B[4]; | |
8340 | + group fa,fb,fc,fd,fe; | |
8341 | + group s1a,s1b,s2a,s2b,s3a,s3b,s4a,s4b,s5a,s5b,s6a,s6b,s7a,s7b; | |
8342 | + group next_E[4]; | |
8343 | + group tmp0,tmp1,tmp2,tmp3,tmp4; | |
8344 | +#ifdef STREAM_INIT | |
8345 | + group *sb_g=(group *)sb; | |
8346 | +#endif | |
8347 | +#ifdef STREAM_NORMAL | |
8348 | + group *cb_g=(group *)cb; | |
8349 | +#endif | |
8350 | + int aboff; | |
8351 | + int i,j,k,b; | |
8352 | + int dbg; | |
8353 | + | |
8354 | +#ifdef STREAM_INIT | |
8355 | + DBG(fprintf(stderr,":::::::::: BEGIN STREAM INIT\n")); | |
8356 | +#endif | |
8357 | +#ifdef STREAM_NORMAL | |
8358 | + DBG(fprintf(stderr,":::::::::: BEGIN STREAM NORMAL\n")); | |
8359 | +#endif | |
8360 | +#ifdef STREAM_INIT | |
8361 | +for(j=0;j<64;j++){ | |
8362 | + DBG(fprintf(stderr,"precall prerot stream_in[%2i]=",j)); | |
8363 | + DBG(dump_mem("",sb+BYPG*j,BYPG,BYPG)); | |
8364 | +} | |
8365 | + | |
8366 | +DBG(dump_mem("stream_prerot ",sb,GROUP_PARALLELISM*8,BYPG)); | |
8367 | +#if GROUP_PARALLELISM==32 | |
8368 | +trasp64_32_88ccw(sb); | |
8369 | +#endif | |
8370 | +#if GROUP_PARALLELISM==64 | |
8371 | +trasp64_64_88ccw(sb); | |
8372 | +#endif | |
8373 | +#if GROUP_PARALLELISM==128 | |
8374 | +trasp64_128_88ccw(sb); | |
8375 | +#endif | |
8376 | +DBG(dump_mem("stream_postrot",sb,GROUP_PARALLELISM*8,BYPG)); | |
8377 | + | |
8378 | +for(j=0;j<64;j++){ | |
8379 | + DBG(fprintf(stderr,"precall stream_in[%2i]=",j)); | |
8380 | + DBG(dump_mem("",sb+BYPG*j,BYPG,BYPG)); | |
8381 | +} | |
8382 | +#endif | |
8383 | + | |
8384 | + aboff=32; | |
8385 | + | |
8386 | +#ifdef STREAM_INIT | |
8387 | + // load first 32 bits of ck into A[aboff+0]..A[aboff+7] | |
8388 | + // load last 32 bits of ck into B[aboff+0]..B[aboff+7] | |
8389 | + // all other regs = 0 | |
8390 | + for(i=0;i<8;i++){ | |
8391 | + for(b=0;b<4;b++){ | |
8392 | +DBG(fprintf(stderr,"dbg from iA A[%i][%i]=",i,b)); | |
8393 | +DBG(dump_mem("",(unsigned char *)&iA[i][b],BYPG,BYPG)); | |
8394 | +DBG(fprintf(stderr," dbg from iB B[%i][%i]=",i,b)); | |
8395 | +DBG(dump_mem("",(unsigned char *)&iB[i][b],BYPG,BYPG)); | |
8396 | + regs->A[aboff+i][b]=iA[i][b]; | |
8397 | + regs->B[aboff+i][b]=iB[i][b]; | |
8398 | + } | |
8399 | + } | |
8400 | + for(b=0;b<4;b++){ | |
8401 | + regs->A[aboff+8][b]=FF0(); | |
8402 | + regs->A[aboff+9][b]=FF0(); | |
8403 | + regs->B[aboff+8][b]=FF0(); | |
8404 | + regs->B[aboff+9][b]=FF0(); | |
8405 | + } | |
8406 | + for(b=0;b<4;b++){ | |
8407 | + regs->X[b]=FF0(); | |
8408 | + regs->Y[b]=FF0(); | |
8409 | + regs->Z[b]=FF0(); | |
8410 | + regs->D[b]=FF0(); | |
8411 | + regs->E[b]=FF0(); | |
8412 | + regs->F[b]=FF0(); | |
8413 | + } | |
8414 | + regs->p=FF0(); | |
8415 | + regs->q=FF0(); | |
8416 | + regs->r=FF0(); | |
8417 | +#endif | |
8418 | + | |
8419 | +for(dbg=0;dbg<4;dbg++){ | |
8420 | + DBG(fprintf(stderr,"dbg A0[%i]=",dbg)); | |
8421 | + DBG(dump_mem("",(unsigned char *)®s->A[aboff+0][dbg],BYPG,BYPG)); | |
8422 | + DBG(fprintf(stderr,"dbg B0[%i]=",dbg)); | |
8423 | + DBG(dump_mem("",(unsigned char *)®s->B[aboff+0][dbg],BYPG,BYPG)); | |
8424 | +} | |
8425 | + | |
8426 | +//////////////////////////////////////////////////////////////////////////////// | |
8427 | + | |
8428 | + // EXTERNAL LOOP - 8 bytes per operation | |
8429 | + for(i=0;i<8;i++){ | |
8430 | + | |
8431 | + DBG(fprintf(stderr,"--BEGIN EXTERNAL LOOP %i\n",i)); | |
8432 | + | |
8433 | +#ifdef STREAM_INIT | |
8434 | + for(b=0;b<4;b++){ | |
8435 | + in1[b]=sb_g[8*i+4+b]; | |
8436 | + in2[b]=sb_g[8*i+b]; | |
8437 | + } | |
8438 | +#endif | |
8439 | + | |
8440 | + // INTERNAL LOOP - 2 bits per iteration | |
8441 | + for(j=0; j<4; j++){ | |
8442 | + | |
8443 | + DBG(fprintf(stderr,"---BEGIN INTERNAL LOOP %i (EXT %i, INT %i)\n",j,i,j)); | |
8444 | + | |
8445 | + // from A0..A9, 35 bits are selected as inputs to 7 s-boxes | |
8446 | + // 5 bits input per s-box, 2 bits output per s-box | |
8447 | + | |
8448 | + // we can select bits with zero masking and shifting operations | |
8449 | + // and synthetize s-boxes with optimized boolean functions. | |
8450 | + // this is the actual reason we do all the crazy transposition | |
8451 | + // stuff to switch between normal and bit slice representations. | |
