1 /* Modified for SILC. -Pekka */
\r
3 /* This is an independent implementation of the encryption algorithm: */
\r
5 /* E2 by Nippon Telegraph and Telephone (NTT) Japan */
\r
7 /* which is a candidate algorithm in the Advanced Encryption Standard */
\r
8 /* programme of the US National Institute of Standards and Technology. */
\r
10 /* Copyright in this implementation is held by Dr B R Gladman but I */
\r
11 /* hereby give permission for its free direct or derivative use subject */
\r
12 /* to acknowledgment of its origin and compliance with any conditions */
\r
13 /* that the originators of the algorithm place on its exploitation. */
\r
15 /* Dr Brian Gladman (gladman@seven77.demon.co.uk) 14th January 1999 */
\r
17 /* In accordance with the wishes of NTT this implementation is made */
\r
18 /* available for academic and study purposes only. I gratefully */
\r
19 /* acknowledge the contributions made by Kazumaro Aoki of NTT Japan */
\r
20 /* for ways to increase the speed of this implementation. */
\r
22 /* Timing data for E2 (e28.c)
\r
24 Core timing without I/O endian conversion:
\r
27 Key Setup: 9473 cycles
\r
28 Encrypt: 687 cycles = 37.3 mbits/sec
\r
29 Decrypt: 691 cycles = 37.0 mbits/sec
\r
30 Mean: 689 cycles = 37.2 mbits/sec
\r
33 Key Setup: 9540 cycles
\r
34 Encrypt: 696 cycles = 36.8 mbits/sec
\r
35 Decrypt: 693 cycles = 36.9 mbits/sec
\r
36 Mean: 695 cycles = 36.9 mbits/sec
\r
39 Key Setup: 9913 cycles
\r
40 Encrypt: 691 cycles = 37.0 mbits/sec
\r
41 Decrypt: 706 cycles = 36.3 mbits/sec
\r
42 Mean: 699 cycles = 36.6 mbits/sec
\r
44 Full timing with I/O endian conversion:
\r
47 Key Setup: 9598 cycles
\r
48 Encrypt: 730 cycles = 35.1 mbits/sec
\r
49 Decrypt: 723 cycles = 35.4 mbits/sec
\r
50 Mean: 727 cycles = 35.2 mbits/sec
\r
53 Key Setup: 9393 cycles
\r
54 Encrypt: 730 cycles = 35.1 mbits/sec
\r
55 Decrypt: 720 cycles = 35.6 mbits/sec
\r
56 Mean: 725 cycles = 35.3 mbits/sec
\r
59 Key Setup: 9720 cycles
\r
60 Encrypt: 727 cycles = 35.2 mbits/sec
\r
61 Decrypt: 721 cycles = 35.5 mbits/sec
\r
62 Mean: 724 cycles = 35.4 mbits/sec
\r
67 #include <sys/types.h>
\r
68 #include "e2_internal.h"
\r
72 0xe1, 0x42, 0x3e, 0x81, 0x4e, 0x17, 0x9e, 0xfd, 0xb4, 0x3f, 0x2c, 0xda,
\r
73 0x31, 0x1e, 0xe0, 0x41, 0xcc, 0xf3, 0x82, 0x7d, 0x7c, 0x12, 0x8e, 0xbb,
\r
74 0xe4, 0x58, 0x15, 0xd5, 0x6f, 0xe9, 0x4c, 0x4b, 0x35, 0x7b, 0x5a, 0x9a,
\r
75 0x90, 0x45, 0xbc, 0xf8, 0x79, 0xd6, 0x1b, 0x88, 0x02, 0xab, 0xcf, 0x64,
\r
76 0x09, 0x0c, 0xf0, 0x01, 0xa4, 0xb0, 0xf6, 0x93, 0x43, 0x63, 0x86, 0xdc,
\r
77 0x11, 0xa5, 0x83, 0x8b, 0xc9, 0xd0, 0x19, 0x95, 0x6a, 0xa1, 0x5c, 0x24,
\r
