Damien Miller | 1e12426 | 2013-11-04 08:26:52 +1100 | [diff] [blame] | 1 | /* $OpenBSD: smult_curve25519_ref.c,v 1.2 2013/11/02 22:02:14 markus Exp $ */ |
| 2 | /* |
| 3 | version 20081011 |
| 4 | Matthew Dempsky |
| 5 | Public domain. |
| 6 | Derived from public domain code by D. J. Bernstein. |
| 7 | */ |
| 8 | |
| 9 | int crypto_scalarmult_curve25519(unsigned char *, const unsigned char *, const unsigned char *); |
| 10 | |
| 11 | static void add(unsigned int out[32],const unsigned int a[32],const unsigned int b[32]) |
| 12 | { |
| 13 | unsigned int j; |
| 14 | unsigned int u; |
| 15 | u = 0; |
| 16 | for (j = 0;j < 31;++j) { u += a[j] + b[j]; out[j] = u & 255; u >>= 8; } |
| 17 | u += a[31] + b[31]; out[31] = u; |
| 18 | } |
| 19 | |
| 20 | static void sub(unsigned int out[32],const unsigned int a[32],const unsigned int b[32]) |
| 21 | { |
| 22 | unsigned int j; |
| 23 | unsigned int u; |
| 24 | u = 218; |
| 25 | for (j = 0;j < 31;++j) { |
| 26 | u += a[j] + 65280 - b[j]; |
| 27 | out[j] = u & 255; |
| 28 | u >>= 8; |
| 29 | } |
| 30 | u += a[31] - b[31]; |
| 31 | out[31] = u; |
| 32 | } |
| 33 | |
| 34 | static void squeeze(unsigned int a[32]) |
| 35 | { |
| 36 | unsigned int j; |
| 37 | unsigned int u; |
| 38 | u = 0; |
| 39 | for (j = 0;j < 31;++j) { u += a[j]; a[j] = u & 255; u >>= 8; } |
| 40 | u += a[31]; a[31] = u & 127; |
| 41 | u = 19 * (u >> 7); |
| 42 | for (j = 0;j < 31;++j) { u += a[j]; a[j] = u & 255; u >>= 8; } |
| 43 | u += a[31]; a[31] = u; |
| 44 | } |
| 45 | |
| 46 | static const unsigned int minusp[32] = { |
| 47 | 19, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128 |
| 48 | } ; |
| 49 | |
| 50 | static void freeze(unsigned int a[32]) |
| 51 | { |
| 52 | unsigned int aorig[32]; |
| 53 | unsigned int j; |
| 54 | unsigned int negative; |
| 55 | |
| 56 | for (j = 0;j < 32;++j) aorig[j] = a[j]; |
| 57 | add(a,a,minusp); |
| 58 | negative = -((a[31] >> 7) & 1); |
| 59 | for (j = 0;j < 32;++j) a[j] ^= negative & (aorig[j] ^ a[j]); |
| 60 | } |
| 61 | |
| 62 | static void mult(unsigned int out[32],const unsigned int a[32],const unsigned int b[32]) |
| 63 | { |
| 64 | unsigned int i; |
| 65 | unsigned int j; |
| 66 | unsigned int u; |
| 67 | |
| 68 | for (i = 0;i < 32;++i) { |
| 69 | u = 0; |
| 70 | for (j = 0;j <= i;++j) u += a[j] * b[i - j]; |
| 71 | for (j = i + 1;j < 32;++j) u += 38 * a[j] * b[i + 32 - j]; |
| 72 | out[i] = u; |
| 73 | } |
| 74 | squeeze(out); |
| 75 | } |
| 76 | |
| 77 | static void mult121665(unsigned int out[32],const unsigned int a[32]) |
| 78 | { |
| 79 | unsigned int j; |
| 80 | unsigned int u; |
| 81 | |
| 82 | u = 0; |
| 83 | for (j = 0;j < 31;++j) { u += 121665 * a[j]; out[j] = u & 255; u >>= 8; } |
| 84 | u += 121665 * a[31]; out[31] = u & 127; |
| 85 | u = 19 * (u >> 7); |
| 86 | for (j = 0;j < 31;++j) { u += out[j]; out[j] = u & 255; u >>= 8; } |
| 87 | u += out[j]; out[j] = u; |
| 88 | } |
| 89 | |
| 90 | static void square(unsigned int out[32],const unsigned int a[32]) |