8452 | + // this code really flies. | |
8453 | + | |
8454 | + fe=regs->A[aboff+3][0];fa=regs->A[aboff+0][2];fb=regs->A[aboff+5][1];fc=regs->A[aboff+6][3];fd=regs->A[aboff+8][0]; | |
8455 | +/* 1000 1110 1110 0001 : lev 7: */ //tmp0=( fa^( fb^( ( ( ( fa|fb )^fc )|( fc^fd ) )^ALL_ONES ) ) ); | |
8456 | +/* 1110 0010 0011 0011 : lev 6: */ //tmp1=( ( fa|fb )^( ( fc&( fa|( fb^fd ) ) )^ALL_ONES ) ); | |
8457 | +/* 0011 0110 1000 1101 : lev 5: */ //tmp2=( fa^( ( fb&fd )^( ( fa&fd )|fc ) ) ); | |
8458 | +/* 0101 0101 1001 0011 : lev 5: */ //tmp3=( ( fa&fc )^( fa^( ( fa&fb )|fd ) ) ); | |
8459 | +/* 1000 1110 1110 0001 : lev 7: */ tmp0=FFXOR(fa,FFXOR(fb,FFXOR(FFOR(FFXOR(FFOR(fa,fb),fc),FFXOR(fc,fd)),FF1()))); | |
8460 | +/* 1110 0010 0011 0011 : lev 6: */ tmp1=FFXOR(FFOR(fa,fb),FFXOR(FFAND(fc,FFOR(fa,FFXOR(fb,fd))),FF1())); | |
8461 | +/* 0011 0110 1000 1101 : lev 5: */ tmp2=FFXOR(fa,FFXOR(FFAND(fb,fd),FFOR(FFAND(fa,fd),fc))); | |
8462 | +/* 0101 0101 1001 0011 : lev 5: */ tmp3=FFXOR(FFAND(fa,fc),FFXOR(fa,FFOR(FFAND(fa,fb),fd))); | |
8463 | + s1a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
8464 | + s1b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
8465 | +//dump_mem("s1as1b-fe",&fe,BYPG,BYPG); | |
8466 | +//dump_mem("s1as1b-fa",&fa,BYPG,BYPG); | |
8467 | +//dump_mem("s1as1b-fb",&fb,BYPG,BYPG); | |
8468 | +//dump_mem("s1as1b-fc",&fc,BYPG,BYPG); | |
8469 | +//dump_mem("s1as1b-fd",&fd,BYPG,BYPG); | |
8470 | + | |
8471 | + fe=regs->A[aboff+1][1];fa=regs->A[aboff+2][2];fb=regs->A[aboff+5][3];fc=regs->A[aboff+6][0];fd=regs->A[aboff+8][1]; | |
8472 | +/* 1001 1110 0110 0001 : lev 6: */ //tmp0=( fa^( ( fb&( fc|fd ) )^( fc^( fd^ALL_ONES ) ) ) ); | |
8473 | +/* 0000 0011 0111 1011 : lev 5: */ //tmp1=( ( fa&( fb^fd ) )|( ( fa|fb )&fc ) ); | |
8474 | +/* 1100 0110 1101 0010 : lev 6: */ //tmp2=( ( fb&fd )^( ( fa&fd )|( fb^( fc^ALL_ONES ) ) ) ); | |
8475 | +/* 0001 1110 1111 0101 : lev 5: */ //tmp3=( ( fa&fd )|( fa^( fb^( fc&fd ) ) ) ); | |
8476 | +/* 1001 1110 0110 0001 : lev 6: */ tmp0=FFXOR(fa,FFXOR(FFAND(fb,FFOR(fc,fd)),FFXOR(fc,FFXOR(fd,FF1())))); | |
8477 | +/* 0000 0011 0111 1011 : lev 5: */ tmp1=FFOR(FFAND(fa,FFXOR(fb,fd)),FFAND(FFOR(fa,fb),fc)); | |
8478 | +/* 1100 0110 1101 0010 : lev 6: */ tmp2=FFXOR(FFAND(fb,fd),FFOR(FFAND(fa,fd),FFXOR(fb,FFXOR(fc,FF1())))); | |
8479 | +/* 0001 1110 1111 0101 : lev 5: */ tmp3=FFOR(FFAND(fa,fd),FFXOR(fa,FFXOR(fb,FFAND(fc,fd)))); | |
8480 | + s2a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
8481 | + s2b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
8482 | + | |
8483 | + fe=regs->A[aboff+0][3];fa=regs->A[aboff+1][0];fb=regs->A[aboff+4][1];fc=regs->A[aboff+4][3];fd=regs->A[aboff+5][2]; | |
8484 | +/* 0100 1011 1001 0110 : lev 5: */ //tmp0=( fa^( fb^( ( fc&( fa|fd ) )^fd ) ) ); | |
8485 | +/* 1101 0101 1000 1100 : lev 7: */ //tmp1=( ( fa&fc )^( ( fa^fd )|( ( fb|fc )^( fd^ALL_ONES ) ) ) ); | |
8486 | +/* 0010 0111 1101 1000 : lev 4: */ //tmp2=( fa^( ( ( fb^fc )&fd )^fc ) ); | |
8487 | +/* 1111 1111 1111 1111 : lev 0: */ //tmp3=ALL_ONES; | |
8488 | +/* 0100 1011 1001 0110 : lev 5: */ tmp0=FFXOR(fa,FFXOR(fb,FFXOR(FFAND(fc,FFOR(fa,fd)),fd))); | |
8489 | +/* 1101 0101 1000 1100 : lev 7: */ tmp1=FFXOR(FFAND(fa,fc),FFOR(FFXOR(fa,fd),FFXOR(FFOR(fb,fc),FFXOR(fd,FF1())))); | |
8490 | +/* 0010 0111 1101 1000 : lev 4: */ tmp2=FFXOR(fa,FFXOR(FFAND(FFXOR(fb,fc),fd),fc)); | |
8491 | +/* 1111 1111 1111 1111 : lev 0: */ tmp3=FF1(); | |
8492 | + s3a=FFXOR(tmp0,FFAND(FFNOT(fe),tmp1)); | |
8493 | + s3b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
8494 | + | |
8495 | + fe=regs->A[aboff+2][3];fa=regs->A[aboff+0][1];fb=regs->A[aboff+1][3];fc=regs->A[aboff+3][2];fd=regs->A[aboff+7][0]; | |
8496 | +/* 1011 0101 0100 1001 : lev 7: */ //tmp0=( fa^( ( fc&( fa^fd ) )|( fb^( fc|( fd^ALL_ONES ) ) ) ) ); | |
8497 | +/* 0010 1101 0110 0110 : lev 6: */ //tmp1=( ( fa&fb )^( fb^( ( ( fa|fc )&fd )^fc ) ) ); | |
8498 | +/* 0110 0111 1101 0000 : lev 7: */ //tmp2=( fa^( ( fb&fc )|( ( ( fa&( fb^fd ) )|fc )^fd ) ) ); | |
8499 | +/* 1111 1111 1111 1111 : lev 0: */ //tmp3=ALL_ONES; | |
8500 | +/* 1011 0101 0100 1001 : lev 7: */ tmp0=FFXOR(fa,FFOR(FFAND(fc,FFXOR(fa,fd)),FFXOR(fb,FFOR(fc,FFXOR(fd,FF1()))))); | |
