1/* $OpenBSD: skipjack.c,v 1.3 2001/05/05 00:31:34 angelos Exp $ */ 2/*- 3 * Further optimized test implementation of SKIPJACK algorithm 4 * Mark Tillotson <markt@chaos.org.uk>, 25 June 98 5 * Optimizations suit RISC (lots of registers) machine best. 6 * 7 * based on unoptimized implementation of 8 * Panu Rissanen <bande@lut.fi> 960624 9 * 10 * SKIPJACK and KEA Algorithm Specifications 11 * Version 2.0 12 * 29 May 1998 13*/ 14 15#include <sys/cdefs.h> 16__FBSDID("$FreeBSD$"); 17 18#include <sys/param.h> 19 20#include <opencrypto/skipjack.h> 21 22static const u_int8_t ftable[0x100] = 23{ 24 0xa3, 0xd7, 0x09, 0x83, 0xf8, 0x48, 0xf6, 0xf4, 25 0xb3, 0x21, 0x15, 0x78, 0x99, 0xb1, 0xaf, 0xf9, 26 0xe7, 0x2d, 0x4d, 0x8a, 0xce, 0x4c, 0xca, 0x2e, 27 0x52, 0x95, 0xd9, 0x1e, 0x4e, 0x38, 0x44, 0x28, 28 0x0a, 0xdf, 0x02, 0xa0, 0x17, 0xf1, 0x60, 0x68, 29 0x12, 0xb7, 0x7a, 0xc3, 0xe9, 0xfa, 0x3d, 0x53, 30 0x96, 0x84, 0x6b, 0xba, 0xf2, 0x63, 0x9a, 0x19, 31 0x7c, 0xae, 0xe5, 0xf5, 0xf7, 0x16, 0x6a, 0xa2, 32 0x39, 0xb6, 0x7b, 0x0f, 0xc1, 0x93, 0x81, 0x1b, 33 0xee, 0xb4, 0x1a, 0xea, 0xd0, 0x91, 0x2f, 0xb8, 34 0x55, 0xb9, 0xda, 0x85, 0x3f, 0x41, 0xbf, 0xe0, 35 0x5a, 0x58, 0x80, 0x5f, 0x66, 0x0b, 0xd8, 0x90, 36 0x35, 0xd5, 0xc0, 0xa7, 0x33, 0x06, 0x65, 0x69, 37 0x45, 0x00, 0x94, 0x56, 0x6d, 0x98, 0x9b, 0x76, 38 0x97, 0xfc, 0xb2, 0xc2, 0xb0, 0xfe, 0xdb, 0x20, 39 0xe1, 0xeb, 0xd6, 0xe4, 0xdd, 0x47, 0x4a, 0x1d, 40 0x42, 0xed, 0x9e, 0x6e, 0x49, 0x3c, 0xcd, 0x43, 41 0x27, 0xd2, 0x07, 0xd4, 0xde, 0xc7, 0x67, 0x18, 42 0x89, 0xcb, 0x30, 0x1f, 0x8d, 0xc6, 0x8f, 0xaa, 43 0xc8, 0x74, 0xdc, 0xc9, 0x5d, 0x5c, 0x31, 0xa4, 44 0x70, 0x88, 0x61, 0x2c, 0x9f, 0x0d, 0x2b, 0x87, 45 0x50, 0x82, 0x54, 0x64, 0x26, 0x7d, 0x03, 0x40, 46 0x34, 0x4b, 0x1c, 0x73, 0xd1, 0xc4, 0xfd, 0x3b, 47 0xcc, 0xfb, 0x7f, 0xab, 0xe6, 0x3e, 0x5b, 0xa5, 48 0xad, 0x04, 0x23, 0x9c, 0x14, 0x51, 0x22, 0xf0, 49 0x29, 0x79, 0x71, 0x7e, 0xff, 0x8c, 0x0e, 0xe2, 50 0x0c, 0xef, 0xbc, 0x72, 0x75, 0x6f, 0x37, 0xa1, 51 0xec, 0xd3, 0x8e, 0x62, 0x8b, 0x86, 0x10, 0xe8, 52 0x08, 0x77, 0x11, 0xbe, 0x92, 0x4f, 0x24, 0xc5, 53 0x32, 0x36, 0x9d, 0xcf, 0xf3, 0xa6, 0xbb, 0xac, 54 0x5e, 0x6c, 0xa9, 0x13, 0x57, 0x25, 0xb5, 0xe3, 55 0xbd, 0xa8, 0x3a, 0x01, 0x05, 0x59, 0x2a, 0x46 56}; 57 58/* 59 * For each key byte generate a table to represent the function 60 * ftable [in ^ keybyte] 61 * 62 * These tables used to save an XOR in each stage of the G-function 63 * the tables are hopefully pointed to by register allocated variables 64 * k0, k1..k9 65 */ 66 67void 68subkey_table_gen (u_int8_t *key, u_int8_t **key_tables) 69{ 70 int i, k; 71 72 for (k = 0; k < 10; k++) { 73 u_int8_t key_byte = key [k]; 74 u_int8_t * table = key_tables[k]; 75 for (i = 0; i < 0x100; i++) 76 table [i] = ftable [i ^ key_byte]; 77 } 78} 79 80 81#define g(k0, k1, k2, k3, ih, il, oh, ol) \ 82{ \ 83 oh = k##k0 [il] ^ ih; \ 84 ol = k##k1 [oh] ^ il; \ 85 oh = k##k2 [ol] ^ oh; \ 86 ol = k##k3 [oh] ^ ol; \ 87} 88 89#define g0(ih, il, oh, ol) g(0, 1, 2, 3, ih, il, oh, ol) 90#define g4(ih, il, oh, ol) g(4, 5, 6, 7, ih, il, oh, ol) 91#define g8(ih, il, oh, ol) g(8, 9, 0, 1, ih, il, oh, ol) 92#define g2(ih, il, oh, ol) g(2, 3, 4, 5, ih, il, oh, ol) 93#define g6(ih, il, oh, ol) g(6, 7, 8, 9, ih, il, oh, ol) 94 95 96#define g_inv(k0, k1, k2, k3, ih, il, oh, ol) \ 97{ \ 98 ol = k##k3 [ih] ^ il; \ 99 oh = k##k2 [ol] ^ ih; \ 100 ol = k##k1 [oh] ^ ol; \ 101 oh = k##k0 [ol] ^ oh; \ 102} 103 104 105#define g0_inv(ih, il, oh, ol) g_inv(0, 1, 2, 3, ih, il, oh, ol) 106#define g4_inv(ih, il, oh, ol) g_inv(4, 5, 6, 7, ih, il, oh, ol) 107#define g8_inv(ih, il, oh, ol) g_inv(8, 9, 0, 1, ih, il, oh, ol) 108#define g2_inv(ih, il, oh, ol) g_inv(2, 3, 4, 5, ih, il, oh, ol) 109#define g6_inv(ih, il, oh, ol) g_inv(6, 7, 8, 9, ih, il, oh, ol) 110 111/* optimized version of Skipjack algorithm 112 * 113 * the appropriate g-function is inlined for each round 114 * 115 * the data movement is minimized by rotating the names of the 116 * variables w1..w4, not their contents (saves 3 moves per round) 117 * 118 * the loops are completely unrolled (needed to staticize choice of g) 119 * 120 * compiles to about 470 instructions on a Sparc (gcc -O) 121 * which is about 58 instructions per byte, 14 per round. 122 * gcc seems to leave in some unnecessary and with 0xFF operations 123 * but only in the latter part of the functions. Perhaps it 124 * runs out of resources to properly optimize long inlined function? 125 * in theory should get about 11 instructions per round, not 14 126 */ 127 128void 129skipjack_forwards(u_int8_t *plain, u_int8_t *cipher, u_int8_t **key_tables) 130{ 131 u_int8_t wh1 = plain[0]; u_int8_t wl1 = plain[1]; 132 u_int8_t wh2 = plain[2]; u_int8_t wl2 = plain[3]; 133 u_int8_t wh3 = plain[4]; u_int8_t wl3 = plain[5]; 134 u_int8_t wh4 = plain[6]; u_int8_t wl4 = plain[7]; 135 136 u_int8_t * k0 = key_tables [0]; 137 u_int8_t * k1 = key_tables [1]; 138 u_int8_t * k2 = key_tables [2]; 139 u_int8_t * k3 = key_tables [3]; 140 u_int8_t * k4 = key_tables [4]; 141 u_int8_t * k5 = key_tables [5]; 142 u_int8_t * k6 = key_tables [6]; 143 u_int8_t * k7 = key_tables [7]; 144 