78 0x6e, 0x50, 0x21, 0x80, 0x2f, 0xe7, 0x53, 0x0f, 0x91, 0x22, 0x04, 0xed,
\r
79 0xa6, 0x48, 0x49, 0x67, 0xec, 0xf7, 0xc0, 0x39, 0xce, 0xf2, 0x2d, 0xbe,
\r
80 0x5d, 0x1c, 0xe3, 0x87, 0x07, 0x0d, 0x7a, 0xf4, 0xfb, 0x32, 0xf5, 0x8c,
\r
81 0xdb, 0x8f, 0x25, 0x96, 0xa8, 0xea, 0xcd, 0x33, 0x65, 0x54, 0x06, 0x8d,
\r
82 0x89, 0x0a, 0x5e, 0xd9, 0x16, 0x0e, 0x71, 0x6c, 0x0b, 0xff, 0x60, 0xd2,
\r
83 0x2e, 0xd3, 0xc8, 0x55, 0xc2, 0x23, 0xb7, 0x74, 0xe2, 0x9b, 0xdf, 0x77,
\r
84 0x2b, 0xb9, 0x3c, 0x62, 0x13, 0xe5, 0x94, 0x34, 0xb1, 0x27, 0x84, 0x9f,
\r
85 0xd7, 0x51, 0x00, 0x61, 0xad, 0x85, 0x73, 0x03, 0x08, 0x40, 0xef, 0x68,
\r
86 0xfe, 0x97, 0x1f, 0xde, 0xaf, 0x66, 0xe8, 0xb8, 0xae, 0xbd, 0xb3, 0xeb,
\r
87 0xc6, 0x6b, 0x47, 0xa9, 0xd8, 0xa7, 0x72, 0xee, 0x1d, 0x7e, 0xaa, 0xb6,
\r
88 0x75, 0xcb, 0xd4, 0x30, 0x69, 0x20, 0x7f, 0x37, 0x5b, 0x9d, 0x78, 0xa3,
\r
89 0xf1, 0x76, 0xfa, 0x05, 0x3d, 0x3a, 0x44, 0x57, 0x3b, 0xca, 0xc7, 0x8a,
\r
90 0x18, 0x46, 0x9c, 0xbf, 0xba, 0x38, 0x56, 0x1a, 0x92, 0x4d, 0x26, 0x29,
\r
91 0xa2, 0x98, 0x10, 0x99, 0x70, 0xa0, 0xc5, 0x28, 0xc1, 0x6d, 0x14, 0xac,
\r
92 0xf9, 0x5f, 0x4f, 0xc4, 0xc3, 0xd1, 0xfc, 0xdd, 0xb2, 0x59, 0xe6, 0xb5,
\r
93 0x36, 0x52, 0x4a, 0x2a
\r
96 u4byte l_box[4][256];
\r
99 #define v_0 0x67452301
\r
100 #define v_1 0xefcdab89
\r
102 /* s_fun(s_fun(s_fun(v))) */
\r
104 #define k2_0 0x30d32e58
\r
105 #define k2_1 0xb89e4984
\r
107 /* s_fun(s_fun(s_fun(s_fun(v)))) */
\r
109 #define k3_0 0x0957cfec
\r
110 #define k3_1 0xd800502e
\r
112 #define bp_fun(a,b,c,d,e,f,g,h) \
\r
113 u = (e ^ g) & 0x00ffff00; \
\r
114 v = (f ^ h) & 0x0000ffff; \
\r
115 a = e ^ u; c = g ^ u; \
\r
116 b = f ^ v; d = h ^ v
\r
118 #define ibp_fun(a,b,c,d,e,f,g,h) \
\r
119 u = (e ^ g) & 0xff0000ff; \
\r
120 v = (f ^ h) & 0xffff0000; \
\r
121 a = e ^ u; c = g ^ u; \
\r
122 b = f ^ v; d = h ^ v
\r
124 #define bp2_fun(x,y) \
\r
125 w = (x ^ y) & 0x00ff00ff; \
\r
128 #define s_fun(x,y) \
\r
129 p = x; q = x >> 8; \
\r
130 r = y; s = y >> 8; \
\r
131 x = l_box[0][r & 255]; \
\r
132 y = l_box[0][p & 255]; \
\r
133 p >>= 16; r >>= 16; \
\r
134 x |= l_box[1][q & 255]; \
\r
135 y |= l_box[1][s & 255]; \
\r
136 x |= l_box[2][r & 255]; \
\r
137 y |= l_box[2][p & 255]; \
\r
138 x |= l_box[3][p >> 8]; \
\r
139 y |= l_box[3][r >> 8]
\r
141 #define sx_fun(x,y) \
\r
144 x = l_box[0][x & 255]; \
\r
145 x |= l_box[1][p & 255]; \
\r
146 x |= l_box[2][q & 255]; \
\r
147 x |= l_box[3][q >> 8]; \
\r
150 y = l_box[0][y & 255]; \
\r
151 y |= l_box[1][p & 255]; \
\r
152 y |= l_box[2][q & 255]; \
\r
153 y |= l_box[3][q >> 8]
\r
155 #define spx_fun(x,y) \
\r
158 x ^= rotr(y, 16); \
\r
162 #define sp_fun(x,y) \
\r
165 x ^= rotr(y, 16); \
\r