8501 | +/* 0010 1101 0110 0110 : lev 6: */ tmp1=FFXOR(FFAND(fa,fb),FFXOR(fb,FFXOR(FFAND(FFOR(fa,fc),fd),fc))); | |
8502 | +/* 0110 0111 1101 0000 : lev 7: */ tmp2=FFXOR(fa,FFOR(FFAND(fb,fc),FFXOR(FFOR(FFAND(fa,FFXOR(fb,fd)),fc),fd))); | |
8503 | +/* 1111 1111 1111 1111 : lev 0: */ tmp3=FF1(); | |
8504 | + s4a=FFXOR(tmp0,FFAND(fe,FFXOR(tmp1,tmp0))); | |
8505 | + s4b=FFXOR(FFXOR(s4a,tmp2),FFAND(fe,tmp3)); | |
8506 | + | |
8507 | + fe=regs->A[aboff+4][2];fa=regs->A[aboff+3][3];fb=regs->A[aboff+5][0];fc=regs->A[aboff+7][1];fd=regs->A[aboff+8][2]; | |
8508 | +/* 1000 1111 0011 0010 : lev 7: */ //tmp0=( ( ( fa&( fb|fc ) )^fb )|( ( ( fa^fc )|fd )^ALL_ONES ) ); | |
8509 | +/* 0110 1011 0000 1011 : lev 6: */ //tmp1=( fb^( ( fc^fd )&( fc^( fb|( fa^fd ) ) ) ) ); | |
8510 | +/* 0001 1010 0111 1001 : lev 6: */ //tmp2=( ( fa&fc )^( fb^( ( fb|( fa^fc ) )&fd ) ) ); | |
8511 | +/* 0101 1101 1101 0101 : lev 4: */ //tmp3=( ( ( fa^fb )&( fc^ALL_ONES ) )|fd ); | |
8512 | +/* 1000 1111 0011 0010 : lev 7: */ tmp0=FFOR(FFXOR(FFAND(fa,FFOR(fb,fc)),fb),FFXOR(FFOR(FFXOR(fa,fc),fd),FF1())); | |
8513 | +/* 0110 1011 0000 1011 : lev 6: */ tmp1=FFXOR(fb,FFAND(FFXOR(fc,fd),FFXOR(fc,FFOR(fb,FFXOR(fa,fd))))); | |
8514 | +/* 0001 1010 0111 1001 : lev 6: */ tmp2=FFXOR(FFAND(fa,fc),FFXOR(fb,FFAND(FFOR(fb,FFXOR(fa,fc)),fd))); | |
8515 | +/* 0101 1101 1101 0101 : lev 4: */ tmp3=FFOR(FFAND(FFXOR(fa,fb),FFXOR(fc,FF1())),fd); | |
8516 | + s5a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
8517 | + s5b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
8518 | + | |
8519 | + fe=regs->A[aboff+2][1];fa=regs->A[aboff+3][1];fb=regs->A[aboff+4][0];fc=regs->A[aboff+6][2];fd=regs->A[aboff+8][3]; | |
8520 | +/* 0011 0110 0010 1101 : lev 6: */ //tmp0=( ( ( fa&fc )&fd )^( ( fb&( fa|fd ) )^fc ) ); | |
8521 | +/* 1110 1110 1011 1011 : lev 3: */ //tmp1=( ( ( fa^fc )&fd )^ALL_ONES ); | |
8522 | +/* 0101 1000 0110 0111 : lev 6: */ //tmp2=( ( fa&( fb|fc ) )^( fb^( ( fb&fc )|fd ) ) ); | |
8523 | +/* 0001 0011 0000 0001 : lev 5: */ //tmp3=( fc&( ( fa&( fb^fd ) )^( fb|fd ) ) ); | |
8524 | +/* 0011 0110 0010 1101 : lev 6: */ tmp0=FFXOR(FFAND(FFAND(fa,fc),fd),FFXOR(FFAND(fb,FFOR(fa,fd)),fc)); | |
8525 | +/* 1110 1110 1011 1011 : lev 3: */ tmp1=FFXOR(FFAND(FFXOR(fa,fc),fd),FF1()); | |
8526 | +/* 0101 1000 0110 0111 : lev 6: */ tmp2=FFXOR(FFAND(fa,FFOR(fb,fc)),FFXOR(fb,FFOR(FFAND(fb,fc),fd))); | |
8527 | +/* 0001 0011 0000 0001 : lev 5: */ tmp3=FFAND(fc,FFXOR(FFAND(fa,FFXOR(fb,fd)),FFOR(fb,fd))); | |
8528 | + s6a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
8529 | + s6b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
8530 | + | |
8531 | + fe=regs->A[aboff+1][2];fa=regs->A[aboff+2][0];fb=regs->A[aboff+6][1];fc=regs->A[aboff+7][2];fd=regs->A[aboff+7][3]; | |
8532 | +/* 0111 1000 1001 0110 : lev 5: */ //tmp0=( fb^( ( fc&fd )|( fa^( fc^fd ) ) ) ); | |
8533 | +/* 0100 1001 0101 1011 : lev 6: */ //tmp1=( ( fb|fd )&( ( fa&fc )|( fb^( fc^fd ) ) ) ); | |
8534 | +/* 0100 1001 1011 1001 : lev 5: */ //tmp2=( ( fa|fb )^( ( fc&( fb|fd ) )^fd ) ); | |
8535 | +/* 1111 1111 1101 1101 : lev 3: */ //tmp3=( fd|( ( fa&fc )^ALL_ONES ) ); | |
8536 | +/* 0111 1000 1001 0110 : lev 5: */ tmp0=FFXOR(fb,FFOR(FFAND(fc,fd),FFXOR(fa,FFXOR(fc,fd)))); | |
8537 | +/* 0100 1001 0101 1011 : lev 6: */ tmp1=FFAND(FFOR(fb,fd),FFOR(FFAND(fa,fc),FFXOR(fb,FFXOR(fc,fd)))); | |
8538 | +/* 0100 1001 1011 1001 : lev 5: */ tmp2=FFXOR(FFOR(fa,fb),FFXOR(FFAND(fc,FFOR(fb,fd)),fd)); | |
8539 | +/* 1111 1111 1101 1101 : lev 3: */ tmp3=FFOR(fd,FFXOR(FFAND(fa,fc),FF1())); | |
8540 | + s7a=FFXOR(tmp0,FFAND(fe,tmp1)); | |
8541 | + s7b=FFXOR(tmp2,FFAND(fe,tmp3)); | |
8542 | + | |
8543 | + | |
8544 | +/* | |
8545 | + we have just done this: | |
8546 | + | |
8547 | + int sbox1[0x20] = {2,0,1,1,2,3,3,0, 3,2,2,0,1,1,0,3, 0,3,3,0,2,2,1,1, 2,2,0,3,1,1,3,0}; | |
8548 | + int sbox2[0x20] = {3,1,0,2,2,3,3,0, 1,3,2,1,0,0,1,2, 3,1,0,3,3,2,0,2, 0,0,1,2,2,1,3,1}; | |
8549 | + int sbox3[0x20] = {2,0,1,2,2,3,3,1, 1,1,0,3,3,0,2,0, 1,3,0,1,3,0,2,2, 2,0,1,2,0,3,3,1}; | |
8550 | + int sbox4[0x20] = {3,1,2,3,0,2,1,2, 1,2,0,1,3,0,0,3, 1,0,3,1,2,3,0,3, 0,3,2,0,1,2,2,1}; | |
8551 | + int sbox5[0x20] = {2,0,0,1,3,2,3,2, 0,1,3,3,1,0,2,1, 2,3,2,0,0,3,1,1, 1,0,3,2,3,1,0,2}; | |