| 91 | { |
| 92 | unsigned int i; |
| 93 | unsigned int j; |
| 94 | unsigned int u; |
| 95 | |
| 96 | for (i = 0;i < 32;++i) { |
| 97 | u = 0; |
| 98 | for (j = 0;j < i - j;++j) u += a[j] * a[i - j]; |
| 99 | for (j = i + 1;j < i + 32 - j;++j) u += 38 * a[j] * a[i + 32 - j]; |
| 100 | u *= 2; |
| 101 | if ((i & 1) == 0) { |
| 102 | u += a[i / 2] * a[i / 2]; |
| 103 | u += 38 * a[i / 2 + 16] * a[i / 2 + 16]; |
| 104 | } |
| 105 | out[i] = u; |
| 106 | } |
| 107 | squeeze(out); |
| 108 | } |
| 109 | |
| 110 | static void select(unsigned int p[64],unsigned int q[64],const unsigned int r[64],const unsigned int s[64],unsigned int b) |
| 111 | { |
| 112 | unsigned int j; |
| 113 | unsigned int t; |
| 114 | unsigned int bminus1; |
| 115 | |
| 116 | bminus1 = b - 1; |
| 117 | for (j = 0;j < 64;++j) { |
| 118 | t = bminus1 & (r[j] ^ s[j]); |
| 119 | p[j] = s[j] ^ t; |
| 120 | q[j] = r[j] ^ t; |
| 121 | } |
| 122 | } |
| 123 | |
| 124 | static void mainloop(unsigned int work[64],const unsigned char e[32]) |
| 125 | { |
| 126 | unsigned int xzm1[64]; |
| 127 | unsigned int xzm[64]; |
| 128 | unsigned int xzmb[64]; |
| 129 | unsigned int xzm1b[64]; |
| 130 | unsigned int xznb[64]; |
| 131 | unsigned int xzn1b[64]; |
| 132 | unsigned int a0[64]; |
| 133 | unsigned int a1[64]; |
| 134 | unsigned int b0[64]; |
| 135 | unsigned int b1[64]; |
| 136 | unsigned int c1[64]; |
| 137 | unsigned int r[32]; |
| 138 | unsigned int s[32]; |
| 139 | unsigned int t[32]; |
| 140 | unsigned int u[32]; |
| 141 | unsigned int j; |
| 142 | unsigned int b; |
| 143 | int pos; |
| 144 | |
| 145 | for (j = 0;j < 32;++j) xzm1[j] = work[j]; |
| 146 | xzm1[32] = 1; |
| 147 | for (j = 33;j < 64;++j) xzm1[j] = 0; |
| 148 | |
| 149 | xzm[0] = 1; |
| 150 | for (j = 1;j < 64;++j) xzm[j] = 0; |
| 151 | |
| 152 | for (pos = 254;pos >= 0;--pos) { |
| 153 | b = e[pos / 8] >> (pos & 7); |
| 154 | b &= 1; |
| 155 | select(xzmb,xzm1b,xzm,xzm1,b); |
| 156 | add(a0,xzmb,xzmb + 32); |
| 157 | sub(a0 + 32,xzmb,xzmb + 32); |
| 158 | add(a1,xzm1b,xzm1b + 32); |
| 159 | sub(a1 + 32,xzm1b,xzm1b + 32); |
| 160 | square(b0,a0); |
| 161 | square(b0 + 32,a0 + 32); |
| 162 | mult(b1,a1,a0 + 32); |
| 163 | mult(b1 + 32,a1 + 32,a0); |
| 164 | add(c1,b1,b1 + 32); |
| 165 | sub(c1 + 32,b1,b1 + 32); |
| 166 | square(r,c1 + 32); |
| 167 | sub(s,b0,b0 + 32); |
| 168 | mult121665(t,s); |
| 169 | add(u,t,b0); |
| 170 | mult(xznb,b0,b0 + 32); |
| 171 | mult(xznb + 32,s,u); |
| 172 | square(xzn1b,c1); |
| 173 | mult(xzn1b + 32,r,work); |
| 174 | select(xzm,xzm1,xznb,xzn1b,b); |
| 175 | } |
| 176 | |
| 177 | for (j = 0;j < 64;++j) work[j] = xzm[j]; |
| 178 | } |
| 179 | |
| 180 | static void recip(unsigned int out[32],const unsigned int z[32]) |
| 181 | { |
| 182 | unsigned int z2[32]; |
| 183 | unsigned int z9[32]; |
| 184 | unsigned int z11[32]; |
| 185 | unsigned int z2_5_0[32]; |
| 186 | unsigned int z2_10_0[32]; |
| 187 | unsigned int z2_20_0[32]; |
| 188 | unsigned int z2_50_0[32]; |