169 #define sr_fun(x,y) \
\r
170 p = x; q = x >> 8; \
\r
171 r = y; s = y >> 8; \
\r
172 y = l_box[1][p & 255]; \
\r
173 x = l_box[1][r & 255]; \
\r
174 p >>= 16; r >>= 16; \
\r
175 x |= l_box[2][q & 255]; \
\r
176 y |= l_box[2][s & 255]; \
\r
177 y |= l_box[3][p & 255]; \
\r
178 x |= l_box[3][r & 255]; \
\r
179 x |= l_box[0][r >> 8]; \
\r
180 y |= l_box[0][p >> 8]
\r
182 #define f_fun(a,b,c,d,k) \
\r
183 u = c ^ *(k); v = d ^ *(k + 1); \
\r
185 u ^= *(k + 2); v ^= *(k + 3); \
\r
190 #define byte_adr(x,n) *(((u1byte*)&x)+n)
\r
192 u4byte mod_inv(u4byte x)
\r
193 { u4byte y1, y2, a, b, q;
\r
195 y1 = ~((-x) / x); y2 = 1;
\r
203 if((a -= q * b) == 0)
\r
205 return (x * y1 == 1 ? y1 : -y1);
\r
211 if((b -= q * a) == 0)
\r
213 return (x * y2 == 1 ? y2 : -y2);
\r
219 void g_fun(u4byte y[8], u4byte l[8], u4byte v[2])
\r
222 spx_fun(y[0], y[1]); spx_fun(v[0], v[1]);
\r
223 l[0] = v[0] ^= y[0]; l[1] = v[1] ^= y[1];
\r
225 spx_fun(y[2], y[3]); spx_fun(v[0], v[1]);
\r
226 l[2] = v[0] ^= y[2]; l[3] = v[1] ^= y[3];
\r
228 spx_fun(y[4], y[5]); spx_fun(v[0], v[1]);
\r
229 l[4] = v[0] ^= y[4]; l[5] = v[1] ^= y[5];
\r
231 spx_fun(y[6], y[7]); spx_fun(v[0], v[1]);
\r
232 l[6] = v[0] ^= y[6]; l[7] = v[1] ^= y[7];
\r
235 u4byte *e2_set_key(E2Context *ctx,
\r
236 const u4byte in_key[], const u4byte key_len)
\r
238 u4byte lk[8], v[2], lout[8];
\r
240 u4byte *l_key = ctx->l_key;
\r
244 for(i = 0; i < 256; ++i)
\r
246 l_box[0][i] = ((u4byte)(s_box[i]));
\r
247 l_box[1][i] = ((u4byte)(s_box[i])) << 8;
\r
248 l_box[2][i] = ((u4byte)(s_box[i])) << 16;
\r
249 l_box[3][i] = ((u4byte)(s_box[i])) << 24;
\r
255 v[0] = bswap(v_0); v[1] = bswap(v_1);
\r
257 lk[0] = io_swap(in_key[0]); lk[1] = io_swap(in_key[1]);
\r
258 lk[2] = io_swap(in_key[2]); lk[3] = io_swap(in_key[3]);
\r
260 lk[4] = io_swap(key_len > 128 ? in_key[4] : k2_0);
\r
261 lk[5] = io_swap(key_len > 128 ? in_key[5] : k2_1);
\r
263 lk[6] = io_swap(key_len > 192 ? in_key[6] : k3_0);
\r
264 lk[7] = io_swap(key_len > 192 ? in_key[7] : k3_1);
\r
266 g_fun(lk, lout, v);
\r
268 for(i = 0; i < 8; ++i)
\r
270 g_fun(lk, lout, v);
\r
272 for(j = 0; j < 4; ++j)
\r
274 // this is complex because of a byte swap in each 32 bit output word
\r
276 k = 2 * (48 - 16 * j + 2 * (i / 2) - i % 2);
\r
278 ((u1byte*)l_key)[k + 3] = ((u1byte*)lout)[j];
\r
279 ((u1byte*)l_key)[k + 2] = ((u1byte*)lout)[j + 16];
\r
281 ((u1byte*)l_key)[k + 19] = ((u1byte*)lout)[j + 8];
\r
282 ((u1byte*)l_key)[k + 18] = ((u1byte*)lout)[j + 24];
\r
284 ((u1byte*)l_key)[k + 131] = ((u1byte*)lout)[j + 4];
\r
285 ((u1byte*)l_key)[k + 130] = ((u1byte*)lout)[j + 20];
\r
287 ((u1byte*)l_key)[k + 147] = ((u1byte*)lout)[j + 12];