8552 | + int sbox6[0x20] = {0,1,2,3,1,2,2,0, 0,1,3,0,2,3,1,3, 2,3,0,2,3,0,1,1, 2,1,1,2,0,3,3,0}; | |
8553 | + int sbox7[0x20] = {0,3,2,2,3,0,0,1, 3,0,1,3,1,2,2,1, 1,0,3,3,0,1,1,2, 2,3,1,0,2,3,0,2}; | |
8554 | + | |
8555 | + s12 = sbox1[ (((A3>>0)&1)<<4) | (((A0>>2)&1)<<3) | (((A5>>1)&1)<<2) | (((A6>>3)&1)<<1) | (((A8>>0)&1)<<0) ] | |
8556 | + |sbox2[ (((A1>>1)&1)<<4) | (((A2>>2)&1)<<3) | (((A5>>3)&1)<<2) | (((A6>>0)&1)<<1) | (((A8>>1)&1)<<0) ]; | |
8557 | + s34 = sbox3[ (((A0>>3)&1)<<4) | (((A1>>0)&1)<<3) | (((A4>>1)&1)<<2) | (((A4>>3)&1)<<1) | (((A5>>2)&1)<<0) ] | |
8558 | + |sbox4[ (((A2>>3)&1)<<4) | (((A0>>1)&1)<<3) | (((A1>>3)&1)<<2) | (((A3>>2)&1)<<1) | (((A7>>0)&1)<<0) ]; | |
8559 | + s56 = sbox5[ (((A4>>2)&1)<<4) | (((A3>>3)&1)<<3) | (((A5>>0)&1)<<2) | (((A7>>1)&1)<<1) | (((A8>>2)&1)<<0) ] | |
8560 | + |sbox6[ (((A2>>1)&1)<<4) | (((A3>>1)&1)<<3) | (((A4>>0)&1)<<2) | (((A6>>2)&1)<<1) | (((A8>>3)&1)<<0) ]; | |
8561 | + s7 = sbox7[ (((A1>>2)&1)<<4) | (((A2>>0)&1)<<3) | (((A6>>1)&1)<<2) | (((A7>>2)&1)<<1) | (((A7>>3)&1)<<0) ]; | |
8562 | +*/ | |
8563 | + | |
8564 | + // use 4x4 xor to produce extra nibble for T3 | |
8565 | + | |
8566 | + extra_B[3]=FFXOR(FFXOR(FFXOR(regs->B[aboff+2][0],regs->B[aboff+5][1]),regs->B[aboff+6][2]),regs->B[aboff+8][3]); | |
8567 | + extra_B[2]=FFXOR(FFXOR(FFXOR(regs->B[aboff+5][0],regs->B[aboff+7][1]),regs->B[aboff+2][3]),regs->B[aboff+3][2]); | |
8568 | + extra_B[1]=FFXOR(FFXOR(FFXOR(regs->B[aboff+4][3],regs->B[aboff+7][2]),regs->B[aboff+3][0]),regs->B[aboff+4][1]); | |
8569 | + extra_B[0]=FFXOR(FFXOR(FFXOR(regs->B[aboff+8][2],regs->B[aboff+5][3]),regs->B[aboff+2][1]),regs->B[aboff+7][0]); | |
8570 | +for(dbg=0;dbg<4;dbg++){ | |
8571 | + DBG(fprintf(stderr,"extra_B[%i]=",dbg)); | |
8572 | + DBG(dump_mem("",(unsigned char *)&extra_B[dbg],BYPG,BYPG)); | |
8573 | +} | |
8574 | + | |
8575 | + // T1 = xor all inputs | |
8576 | + // in1, in2, D are only used in T1 during initialisation, not generation | |
8577 | + for(b=0;b<4;b++){ | |
8578 | + regs->A[aboff-1][b]=FFXOR(regs->A[aboff+9][b],regs->X[b]); | |
8579 | + } | |
8580 | + | |
8581 | +#ifdef STREAM_INIT | |
8582 | + for(b=0;b<4;b++){ | |
8583 | + regs->A[aboff-1][b]=FFXOR(FFXOR(regs->A[aboff-1][b],regs->D[b]),((j % 2) ? in2[b] : in1[b])); | |
8584 | + } | |
8585 | +#endif | |
8586 | + | |
8587 | +for(dbg=0;dbg<4;dbg++){ | |
8588 | + DBG(fprintf(stderr,"next_A0[%i]=",dbg)); | |
8589 | + DBG(dump_mem("",(unsigned char *)®s->A[aboff-1][dbg],BYPG,BYPG)); | |
8590 | +} | |
8591 | + | |
8592 | + // T2 = xor all inputs | |
8593 | + // in1, in2 are only used in T1 during initialisation, not generation | |
8594 | + // if p=0, use this, if p=1, rotate the result left | |
8595 | + for(b=0;b<4;b++){ | |
8596 | + regs->B[aboff-1][b]=FFXOR(FFXOR(regs->B[aboff+6][b],regs->B[aboff+9][b]),regs->Y[b]); | |
8597 | + } | |
8598 | + | |
8599 | +#ifdef STREAM_INIT | |
8600 | + for(b=0;b<4;b++){ | |
8601 | + regs->B[aboff-1][b]=FFXOR(regs->B[aboff-1][b],((j % 2) ? in1[b] : in2[b])); | |
8602 | + } | |
8603 | +#endif | |
8604 | + | |
8605 | +for(dbg=0;dbg<4;dbg++){ | |
8606 | + DBG(fprintf(stderr,"next_B0[%i]=",dbg)); | |
8607 | + DBG(dump_mem("",(unsigned char *)®s->B[aboff-1][dbg],BYPG,BYPG)); | |
8608 | +} | |
8609 | + | |
8610 | + // if p=1, rotate left (yes, this is what we're doing) | |
8611 | + tmp3=regs->B[aboff-1][3]; | |
8612 | + regs->B[aboff-1][3]=FFXOR(regs->B[aboff-1][3],FFAND(FFXOR(regs->B[aboff-1][3],regs->B[aboff-1][2]),regs->p)); | |
8613 | + regs->B[aboff-1][2]=FFXOR(regs->B[aboff-1][2],FFAND(FFXOR(regs->B[aboff-1][2],regs->B[aboff-1][1]),regs->p)); | |
8614 | + regs->B[aboff-1][1]=FFXOR(regs->B[aboff-1][1],FFAND(FFXOR(regs->B[aboff-1][1],regs->B[aboff-1][0]),regs->p)); | |
8615 | + regs->B[aboff-1][0]=FFXOR(regs->B[aboff-1][0],FFAND(FFXOR(regs->B[aboff-1][0],tmp3),regs->p)); | |
8616 | + | |
8617 | +for(dbg=0;dbg<4;dbg++){ | |
8618 | + DBG(fprintf(stderr,"next_B0[%i]=",dbg)); | |
8619 | + DBG(dump_mem("",(unsigned char *)®s->B[aboff-1][dbg],BYPG,BYPG)); | |
8620 | +} | |
8621 | + | |
8622 | + // T3 = xor all inputs | |
8623 | + for(b=0;b<4;b++){ | |
8624 | + regs->D[b]=FFXOR(FFXOR(regs->E[b],regs->Z[b]),extra_B[b]); | |