u_int8_t * k8 = key_tables [8]; 145 u_int8_t * k9 = key_tables [9]; 146 147 /* first 8 rounds */ 148 g0 (wh1,wl1, wh1,wl1); wl4 ^= wl1 ^ 1; wh4 ^= wh1; 149 g4 (wh4,wl4, wh4,wl4); wl3 ^= wl4 ^ 2; wh3 ^= wh4; 150 g8 (wh3,wl3, wh3,wl3); wl2 ^= wl3 ^ 3; wh2 ^= wh3; 151 g2 (wh2,wl2, wh2,wl2); wl1 ^= wl2 ^ 4; wh1 ^= wh2; 152 g6 (wh1,wl1, wh1,wl1); wl4 ^= wl1 ^ 5; wh4 ^= wh1; 153 g0 (wh4,wl4, wh4,wl4); wl3 ^= wl4 ^ 6; wh3 ^= wh4; 154 g4 (wh3,wl3, wh3,wl3); wl2 ^= wl3 ^ 7; wh2 ^= wh3; 155 g8 (wh2,wl2, wh2,wl2); wl1 ^= wl2 ^ 8; wh1 ^= wh2; 156 157 /* second 8 rounds */ 158 wh2 ^= wh1; wl2 ^= wl1 ^ 9 ; g2 (wh1,wl1, wh1,wl1); 159 wh1 ^= wh4; wl1 ^= wl4 ^ 10; g6 (wh4,wl4, wh4,wl4); 160 wh4 ^= wh3; wl4 ^= wl3 ^ 11; g0 (wh3,wl3, wh3,wl3); 161 wh3 ^= wh2; wl3 ^= wl2 ^ 12; g4 (wh2,wl2, wh2,wl2); 162 wh2 ^= wh1; wl2 ^= wl1 ^ 13; g8 (wh1,wl1, wh1,wl1); 163 wh1 ^= wh4; wl1 ^= wl4 ^ 14; g2 (wh4,wl4, wh4,wl4); 164 wh4 ^= wh3; wl4 ^= wl3 ^ 15; g6 (wh3,wl3, wh3,wl3); 165 wh3 ^= wh2; wl3 ^= wl2 ^ 16; g0 (wh2,wl2, wh2,wl2); 166 167 /* third 8 rounds */ 168 g4 (wh1,wl1, wh1,wl1); wl4 ^= wl1 ^ 17; wh4 ^= wh1; 169 g8 (wh4,wl4, wh4,wl4); wl3 ^= wl4 ^ 18; wh3 ^= wh4; 170 g2 (wh3,wl3, wh3,wl3); wl2 ^= wl3 ^ 19; wh2 ^= wh3; 171 g6 (wh2,wl2, wh2,wl2); wl1 ^= wl2 ^ 20; wh1 ^= wh2; 172 g0 (wh1,wl1, wh1,wl1); wl4 ^= wl1 ^ 21; wh4 ^= wh1; 173 g4 (wh4,wl4, wh4,wl4); wl3 ^= wl4 ^ 22; wh3 ^= wh4; 174 g8 (wh3,wl3, wh3,wl3); wl2 ^= wl3 ^ 23; wh2 ^= wh3; 175 g2 (wh2,wl2, wh2,wl2); wl1 ^= wl2 ^ 24; wh1 ^= wh2; 176 177 /* last 8 rounds */ 178 wh2 ^= wh1; wl2 ^= wl1 ^ 25; g6 (wh1,wl1, wh1,wl1); 179 wh1 ^= wh4; wl1 ^= wl4 ^ 26; g0 (wh4,wl4, wh4,wl4); 180 wh4 ^= wh3; wl4 ^= wl3 ^ 27; g4 (wh3,wl3, wh3,wl3); 181 wh3 ^= wh2; wl3 ^= wl2 ^ 28; g8 (wh2,wl2, wh2,wl2); 182 wh2 ^= wh1; wl2 ^= wl1 ^ 29; g2 (wh1,wl1, wh1,wl1); 183 wh1 ^= wh4; wl1 ^= wl4 ^ 30; g6 (wh4,wl4, wh4,wl4); 184 wh4 ^= wh3; wl4 ^= wl3 ^ 31; g0 (wh3,wl3, wh3,wl3); 185 wh3 ^= wh2; wl3 ^= wl2 ^ 32; g4 (wh2,wl2, wh2,wl2); 186 187 /* pack into byte vector */ 188 cipher [0] = wh1; cipher [1] = wl1; 189 cipher [2] = wh2; cipher [3] = wl2; 190 cipher [4] = wh3; cipher [5] = wl3; 191 cipher [6] = wh4; cipher [7] = wl4; 192} 193 194 195void 196skipjack_backwards (u_int8_t *cipher, u_int8_t *plain, u_int8_t **key_tables) 197{ 198 /* setup 4 16-bit portions */ 199 u_int8_t wh1 = cipher[0]; u_int8_t wl1 = cipher[1]; 200 u_int8_t wh2 = cipher[2]; u_int8_t wl2 = cipher[3]; 201 u_int8_t wh3 = cipher[4]; u_int8_t wl3 = cipher[5]; 202 u_int8_t wh4 = cipher[6]; u_int8_t wl4 = cipher[7]; 203 204 u_int8_t * k0 = key_tables [0]; 205 u_int8_t * k1 = key_tables [1]; 206 u_int8_t * k2 = key_tables [2]; 207 u_int8_t * k3 = key_tables [3]; 208 u_int8_t * k4 = key_tables [4]; 209 u_int8_t * k5 = key_tables [5]; 210 u_int8_t * k6 = key_tables [6]; 211 u_int8_t * k7 = key_tables [7]; 212 u_int8_t * k8 = key_tables [8]; 213 u_int8_t * k9 = key_tables [9]; 214 215 /* first 8 rounds */ 216 g4_inv (wh2,wl2, wh2,wl2); wl3 ^= wl2 ^ 32; wh3 ^= wh2; 217 g0_inv (wh3,wl3, wh3,wl3); wl4 ^= wl3 ^ 31; wh4 ^= wh3; 218 g6_inv (wh4,wl4, wh4,wl4); wl1 ^= wl4 ^ 30; wh1 ^= wh4; 219 g2_inv (wh1,wl1, wh1,wl1); wl2 ^= wl1 ^ 29; wh2 ^= wh1; 220 g8_inv (wh2,wl2, wh2,wl2); wl3 ^= wl2 ^ 28; wh3 ^= wh2; 221 g4_inv (wh3,wl3, wh3,wl3); wl4 ^= wl3 ^ 27; wh4 ^= wh3; 222 g0_inv (wh4,wl4, wh4,wl4); wl1 ^= wl4 ^ 26; wh1 ^= wh4; 223 g6_inv (wh1,wl1, wh1,wl1); wl2 ^= wl1 ^ 25; wh2 ^= wh1; 224 225 /* second 8 rounds */ 226 wh1 ^= wh2; wl1 ^= wl2 ^ 24; g2_inv (wh2,wl2, wh2,wl2); 227 wh2 ^= wh3; wl2 ^= wl3 ^ 23; g8_inv (wh3,wl3, wh3,wl3); 228 wh3 ^= wh4; wl3 ^= wl4 ^ 22; g4_inv (wh4,wl4, wh4,wl4); 229 wh4 ^= wh1; wl4 ^= wl1 ^ 21; g0_inv (wh1,wl1, wh1,wl1); 230 wh1 ^= wh2; wl1 ^= wl2 ^ 20; g6_inv (wh2,wl2, wh2,wl2); 231 wh2 ^= wh3; wl2 ^= wl3 ^ 19; g2_inv (wh3,wl3, wh3,wl3); 232 wh3 ^= wh4; wl3 ^= wl4 ^ 18; g8_inv (wh4,wl4, wh4,wl4); 233 wh4 ^= wh1; wl4 ^= wl1 ^ 17; g4_inv (wh1,wl1, wh1,wl1); 234 235 /* third 8 rounds */ 236 g0_inv (wh2,wl2, wh2,wl2); wl3 ^= wl2 ^ 16; wh3 ^= wh2; 237 g6_inv (wh3,wl3, wh3,wl3); wl4 ^= wl3 ^ 15; wh4 ^= wh3; 238 g2_inv (wh4,wl4, wh4,wl4); wl1 ^= wl4 ^ 14; wh1 ^= wh4; 239 g8_inv (wh1,wl1, wh1,wl1); wl2 ^= wl1 ^ 13; wh2 ^= wh1; 240 g4_inv (wh2,wl2, wh2,wl2); wl3 ^= wl2 ^ 12; wh3 ^= wh2; 241 g0_inv (wh3,wl3, wh3,wl3); wl4 ^= wl3 ^ 11; wh4 ^= wh3; 242 g6_inv (wh4,wl4, wh4,wl4); wl1 ^= wl4 ^ 10; wh1 ^= wh4; 243 g2_inv (wh1,wl1, wh1,wl1); wl2 ^= wl1 ^ 9; wh2 ^= wh1; 244 245 /* last 8 rounds */ 246 wh1 ^= wh2; wl1 ^= wl2 ^ 8; g8_inv (wh2,wl2, wh2,wl2); 247 wh2 ^= wh3; wl2 ^= wl3 ^ 7; g4_inv (wh3,wl3, wh3,wl3); 248 wh3 ^= wh4; wl3 ^= wl4 ^ 6; g0_inv (wh4,wl4, wh4,wl4); 249 wh4 ^= wh1; wl4 ^= wl1 ^ 5; g6_inv (wh1,wl1, wh1,wl1); 250 wh1 ^= wh2; wl1 ^= wl2 ^ 4; g2_inv (wh2,wl2, wh2,wl2); 251 wh2 ^= wh3; wl2 ^= wl3 ^ 3; g8_inv (wh3,wl3, wh3,wl3); 252 wh3 ^= wh4; wl3 ^= wl4 ^ 2; g4_inv (wh4,wl4, wh4,wl4); 253 wh4 ^= wh1; wl4 ^= wl1 ^ 1; g0_inv (wh1,wl1, wh1,wl1); 254 255 /* pack into byte vector */ 256 plain [0] = wh1; plain [1] = wl1; 257 plain [2] = wh2; plain [3] = wl2; 258 plain [4] = wh3; plain [5] = wl3; 259 plain [6] = wh4; plain [7] = wl4; 260} 261