\r
288 ((u1byte*)l_key)[k + 146] = ((u1byte*)lout)[j + 28];
\r
292 for(i = 52; i < 60; ++i)
\r
294 l_key[i] |= 1; l_key[i + 12] = mod_inv(l_key[i]);
\r
297 for(i = 0; i < 48; i += 4)
\r
299 bp2_fun(l_key[i], l_key[i + 1]);
\r
302 return (u4byte*)&l_key;
\r
305 void e2_encrypt(E2Context *ctx, const u4byte in_blk[4], u4byte out_blk[4])
\r
307 u4byte a,b,c,d,p,q,r,s,u,v;
\r
308 u4byte *l_key = ctx->l_key;
\r
310 p = io_swap(in_blk[0]); q = io_swap(in_blk[1]);
\r
311 r = io_swap(in_blk[2]); s = io_swap(in_blk[3]);
\r
313 p ^= l_key[48]; q ^= l_key[49]; r ^= l_key[50]; s ^= l_key[51];
\r
314 p *= l_key[52]; q *= l_key[53]; r *= l_key[54]; s *= l_key[55];
\r
316 bp_fun(a, b, c, d, p, q, r, s);
\r
318 f_fun(a, b, c, d, l_key);
\r
319 f_fun(c, d, a, b, l_key + 4);
\r
320 f_fun(a, b, c, d, l_key + 8);
\r
321 f_fun(c, d, a, b, l_key + 12);
\r
322 f_fun(a, b, c, d, l_key + 16);
\r
323 f_fun(c, d, a, b, l_key + 20);
\r
324 f_fun(a, b, c, d, l_key + 24);
\r
325 f_fun(c, d, a, b, l_key + 28);
\r
326 f_fun(a, b, c, d, l_key + 32);
\r
327 f_fun(c, d, a, b, l_key + 36);
\r
328 f_fun(a, b, c, d, l_key + 40);
\r
329 f_fun(c, d, a, b, l_key + 44);
\r
331 ibp_fun(p, q, r, s, a, b, c, d);
\r
333 p *= l_key[68]; q *= l_key[69]; r *= l_key[70]; s *= l_key[71];
\r
334 p ^= l_key[60]; q ^= l_key[61]; r ^= l_key[62]; s ^= l_key[63];
\r
336 out_blk[0] = io_swap(p); out_blk[1] = io_swap(q);
\r
337 out_blk[2] = io_swap(r); out_blk[3] = io_swap(s);
\r
340 void e2_decrypt(E2Context *ctx, const u4byte in_blk[4], u4byte out_blk[4])
\r
342 u4byte a,b,c,d,p,q,r,s,u,v;
\r
343 u4byte *l_key = ctx->l_key;
\r
345 p = io_swap(in_blk[0]); q = io_swap(in_blk[1]);
\r
346 r = io_swap(in_blk[2]); s = io_swap(in_blk[3]);
\r
348 p ^= l_key[60]; q ^= l_key[61]; r ^= l_key[62]; s ^= l_key[63];
\r
349 p *= l_key[56]; q *= l_key[57]; r *= l_key[58]; s *= l_key[59];
\r
351 bp_fun(a, b, c, d, p, q, r, s);
\r
353 f_fun(a, b, c, d, l_key + 44);
\r
354 f_fun(c, d, a, b, l_key + 40);
\r
356 f_fun(a, b, c, d, l_key + 36);
\r
357 f_fun(c, d, a, b, l_key + 32);
\r
359 f_fun(a, b, c, d, l_key + 28);
\r
360 f_fun(c, d, a, b, l_key + 24);
\r
362 f_fun(a, b, c, d, l_key + 20);
\r
363 f_fun(c, d, a, b, l_key + 16);
\r
365 f_fun(a, b, c, d, l_key + 12);
\r
366 f_fun(c, d, a, b, l_key + 8);
\r
368 f_fun(a, b, c, d, l_key + 4);
\r
369 f_fun(c, d, a, b, l_key);
\r
371 ibp_fun(p, q, r, s, a, b, c, d);
\r
373 p *= l_key[64]; q *= l_key[65]; r *= l_key[66]; s *= l_key[67];
\r
374 p ^= l_key[48]; q ^= l_key[49]; r ^= l_key[50]; s ^= l_key[51];
\r
376 out_blk[0] = io_swap(p); out_blk[1] = io_swap(q);
\r
377 out_blk[2] = io_swap(r); out_blk[3] = io_swap(s);
\r