8625 | + } | |
8626 | + | |
8627 | +for(dbg=0;dbg<4;dbg++){ | |
8628 | + DBG(fprintf(stderr,"D[%i]=",dbg)); | |
8629 | + DBG(dump_mem("",(unsigned char *)®s->D[dbg],BYPG,BYPG)); | |
8630 | +} | |
8631 | + | |
8632 | + // T4 = sum, carry of Z + E + r | |
8633 | + for(b=0;b<4;b++){ | |
8634 | + next_E[b]=regs->F[b]; | |
8635 | + } | |
8636 | + | |
8637 | + tmp0=FFXOR(regs->Z[0],regs->E[0]); | |
8638 | + tmp1=FFAND(regs->Z[0],regs->E[0]); | |
8639 | + regs->F[0]=FFXOR(regs->E[0],FFAND(regs->q,FFXOR(regs->Z[0],regs->r))); | |
8640 | + tmp3=FFAND(tmp0,regs->r); | |
8641 | + tmp4=FFOR(tmp1,tmp3); | |
8642 | + | |
8643 | + tmp0=FFXOR(regs->Z[1],regs->E[1]); | |
8644 | + tmp1=FFAND(regs->Z[1],regs->E[1]); | |
8645 | + regs->F[1]=FFXOR(regs->E[1],FFAND(regs->q,FFXOR(regs->Z[1],tmp4))); | |
8646 | + tmp3=FFAND(tmp0,tmp4); | |
8647 | + tmp4=FFOR(tmp1,tmp3); | |
8648 | + | |
8649 | + tmp0=FFXOR(regs->Z[2],regs->E[2]); | |
8650 | + tmp1=FFAND(regs->Z[2],regs->E[2]); | |
8651 | + regs->F[2]=FFXOR(regs->E[2],FFAND(regs->q,FFXOR(regs->Z[2],tmp4))); | |
8652 | + tmp3=FFAND(tmp0,tmp4); | |
8653 | + tmp4=FFOR(tmp1,tmp3); | |
8654 | + | |
8655 | + tmp0=FFXOR(regs->Z[3],regs->E[3]); | |
8656 | + tmp1=FFAND(regs->Z[3],regs->E[3]); | |
8657 | + regs->F[3]=FFXOR(regs->E[3],FFAND(regs->q,FFXOR(regs->Z[3],tmp4))); | |
8658 | + tmp3=FFAND(tmp0,tmp4); | |
8659 | + regs->r=FFXOR(regs->r,FFAND(regs->q,FFXOR(FFOR(tmp1,tmp3),regs->r))); // ultimate carry | |
8660 | + | |
8661 | +/* | |
8662 | + we have just done this: (believe it or not) | |
8663 | + | |
8664 | + if (q) { | |
8665 | + F = Z + E + r; | |
8666 | + r = (F >> 4) & 1; | |
8667 | + F = F & 0x0f; | |
8668 | + } | |
8669 | + else { | |
8670 | + F = E; | |
8671 | + } | |
8672 | +*/ | |
8673 | + for(b=0;b<4;b++){ | |
8674 | + regs->E[b]=next_E[b]; | |
8675 | + } | |
8676 | +for(dbg=0;dbg<4;dbg++){ | |
8677 | + DBG(fprintf(stderr,"F[%i]=",dbg)); | |
8678 | + DBG(dump_mem("",(unsigned char *)®s->F[dbg],BYPG,BYPG)); | |
8679 | +} | |
8680 | +DBG(fprintf(stderr,"r=")); | |
8681 | +DBG(dump_mem("",(unsigned char *)®s->r,BYPG,BYPG)); | |
8682 | +for(dbg=0;dbg<4;dbg++){ | |
8683 | + DBG(fprintf(stderr,"E[%i]=",dbg)); | |
8684 | + DBG(dump_mem("",(unsigned char *)®s->E[dbg],BYPG,BYPG)); | |
8685 | +} | |
8686 | + | |
8687 | + // this simple instruction is virtually shifting all the shift registers | |
8688 | + aboff--; | |
8689 | + | |
8690 | +/* | |
8691 | + we've just done this: | |
8692 | + | |
8693 | + A9=A8;A8=A7;A7=A6;A6=A5;A5=A4;A4=A3;A3=A2;A2=A1;A1=A0;A0=next_A0; | |
8694 | + B9=B8;B8=B7;B7=B6;B6=B5;B5=B4;B4=B3;B3=B2;B2=B1;B1=B0;B0=next_B0; | |
8695 | +*/ | |
8696 | + | |
8697 | + regs->X[0]=s1a; | |
8698 | + regs->X[1]=s2a; | |
8699 | + regs->X[2]=s3b; | |
8700 | + regs->X[3]=s4b; | |
8701 | + regs->Y[0]=s3a; | |
8702 | + regs->Y[1]=s4a; | |
8703 | + regs->Y[2]=s5b; | |
8704 | + regs->Y[3]=s6b; | |
8705 | + regs->Z[0]=s5a; | |
8706 | + regs->Z[1]=s6a; | |
8707 | + regs->Z[2]=s1b; | |
8708 | + regs->Z[3]=s2b; | |
8709 | + regs->p=s7a; | |
8710 | + regs->q=s7b; | |
8711 | +for(dbg=0;dbg<4;dbg++){ | |
8712 | + DBG(fprintf(stderr,"X[%i]=",dbg)); | |
8713 | + DBG(dump_mem("",(unsigned char *)®s->X[dbg],BYPG,BYPG)); | |
8714 | +} | |
8715 | +for(dbg=0;dbg<4;dbg++){ | |
8716 | + DBG(fprintf(stderr,"Y[%i]=",dbg)); | |
8717 | + DBG(dump_mem("",(unsigned char *)®s->Y[dbg],BYPG,BYPG)); | |
8718 | +} | |
8719 | +for(dbg=0;dbg<4;dbg++){ | |
8720 | + DBG(fprintf(stderr,"Z[%i]=",dbg)); | |
8721 | + DBG(dump_mem("",(unsigned char *)®s->Z[dbg],BYPG,BYPG)); | |
8722 | +} | |
8723 | +DBG(fprintf(stderr,"p=")); | |
8724 | +DBG(dump_mem("",(unsigned char *)®s->p,BYPG,BYPG)); | |
8725 | +DBG(fprintf(stderr,"q=")); | |
8726 | +DBG(dump_mem("",(unsigned char *)®s->q,BYPG,BYPG)); | |
8727 | + | |
8728 | +#ifdef STREAM_NORMAL | |
8729 | + // require 4 loops per output byte | |
8730 | + // 2 output bits are a function of the 4 bits of D | |
8731 | + // xor 2 by 2 | |
8732 | + cb_g[8*i+7-2*j]=FFXOR(regs->D[2],regs->D[3]); | |
8733 | + cb_g[8*i+6-2*j]=FFXOR(regs->D[0],regs->D[1]); | |
8734 | +for(dbg=0;dbg<8;dbg++){ | |
8735 | + DBG(fprintf(stderr,"op[%i]=",dbg)); | |