| 189 | unsigned int z2_100_0[32]; |
| 190 | unsigned int t0[32]; |
| 191 | unsigned int t1[32]; |
| 192 | int i; |
| 193 | |
| 194 | /* 2 */ square(z2,z); |
| 195 | /* 4 */ square(t1,z2); |
| 196 | /* 8 */ square(t0,t1); |
| 197 | /* 9 */ mult(z9,t0,z); |
| 198 | /* 11 */ mult(z11,z9,z2); |
| 199 | /* 22 */ square(t0,z11); |
| 200 | /* 2^5 - 2^0 = 31 */ mult(z2_5_0,t0,z9); |
| 201 | |
| 202 | /* 2^6 - 2^1 */ square(t0,z2_5_0); |
| 203 | /* 2^7 - 2^2 */ square(t1,t0); |
| 204 | /* 2^8 - 2^3 */ square(t0,t1); |
| 205 | /* 2^9 - 2^4 */ square(t1,t0); |
| 206 | /* 2^10 - 2^5 */ square(t0,t1); |
| 207 | /* 2^10 - 2^0 */ mult(z2_10_0,t0,z2_5_0); |
| 208 | |
| 209 | /* 2^11 - 2^1 */ square(t0,z2_10_0); |
| 210 | /* 2^12 - 2^2 */ square(t1,t0); |
| 211 | /* 2^20 - 2^10 */ for (i = 2;i < 10;i += 2) { square(t0,t1); square(t1,t0); } |
| 212 | /* 2^20 - 2^0 */ mult(z2_20_0,t1,z2_10_0); |
| 213 | |
| 214 | /* 2^21 - 2^1 */ square(t0,z2_20_0); |
| 215 | /* 2^22 - 2^2 */ square(t1,t0); |
| 216 | /* 2^40 - 2^20 */ for (i = 2;i < 20;i += 2) { square(t0,t1); square(t1,t0); } |
| 217 | /* 2^40 - 2^0 */ mult(t0,t1,z2_20_0); |
| 218 | |
| 219 | /* 2^41 - 2^1 */ square(t1,t0); |
| 220 | /* 2^42 - 2^2 */ square(t0,t1); |
| 221 | /* 2^50 - 2^10 */ for (i = 2;i < 10;i += 2) { square(t1,t0); square(t0,t1); } |
| 222 | /* 2^50 - 2^0 */ mult(z2_50_0,t0,z2_10_0); |
| 223 | |
| 224 | /* 2^51 - 2^1 */ square(t0,z2_50_0); |
| 225 | /* 2^52 - 2^2 */ square(t1,t0); |
| 226 | /* 2^100 - 2^50 */ for (i = 2;i < 50;i += 2) { square(t0,t1); square(t1,t0); } |
| 227 | /* 2^100 - 2^0 */ mult(z2_100_0,t1,z2_50_0); |
| 228 | |
| 229 | /* 2^101 - 2^1 */ square(t1,z2_100_0); |
| 230 | /* 2^102 - 2^2 */ square(t0,t1); |
| 231 | /* 2^200 - 2^100 */ for (i = 2;i < 100;i += 2) { square(t1,t0); square(t0,t1); } |
| 232 | /* 2^200 - 2^0 */ mult(t1,t0,z2_100_0); |
| 233 | |
| 234 | /* 2^201 - 2^1 */ square(t0,t1); |
| 235 | /* 2^202 - 2^2 */ square(t1,t0); |
| 236 | /* 2^250 - 2^50 */ for (i = 2;i < 50;i += 2) { square(t0,t1); square(t1,t0); } |
| 237 | /* 2^250 - 2^0 */ mult(t0,t1,z2_50_0); |
| 238 | |
| 239 | /* 2^251 - 2^1 */ square(t1,t0); |
| 240 | /* 2^252 - 2^2 */ square(t0,t1); |
| 241 | /* 2^253 - 2^3 */ square(t1,t0); |
| 242 | /* 2^254 - 2^4 */ square(t0,t1); |
| 243 | /* 2^255 - 2^5 */ square(t1,t0); |
| 244 | /* 2^255 - 21 */ mult(out,t1,z11); |
| 245 | } |
| 246 | |
| 247 | int crypto_scalarmult_curve25519(unsigned char *q, |
| 248 | const unsigned char *n, |
| 249 | const unsigned char *p) |
| 250 | { |
| 251 | unsigned int work[96]; |
| 252 | unsigned char e[32]; |
| 253 | unsigned int i; |
| 254 | for (i = 0;i < 32;++i) e[i] = n[i]; |
| 255 | e[0] &= 248; |
| 256 | e[31] &= 127; |
| 257 | e[31] |= 64; |
| 258 | for (i = 0;i < 32;++i) work[i] = p[i]; |
| 259 | mainloop(work,e); |
| 260 | recip(work + 32,work + 32); |
| 261 | mult(work + 64,work,work + 32); |
| 262 | freeze(work + 64); |
| 263 | for (i = 0;i < 32;++i) q[i] = work[64 + i]; |
| 264 | return 0; |
| 265 | } |