8736 | + DBG(dump_mem("",(unsigned char *)&cb_g[8*i+dbg],BYPG,BYPG)); | |
8737 | +} | |
8738 | +#endif | |
8739 | + | |
8740 | +DBG(fprintf(stderr,"---END INTERNAL LOOP\n")); | |
8741 | + | |
8742 | + } // INTERNAL LOOP | |
8743 | + | |
8744 | +DBG(fprintf(stderr,"--END EXTERNAL LOOP\n")); | |
8745 | + | |
8746 | + } // EXTERNAL LOOP | |
8747 | + | |
8748 | + // move 32 steps forward, ready for next call | |
8749 | + for(k=0;k<10;k++){ | |
8750 | + for(b=0;b<4;b++){ | |
8751 | +DBG(fprintf(stderr,"moving forward AB k=%i b=%i\n",k,b)); | |
8752 | + regs->A[32+k][b]=regs->A[k][b]; | |
8753 | + regs->B[32+k][b]=regs->B[k][b]; | |
8754 | + } | |
8755 | + } | |
8756 | + | |
8757 | + | |
8758 | +//////////////////////////////////////////////////////////////////////////////// | |
8759 | + | |
8760 | +#ifdef STREAM_NORMAL | |
8761 | +for(j=0;j<64;j++){ | |
8762 | + DBG(fprintf(stderr,"postcall prerot cb[%2i]=",j)); | |
8763 | + DBG(dump_mem("",(unsigned char *)(cb+BYPG*j),BYPG,BYPG)); | |
8764 | +} | |
8765 | + | |
8766 | +#if GROUP_PARALLELISM==32 | |
8767 | +trasp64_32_88cw(cb); | |
8768 | +#endif | |
8769 | +#if GROUP_PARALLELISM==64 | |
8770 | +trasp64_64_88cw(cb); | |
8771 | +#endif | |
8772 | +#if GROUP_PARALLELISM==128 | |
8773 | +trasp64_128_88cw(cb); | |
8774 | +#endif | |
8775 | + | |
8776 | +for(j=0;j<64;j++){ | |
8777 | + DBG(fprintf(stderr,"postcall postrot cb[%2i]=",j)); | |
8778 | + DBG(dump_mem("",(unsigned char *)(cb+BYPG*j),BYPG,BYPG)); | |
8779 | +} | |
8780 | +#endif | |
8781 | + | |
8782 | +#ifdef STREAM_INIT | |
8783 | + DBG(fprintf(stderr,":::::::::: END STREAM INIT\n")); | |
8784 | +#endif | |
8785 | +#ifdef STREAM_NORMAL | |
8786 | + DBG(fprintf(stderr,":::::::::: END STREAM NORMAL\n")); | |
8787 | +#endif | |
8788 | + | |
8789 | +} | |
8790 | + | |
8791 | Index: libs/libmythtv/dvbrecorder.cpp | |
8792 | =================================================================== | |
8793 | --- libs/libmythtv/dvbrecorder.cpp.orig 2006-06-20 17:34:53.000000000 -0400 | |
8794 | +++ libs/libmythtv/dvbrecorder.cpp 2006-06-20 17:39:37.000000000 -0400 | |
8795 | @@ -67,6 +67,11 @@ | |
8796 | #include "../libavformat/avformat.h" | |
8797 | #include "../libavformat/mpegts.h" | |
8798 | ||
8799 | +#include "FFdecsa/FFdecsa.h" | |
8800 | +static pthread_mutex_t csalock=PTHREAD_MUTEX_INITIALIZER; | |
8801 | +static pthread_mutex_t allkeylock=PTHREAD_MUTEX_INITIALIZER; | |
8802 | +static QMap<int, struct decsaKey *> decsaMap; | |
8803 | + | |
8804 | const int DVBRecorder::PMT_PID = 0x1700; ///< PID for rewritten PMT | |
8805 | const int DVBRecorder::TSPACKETS_BETWEEN_PSIP_SYNC = 2000; | |
8806 | const int DVBRecorder::POLL_INTERVAL = 50; // msec | |
8807 | @@ -111,6 +116,9 @@ | |
8808 | ||
8809 | _buffer = new unsigned char[_buffer_size]; | |
8810 | bzero(_buffer, _buffer_size); | |
8811 | + | |
8812 | + initDeCSA(); | |
8813 | + | |
8814 | } | |
8815 | ||
8816 | DVBRecorder::~DVBRecorder() | |
8817 | @@ -190,6 +198,8 @@ | |
8818 | _reset_pid_filters = true; | |
8819 | ||
8820 | bzero(_ps_rec_buf, sizeof(unsigned char) * 3); | |
8821 | + _csa_softKey = DVBRecorder::UpdateDeCSAKeys( | |
8822 | + _card_number_option, 'G', 0, NULL, 0); //'G' = 'Get' | |
8823 | } | |
8824 | ||
8825 | bool DVBRecorder::Open(void) | |
8826 | @@ -411,6 +421,8 @@ | |
8827 | ||
8828 | void DVBRecorder::StartRecording(void) | |
8829 | { | |
8830 | + uint bufferOffset = 0; | |
8831 | + | |
8832 | if (!Open()) | |
8833 | { | |
8834 | _error = true; | |
8835 | @@ -467,12 +479,14 @@ | |
8836 | if (Poll()) | |
8837 | { | |
8838 | #ifdef USE_DRB | |
8839 | - ssize_t len = _drb->Read(_buffer, _buffer_size); | |
8840 | + ssize_t len = _drb->Read(_buffer + bufferOffset, | |
8841 | + _buffer_size - bufferOffset); | |
8842 | #else // if !USE_DRB | |
8843 | - ssize_t len = safe_read(_stream_fd, _buffer, _buffer_size); | |
8844 | + ssize_t len = safe_read(_stream_fd, _buffer + bufferOffset, | |
8845 | + _buffer_size - bufferOffset); | |
8846 | #endif // !USE_DRB | |
8847 | - if (len > 0) | |
8848 | - ProcessDataTS(_buffer, len); | |
8849 | + if (len + bufferOffset > 0) | |
8850 | + bufferOffset = ProcessDataTS(_buffer, len + bufferOffset); | |
8851 | } | |
8852 | ||
8853 | #ifdef USE_DRB | |
8854 | @@ -690,14 +704,20 @@ | |
8855 | return len; | |
8856 | ||
8857 | uint pos = 0; | |
8858 | - uint end = len - TSPacket::SIZE; | |
8859 | + int end = len - TSPacket::SIZE; | |
8860 | + | |
8861 | + // we may not decode the entire buffer | |
8862 | + if ((end = PreProcessDataTS(buffer, len)) < 0) | |
8863 | + return 0; | |
8864 | while (pos <= end) | |
8865 | { | |
8866 | const TSPacket *pkt = reinterpret_cast<const TSPacket*>(&buffer[pos]); | |
8867 | ProcessTSPacket(*pkt); | |
8868 | pos += TSPacket::SIZE; | |
8869 | } | |
8870 | - return len - pos; | |
8871 | + | |
8872 | + // At this point pos is the start of the first packet not handled | |
8873 | + return PostProcessDataTS(buffer, pos, len - pos); | |
8874 | } | |
8875 | ||
8876 | bool DVBRecorder::ProcessTSPacket(const TSPacket& tspacket) | |
8877 | @@ -1205,3 +1225,203 @@ | |
8878 | int cardnum = _card_number_option; | |
8879 | GENERAL(debugmsg); | |
8880 | } | |
8881 | + | |
8882 | +void DVBRecorder::initDeCSA() | |
8883 | +{ | |
8884 | + delete _buffer; | |
8885 | + | |
8886 | + _csa_cluster_size = get_suggested_cluster_size(); | |
8887 | + _buffer_size = MPEG_TS_PKT_SIZE * _csa_cluster_size; | |
8888 | + _buffer = new unsigned char[_buffer_size]; | |
8889 | + bzero(_buffer, _buffer_size); | |
8890 | + _csa_cluster = new unsigned char *[2 * _csa_cluster_size]; // start / end ptr for each packet | |
8891 | + _csa_clusterptr = _csa_cluster; | |
8892 | + _csa_pkt_buf_count = 0; | |
8893 | +#if 0 | |
8894 | + _csa_even_pkt_count = 0; | |
8895 | + _csa_odd_pkt_count = 0; | |
8896 | +#endif | |
8897 | + | |
8898 | + memset(_csa_even_ck, 0, 8); | |
8899 | + memset(_csa_odd_ck, 0, 8); | |
8900 | + _csa_softKey = NULL; | |
8901 | +} | |
8902 | + | |
8903 | +struct decsaKey *DVBRecorder::UpdateDeCSAKeys( | |
8904 | + int cardnum, unsigned char keytype, | |
8905 | + int index, unsigned char *key, int pid) | |
8906 | +{ | |
8907 | + struct decsaKey *decsaPtr = NULL; | |
8908 | + | |
8909 | + VERBOSE(VB_RECORD, QString("Got Key type(%1) idx: %2, pid: %3") | |
8910 | + .arg(keytype).arg(index).arg(pid)); | |
8911 | + | |
8912 | + pthread_mutex_lock(&allkeylock); | |
8913 | + QMap<int, struct decsaKey *>::Iterator it = decsaMap.begin(); | |
8914 | + while (it != decsaMap.end()) | |
8915 | + { | |
8916 | + if (it.key() == cardnum) | |
8917 | + { | |
8918 | + decsaPtr = it.data(); | |
8919 | + break; | |
8920 | + } | |
8921 | + else | |
8922 | + it++; | |
8923 | + } | |
8924 | + if (keytype == 'I') | |
8925 | + { | |
8926 | + if (decsaPtr) | |
8927 | + { | |
8928 | + pthread_mutex_lock(&decsaPtr->keylock); | |
8929 | + decsaPtr->valid_keys = 0x00; | |
8930 | + pthread_mutex_unlock(&decsaPtr->keylock); | |
8931 | + } | |
8932 | + pthread_mutex_unlock(&allkeylock); | |
8933 | + return decsaPtr; | |
8934 | + } | |
8935 | + if (!decsaPtr) { | |
8936 | + decsaPtr = new struct decsaKey; | |
8937 | + memset(decsaPtr,0, sizeof(struct decsaKey)); | |
8938 | + pthread_mutex_init(&decsaPtr->keylock, NULL); | |
8939 | + decsaMap[cardnum] = decsaPtr; | |
8940 | + memset(decsaPtr->keys, 0, sizeof(decsaPtr->keys)); | |
8941 | + memset(decsaPtr->pidmap, 0, sizeof(decsaPtr->pidmap)); | |
8942 | + } | |
8943 | + pthread_mutex_lock(&decsaPtr->keylock); | |
8944 | + pthread_mutex_unlock(&allkeylock); | |
8945 | + if (keytype == 'G') | |
8946 | + { | |
8947 | + pthread_mutex_unlock(&decsaPtr->keylock); | |
8948 | + return decsaPtr; | |
8949 | + } | |
8950 | + if (! decsaPtr->use_decsa) | |
8951 | + decsaPtr->use_decsa = true; | |
8952 | + if (! decsaPtr->keys[index]) | |
8953 | + decsaPtr->keys[index] = get_key_struct(); | |
8954 | + | |
8955 | + if (keytype == 'P') // PID | |
8956 | + decsaPtr->pidmap[pid] = index; | |
8957 | + else if (keytype == 'F') //Force PID | |
8958 | + memset(decsaPtr->pidmap, index, sizeof(decsaPtr->pidmap)); | |
8959 | + else if (keytype == 'E') //Even | |
8960 | + { | |
8961 | + decsaPtr->valid_keys |= 0x01; | |
8962 | + set_even_control_word(decsaPtr->keys[index], key); | |
8963 | + } | |
8964 | + else if (keytype == 'O') //Odd | |
8965 | + { | |
8966 | + decsaPtr->valid_keys |= 0x02; | |
8967 | + set_odd_control_word(decsaPtr->keys[index], key); | |
8968 | + } | |
8969 | + | |
8970 | + pthread_mutex_unlock(&decsaPtr->keylock); | |
8971 | + return decsaPtr; | |
8972 | +} | |
8973 | + | |
8974 | +int DVBRecorder::PreProcessDataTS(unsigned char *buffer, uint len) | |
8975 | +{ | |
8976 | + int decodedPackets = 0, unencrypted = 0; | |
8977 | + bool new_range = true; | |
8978 | + int curr_idx = -1; | |
8979 | + int offset = -2; | |
8980 | + | |
8981 | + // packets already added to the cluster array can be skipped | |
8982 | + uint pos = _csa_pkt_buf_count * TSPacket::SIZE; | |
8983 | + uint end = len - TSPacket::SIZE; | |
8984 | + | |
8985 | + // walk the packets to setup the decrypt cluster array | |
8986 | + while (pos <= end) | |
8987 | + { | |
8988 | + if (buffer[pos + 3] & 0xC0) | |
8989 | + { | |
8990 | + // encrypted | |
8991 | + int index = _csa_softKey->pidmap[((buffer[pos + 1] << 8) + | |
8992 | + (buffer[pos + 2])) & (MAX_CSA_PIDS-1)]; | |
8993 | + if (curr_idx < 0 || index == curr_idx) | |
8994 | + { | |
8995 | + //same or no index | |
8996 | + curr_idx = index; | |
8997 | + | |
8998 | + if (new_range) | |
8999 | + { | |
9000 | + // the buffer is allocated to never be | |
9001 | + // bigger than the cluster array | |
9002 | + new_range = false; | |
9003 | + offset += 2; | |
9004 | + _csa_clusterptr[offset] = buffer + pos; | |
9005 | + } | |
9006 | + _csa_clusterptr[ offset + 1] = buffer + pos + TSPacket::SIZE; | |
9007 | + // advance packet count and packet position | |
9008 | + _csa_pkt_buf_count++; | |
9009 | + } | |
9010 | + else | |
9011 | + new_range = true; | |
9012 | + } | |
9013 | + else if (! new_range) | |
9014 | + { | |
9015 | + _csa_clusterptr[ offset + 1] = buffer + pos + TSPacket::SIZE; | |
9016 | + _csa_pkt_buf_count++; | |
9017 | + } | |
9018 | + else if (_csa_pkt_buf_count == 0) | |
9019 | + unencrypted++; | |
9020 | + pos += TSPacket::SIZE; | |
9021 | + } | |
9022 | + _csa_clusterptr += offset + 2; | |
9023 | + | |
9024 | + // terminate the cluster array | |
9025 | + _csa_clusterptr[0] = NULL; | |
9026 | + | |
9027 | + // set the keys (if needed) and decrypt | |
9028 | + pthread_mutex_lock(&_csa_softKey->keylock); | |
9029 | + if(_csa_softKey->use_decsa && _csa_pkt_buf_count) { | |
9030 | + if (_csa_softKey->valid_keys && curr_idx > 0) { | |
9031 | + pthread_mutex_lock(&csalock); | |
9032 | + decodedPackets = decrypt_packets(_csa_softKey->keys[curr_idx], | |
9033 | + _csa_cluster); | |
9034 | + pthread_mutex_unlock(&csalock); | |
9035 | + pthread_mutex_unlock(&_csa_softKey->keylock); | |
9036 | + } | |
9037 | + else | |
9038 | + { | |
9039 | + // Clear all queued packets since we don't have a valid key | |
9040 | + pthread_mutex_unlock(&_csa_softKey->keylock); | |
9041 | + _csa_pkt_buf_count = 0; | |
9042 | + _csa_clusterptr = _csa_cluster; | |
9043 | + return (unencrypted - 1) * TSPacket::SIZE; | |
9044 | + } | |
9045 | + } else { | |
9046 | + // Not using decsa so pass through all the packets | |
9047 | + pthread_mutex_unlock(&_csa_softKey->keylock); | |
9048 | + _csa_cluster[0] = NULL; // ensure the cluster table doesn't overflow when not using decsa | |
9049 | + return len - TSPacket::SIZE; | |
9050 | + } | |
9051 | + | |
9052 | + _csa_pkt_buf_count -= decodedPackets; | |
9053 | + | |
9054 | + // Return the start of the last available packet | |
9055 | + return (unencrypted + decodedPackets - 1) * TSPacket::SIZE; | |
9056 | +} | |
9057 | + | |
9058 | +uint DVBRecorder::PostProcessDataTS(unsigned char *buffer, uint offset, uint len) | |
9059 | +{ | |
9060 | + if (len == 0) { | |
9061 | + _csa_clusterptr = _csa_cluster; | |
9062 | + return 0; | |
9063 | + } | |
9064 | + | |
9065 | + // relocate the unhandled packets | |
9066 | + memmove(buffer, buffer + offset, len); | |
9067 | + | |
9068 | + // fix up the cluster array to account for the relocation | |
9069 | + unsigned char **clusterWalk = _csa_cluster; | |
9070 | + while (clusterWalk[0]) { | |
9071 | + clusterWalk[0] -= offset; | |
9072 | + clusterWalk[1] -= offset; | |
9073 | + clusterWalk += 2; | |
9074 | + } | |
9075 | + | |
9076 | + // save the address for the next time we add to the cluster array | |
9077 | + _csa_clusterptr = clusterWalk; | |
9078 | + | |
9079 | + return len; | |
9080 | +} |